The Relative Clause in Biblical Hebrew 9781575064208

This book is the result of 15 years of research on the ancient Hebrew relative clause as well as the effective applicati

218 22 4MB

English Pages 472 [473] Year 2016

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

The Relative Clause in Biblical Hebrew
 9781575064208

Citation preview

The Relative Clause in Biblical Hebrew

Linguistic Studies in Ancient West Semitic edited by

Cynthia L. Miller-Naudé and Jacobus A. Naudé The series Linguistic Studies in Ancient West Semitic is devoted to the ancient West Semitic languages, including Hebrew, Aramaic, Ugaritic, and their near congeners. It includes monographs, collections of essays, and text editions informed by the approaches of linguistic science. The material studied will span from the earliest texts to the rise of Islam.  1. The Verbless Clause in Biblical Hebrew: Linguistic Approaches, edited by Cynthia L. Miller  2. Phonology and Morphology of Biblical Hebrew: An Introduction, by Joshua Blau  3. A Manual of Ugaritic, by Pierre Bordreuil and Dennis Pardee  4. Word Order in the Biblical Hebrew Finite Clause: A Syntactic and Pragmatic Analysis of Preposing, by Adina Moshavi  5. Oath Formulas in Biblical Hebrew, by Blane Conklin  6. Biblical Hebrew Grammar Visualized, by Francis I. Andersen and A. Dean Forbes  7. Time and the Biblical Hebrew Verb: The Expression of Tense, Aspect, and Modality in Biblical Hebrew, by John A. Cook  8. Diachrony in Biblical Hebrew, edited by Cynthia L. Miller-Naudé and Ziony Zevit  9. The Syntax of Volitives in Biblical Hebrew and Amarna Canaanite Prose, by Hélène Dallaire 10.  The Relative Clause in Biblical Hebrew, by Robert D. Holmstedt 11.  Language Change in the Wake of Empire: Syriac in Its Greco-Roman Context, by Aaron Michael Butts

The Relative Clause in Biblical Hebrew

R obert D. H olmstedt

Winona Lake, Indiana Eisenbrauns 2016

Copyright © 2016 Eisenbrauns All rights reserved. Printed in the United States of America. www.eisenbrauns.com

Library of Congress Cataloging-in-Publication Data Names: Holmstedt, Robert D., author. Title: The relative clause in biblical Hebrew / Robert D. Holmstedt. Description: Winona Lake, Indiana : Eisenbrauns, [2016] | 2016 | Series: Linguistic studies in ancient West Semitic ; 10 | Includes bibliographical references and indexes. | Description based on print version record and CIP data provided by publisher; resource not viewed. Identifiers: LCCN 2015042928 (print) | LCCN 2015042032 (ebook) | ISBN 9781575064208 (pdf) | ISBN 9781575064192 (hardback : alk. paper) Subjects: LCSH: Hebrew language—Clauses. | Bible Old Testament—Language, style. Classification: LCC PJ4717 (print) | LCC PJ4717 .H65 2016 (ebook) | DDC 492.4/55—dc23 LC record available at http://lccn.loc.gov/2015042928

The paper used in this publication meets the minimum requirements of the American National Standard for Information Sciences—Permanence of Paper for Printed Library Materials. ANSI Z39.48-1984.♾™

Contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .   x Transliteration of Semitic Characters . . . . . . . . . . . . . . . . . . . xiii Consonants xiii Vowels xiii Abbreviations and Symbols . . . . . . . . . . . . . . . . . . . . . . . . xv Symbols xv General Abbreviations  xv Reference Works  xvi Chapter 1.  Relative Clauses: Orientation . . . . . . . . . . . . . . . . .   1 1.1.  Relative Typology  2 1.1.1.  Relative Definition  5 1.1.2.  The Cline of Embeddedness  9 1.1.3.  The Cline of Internal Complexity  10 1.2.  Previous Scholarship  16 1.3.  Corpus and Conventions for the Representation and Translation of Hebrew  17 Chapter 2.  Issues of Method and Theory . . . . . . . . . . . . . . . . . 19 2.1.  Methodological Challenges for the Biblical Hebrew Linguist 19 2.1.1.  Philology and Linguistics  20 2.1.2.  Analyzing No-Longer-Spoken Languages  21 2.1.3.  The Linguistic Status of Biblical Hebrew  31 2.2.  The Linguistic Framework of This Study  36 2.2.1.  Hierarchical and Binary-Branching Phrase Structure  37 2.2.2.  Constituent Movement  41 2.2.3.  Null Constituents  43 2.3.  A Brief Orientation to Biblical Hebrew Syntax  44 2.3.1.  A Typological Description of Hebrew Constituent Order  45 2.3.2.  A Generative Description of Hebrew Phrase Structure 50 2.3.3. Summary  55 v

vi

Contents

Chapter 3.  Relative Words . . . . . . . . . . . . . . . . . . . . . . . . 57 3.1.  Relative Word Typology  57 3.2.  Hebrew Relative Elements  61 3.2.1.  ‫ז‬-Relatives 64 3.2.2.  ‫אׁשר‬- and ‫ׁש‬-Relatives 68 3.2.3.  ‫ה‬-Relatives 69 3.2.4.  ‫מ‬-Relatives 77 3.2.5. Zero-Relatives  81 3.2.6.  Combination of Hebrew Relative Elements  83 3.3.  The Etymologies of ‫ אׁשר‬and ‫ ׁש‬85 3.3.1.  ‫ אׁשר‬and Its Semitic Relatives  86 3.3.2.  ‫ שׁ‬and Its Semitic Relatives  91 3.3.3. Are ‫ אׁשר‬and ‫ שׁ‬Related? 93 3.4. Summary  101 Chapter 4.  Relative Heads . . . . . . . . . . . . . . . . . . . . . . . 102 4.1.  Relative Head Typology  104 4.1.1.  Overt and Null Heads  104 4.1.2.  Hierarchical Position of the Head  105 4.1.3.  Linear Order of the Head and Relative Clause  106 4.1.4.  Position of Determiner and Case-Marker  108 4.2.  Overt Relative Heads in Hebrew  109 4.3.  Null Relative Heads in Hebrew  113 4.3.1.  Excursus on Hebraist Terminology for Null-Head Relatives  118 4.3.2.  Preposition/Conjunction + Relative Word Combinations  119 4.4.  Hierarchy, Linear Order, Determiners, and Case-Marking in Hebrew Relatives  128 Chapter 5.  Relative Syntax . . . . . . . . . . . . . . . . . . . . . . . 129 5.1.  Relative Syntax  129 5.1.1.  Typological Features  129 5.1.2.  Phrase Structure Features  132 5.1.2.1. Resumption  135 5.1.2.2. Extraposition  141 5.2.  Relative Syntax in Hebrew  142 5.2.1.  Syntax of Null-Head Relative Clauses  154 5.2.2.  Preposition Stranding, Pied-Piping, and Stacking  155 5.2.3. Resumption  158 5.2.3.1.  Previous Studies on Hebrew Relative Resumption 158

Contents

vii

5.2.3.2.  Resumption and the Noun Phrase Accessibility Hierarchy 169 5.2.3.3. Function  175 5.2.4. Extraposition  186 5.2.4.1. Function  191 Chapter 6.  Relative Semantics . . . . . . . . . . . . . . . . . . . . . 193 6.1.  Relative Semantic Typology  193 6.1.1.  Verbal Inflection within the Relative  193 6.1.2.  Restrictive, Nonrestrictive, and Maximalizing Relatives 195 6.1.2.1.  Syntax of Nonrestrictive Relatives  196 6.2.  Hebrew Relatives and Verbal Semantics  203 6.3.  Hebrew Relatives and Restrictiveness  205 6.3.1.  How to Identify Relative Restrictiveness in Hebrew  209 Chapter 7.  “Relative” Diachrony . . . . . . . . . . . . . . . . . . . . 215 7.1.  Nominalization and Relativization  215 7.2.  The Story of the ‫ אׁשר‬Clauses 216 7.2.1.  Pre-Hellenistic Period Nonbiblical Data  216 7.2.2.  Biblical Data  219 7.2.3.  Hellenistic and Roman Period Nonbiblical Data  221 7.3.  The Story of ‫ ׁש‬Clauses 225 7.4.  Sorting Out the Many Proposed Functions of ‫ אׁשר‬229 7.5. Summary  244 Chapter 8.  The Semitic Relative . . . . . . . . . . . . . . . . . . . . 248 8.1.  Brief Introduction to Semitic Languages  248 8.2. Akkadian  249 8.3.  Edomite, Ammonite, Moabite, and Philistine  254 8.4. Phoenician  256 8.5. Aramaic  263 8.6. Ugaritic  267 8.7.  Classical Arabic  270 8.8.  Epigraphic South Arabian (ESA)  272 8.9.  Classical Ethiopic (Geʿez)  277 8.10.  Summary and Final Matters  281 Appendix A.  Relative Clause Data . . . . . . . . . . . . . . . . . . . 286 1.  ‫ׁש‬-Relative Data (99×)  286 2.  ‫ׁש‬-Complementizer Data (40×)  295 3.  ‫ז‬-Relative Data (24×)  298 4.  ‫מ‬-Pronoun Relative Data (37×)  300

viii

Contents

5.  ‫ה‬-Relative Data (nonparticipial/nonadjectival) (23×)  303 6.  Zero-Relative Data (299×)  305 7.  Extraposed Relative Data (133×)  325 8.  ‫אׁשר‬-Nonrelative, Nominalizer Data (386×)  340 8a.  ‫אשׁר‬-Nominalized Arguments  340 8b.  Preposition + ‫ אשׁר‬345 8c.  Conjunction + ‫ אשׁר‬356 Appendix B.  Difficult Relative Clauses in the Hebrew Bible . . . . . . 365 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382 Indexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411 Index of Authors  411 Index of Scripture  416 Index of Other Ancient Sources  438 Index of Subjects  445

Contents

ix

List of Tables Table 1.1.  The Parameters of Relative Clause Formation . . . . . . .    5 Table 1.2.  Syntactic Variety of Relative Clauses . . . . . . . . . . . .    6 Table 2.1.  Typological Illustration of English Word Order . . . . . . .   46 Table 3.1.  Combination of Relative Elements . . . . . . . . . . . . .   61 Table 3.2.  Distinctions between ‫ ׁש‬and ‫ה‬/Semi-Relatives in   Modern Hebrew . . . . . . . . . . . . . . . . . . . . . . .   72 Table 4.1.  Linear Order of Head and Relative Clause . . . . . . . . .   107 Table 6.1.  Criteria for Determining Restrictiveness . . . . . . . . . .   209 Table 7.1.  Non-nominalizing Analyses of ‫ אׁשר‬in Representative   Grammars . . . . . . . . . . . . . . . . . . . . . . . . . .   233 Table 7.2.  ‫ אׁשר‬and ‫ ׁש‬in Ancient Hebrew . . . . . . . . . . . . . . .   243 Table 8.1.  The Old Akkadian Relative Marker . . . . . . . . . . . . .   250 Table 8.2.  The Ugaritic Relative Marker (First View) . . . . . . . . .   268 Table 8.3.  The Ugaritic Relative Marker (Second View) . . . . . . . .   269 Table 8.4.  The Classical Arabic “Determinative-Relative” Marker . .   271 Table 8.5.  The Classical Arabic Relative Marker . . . . . . . . . . .   272 Table 8.6.  The Epigraphic South Arabian Relative Marker . . . . . .   273 List of Figures Figure 3.1.  Relative Elements . . . . . . . . . . . . . . . . . . . . .   57 Figure 7.1.  The Diffusion of ‫ ׁש‬on an Idealized S-Curve . . . . . . . .   244 Figure 8.1.  The Semitic Language Family Tree . . . . . . . . . . . .   249

Contents

ix

List of Tables Table 1.1.  The Parameters of Relative Clause Formation . . . . . . .    5 Table 1.2.  Syntactic Variety of Relative Clauses . . . . . . . . . . . .    6 Table 2.1.  Typological Illustration of English Word Order . . . . . . .   46 Table 3.1.  Combination of Relative Elements . . . . . . . . . . . . .   61 Table 3.2.  Distinctions between ‫ ׁש‬and ‫ה‬/Semi-Relatives in   Modern Hebrew . . . . . . . . . . . . . . . . . . . . . . .   72 Table 4.1.  Linear Order of Head and Relative Clause . . . . . . . . .   107 Table 6.1.  Criteria for Determining Restrictiveness . . . . . . . . . .   209 Table 7.1.  Non-nominalizing Analyses of ‫ אׁשר‬in Representative   Grammars . . . . . . . . . . . . . . . . . . . . . . . . . .   233 Table 7.2.  ‫ אׁשר‬and ‫ ׁש‬in Ancient Hebrew . . . . . . . . . . . . . . .   243 Table 8.1.  The Old Akkadian Relative Marker . . . . . . . . . . . . .   250 Table 8.2.  The Ugaritic Relative Marker (First View) . . . . . . . . .   268 Table 8.3.  The Ugaritic Relative Marker (Second View) . . . . . . . .   269 Table 8.4.  The Classical Arabic “Determinative-Relative” Marker . .   271 Table 8.5.  The Classical Arabic Relative Marker . . . . . . . . . . .   272 Table 8.6.  The Epigraphic South Arabian Relative Marker . . . . . .   273 List of Figures Figure 3.1.  Relative Elements . . . . . . . . . . . . . . . . . . . . .   57 Figure 7.1.  The Diffusion of ‫ ׁש‬on an Idealized S-Curve . . . . . . . .   244 Figure 8.1.  The Semitic Language Family Tree . . . . . . . . . . . .   249

Preface This book picks up a conversation I began over 15 years ago. The first significant result of this conversation was my 2002 doctoral thesis, completed at the University of Wisconsin–Madison under Cynthia L. Miller-Naudé and Michael V. Fox. I have contributed to this conversation intermittently over the last 13 years in a variety of ways. In some cases, I worked through focused studies and presented them as conference papers, published them as journal and volume articles, or tested them on graduate students in various seminars. In other cases, I worked my grammatical ideas out by ensuring that they make good sense of biblical texts, both in teaching contexts and in three grammatical commentaries, most notably on Ruth (Holmstedt 2010), Esther (Screnock and Holmstedt 2015), and Ecclesiastes (Holmstedt, Cook, and Marshall 2016). Although my original thesis was accepted for publication in Eisenbrauns’ LSAWS series in 2003, two concerns kept me from publishing it until now. The first concern was that I wanted to expand my research into areas relevant to the analysis of the relative clause and the Hebrew words ‫ אשׁר‬and ‫שׁ‬, and this research led me deeply into the complex areas of Hebrew diachrony and comparative Semitic syntax. My second concern was that the thesis made sense as a thesis but less so as a monograph. As a work situated between two worlds, it was a successful thesis but would have been a tragic book. To wit, in November 2003, the inimitable Michael O’Connor reviewed my thesis, which I had submitted to the LSAWS series, and summarized the overall effect of the work as a collision of two trains (the Hebrew relative clause and Hebrew word order) and a bicycle (the functions of ‫)אשׁר‬. O’Connor’s keen sense of humor aside (even I laughed at his assessment), the image reflects an accurate assessment of the thesis. However, while the particular issues may differ, the basic point is generally true for most doctoral theses—it is the rare thesis that, without significant revision, avoids becoming a tragic book. I hope that this book avoids tragedy, though it is up to the reader to decide. In fact, this monograph is not simply a revision of my doctoral thesis. I have both rewritten all the material I kept and excised a good bit that I deemed either irrelevant to my intended audience or no longer concordant with my grammatical views. But more than rewriting, I have added a great deal of new research beyond what I covered for my 2002 thesis. For example, new work includes the chapters on philology, linguistic typology, Hebrew diachrony, and x

Preface

xi

the Semitic context, as well as appendix A, in which I provide lists of data that I used in the book, thereby satisfying my desire to be fully transparent in my analyses. This desire for transparency no doubt opens the door to unending disagreements on this verse or that one. It is certainly the reader’s right to disagree with my brilliant analyses, but to paraphrase Maxwell Smart, “I’ll ask you not to tell me that!” The main methodological goal of the thesis was to illustrate the advantage of cross-disciplinary research for biblical studies, specifically the analysis of one syntactic issue, the relative clause, within an explicit theoretical linguistic framework. This book moves beyond this goal in two ways. First, it is not simply an illustration of a process or methodology but a full description of a linguistic phenomenon in which the methodological and theoretical roots have deepened significantly. Second, this work represents over a decade of teaching (general linguistics, Hebrew and Northwest Semitics, comparative Semitics, Bible, or ancient Israelite history), research, and graduate supervision, all of which have resulted in what I consider a more mature expression of my vision for Hebrew grammatical study. I believe that there is in this work something for everyone who seeks input on Hebrew grammar, ancient language philology, close reading of biblical texts, comparative Semitics, and linguistic typology (and even a little theory). No relevant subject goes unexplored; few previous analyses escape mention in some way. I am, after all, an equal opportunity critic and provocateur. I fully acknowledge the powerful shoulders of past Hebrew philologists upon which I stand; even so, given the ground I cover in this work from data to theory, I can say with Kulamuwa, mʾš pʿlt bl pʿl hlpny hm. My audience has also shifted. Whereas the thesis did not assume that the reader could read Hebrew (and therefore did not use Hebrew script), and I intentionally addressed both a Hebrew and general linguistic audience, this book is aimed primarily at Hebraists. For this reason, I have broadened my methodological discussion to address the distinction between philology and linguistics and the use of ancient texts in linguistic analysis. Moreover, I introduce the theoretical background of my argument—linguistic typology and generative linguistics—with the nonspecialist reader in mind. And throughout the work I have included many more examples to support my points, with footnotes filled with additional relevant examples (and an appendix containing the examples with translation). My hope is that the product is a readable and searchable reference work for the wider audience of Hebrew and biblical studies. I have also tried to be as clear and non-boring as possible, though given the subject matter, I apologize in advance. I will draw this preface to a close by stating three principles that I have come to see as both critical and generally unexpressed in our field (at least within the current generation of scholarship). First, studying ancient languages

xii

Preface

requires textual sensitivity. This is not often exhibited in regard to historicalcomparative Semitic works (e.g., see my discussion of ‫ז‬-relatives in §3.2.1). Second, the linguistic data cannot be studied isolated from history, geography, and intercultural contact, since the languages and the people who used them did not appear and exist in a vacuum (e.g., see my discussion of the etymologies of ‫ אשׁר‬and ‫ שׁ‬in §3.3). Third and finally, there is no such thing as analysis sans theory. To assert, as I have personally heard, that someone carrying out a historical linguistic study “has as much a theory as a generative linguist” is facile and vacuous, if the assertion presupposes that the activity of historical-comparative linguistic inquiry itself is somehow inherently theory-based. This confuses method with theory. Tracing the historical change or the comparative manifestations of “feature X” is not inherently theoretical. Theory in the realm of language study refers to a coherent view of the system of language as it relates to its external world (e.g., the human mind, the context of communication). Whether one chooses some sort of formalist or functionalist linguistic theory, any work performed without some theoretical awareness runs the severe risk of being theoretically anchor-less and therefore probably useless. Although it is a daunting task if we want to do it well, the linguistic analysis of an ancient language such as Biblical Hebrew demands theoretical awareness, descriptive clarity, textual sensitivity, and attention to historical, geographical, and cultural context. I must acknowledge multiple influences on this project in its 15-year history. First, my beloved Rachel has watched me deal with various aspects of this work since the second year of our marriage. We both wonder about life after this strange ménage à trois is reduced to two. Second, my father and mother have been an indefatigable support, encouraging me for many years to “just finish the [colorful adjective] book.” Look, folks—it’s done! Academically, I owe many thanks to Cynthia Miller-Naudé (my Doktormutter), John A. Cook (“as iron sharpens iron”), Michael O’Connor (requiescat in pace), Mark Graham, Andrew Jones, André Arsenault, and many others who have encouraged me over the years, often simply by mentioning that they have read my work and found it stimulating. I also appreciate the patience shown by Jim Eisenbraun, who waited for 12 years for a book that he contracted in 2003. Finally, all opinions in this work are solely mine—for who else would dare promote them?

Transliteration of Semitic Characters The majority of Hebrew in this book is presented in Hebrew script. When transcription is used for Hebrew in left-to-right linguistic examples or to present some of the other Semitic language data, the scheme closely follows the style presented in The SBL Handbook of Style, with only a few exceptions concerning the representation of vowels (see below). Consonants Hebrew

Transliteration

Hebrew

Transliteration

‫א‬

ʾ

‫ל‬

l

‫ּב‬

b

‫מ‬

m

‫ב‬

b

‫נ‬

n

‫ּג‬

g

‫ס‬

s

‫ג‬

g

‫ע‬

ʿ

‫ּד‬

d

‫ּפ‬

p

‫ד‬

d

‫פ‬

p

‫ה‬

h

‫צ‬



‫ו‬

w

‫ק‬

q

‫ז‬

z

‫ר‬

r

‫ח‬



‫ׂש‬

ś

‫ט‬



‫ׁש‬

š

‫י‬

y

‫ּת‬

t

‫ּכ‬

k

‫ת‬

t

‫כ‬

k

Vowels The common transliteration of Biblical Hebrew vowels is based on the reconstruction of a system that is closely related to the common Semitic threequality, two-quantity sytem (a, ā, i, ī, u, ū). For convenience, I have followed this basic reconstruction, though I depart from the The SBL Handbook of xiii

xiv

Transliteration of Semitic Characters

Style’s system on two points. First, I do not consider it necessary to distinguish between a short and long vowel when there is not distinction in the orthography. For example, the SBL scheme contrasts short ḥireq (i) with long ḥireq (ī), whereas I do not. I only contrast the ḥireq written defectively (without a Hebrew vowel letter), i; and the ḥireq written with a ‫ י‬for a mater lectionis, î. Second, I use ə for the vocal šewa, since the ĕ in the SBL scheme is also used for the ḥatep segol, and is thus confusing. Hebrew

Transliteration

Hebrew

Transliteration ŏ

ִ

i ֳ

‫ִי‬

î ֻ

u

ֵ

ē

‫ּו‬

û

‫ֵי‬

ê ֹ

ō

ֶ

e

‫ֹו‬

ô

‫ֶי‬

ê

  ָ*

o

ֱ

ĕ ְ

ə

ַ

a ֲ

ă ָ

ā

‫ָה‬

â

*Qameṣ-ḥatup.

Abbreviations and Symbols Symbols .

(period) to indicate that a single morph corresponds to multiple features (e.g., 3.sg glosses a morph that carries both third-person and singular features). - (hyphen) to mark clitic boundaries ___ “gap”; see also t(race)) * ungrammaticality ** recursion ∅ phonologically null constituent (e.g., null head, null relative word, null resumptive) x i co-indexation (indicated by subscript) ^  ^ indicates supralinear correction in the manuscript made by the ancient copyist/scribe

General Abbreviations

1 2 3

1st person 2nd person 3rd person acc accusative adj/ap adjective/adjective phrase adv/advp adverb/adverb phrase c common (used for epicene forms) c/comp complementizer compl completive cp complementizer phrase d/dp determiner/determiner phrase dat dative def definite marker dem demonstrative det determiner du dual dur durative evid evidential exst existential f feminine indet indeterminate inf infinitive ipfv imperfective lf Logical Form m masculine

xv

xvi

Abbreviations and Symbols

n neuter nas95

The New American Standard, 1995 negative niv The New International Version njps The New Jewish Publication Society Tanakh nkjv The New King James Version nmlz nominalizer nom nominative np noun phrase NR nonrelative option nrsv The New Revised Standard Version p/pl plural pass passive past past/preterite perf perfect pfv perfective poss possessive pp prepositional phrase ptcp/ptcpp participle/participle phrase rel relative word/complementizer res resultative s/sg singular subj subjunctive t “trace” of a moved constituent v./vv.  verse/verses vp verb phrase wh wh-word (i.e., interrogative or relative words such as when, what, which,   who) x any head constituent xp any phrasal constituent neg



Reference Works

Note: The abbreviations used in this volume follow The SBL Handbook of Style for Biblical Studies and Related Disciplines, 2nd ed. Atlanta: Society of Biblical Literature Press, 2014. In addition, the following abbreviations are used:

BHRG GBH SAB TAD

Merwe, Christo H. J. van der, Jackie A. Naudé, and Jan H. Kroeze. A Biblical Hebrew Reference Grammar. Sheffield: Sheffield Academic Press, 1999 Joüon, Paul, and Takamitsu Muraoka. A Grammar of Biblical Hebrew. Rev. ed. SuBib 27. Rome: Pontifical Biblical Institute, 2006 Kienast, Burkhart, and Konrad Volk. Die sumerischen und akkadischen Briefe des III. Jahrtausends aus der Zeit vor der III. Dynastie von Ur. Stuttgart: Franz Steiner, 1995 Porten, B., and A. Yardeni, editors. Textbook of Aramaic Documents from Ancient Egypt. 4 vols. Jerusalem: Hebrew University Dept. of the History of the Jewish People, 1986–99

Chapter 1

Relative Clauses: Orientation “All languages make use of restrictive relative clauses” (Downing 1978: 381) The linguistic evidence suggests that deep within the human psyche there exists a need to elaborate, to take what is simple and plain and add various types and degrees of description. If this were not so, why does one rarely find a bare noun phrase (NP) subject, a bare verb, and a bare NP object in the context of normal language use? 1 Rather, nouns and verbs are almost always modified by adjectives, adverbs, prepositional phrases (PPs), and subordinate clauses. And of this last category, one of the most common types is the relative clause, 2 the only subordinate clause type that modifies a nominal head 3 rather than a verbal or clausal head. Ancient Hebrew is no exception. If we take the largest corpus of extant data, the Hebrew Bible, in illustration, we find that out of approximately 70,000 total clauses 4 there are over 7,500 relative clauses (introduced by an overt relative word or not). It is thus quite fitting, and representative, to find a relative clause within seven verses of the beginning of the Hebrew Bible (it is the 18th clause), provided in (1): (1)  ‫ּובין ַה ַּמיִ ם ֲא ֶׁשר ֵמ ַעל ָל ָר ִק ַיע‬ ֵ ‫וַ ְּיַב ֵּדל ֵּבין ַה ַּמיִ ם ֲא ֶׁשר ִמ ַּת ַחת ָל ָר ִק ַיע‬ ‘and he divided between the water that was under the firmament and the water that was over the firmament’ (Gen 1:7) 1.  We find such bare clauses quite often in linguistic studies, but these are often contrived for the purposes of grammatical illustration and do not represent natural language usage. 2.  Through this work, I use both relative clause and relative interchangeably for the sake of variety. and the abbreviation “RC” for economy in tables, lists, and example headings. 3.  Head is the standard linguistic term for what is more often called the antecedent in traditional grammar. 4.  The number is an extrapolation from the current stage of the Hebrew Bible syntax within the Holmstedt-Abegg Syntactic Database of Ancient Hebrew (http://individual .utoronto.ca/holmstedt/Ancient_Hebrew_Syntax_Database.html; also see http://www .accordancebible.com/store/details/?pid=HMT-W4.syntax). In the analyzed texts (20,045 verses) there are 60,789 clauses, which is on average 3.03 clauses per verse. There are 3,168 verses remaining to be analyzed for the database (Isaiah, Jeremiah, Hosea, Micah, and Zechariah), which will result in approximately 9,599 clauses.

1

2

Chapter 1

Within the various linguistic frameworks of the twentieth and twenty-first centuries, the relative clause has been the object of as much, if not more, scrutiny than any other clause type. It has a high frequency of usage, independent of text or register type, and its specific features, such as the placement of the head vis-à-vis the relative clause itself, often provide access to the basic structural properties of the language under investigation. Additionally, since the relative clause is apparently used by all known languages (Downing 1978: 381), studying this clause type has the potential for both insightful languagespecific and cross-linguistic generalizations. Likewise, it stands to reason that a thorough study of a commonly used linguistic construction such as the relative clause would be of interest not only to linguists and grammarians but also to those focusing on texts and their interpretation. And yet, the Biblical Hebrew 5 relative clause has been studied in depth only four times in the last 150 years (see below, §1.2) and primarily from a strictly classical grammatical perspective (i.e., before the frameworks of modern linguistics). The insights of previous studies notwithstanding, I will demonstrate in this study that more—much more—can be said about the relative clause and the function words that introduce relatives, ‫אשׁר‬‎, ‫שׁ‬‎, ‫ה‬‎, ‫( זה‬and similar forms), and ‫( מה‬and similar forms). Because the relative clause is so frequently used in the Hebrew Bible and due to the textual distribution of the Hebrew relative words, this topic opens an unusually wide window on the syntax, semantics, and linguistic history of Hebrew. From word order and discourse structure to the interaction of subordination with verb types and the development of Hebrew in the Second Temple Period—this study will demonstrate how the grammar of the relative clause intersects with them all. 1.1.  Relative Typology After word order variation, the relative clause may be the most studied feature of syntax within typological scholarship. No doubt the reason for this relates to one of the typological generalizations that emerged early, that “all languages make use of restrictive relative clauses” (Downing 1978: 381). But what is the typological approach and how does it aid the study of the Hebrew relative clause? In the next few pages, I briefly introduce linguistic typology and the relevant results for the analysis of relative clauses. The features of the relative clause described in this section serve to structure my analysis of the Hebrew relative in chaps. 3–6, the synchronic descriptive core of this work. The assumption behind linguistic typology and cross-linguistic comparison is that the structures of human languages do not vary infinitely but exhibit constraints on the range and type of structural variation (Croft 2003: 5). That is, if enough languages are studied to provide a statistically legitimate repre5.  Henceforth, I will mostly refer to Biblical Hebrew simply as “Hebrew.” For other historical stages (e.g., modern) or corpora (e.g., Mishnaic, Qumran), I will use the appropriate modifier.

Relative Clauses: Orientation

3

sentation, there would emerge a limited number of grammatical feature sets that would account for the observable variation. To that end, typological studies have focused on data collection in significant quantity across numerous languages in order to identify linguistic “universals,” which are statements “intended to represent inductive generalizations across languages” (Croft 2003: 5). The generalizations that emerge from the cross-linguistic comparative approach help to establish the salient features of a given construction in a given language, and so hopefully, a more detailed and accurate description of the phenomenon. Typologists have settled on four basic cross-linguistic generalizations (2), which are determined by whether they hold in all languages for which we have relevant data: (2)  Typological Generalizations a. Absolute universals (AU): “All languages have property X.” b. Implicational universals (IU): “If language A has property X, it will also have property Y.” c. General tendencies (GT): “Most languages have property X.” d. Implicational tendencies (IT): “If language A has property X, it is likely to have property Y.” In his seminal typological study of the relative clause, Downing (1978) lists 1 absolute universal (“all languages make use of restrictive relative clauses”), 25 implicational universals, and 9 implicational tendencies. Based on data collected since Downing’s study, de Vries (2002; 2005) adds 2 absolute universals but reduces the number of implicational universals to 12 by downgrading the remaining 13 to implicational tendencies. The list in (3) provides de Vries’s absolute and implicational universals for relatives (2005: 155; see also de Vries 2002: 35–39), excluding statements that are irrelevant for Hebrew (e.g., implicationals relating to circumnominal relatives). (3)  Typological Universals of Relative Clauses (de Vries 2005, with modification) a. All languages have RCs (AU). b. All languages have semantically headed RCs (AU). c. All languages have restrictive or maximalizing RCs (AU). d. If an RC is semantically nonrestrictive, it is syntactically postnominal (IU). e. The relative gap of a restrictive or maximalizing RC cannot be filled by a lexical copy of the head NP (IU). f. If a language has postnominal RCs, relative complementizers are clause-initial (IU). g. The use of a relative pronoun excludes a resumptive pronoun or clitic, and vice versa (IU).

4

Chapter 1 h. The use of a relative complementizer excludes a relative particle, and vice versa (IU). i. A nonrestrictive RC must contain a relative element (IU).

The list in (4) presents the relevant tendencies, general and implicational, of Hebrew relatives (Downing 1978; de Vries 2002: 35–39): (4)  Typological Tendencies of Relative Clauses a. Most languages use null-head relatives (GT). b. Most languages that have postnominal RCs also have VO word order (IT). c. A subject RC is marked by a clause-initial element (IT). d. For most languages, there is an incomplete participial relativization strategy that is restricted to subjects (GT). While these tendencies do not carry the weight of universals, they do allow us to see how a language such as Hebrew compares with other languages within similar structural features. Given the typological goal of identifying the constraints on variation, it is encouraging that the generalizations emerging from the study of the relative clause can be further sorted into smaller sets of syntactic options and semantic characteristics, such as the word order within relatives, the placement of the relative vis-à-vis the relative head, and the morphology of the relative word. These features are listed in table 1.1 as the parameters 6 of relative clause formation (de Vries 2002: 17; see also Nichols 1984; Keenan 1985; Comrie 1998; Andrews 2007). The list of parameters in table 1.1 accounts for the variety of known relativization strategies manifested in language. Moreover, they provide a convenient and typologically grounded starting point for the analysis given in the corresponding chapters (listed in the table for each category) in this book. However, before we can consider these specific features and how they are manifested in Hebrew, we must consider the relative clause in general. In the next three sections, I will offer a definition of the relative clause with some basic examples and suggest two linguistic clines that isolate the relative clause within the larger grammatical categories to which it belongs. 6. The linguistic concept of parameters arises from the generative approach to linguistic variation. Within generative theory, the initial state of language—its innate, genetic condition within the human brain prior to language acquisition, and its characteristics—is referred to as Universal Grammar. Universal Grammar consists of principles and their parameters (hence, the common name used for generative theory, Principles and Parameters Theory). The principles are the essential rules shared by all humans, and language variation is accounted for by positing that the principles have variable settings, their parameters. For an unimposing introduction, see Carnie 2006: 4–24; also Boeckx 2006: 53–60.

Relative Clauses: Orientation

5

Table 1.1.  The Parameters of Relative Clause Formation Relative Element (chap. 3) a. Presence of relative pronoun b. Presence of complementizer c. Presence of resumptive pronoun

yes/no yes/no yes/no

Relative Head (chap. 4) d. Presence of head e. Hierarchical position of head f. Linear order of head and RC g. Position of determiner w.r.t. N and RC h. Position of Case marker, if any

overt/null externally headed/internally headed head-initial (postnominal)/head-final (prenominal) initial/middle/final on N, on N and RC

Syntax (chap. 5) i. Hierarchical status of RC j. Resumption

embedded within NP/correlative yes/no

Semantics (chap. 6) k. Inflectional completeness of RC l. Kind of modification/relation

finite/non-finite restrictive/nonrestrictive/maximalizing/ free choice

1.1.1.  Relative Definition The first issue to address is a basic cross-linguistic description of the relative clause. Discussing the relative clause as a grammatical item implies that it can be distinguished by some features that exist in and only in all relatives in all languages. Determining the universal features of the relative is not an easy task, however: as table 1.2 illustrates, the syntactic features exhibited by relative clauses in a variety of languages are diverse (Downing 1978: 377). Clearly a universal characterization of the syntax of relative clauses is impossible, since some of these features are mutually exclusive (a vs. b; c vs. d; f vs. g). “These facts,” asserts Downing (1978: 378), “suggest that a universal characterization of the notion ‘relative clause’ can only be given in semantic terms.” The semantic terms that Downing proposes are co-reference, assertion, and modification. The first two apply to all relative clauses, whereas the third feature applies only to restrictive (versus nonrestrictive or appositive) relatives. To illustrate these three semantic characteristics, consider the relative in (5): 7 7.  Note that the head, the relative word, and the gap corresponding to the syntactic role of the relative head within the relative clause are all marked with subscript letters to signal their semantic link. This is referred to as co-indexation, that is, the process whereby two

6

Chapter 1

Table 1.2.  Syntactic Variety of Relative Clauses a. A RC contains a finite verb. b. The verb of a RC assumes a distinctive non-finite form. c. A RC contains a pronoun co-referential with a noun that immediately precedes (or follows) the RC. d. No nominal in the RC is co-referential with a preceding (or following) noun. e. A RC together with a nominal expression forms a noun phrase (NP) constituent. f. A RC is the sole constituent of an NP. g. A RC is not a constituent of an NP. h. A RC begins (or ends) with a distinctive marker. i. A RC contains a marker that is linked by co-occurrence with a nominal marker outside the clause. j. The internal structure of a RC is indistinguishable from that of (some) nonrelative clauses.

(5)  I saw the dogi thati ___ i was black In (5), the NP 8 head the dog is followed by a restrictive relative (unambiguously marked in English by that) that contains a gap (signaled by the ___ in the examples) with which the head is co-referential. The relative in (5) also makes an assertion about the dog, in this case that “(it) was black.” Additionally, the restrictive relative that was black defines its head, the dog, in much the same way that an attributive adjective would (e.g., the black dog). Both adjectives and relative clauses provide information (the assertion) about the modified item that “restrict[s] the reference of [the head] to those possible referents of which that assertion is believed to be true” (Downing 1978: 379; see also Heim and Kratzer 1998: 86–88 on the “property of modification”). Thus, the restrictive relative enables a listener/reader to distinguish the head from other possible or real items in the field of discourse. In (5), the relative that was black restricts the semantic domain covered by the head constituent dog, narrowing the referent from “any dog” to “the black dog.” Finally, clausal status sets the relative apart from other modificational strategies, such as adjectival modification: a relative includes its own predication (separate from the predication of its matrix clause). Examples (6)–(7) illustrate constituents (e.g., nouns, NPs) are marked with the same subscript letter or numeral (the first letter used is usually i, which stands for index, although when multiple pairs of constituents are marked, the letters on each side of i are often used, e.g., h, j). Co-indexation is usually used to indicate that the two constituents marked with the same subscript share the same reference (see Crystal 2008: 71–72, s.v. “chain”). 8.  Currently in generative grammar, the determiner is analyzed as the head of its own phrase (i.e., Determiner Phrase, or DP) that takes an NP as a complement (Carnie 2006: 195–99). However, for simplicity’s sake, I will often use NP in this work, even when DP is more accurate.

Relative Clauses: Orientation

7

the basic differences between an adjectival phrase and a relative in English. Note the embedded predication, X was damaged, in the relative in (7). (6)  Adjectival Modification of NP: The damaged van is his only vehicle. (7)  Clausal Modification of NP: The vani thati ___i was damaged is his only vehicle. Downing’s three semantic characteristics arguably provide an accurate description of relative clauses, but as de Vries notes, co-reference and assertion are not limited to relatives and thus do not adequately delimit the relative clause. For instance, the two juxtaposed clauses in (8) also exhibit co-reference and assertion (de Vries 2002: 14): (8)  I saw the dogi. Iti was black. We can, however, define the relative clause by combining two properties that are both syntactic and semantic in nature: (1) they are subordinate clauses, and (2) they are connected to the matrix clause by a pivot constituent (de Vries 2002: 14; see also Grosu 2002: 145; de Vries 2005: 127–28). As subordinate clauses, relatives are syntactically nonobligatory, like any other type of phrasal or clausal adjunct (e.g., adjectives, adverbs), even though the content of relatives may be semantically necessary for the identification of their head within the discourse. Relative clauses (9), as adjuncts, thus differ from complements (10), which are syntactically obligatory. (9)  I saw the dogi thati ___i was black (10)  I saw that the dog was black In (9), the main clause I saw the dog would be acceptable without the subordinate relative, whereas in (10) the omission of the complement clause that the dog was black would result in an incomplete clause. Turning to the pivot, notice a second difference between relative and complement clauses: in the relative in (9), the head NP, the dog, plays two roles, one within the matrix clause (as the direct object of the verb saw) and one as the head of the relative; in (10), the NP the dog plays only one role, as the subject of the predication was black. This contrast is accurate for Hebrew as well: (11)  Relative Clause ‫אתיָך ֵמ ֶא ֶרץ ִמ ְצ ַריִ ם ִמ ֵּבית ֲע ָב ִדים‬ ִ ‫הֹוצ‬ ֵ ‫ֹלהיָך ֲא ֶׁשר‬ ֶ ‫ָאנ ִֹכי יְ הוָ ה ֱא‬ ‘I am Yhwhi, your god, whoi ___i brought 9 you out from the land of Egypt, from a house of slaves’. (Deut 5:6) 9.  Although the verb within the relative clause carries a first-person agreement feature, the relative head should be identified as ‫יהוה‬, which carries 3ms agreement features. The

8

Chapter 1 (12)  Complement Clause ‫ֹלהיָך‬ ֶ ‫ית ֲא ֶׁשר נְ ָׂש ֲאָך יְ הוָ ה ֱא‬ ָ ‫ָר ִא‬ ‘you saw that Yhwh, your god, carried you’ (Deut 1:31)

In (11), the main clause ‫ אנכי יהוה‬would be acceptable without the subordinate relative clause, whereas in (12) the omission of the complement clause ‫אשׁר‬ ‫ נשׂאך יהוה אלהיך‬would result in an incomplete clause. Turning to the pivot, notice a second difference between relative and complement clauses: in the relative in (11), the head NP ‫ יהוה‬plays two roles, one within the matrix clause (as the complement of the null copula ‘am’) and one as the head of the non­ restrictive relative; in (12), the NP ‫ יהוה‬plays only one role, as the subject of the verb ‫ נשׂאך‬within the complement clause. Another way to characterize the pivot constituent is, as Downing notes, that relative clauses “have the form of clauses from which [a copy of the antecedent] . . . has been deleted” (Downing 1978: 379). In other words, the head plays a role within the relative clause because it has been extracted and promoted to the higher, matrix clause. Consider the English examples in (13) and (14). (13)  a. Don’s vani whichi ___i was damaged b. Don’s van was damaged (14)  a. The dogi whichi I bought ___i b. I bought the dog In each example, the gap ( __ ) that I have marked illustrates where there is a missing constituent: in (13a), there is no overt subject within the relative for the predication was damaged; in (14a), there is no overt object within the relative for the transitive verb bought. When the relatives are compared to the nonrelative examples in (13b) and (14b), we see that the missing constituent corresponds in some way to the head noun that is separated from the clause by the relative element. This fact, in addition to the agreement between the head and the relative word noted above suggests that the head, the relative element, and the gap within the relative have the same semantic reference (i.e., they are co-referential and so co-indexed in the examples). Hebrew also shows the promotion and gap formation of relatives. In (15), I have indicated the extraction site with the gap. (15)  Hebrew Relativization with a Gap ‫ר־ּת ְר ִּתי ָל ֶהם‬ ַ ‫ֶא ֶרץ ֲא ֶׁש‬ ‘a landi thati I explored ___i for them’ (Ezek 20:6) The promotion of the relativized constituent, in this case the bare NP ‫ארץ‬, leaves a gap within the relative at the point of extraction. This gap is technifirst-person voice of the statement, indicated by the identificational clause ‘I am Yhwh’, has determined the deixis and thus the person agreement within the relative.

Relative Clauses: Orientation

9

cally not “empty” but contains either a co-indexed trace of the raised constituent or a copy (that is later deleted at the point of interpretation). 10 Thus, a more accurate representation of relativization would be (16), where I have left a “deleted” copy of the relativized noun to indicate its null syntactic presence (see below, §5.1.2, for further discussion). (16)  (= [15]) [NP ʾereṣi [RC ʾăšer tartî ʾereṣi lāhem]] In relatives that have a resumptive constituent, the gap has been overwritten lexically with a co-referential item, such as the clitic pronoun on the accusative marker in (17). (17)  Hebrew Relativization with a Resumptive Element ‫ָה ָא ֶרץ ֲא ֶׁשר ָּתרּו א ָֹתּה‬ ‘the landi thati they explored iti’ (Num 13:32) In this section, a definition of the relative clause was offered based on its basic syntactic and semantic features: it is a subordinate clause, and it is connected to a higher clause by a nominal pivot constituent. These two features allow us to offer not only a descriptive definition but also a comparative definition: How does the relative differ from other subordinate clauses? And how does it differ from other constructions that modify a noun? Both questions lead us to situate the relative clause on two linguistic clines, a general one concerning the degree to which a subordinate clause is embedded and a more specific one concerning the internal complexity of noun modification strategies. 1.1.2.  The Cline of Embeddedness Clauses that do not have syntactic independence but must be contained within a larger constituent can be characterized in terms of their degree of subordination (Aarts 2006: 252). This degree of subordination is often described in terms of embeddedness. At the embedded end are clauses that are arguments (i.e., obligatory constituents) within a larger constituent (e.g., a complement clause). At the nonembedded end are clauses that are paratactically linked (i.e., juxtaposed) with another clause. 11 Subordinate constituents can thus be placed on a cline of embeddedness, with three primary gradations: nominal, adnominal, and adverbial. These three gradations and their matching clauses types are given in (18) (see Thompson, Longacre, and Hwang 2007: 237–38). 10.  The point of interpretation is called the “logical form” (LF) in Chomskyan minimalism. On the status of traces versus copies within current generative syntax, see Boeckx 2006: 105–9; and Takahashi 2010a; 2010b. 11. As the markedness of subordination decreases (i.e., becomes more paratactic), the way that the subordination of one clause to the other is signaled relates to two factors: (1) the level of co-referentiality between the two clauses (that is, do they concern the same agents and patients?), and (2) the contextual interpretation (that is, does the clause with coreferential constituents, such as pronouns, provide a semantically subordinate assertion, such as cause or purpose?). See also Thompson, Longacre, and Hwang 2007: 237–38.

10

Chapter 1 (18)  Nominal—Adnominal—Adverbial (complement)—(relative)—(causal, purpose, temporal, conditional, etc.)

Nominal (complement) subordinate clauses are embedded clauses that function as the complement of a verb (19) or verbal noun (20). In Hebrew, this mostly occurs with verbs, as in (21), repeated from (12). (19)  Jack saw that snow was falling. (20)  Seeing that snow was falling so early was disturbing. (21)  Complement Clause ‫ֹלהיָך‬ ֶ ‫ית ֲא ֶׁשר נְ ָׂש ֲאָך יְ הוָ ה ֱא‬ ָ ‫ָר ִא‬ ‘you saw that Yhwh, your god, carried you’ (Deut 1:31) Subordinate nominal/complement clauses may be introduced with or without function words that overtly signal the subordination, such as English that, or Hebrew ‫אשׁר‬‎, ‫שׁ‬‎, and ‫( כי‬as with the ‫ אשׁר‬in [21]; see also §7.2). Clauses that are not nominal themselves but modify a nominal constituent (hence, “adnominal”) fall into a single category, the relative clause. Whether introduced by a relative word or not, such as English who(m) (22) or Hebrew ‫אשׁר‬‎(23), the relative clause is subordinate to, and so describes, a nominal head. (22)  The graduate student (who[m]) I taught last term was excellent. (23)  ‫אתי ֵמ ַעל ְּפנֵ י ָה ֲא ָד ָמה‬ ִ ‫ר־ּב ָר‬ ָ ‫ת־ה ָא ָדם ֲא ֶׁש‬ ָ ‫ֶא ְמ ֶחה ֶא‬ ‘I will wipe the man that I created from upon the surface of the land’ (Gen 6:7) Finally, adverbial subordinate clauses modify an event or activity rather than an entity. These subordinate clauses provide a cause, time, purpose, condition, etc., that situates or qualifies the event or action described by the main verb. These clauses are often introduced by overt function words, such as English because, when, if (24), or Hebrew ‫כי‬‎ and ‫אם‬‎ (25). (24)  They went inside because it had become dark. (25)  ‫ל־ה ָא ֶרץ‬ ָ ‫ל־ּפנֵ י ָכ‬ ְ ‫י־מיִ ם ַע‬ ַ ‫ל־ה ֵּת ָבה ִּכ‬ ַ ‫‏וַ ָּת ָׁשב ֵא ָליו ֶא‬ ‘and it [the dove] returned to him, to the ark, because water was upon the surface of the whole earth’ (Gen 8:9) 1.1.3.  The Cline of Internal Complexity As a clause subordinate to a nominal head, the “pivot,” the relative stands on a second cline, which consists of strategies for nominal modification. In this cline, the relative is also positioned at one end, since it is the strategy that is the most internally complex and so contrasted with a bare adjective, which stands at the opposite end. This cline is illustrated in (26).

Relative Clauses: Orientation

11

(26)  adjective—PP—appositive—relative clause (internally) simple > (internally) complex Hebrew examples of each type of nominal modification are provided below: adjectival in (27), prepositional in (28), appositional in (29), and relatival in (30). (27)  ‫ל־מקֹום ֶא ָחד‬ ָ ‫ֹלהים יִ ָּקוּו ַה ַּמיִ ם ִמ ַּת ַחת ַה ָּׁש ַמיִ ם ֶא‬ ִ ‫אמר ֱא‬ ֶ ֹ ‫‏וַ ּי‬ ‘and God said: Let the waters under the heavens be gathered to one place’ (Gen 1:9) (28)  ‫ל־מקֹום ֶא ָחד‬ ָ ‫ֹלהים יִ ָּקוּו ַה ַּמיִ ם ִמ ַּת ַחת ַה ָּׁש ַמיִ ם ֶא‬ ִ ‫אמר ֱא‬ ֶ ֹ ‫‏וַ ּי‬ ‘and God said: Let the waters under the heavens be gathered to one place’ (Gen 1:9) (29)  ‫ן־א ִחיו‬ ָ ‫ת־ׂש ַרי ִא ְׁשּתֹו וְ ֶאת־לֹוט ֶּב‬ ָ ‫‏וַ ּיִ ַּקח ַא ְב ָרם ֶא‬ ‘and Abram took Sarai, his wife, and Lot, the son of his brother’   (Gen 12:5) (30)  (= [1]) ‫ּובין ַה ַּמיִ ם ֲא ֶׁשר ֵמ ַעל ָל ָר ִק ַיע‬ ֵ ‫וַ ְּיַב ֵּדל ֵּבין ַה ַּמיִ ם ֲא ֶׁשר ִמ ַּת ַחת ָל ָר ִק ַיע‬ and he divided between the water that was under the firmament and the water that was over the firmament (Gen 1:7) All four primary strategies lie somewhere within the phrase structure headed by a noun. 12 And each strategy can accommodate significant internal complexity: for example, a noun can be modified by a bare adjective (31) or an adjective that has its own modification (32)–(33). (31)  ‫נ ַֹח ִאיׁש ַצ ִּדיק‬ ‘Noah (was) a righteous man’ (Gen 6:9) (32)  ‫יפת ַמ ְר ֶאה‬ ַ ִ‫‏‏‏וְ ָר ֵחל ָהיְ ָתה יְ ַפת־ּת ַֹאר ו‬ ‘and Rachel was beautiful of shape and beautiful of looks’   (Gen 29:17) (33)  ‫ל־ּביתֹו ְב ִמ ְכ ַסת נְ ָפשׁ ֹת‬ ֵ ‫ּוׁש ֵכנֹו ַה ָּקר ֹב ֶא‬ ְ ‫וְ ָל ַקח הּוא‬ ‘and he and his neighbor close to his house shall take [a lamb] according to the number of people’ (Exod 12:4) In (32), the two cases of the adjective ‫‘ יפה‬beautiful’ are fs to agree with the fs noun ‫ רחל‬and are also in their bound form and so the clitic hosts of the nouns ‫ תאר‬and ‫מראה‬. In (33), the adjective ‫‘ קרב‬near’ modifies the noun ‫שׁכנו‬ 12.  The precise position of each type of modifier with regard to the NP not only differs by language (e.g., English uses prenominal adjectives, Hebrew uses postnominal) but also according to whether the modifier is restrictive or nonrestrictive (see §6.3).

12

Chapter 1

and is itself modified by an internal PP, ‫אל ביתו‬, which clarifies the nature of “near-ness” that is intended. In each of the cases, the complexity of the phrase structure accords to what is required, no more and no less. What distinguishes the relative from adjectives, internal PPs, and apposition is the existence within the relative of its own predication. Appositive phrases, for instance, may be complex and may have a predication but only in an additional subordinate structure (i.e., an appositive NP may be modified by a relative). At its core, an adjective is a nonreferential nominal item juxtaposed to a referential item (a noun), an internal PP is a preposition with its complement adjoined to a noun, and an appositive is typically a noun adjoined to another noun. In contrast, a relative is at its syntactic core a clause adjoined to a noun. The line between relative and nonrelative modification can be quite blurry, however. Consider the English and Hebrew examples in (34)–(37). (34)  A man that was from the house of Levi went out . . . (35)  A man from the house of Levi went out . . . (36)  A man went out from the house of Levi . . . (37)  ‫( וַ ּיֵ ֶלְך ִאיׁש ִמ ֵּבית ֵלוִ י‬Exod 2:1) ‘a man from the house of Levite went out’ In the first English example in (34), the presence of a relative word, that, and a lexical copula, was, clearly signals the relative status of the modifier. In the second example (35), the absence of those two items makes the relative only slightly less clear, since the position of the PP from the house of Levi—immediately following the noun and before the verb—strongly suggests that it modifies the noun and not the verb. But since a relative can lack an overt relative word (e.g., the toy ∅ I bought my daughter versus the toy that I bought my daughter), it is only the absence of the copula that prevents (35) from being a relative. Even more difficult is the third English example in (36), since the postverbal position of the PP allows it to be interpreted either as an extraposed (i.e., displaced to the right) PP modifier of the noun man (compare a man with green eyes appeared and a man appeared with green eyes; see Guéron 1980) or as a verbal adjunct, indicating the origin of the activity described by the motion verb went. While at least the first two English examples provided some structural clues to interpret the status of the modifier, the Hebrew example in (37) combines the challenges of the English examples in (35) and (36). First, the position of ‫ מבית לוי‬allows the PP to modify either the immediately preceding noun ‫ אישׁ‬or the slightly more distant verb ‫( ילך‬for a sketch of Hebrew phrase structure, see below, §2.3.2). Second, Hebrew frequently uses both zero (unmarked) relative clauses (38) and a null copula, whether in a main or subordinate context, such as the relative in (39).

Relative Clauses: Orientation

13

(38)  ‫ת־ה ֶּד ֶרְך יֵ ְלכּו ָבּה‬ ַ ‫הֹוד ְע ָּת ָל ֶהם ֶא‬ ַ ְ‫ו‬ ‘and you shall make known to them the way (that) they should walk in it’ (Exod 18:20) (39)  ‫ל־ה ָּׁש ָמיִ ם‬ ַ ‫ר־ּת ַחת ָּכ‬ ַ ‫ל־ה ָה ִרים ַהּגְ ב ִֹהים ֲא ֶׁש‬ ֶ ‫וַ יְ ֻכּסּו ָּכ‬ ‘and all the high mountains that (were) under all the heavens were covered’ (Gen 7:19) Thus, the syntactic status of the PP ‫ מבית לוי‬in (37) is ambiguous: it could be an NP-internal PP modifying ‫אישׁ‬, an unmarked null-copula relative clause modifying ‫אישׁ‬, or a verbal adjunct to ‫ילך‬. It is only the discourse context that makes one option more felicitous than the others. And in this case, it is much more likely that ‫ מבית לוי‬is an internal PP modifying ‫ אישׁ‬rather than a verbal adjunct. The phrase “house of Levi” is not a locative reference but an affiliative term relating to the man’s tribal association; therefore, the PP is not appropriate as a modifier for the action of “going out.” 13 A similar ambiguity is illustrated by the minimal pair in (40) and (41): (40)  ‫יָמן נִ גָּ ִפים ֵהם ְל ָפנֵ ינוּ ְכּ ָב ִראשׁ ֹנָ ה‬ ִ ְ‫אמרוּ ְבּנֵ י ִבנ‬ ְ ֹ ‫‏וַ יּ‬ ‘and the Benjaminites said, “They are being routed before us as at the first”’ (Judg 20:32) (41)  ‫אמרוּ נָ ִסים ְל ָפנֵ ינוּ ַכּ ֲא ֶשׁר‬ ְ ֹ ‫ן־ה ִעיר ִכּי י‬ ָ ‫אֹותם ִמ‬ ָ ‫אַח ֵרינוּ ַעד ַה ִתּ ֵיקנוּ‬ ֲ ‫וְ יָ ְצאוּ‬ ‫ָבּ ִראשׁ ֹנָ ה‬ ‘and they shall come out after us until we draw them away from the city, because they shall say, “They are fleeing before us like (their fleeing) that was at the first”’ (Josh 8:6) It would be possible to unpack the double preposition ‫ כב‬in ‫ כבראשׁנה‬in (40) so that it has a structure like the ‫ אשׁר‬examples in (41), that is, we could analyze them both as full clausal structures, with a null subject and null copula, ‘like (the routing/fleeing) that it was at first’. In this analysis, the relative in (40) would be unmarked, whereas it is overtly marked by the ‫ אשׁר‬in (41). However, it is also possible to take (40) as a complex PP (as translated), while the ‫אשׁר‬ in (41) forces a more complex clausal structure (as translated). Because it is grounded in linguistic economy, that a structure is only as complex as is required, the second option is preferable—that, although the two examples might appear to be structurally similar, they are in fact distinct, since the double 13.  Determining the relative or nonrelative status of PPs such as that in (37), especially without the subtle signals that one might obtain from native speaker input, is impossible— the head noun is modified either way, and there is no discernible difference in the resulting semantics. So, invoking Occam’s razor—unless internal clues suggest the more complex structure of a relative clause, I take PPs such as that in (37) to be nonrelative.

14

Chapter 1

preposition in (40) allows a phrasal analysis, while the ‫ אשׁר‬in (41) necessitates a clausal analysis. Using the two clines of embeddedness (18) and internal complexity (26), relative clauses can be isolated among the numerous types of subordinate clauses and noun modification strategies. Within the worlds established by these two clines, the Hebrew relative clause displays a bit of syntactic and semantic trickery. First, in Hebrew there is overlap between the relative clause and the nominal clause, since both are introduced by the function words ‫אשׁר‬ and ‫שׁ‬, as in (42) and (43) (see also [11] and [12], above). (42)  ‫ל־ה ַּמ ֲע ִׂשים ֶׁשּנַ ֲעׂשּו ַּת ַחת ַה ָּׁש ֶמׁש‬ ַ ‫ת־ּכ‬ ָ ‫יתי ֶא‬ ִ ‫ָר ִא‬ ‘I have seen all the deeds that have been done under the sun’   (Eccl 1:14) (43)  ‫ן־ה ִּס ְכלּות‬ ַ ‫יתי ָאנִ י ֶׁשּיֵ ׁש יִ ְתרֹון ַל ָח ְכ ָמה ִמ‬ ִ ‫וְ ָר ִא‬ ‘I saw, I, that the profit of wisdom is more than (the profit of) foolishness’ (Eccl 2:13) In (42), the ‫ שׁ‬clause modifies the noun ‫ מעשׂים‬as a relative clause; in (43), the ‫ שׁ‬introduces the clausal complement of the verb ‫ראיתי‬. Second, Hebrew allows relative clause extraposition, making it sometimes difficult to identify the intended relative head. Consider example (44): (44)  ‫יכם‬ ֶ ֵ‫ם־אד ֹנ‬ ֲ ‫יתם ַה ֶח ֶּסד ַהּזֶ ה ִע‬ ֶ ‫יהם ְּב ֻר ִכים ַא ֶּתם ַליהוָ ה ֲא ֶׁשר ֲע ִׂש‬ ֶ ‫אמר ֲא ֵל‬ ֶ ֹ ‫וַ ּי‬ ‫ם־ׁשאּול‬ ָ ‫ִע‬ ‘And he said to them: Blessed are you to Yhwh, who you did this faithful thing with your masters, with Saul’ (2 Sam 2:5) In (44), the relative clause might first be processed so that it modifies the closer NP, ‫יהוה‬, since relatives are typically positioned immediately following their heads. But the 2mp agreement affix on the verb ‫ עשׂיתם‬clarifies that the relative head is really the 2mp pronoun ‫אתם‬, which means that the relative clause itself is positioned at a distance from its head. Third, Hebrew often includes resumption of the relative head within the relative by means of a co-referential pronoun, but identifying a pattern for the presence (45) or absence (46) of resumption has eluded Hebraists. (45)  ‫ֹלהיָך ּבֹו‬ ֶ ‫ַּב ָּמקֹום ֲא ֶׁשר ְיִב ַחר יְ הוָ ה ֱא‬ ‘in the place that Yhwh your god chooses it’ (Deut 12:18) (46)  ‫ר־יִב ַחר יְ הוָ ה‬ ְ ‫ל־ה ָּמקֹום ֲא ֶׁש‬ ַ ‫ֶא‬ ‘to the place that Yhwh chooses ∅/[it]’ (Deut 12:26) In both (45) and (46), the relative head is the same (the NP ‫)מקום‬, the verb is the same (the imperfective 3m s ‫)יבחר‬, and the subject is the same (the proper

Relative Clauses: Orientation

15

noun, ‫)יהוה‬. Yet, in (45), the relative clause resumes the head by means of the clitic 3ms pronoun ‫ ו‬in the PP ‫‘ בו‬in it’, whereas in (46) the head is not resumed (note also that the lack of resumption leaves the transitive verb ‘choose’ without an overt complement). While the ‘when’ and ‘where’ of resumption in the Hebrew relative clause has been noted in most reference grammars, what have been lacking are the ‘what’ (i.e., a structural analysis) and a compelling explanation for ‘why’. For the structural, we must account for how the resumptive constituent relates to the relative head; for the motivation, we must try to determine what provoked an ancient Hebrew speaker/writer to use a resumptive element in one clause and not another. Fourth, while most relative heads in Hebrew exist in their morphological free form, 14 such as above in (45) and (46), some head nouns exhibit their bound 15 form, as in (47): (47)  ‫ל־ה ִּמ ְׁש ָּכן יַ ֲחנּו‬ ַ ‫ָּכל־יְ ֵמי ֲא ֶׁשר יִ ְׁשּכֹן ֶה ָענָ ן ַע‬ ‘all the days-that the cloud would settle over the tabernacle, they would camp’ (Num 9:18) In (47), the noun ‫‘ ימי‬days of’ is the head of the relative clause. This confluence of syntax and morpho-phonology may suggest that relatives with a bound form head exhibit a greater degree of subordination—they are syntactically embedded, similar to complement clauses. Fifth, context unequivocally establishes that Hebrew employs both restrictive (48) and nonrestrictive (49) relatives, though, again, both the distinction itself and any marking strategies have escaped previous attention. (48)  ‫ן־ה ָא ָדם ְל ִא ָּׁשה‬ ָ ‫ר־ל ַקח ִמ‬ ָ ‫ת־ה ֵּצ ָלע ֲא ֶׁש‬ ַ ‫ֹלהים ֶא‬ ִ ‫‏וַ ֶּיִבן יְ הוָ ה ֱא‬ ‘and Yhwh God built the rib that he had taken from the man into a woman’ (Gen 2:22) (49)  ‫יאָך ֵמ ֶא ֶרץ ִמ ְצ ַריִ ם ִמ ֵּבית ֲע ָב ִדים‬ ֲ ‫הֹוצ‬ ִ ‫ן־ּת ְׁש ַּכח ֶאת־יְ הוָ ה ֲא ֶׁשר‬ ִ ‫ִה ָּׁש ֶמר ְלָך ֶּפ‬ ‘Take care, lest you forget Yhwh, who brought you out of the land of Egypt’ (Deut 6:12) The relative clause in (48) unambiguously identifies the ‫ צלע‬that is referenced—it is the ‫ צלע‬that God took from the man (there is no other possible rib that God had access to). In contrast, the relative head in (49) is the proper name

14.  The term free references the morphological shape and distributive restrictions of what has traditionally been called the absolute state of nouns, in contrast to the bound, or construct, forms. See Holmstedt and Dresher 2013. 15.  The term bound references the morphological shape and distributive restrictions of what has traditionally been called the construct state of nouns, which is best understood as a process of cliticization. See Holmstedt and Dresher 2013.

16

Chapter 1

‫יהוה‬, which as a unique referent cannot be further defined; thus the relative is nonrestrictive. The Hebrew examples discussed in this introduction provide just a taste of the complexity of the Hebrew relative clause. Yet, in comparison with the effort previous scholars have spent investigating the etymology of the relative words ‫ אשׁר‬and ‫( שׁ‬see chap. 3), the syntactic, semantic, and discoursepragmatic features of the relative clause have been neglected. 1.2.  Previous Scholarship Since the relative clause is a common subordinate clause type, every introductory or reference grammar for Hebrew includes some description of it, however brief. It is not altogether easy, however, to discern which features of the relative clause grammarians have identified as salient. Even a cursory glance at many grammars reveals an often confusing description and/or order of presentation. For example, the classic reference work, Kautzsch 1909 (German)/1910 (English = GKC), divides the discussion of “relative clause matters” into two sections, one dealing with the relative pronoun (§138) and one dealing with the relative clause (§155). The section on the relative “pronoun” 16 begins clearly, with a discussion of the etymology of ‫אשׁר‬, but then proceeds well beyond etymology to various syntactic issues—material that better belongs to the section on the relative clause. Furthermore, the presentations regarding unselbständige/incomplete (= dependent) versus selbständige/complete (= independent) relative clauses, the presence or absence of Rückbeziehung/ resumption by a pronoun or adverb, and the presence or absence of a relative word all lack organizational clarity. The layout and clarity of most other Hebrew grammars regarding the relative clause suffer similarly (the exception is the chapter on the relative clause in IBHS). The value of sorting through and summarizing the descriptions provided in the reference grammars is questionable, though I did this diligently in Holmstedt 2002 (see that work for an overview). In this volume, I have instead chosen to refer the reader to the appropriate sections as the data are presented and analyzed. I have consulted and often cite the following reference works: Ewald 1891; Davidson 1901; GKC; Joüon 1923; Meyer 1992; Schneider 1974; IBHS; BHRG; and GBH. Similarly, I interact with relevant article-length studies at the appropriate points. Beyond the small sections in the reference works and the meager number of articles that treat the relative clause in some way, there have been only four monographic studies in the last century and half: by Arthur Sperling in 1876; 16.  The relative elements ‫אשׁר‬‎, ‫שׁ‬, and ‫ ה‬are indeclinable, and their distribution does not match that of Hebrew pronouns. Thus, these Hebrew relative elements are not “pronouns,” though the ‫ז‬-series relative elements show evidence of once having been pronominal. See below, §3.2.

Relative Clauses: Orientation

17

Victor Baumann in 1894; Carl Gaenssle in 1915; and Yitzhaq Peretz in 1967. Sperling’s and Baumann’s works are both insightful but also brief and descriptively incomplete. Gaenssle’s study is focused on the function word ‫אשׁר‬ and so considers more clause types than simply the relative. Finally, Peretz’s 1967 monograph (in Hebrew) describes the features of the relative clause (and related structures using ‫ שׁ‬and ‫ )אשׁר‬from Biblical to Modern Hebrew. All four works contribute significantly to our understanding of the nature of this clause type in Hebrew and contain a wealth of data. The first three, however, were written before the perspective and tools of modern linguistics were available, and though Peretz could have availed himself of some linguistic framework current in the 1960s, his work is firmly descriptive. Moreover, the ambitious scope of Peretz’s work is both its greatest strength and its weakness—it provides the historical long view of the Hebrew relative but misses significant features of the biblical construction. 1.3.  Corpus and Conventions for the Representation and Translation of Hebrew The present work is a comprehensive and exhaustive investigation of the relative clause in the Hebrew Bible. It is comprehensive in the sense that I have addressed (or, at least, considered and left unaddressed if irrelevant) all grammatical features that are salient in the analysis of the Hebrew relative clause. This work is exhaustive in that I have analyzed every single relative clause in the Hebrew Bible. For those relatives that are introduced by an overt relative word, I compiled the data using both a recognized concordance (EvenShoshan 1990) and the digital Hebrew Masoretic Text with Westminster Hebrew Morphology (HMT-W4, formerly BHS-W4), v. 4.16, in the Accordance Bible Software. For those relatives that are not introduced by an overt relative word (“zero,” or “unmarked” relatives), I both culled the reference grammars and focused studies, and simply read the Bible (toward the end of this project, the near completion of the syntactic database that I developed and oversaw with Martin G. Abegg made it possible to check my examples and, occasionally, to add a few that I had not seen). 17 Additionally, I isolated and analyzed all the overtly introduced relatives in the extant Hebrew inscriptions, the Hebrew portions of Ben Sira, and the Qumran texts, as well as a number of zero relatives in these corpora (which can only be found by reading the texts). While a relatively small number of relatives (approximately 1 in 20) display features that result in unavoidable syntactic or semantic ambiguity—and even I waffle on how to interpret such examples in their discourse context when I read them anew—their options always lie within the parameters that this volume establishes. Thus, the comprehensive and exhaustive nature of this work means 17.  On the syntax database, see n. 4 above.

18

Chapter 1

that, though the reader may not find that I have analyzed a difficult example about which he or she has questions, the options for how to analyze the difficult clause are set out clearly within. In preparing this volume, I have assumed that the primary audience of this work will be individuals trained in Hebrew studies. Therefore, data from the Hebrew Bible are presented in consonantal Hebrew script when given in the main text and in vocalized Hebrew script in the data examples. For typographic convenience, I omit the Masoretic prosodic system, the ‫טעמים‬, and other textual markers, e.g., the ‫סוף פסוק‬. Occasionally I render a Hebrew example in transliteration in order to use the standard left-to-right layout of linguistic notation (for the transliteration scheme, see pp. x–xi). Other Semitic language data (chapter 8) are presented in conventional transliteration. Non-Semitic languages that are not the common languages of Western scholarship (e.g., Hindi) are presented in Roman script with a line of linguistic glossing between the data and the English translation; for these glosses, I relied on the source of the data as well as the Leipzig Glossing Rules rev. 2008. Finally, I offer here a brief word on my English translations. Unless otherwise noted, all translations are my own. It is important to note, though, that this work is not concerned with translation theory or practice. While I could offer smoother English translations, I often forego style and fluidity in English for a more direct representation of the Hebrew structure, especially when the specific features of the Hebrew that I am discussing are not clearly illustrated by a smooth English translation.

Chapter 2

Issues of Method and Theory 2.1.  Methodological Challenges for the Biblical Hebrew Linguist Any description of a linguistic feature from an ancient language must deal with a number of methodological problems. Moreover, in the face of a veritable buffet of contemporary linguistic theories—few of which are familiar to Hebraists—it is desirable, if not critical to be both theoretically transparent and explicit. This assumes, of course, that the researcher understands the absolute necessity of having a theory of language and following a careful methodology. Without a theory, data are worthless, since they cannot be interpreted; without a clear methodology by which data are interpreted in light of a theory, the conclusions drawn can be nothing but suspect. This chapter, therefore, provides the readers of this work with a road map by which to navigate my analysis of Hebrew syntax in general and the relative clause in specific. It also stands as a standard by which to gauge the coherence of my analyses. And finally, it distills a great deal of my theoretical and methodological reflection from 15 years of attempting to elucidate Hebrew grammar using the tools of modern linguistic theory. This chapter is thus a statement of theory and methodology for this work in particular but also a general programme for future work in the linguistic study of Hebrew grammar. To that end, in §2.1.1, I discuss the distinction between philology and linguistics and situate the present work with regard to both. In §2.1.2, I identify the challenges inherent to the linguistic analysis of no-longer-spoken languages. And in §2.1.3, I provide my assessment of the linguistic status of ancient Hebrew. These three subsections acknowledge the necessity of being philologically aware when approaching Biblical Hebrew grammar. The subsequent sections clarify for the reader my theoretical stance: in §2.2, I introduce the linguistic framework I employ, and in §2.3, I sketch my theoretically grounded approach to Hebrew syntax. The first three discussions are driven by three large questions that face the ancient language scholar: What is it that we do? How can we analyze a language no longer spoken? And, what is the linguistic reality of our object of study? The fourth and fifth discussions are less philosophical and provide the theoretical context for my analysis of the Hebrew relative clause. 19

20

Chapter 2

2.1.1.  Philology and Linguistics The distinction between philology and linguistics can be traced to the early to mid-nineteenth century and the scholarship of August Schleicher (1821– 1868). Schleicher was the first to promote a sharp distinction between Philologie, a historical inquiry, using language as a vehicle in the study of culture, and Linguistik, the scientific investigation of language itself (Koerner 1997: 170). Schleicher asserted that quite often the linguist and philologist need each other, but he envisioned the two as having distinct objects of investigation: “In contrast to a philologist, who could work on the basis of the knowledge of only one language (e.g., Greek), a linguist, in Schleicher’s view . . . needs to know many languages, to the extent that ‘Linguistik’ becomes synonymous with ‘Sprachvergleichung’” (Koerner 1997: 170). 1 Schleicher’s conceptual heirs, the Junggrammatiker, or “neogrammarians,” adopted his essential distinction but did not share the view that the linguist and philologist needed each other. In fact, they defined linguistics in such a way as to exclude Schleicher! It . . . seems strange to us that Berthold Delbrück in his 1880 Einleitung in das Sprachstudium . . . presents Schleicher ‘in the essence of his being’ as a philologist, since thirty years before it had been Schleicher (and no one else) who had clearly set off his work from those of the (classical) philologists. However, if one remembers the ‘eclipsing stance’ which the Young Turks at the University of Leipzig and elsewhere in Germany took vis-à-vis their elders from the mid-1870s to the mid-1880s, one might not be surprised that Delbrück (1842–1922) distorted the facts to suit his argument, namely that the junggrammatische Richtung represented ‘new endeavors’ . . . rather than a continuation of research along established lines. (Koerner 1997: 171)

This association of philology with “older” historical-comparative study and linguistics with “newer” methods does not reflect the earliest uses of the terms (i.e., Schleicher’s) but was a stance adopted soon thereafter, and this eclipsing stance toward philology continued through the twentieth century. Indeed, one can still hear dismissive comments toward philology in the halls of linguistics departments, and for their part, philologists often reciprocate with a hostility of their own toward contemporary linguistic theories and methodologies. I take a stance of complementarity that allows for a functional and productive working relationship between philology and linguistics: the question each approach addresses is distinct from but important to the other. As Faarlund (in the spirit of Schleicher) asserts, “A linguist working on historical material 1.  In this quotation, Koerner references the introductory chapter, “Linguistik und Philologie,” in Schleicher’s 1850 work Die Sprachen Europas in systematischer Uebersicht (Bonn: König; reprinted with introduction by K. Koerner; Amsterdam: Benjamins, 1983).

Issues of Method and Theory

21

depends on a good philologist” (Faarlund 2005: xiii). Hale similarly asserts that the philologist establishes “the attributes of a text, many of which may be relevant for subsequent linguistic analysis” (Hale 2007: 21). He argues that linguists are not interested in ancient texts qua texts but the linguistic structures the texts contain (2007: 22 n. 5) and asserts that “a rigorous linguistic analysis of the text which was philologically established from a given historical artifact would lead, in principle, to a hypothesized linguistic structure for the relevant aspects of that text” (2007: 23). It is from this linguistic text that the linguist deduces the grammar that produced the language “output” represented in the linguistic text. The reconstruction of the grammar represented in the philological text is constrained by grammatical principles arising from theoretical and cross-linguistic research. Where this sequence becomes a cycle is when the reconstructed grammar is then used by the philologist to better reconstruct the artifactual text (see Holmstedt 2012). In this view of philology and linguistics, any Hebraist who investigates the linguistic features of a particular corpus, e.g., a passage or book of the Hebrew Bible, is engaging in philological analysis. In contrast, those who examine linguistic features in light of some linguistic theory in order to make sense of some dialect or stratum of ancient Hebrew as a system are engaging in linguistic analysis. And those of us who examine specific texts or corpora as well as linguistic systems can identify ourselves as both philologists and linguists. This study of the Hebrew relative clause is first a linguistic analysis in that I am concerned with the features of this particular clause type; however, I also occasionally illustrate the implications of my analyses for the interpretation of the ancient Hebrew texts, and at these points my work becomes philological (see also Holmstedt 2006a). 2.1.2.  Analyzing No-Longer-Spoken Languages A central methodological question for the study of ancient, no-longerspoken, or “dead” languages is whether and how the tools of linguistic analysis can be applied (and, if necessary, modified) to the fixed corpus, which will always produce, as Kiss says, an “underdetermined” analysis (Kiss 2005b: 2). Faarlund succinctly summarizes the challenge: Describing the syntax of a dead language is rife with theoretical problems and methodological stumbling blocks. A major question is determining what the description should seem to describe. Traditional, philologically oriented grammars of dead languages are descriptions of finite corpora. Modern generative grammar, on the other hand, aims to account for speakers’ linguistic competence, their internalized grammar. . . . In the absence of live speakers and their intuitions, and in the absence of contemporary syntactic descriptions, our sources of knowledge of the internalized grammar of the speakers are limited to extant texts, besides grammatical theory. (Faarlund 2005: 1)

22

Chapter 2

Despite these challenges, the analysis of no-longer-spoken languages is on the upswing in linguistics, even from within a generative framework. The linguistic analysis of, e.g., Old English, Old Norse, Middle Dutch, Middle French, or Early Modern English has an increasingly respectable place at the theoretical table. Moreover, with the addition of generative-oriented studies on Old and Coptic Egyptian, Sumerian, Akkadian, Biblical Hebrew, IndoEuropean, Classical Greek, Latin, and Sanskrit (see the studies collected in Kiss 2005), 2 this trend is expanding beyond the study of Western European languages. Even so, methodological obstacles loom large, and discussion of these obstacles lags behind the analysis of the data. First among the issues is determining what linguistic information we can and cannot receive from ancient written sources. Campbell aptly describes this initial problem in his textbook on historical linguistics: Very often, what information we can derive from interpreting the structure of the language at the time when the texts were written and extrapolating from that for the understanding of the history of the language is a matter of luck, of what happens to show up in the sources available. In the best case, we may have descriptions or commentaries about the pronunciation at the time, and these can be immensely helpful. In most situations, however, we are not so fortunate as to have worthwhile, readily interpretable phonetic descriptions from the past. Other valuable sources of phonetic information include rhymes, metre, occasional spelling, transliterations of forms from other languages whose phonology is better known, aid from translations from texts known in other languages, and clues from related languages and dialects. (Campbell 2004: 368–70)

Though Campbell’s focus in the quotation above is on phonetics and phonology, the problems are the same for other areas of grammar. To sort through these issues, Lass identifies five critical questions for the linguistic analysis of written historical sources (Lass 1997: 44, with slight modification): 1. How far are the texts “representative” of “languages” and in what way(s)? 2. How do we separate the variations we find in noncodified traditions from genuine scribal errors; how do we distinguish evidence from garbage? 3. How do we decide, in the case of dead languages, what meanings and structures ought to be assigned to the items in a text, regardless of their phonetic interpretation?

2.  It is notable that the studies in Kiss’s volume “aim to demonstrate that descriptive problems which proved to be unsolvable for the traditional, inductive approach to ancient languages can be reduced to the interaction of regular operations and constraints of the hypothetical Universal Grammar” (Kiss 2005b: 3).

Issues of Method and Theory

23

4. How do we arrive at the meanings of lexical items or grammatical forms in languages no longer spoken and lacking—unlike Latin or Greek in the West or Sanskrit in India—a subsequent preservative or antiquarian tradition? 5. How if at all can we recover the (putative) original stylistic effects of particular constructions of lexical choices, etc.?

Some of these questions appear easier to answer than others. For example, to answer questions ##3–4 with regard to Hebrew studies, we can point to the rich preservative tradition for the Hebrew Bible, from the Masoretic layers of information (the vocalization, ‫טעמים‬, and ‫ )מסורות‬and ancient translations, such as the Greek Septuagint (including the important evidence of transliteration of names), to the transliteration information that can be gleaned from the reconstruction of Origen’s Secunda and numerous early commentaries. 3 Unfortunately, for a syntactic study such as the present work, none of the ancient non-Hebrew witnesses to the Hebrew language provides more than marginally interesting information. Even in the case of how the various translators of the Septuagint rendered the relative clauses in the Hebrew Bible into Greek, we must remember that two facts severely mitigate the value of such information: first, the Greek translation reflects the translators’ understanding of the Hebrew text, which may or may not have been accurate, and second, the translation reflects what was grammatically possible and felicitous within the target language and in many cases more likely represents a significant linguistic interpretation rather than a one-to-one gloss of the Hebrew. Thus, despite its rich body of textual tradition, in my analysis of the relative clause in the Hebrew Bible I only rarely appeal to the various components of this non-Hebrew tradition. If we modify Lass’s second question so that it refers to the manuscript variation even within the codified traditions, it becomes particularly relevant to the study of the Hebrew Bible but also to other ancient texts such as Ben Sira and the Qumran texts. How should we understand, for example, the variation between the null-head relative begun by ‫‘ כאשׁר‬like that’ in the Masoretic Text of Eccl 5:14 (50) and the causal ‫‘ כיא‬because’ in the same verse of 4QQoha (51)? (50)  ‫ֲׁשר יָ ָצא ִמ ֶּב ֶטן ִאּמֹו‬ ֶ ‫‏ ַּכא‬ ‘like-that he came out from his mother’s womb’ (Eccl 5:14 MT) (51)  [‫‏כיא [יצא מבטן אמו‬ ‘because [he came out from his mother’s womb]’ (4Q109 [4QQoha] col. 1, frg. 1 i 2; Ulrich 2000: 222) 3. See Sáenz-Badillos (1993: 80–85) for an overview of the core research on reconstructing Hebrew phonology from the Greek and Latin transcriptions. See also Kutscher 1982: 106–7, 143, 145.

24

Chapter 2

Is the variation between (50) and (51) a scribal mistake? Or perhaps some scribe updated the language to reflect his own “grammar”? Similarly, the manuscripts for the book of Ben Sira show four cases of alternation between ‫א‬ ‫ שׁר‬and ‫שׁ‬, presented here in (52)–(55): 4 (52)  Sir 3:22 a. Manuscript A ‫במה שׁהורשׁית התבונן‬ ‘reflect on what that you have been empowered (over it)’ b. manuscript B ‫באשׁר הורשׁיתה התבונן‬ ‘reflect on ∅/(the thing) that you have been empowered (over it)’ (53)  Sir 15:17 a. Manuscript A ‫אשׁר יחפץ ינתן לו‬ ‘what (= [the thing] that) he desires (one) will give to him’ b. Manuscript B ‫וכל שׁיחפץ יתן לו‬ ‘all that he desires (one) will give to him’ (54)  Sir 34:15 a. Manuscript B ‫ובכל שׁשׂנאת התבונן‬ ‘and reflect on all that you hate’ b. Manuscript B, margin ‫כל אשׁר שׂנאת התבונן‬ ‘all that you hate understand’ (55)  Sir 44:9 a. Manuscript B ‫וישׁ מהם אשׁר אין לו זכר‬ ‘and there are some of them that there is no memorial for him’ (i.e., ‘for whom there is no memorial’) b. Manuscript M ‫וישׁ מהם שׁאין לו זכר‬ 4.  All examples from Ben Sira are taken from the 1973 Academy of the Hebrew Language edition, although I have compared them with the newer edition in Beentjes 2003.

Issues of Method and Theory

25

‘and there are some of them that there is no memorial for him’ (i.e., ‘for whom there is no memorial’)

One could make a reasonable case for taking each of these variations, the single example in Eccl 5:14 or the four examples in Ben Sira, to be scribal mistakes and/or to be scribal updating and thus include them as examples of the scribes’ later grammar. Either way, they could on principle be discounted from any grammatical description. At best, they should be identified as marginal constructions. However, marginal does not accurately describe the rare and inexplicable redundancy of the relative word in Eccl 8:17 (56). (56)  Eccl 8:17 ‫ת־ה ַּמ ֲע ֶׂשה ֲא ֶׁשר‬ ַ ‫יּוכל ָה ָא ָדם ִל ְמצֹוא ֶא‬ ַ ‫ֹלהים ִּכי לֹא‬ ִ ‫ל־מ ֲע ֵׂשה ָה ֱא‬ ַ ‫ת־ּכ‬ ָ ‫יתי ֶא‬ ִ ‫וְ ָר ִא‬ ‫יִמ ָצא‬ ְ ‫ת־ה ֶּׁש ֶמׁש ְּב ֶׁשל ֲא ֶׁשר יַ ֲעמֹל ָה ָא ָדם ְל ַב ֵּקׁש וְ לֹא‬ ַ ‫נַ ֲע ָׂשה ַת ַח‬ ‘I saw the whole work of God, that man is not able to “find out” the deed that happens under the sun, in that that man toils to seek but does not find’ (8:17) In (56) the sequence ‫בשׁל אשׁר‬, lit., ‘in that that’, defies syntactic analysis, no matter how we try to parse it. It is thus not surprising that one finds a variety of translations: ὅσα ἂν ‘as much as’ (Septuagint), ‫‘ בד‬concerning that of’ (Targum), et quanto plus ‘and how much more’ (Vulgate), ‘because though’ (kjv), ‘however much’ (nrsv), ‘despite’ (niv), ‘even though’ (nas95). 5 A similar 5.  Ecclesiastes’ use of this collocation is often explained as a calque on Aramaic ‫בדיל ד‬ (see Delitzsch 1875: 345; Whitley 1979: 77; Fredericks 1988: 220; Schoors 1992: 145–46; Seow 1997: 289–90; Fox 1999: 289). However, Gordis’s comment is instructive: In 8:17, ‫ם‬ ‫ ְּב ֶׁשל ֲא ֶׁשר יַ ֲעמֹל ָה ָא ָד ‏‬is regarded as a clumsy, artificial reproduction of the Aramaic ‫ בדיל ד‬and rendered: “Because the man labors.” But the construction has its analogy in two perfectly clear contexts, in Jonah 1:8 ‫י־ה ָר ָעה ַהּזֹאת ָלנּו‏‬ ָ ‫“ ַּב ֲא ֶׁשר ְל ִמ‬For the sake of whom has this evil come upon us” and even more closely in 1:12 ‫ם‬ ‫יכ ‏‬ ֶ ‫“ ְב ֶׁש ִּלי ַה ַּס ַער ַהּגָ דֹול ַהּזֶ ה ֲע ֵל‬For my sake, on my account, this great storm has come upon you.” Hence the idiom in Qohelet means “for the sake of which, on account of which,” and the verse is to be rendered literally: “I saw that a man cannot discover all the work that is done under the sun, for the sake of which a man may labor to seek it out, yet he shall not find it.” If the phrase be rendered “because,” a clumsy locution emerges, as in the Jewish Version: “Because though a man labor, etc.,” for there is no basis for “though” in the text. But even this sense “because” is derived from the literal meaning, “on account of the fact that = seeing that a man labors.” (Gordis 1949: 107)

I agree with Gordis that Hebrew ‫ בשׁל אשׁר‬makes no sense as a calque on Aramaic ‫בדיל ד‬ ‘because’. Additionally, I agree that the first half of the collocation ‫ בשׁל אשׁר‬in Eccl 8:17 has a partial parallel in the two Jonah passages. But after this, Gordis’s analysis fails (as do all others). First, the phrase ‫ בשׁל אשׁר‬as a whole cannot be explained by recourse to Jonah, since in Jonah the redundant second relative word is missing, the lack of which leaves the Jonah examples grammatically explicable (see below, esp. n. 92, and §§3.3.3, 4.3.2 n. 24, and 7.4 n. 42 for further discussion of the Jonah examples). Second, Gordis’s gloss of ‫בשׁל אשׁר יעמל‬ ‫‘ האדם‬on account of the fact that’ = ‘seeing that a man labors’, while better than most, still ignores one or the other of the two relative words. It is also worth noting here that the lack of

26

Chapter 2

sequence occurs four times in the Qumran texts, all in 4QMMT, illustrated in (57): 6 (57)  4Q394 frg. 3_7i:15–16 ‫להזה^י^ר בדבר הזה בשל שלוא י]היו[‏ מסיא]י[ם את העם‬ ̇ ‫‏הכוהנ̇ ]ים[ ̇ר ̇או‬ ‫עוון‬ ‘the priests see to warn in this matter in (the event) that that (?) they do not load the people (with) guilt’ The Ecclesiastes and Qumran examples, reinforced by the repetition of the relative ‫ שׁ‬in a single Qumran line (58), (58)  4Q394 frg. 8 iv 5–6 (= 4Q396 frg. 1–2 ii 7) ‫‏]ו[אף על המוצקות אנחנו אומר]ים[ שהמ שאין בהמ‏ ]ט[הרה‬ ‘concerning streams of liquid we say that they that (?) there is not in them purity’, drive home the issue of manuscript evidence: are these examples of scribal error, “garbage” in Lass’s words, or evidence of the authors’/scribes’ grammar, however marginal or rarely used. It is one of the great, and often forgotten, obstacles in the study of ancient Hebrew. But the linguistic data contained in a text produced by a scribal tradition, particularly a long one like that of the Hebrew Bible, depends on how mechanical the transmission of the text was. In this regard, it is critical to keep in mind that, aside “from simple scribal error, the translator or glossator may have a defective knowledge of the source language; or may have some agenda other than providing simple lexical equivalence; or the lemma in the source language may itself be referentially indeterminate” (Lass 1997: 83). On this issue of the use of texts that are the products of a scribal tradition and dating schemes, Ó Buachalla makes four points about the Irish scribal tradition that are relevant to the study of Biblical Hebrew, which I summarize here: 1. Professional scribes tended to be mere passive transmittors of material, but the greater the distance between the language of the transmitted text and the scribe’s native grammar, the more they interfered (consciously or unconsciously) with it linguistically. The result in most scribal traditions is a text in which it is possible to recognize the various linguistic layers that have accreted. The linguist’s task is not merely extracting linguistic data in a naïve way, but dating the discernible layers and of establishing a linguistic chronology. (Ó Buachalla 1982: 425–26) even one case of ‫ בשׁל שׁ‬in the Mishnah severely decreases the plausibility of ‫( בשׁל אשׁר‬or ‫ב‬ ‫ )שׁל שׁ‬as grammatically acceptable. 6.  See also ‎4Q394 frg. 3–7 i 19, ‎4Q398 frg. 14–17 ii 6 (= 4Q397 frg. 23 2). It is also possible that one more example exists in 4Q395 frg. 1 10, but it has only the first two segments, ‫בשׁ‬, leaving the ‫ ל‬to be reconstructed.

Issues of Method and Theory

27

2. Scribes update the language of the text that they are transmitting and recognize linguistic changes. This suggests that the scribes who engaged in updating had some awareness of their own grammar as well as that of the text. However, “a conscious awareness of grammar is not synonymous with a systematic one and . . . one could have a conscious understanding or awareness without that awareness being in any way systematic.” The distinction between a nonsystematic and systematic awareness of the grammars involved explains why scribes sometimes created back-formations in which they imposed morphophonological or morphological innovation onto earlier forms. (1982: 428) 3. Written materials do not always exhibit the logical and expected diachronic developments. What some call “‘the internal logic of a historical development’ is not always as logical as we would like it to be and may in fact never reach its ‘logical’ conclusion. There is also a danger . . . that we assume that the chronological demarcations we have set up . . . constitute a direct genetic line of descent.” (1982: 428–29) 4. It is necessary to distinguish between innovations and inherited linguistic elements in scribal traditions. There exist cases where an earlier paradigm can only be completed with forms found in later texts, and these forms are often described as “archaisms.” But the label is uninformative, and one must determine whether the form is an inherited linguistic item, which would be a legitimate component in the earlier paradigm, or the form is a scribal innovation based on scribal ingenuity. (1982: 430)

All four questions lie at the heart of the challenge for Hebrew linguists, especially if Hebrew scribes were not all “passive transmittors” but were actively editing the texts they copied. Setting aside issues of content and focusing solely on the linguistic issues, if scribes changed the language of the texts they were transmitting, then we cannot take—without significant qualification— the linguistic data in those texts as representative of the language of the final compositional stage. Consider two scenarios that reflect current debates on the linguistic dating of biblical texts. First, if a scribe updated specific forms to accord more closely with his own grammar (e.g., the replacement of ‫ אשׁר‬with ‫)שׁ‬, then the later form must be excluded from the linguistic profile of the composition and is thus no evidence for a date matching the inserted item’s later historical context. Second, if a scribe inserted an innovative form that was consciously modeled on the scribe’s understanding of the grammar of the text but was not itself a historical form, then this item must also be excluded from the linguistic profile of the composition and cannot be used for any dating, whether early or late. The discovery of the Qumran texts affirmed that some scribes were more passive than others, but it also demonstrated beyond any doubt that other scribes expanded, rearranged, clarified, and updated the texts (and the language of the texts) they were transmitting (Ulrich 1992; see also Holmstedt 2013c; cf. Young, Rezetko, and Ehrensvärd 2008: 341–60).

28

Chapter 2

Lass’s fifth question is more difficult to address. The issues of style, choice, and more basic, semantic interpretation (e.g., the possible scope issues involved in resumption in relative clauses) all relate to our lack of native speakers for these ancient languages. Miller has noted that, if we had a resurrected native speaker for ancient Hebrew, we “could confirm that the orthographic symbol ʿayin (‫ )ע‬was used to represent two sounds in ancient Hebrew—the pharyngeal /ʕ/ as well as the postvelar fricative /ɣ/” (Miller 2004: 292). Without the resurrected native speaker, however, we are limited to determining a phonetic range for each consonant, based on distribution with other consonants and representation in contact languages, such as Akkadian, Aramaic, and Greek. We are similarly deprived of most prosodic features of the language and thus their possible roles as signals for syntax and semantics (e.g., restrictive versus nonrestrictive relative clauses) and pragmatics (e.g., intonation used to mark Focus). 7 At the end of the day, all we can do is collect the data responsibly and tell the story that makes the most sense to us, admitting that no matter how good, coherent, logical, or empirically grounded, it may not be historically accurate. A particular type of native-speaker input is considered crucial by generative linguistics: native-speaker “intuition.” 8 Why? If a language is represented by the infinite diversity of utterances that the grammar residing in the mental language faculty of the ideal listener-speaker is able to generate, it is impossible to describe the entire language or even to specify all possible constructions. However, the linguist can use the intuition of native-speaker informants to discern the boundaries of what is possible. 9 Hence, great importance is assigned to both counterexamples and the so-called ungrammatical examples (those marked with asterisks): they illustrate grammatical boundaries for the 7.  This assumes that the Tiberian system of ‫ טעמים‬reflects the prosody of the biblical text as it was read from the fifth century c.e. onward and that it very well may not reflect the prosody of ancient Hebrew before ca. 200 c.e. For linguistic discussions of the Tiberian ‫טעמים‬, see Aronoff 1985; and Dresher 1994. For a definition of Focus as I understand it, see Holmstedt 2009a; 2014b. 8.  We should not confuse the use of intuition in generative linguistics with the common use of intuition. Often, when we use the word intuition, the connotation is that of guesses and luck, something very “unscientific.” And sometimes generative grammar has been criticized this way. However, this reflects a misunderstanding of the use of intuition in generative linguistics, in which it refers to “tapping into our subconscious knowledge” (Carnie 2006: 13). In this way, linguistic intuition about issues such as grammaticality are no different, and no less scientific, than visual judgments about colors or shapes. 9.  Note that, while native speakers cannot provide direct, conscious information about their mental grammar(s) in the form of linguistic analytical propositions, their judgments and intuitions on the grammaticality/acceptability and felicitousness of data do provide access to their mental grammar(s), although admittedly a step removed from the ideal of competence (given that even judgments reflect performance). See Smith 2004: 25–45 for a clear discussion of these issues.

Issues of Method and Theory

29

purposes of syntax and provide necessary clarification for semantic readings and pragmatic nuances. This allows whatever proposals are put forth to meet the criteria of observational and descriptive adequacy and to predict the acceptability of novel examples. Consider the contrast among the three English examples in (59)–(61). (59)  I didn’t see that the red car went into the ditch (60)  I didn’t see which car went into the ditch (61)  *I didn’t see which car that went into the ditch The first two examples illustrate a complement clause and a preposed whphrase, respectively. The third example, in (61), illustrates that a wh-phrase + that sequence is not acceptable in modern standard English. By this process, we are able to establish a linguistic boundary between the grammatically “good” and “bad.” Unfortunately, we simply do not have access to such judgments for ancient Hebrew. The lack of native speakers puts us at a distinct disadvantage; it is impossible to elicit fresh data or to check the semantic reading or pragmatic nuance of the data we do have. 10 The linguist of an ancient language such as Hebrew must accept two principles of operation. First, nearly all of the extant ancient data is grammatical, interpretable, and pragmatically felicitous within its discourse context. Aside from a relative few arguably bad examples, 11 most of the data must be grammatically acceptable or we would have nothing upon which to reconstruct 10. See Ehlich 1981 for the following (very intriguing) claim: [W]e find the philologist in a position that is parallel to that of the ‘linguist native speaker (LNS)’. The philologist develops a kind of linguistic knowledge of which he makes use by introspection. Since there is no native speaker . . . , he himself is the only one who can really “speak” the language, i.e., who develops a concrete, individualized competence in that language, a competence comprising all of the elements that make a language. This secondary competence of the philologist is open to introspection, and introspection is the main way how the philologist comes to systematic results on structures of his subject. . . .The philologist is his own LNS. (Ehlich 1981: 161)

In my opinion, Ehlich overestimates the “competence” of even the most gifted philologist. Or at least Ehlich’s understanding of competence is not that of the technical sort discussed within generative linguistics, in which competence refers to the mental language faculty of the ideal native speaker-listener. No amount of expertise in a “dead” language, gained from textual remains, could allow one to achieve this technical kind of competence, and thus no philologist could be considered a suitable stand-in for a true native speaker (no matter how much we would like to flatter ourselves!). 11.  For example, I suggested in Holmstedt 2006b that 11 cases of the relative word ‫אשׁר‬ in the MT are not just infelicitous but grammatically unacceptable (Gen 11:7; 34:13; Deut 4:10, 40; 6:3; 11:26–28; 1 Sam 15:15; Ezek 36:27; Eccl 7:21; Dan 1:10; Neh 2:3). I take these examples up again in appendix B.

30

Chapter 2

the grammar! Related to this is the propensity, particularly in Hebrew Bible scholarship, to propose emendations or reconstructions for troublesome texts. Lass’s comment on this issue is apropos: Rather than interpreting the (multiply attested) [hapax] form, or abandoning it as hopeless, they emend the text, and tell us ‘what must have been meant’. This is only one step less criminal than another common text-editor’s ploy: ‘restoring’ lacunae resulting from destruction of text by inventing material that ought to have been there. . . . One obvious motivation for not emending is that emendation is not in fact creative, but destructive; it obliterates part of the record, and substitutes for it an invention of another time, place and culture. Rather than filling an epistemic gap, as reconstruction or interpretation of documents do, it falsifies the record, and produces second-order witnesses not in essence principle-driven, but dependent only on argument (often flabby and aesthetic rather than linguistically controlled). . . . [T]rying to reconstruct an individual’s word-choice is as fatuous as writing texts in a protolanguage, or reading ancient poems aloud in a reconstructed pronunciation ‘expressively’, suprasegments and all. As Herbert Penzl very properly puts it (1991: 62), ‘Our methods permit reconstruction of phonemes, morphemes, phrases, syntactical rules, vocabulary forms, but not of speakers and their language acts, such as e.g. the creation of written texts. Not that this stops people from doing it, but they shouldn’t. (Lass 1997: 100).

I accept Lass’s admonition as valid and thus do not base any part of my analysis on examples that are textually problematic. The second principle that the ancient Hebrew linguist must accept is the necessity of interpreting the ancient data by means of a linguistic theory that has been developed and tested on languages for which native speakers exist. The difficulties of reconstructing the grammar of ancient language resemble the difficulties that a child experiences when reconstructing the grammar of its mother tongue. A child acquiring its mother tongue, too, has access only to a limited—and sometimes defective—set of positive evidence (the correction of a child’s mistakes by adults is by no means a necessary element of language acquisition). If the two processes are similar, then the methodology adopted in the reconstruction of the grammar of an ancient language must also be similar to that employed by a child in the course of language acquisition. What the child does is interpret the data it has access to on the basis of the genetically coded Universal Grammar that it possesses. This is what the linguist setting out to reconstruct the grammar of a dead language must do, as well; he or she must interpret the data available as indications of how the open parameters of Universal Grammar are to be set. (Kiss 2005b: 2–3)

Where this becomes potentially problematic is when the theoretical position conflicts with what appears to be the simplest analysis of the ancient data. Consider the case of left-dislocation. It has been asserted that one of the features

Issues of Method and Theory

31

of left-dislocation in English is that it cannot be embedded, as the contrasts between (62a–c) and (62d) illustrate (Chomsky 1977; Baltin 1982). (62)  a. He loves Miryam (main clause; no LD) b. Because he loves Miryam, he hugged her (embedded clause; no LD) c. Miryam—he loves her (main clause; LD) d. *Because Miryam—he loves her, he hugged her (embedded clause; LD) The examples in (62) illustrate how left-dislocation is prohibited in embedded structures, and apparently this constraint is so well attested that left-dislocation is characterized as a “main-clause phenomenon” (Lambrecht 2001: 1069). But consider the Hebrew example in (63). (63)  ‫ל־ה ִאיׁש ֲא ֶׁשר ָה ַלְך ַא ֲח ֵרי‬ ָ ‫ר־ע ָׂשה יְ הוָ ה ְּב ַב ַעל ְּפעֹור ִּכי ָכ‬ ָ ‫יכם ָהרֹאֹת ֵאת ֲא ֶׁש‬ ֶ ֵ‫‏עינ‬ ֵ ‫ֹלהיָך ִמ ִּק ְר ֶּבָך‬ ֶ ‫ל־ּפעֹור ִה ְׁש ִמידֹו יְ הוָ ה ֱא‬ ְ ‫ַב ַע‬ ‘Your eyes were those that saw how Yhwh acted at Baal Peor, that every man who followed Baal Peor—your God destroyed him from your midst’ (Deut 4:3) The example in (63) appears to be left-dislocation in an embedded context, in this case, a complement clause (see Holmstedt 2014b for more examples and discussion). The critical question is whether we can suggest a modification of what is taken to be a linguistic universal, even though we are unable to check the grammaticality or felicitousness by appealing to native speakers. Must we read ancient linguistic data through the strict lens of principles, parameters, and various derivative conditions built upon modern data? The linguist working on ancient languages faces a considerable problem when such issues are encountered. The safe route is to present the data as clearly as possible and leave any strong conclusions for the future, when additional data may clarify the issues. Thankfully, in the case of left-dislocation, further cross-linguistic investigation has indicated that, while embedded left-dislocation is less preferred and strains discourse processing, some languages do allow it. Thus, we may add Hebrew to the list of languages that allows left-dislocation in certain embedded environments. 2.1.3.  The Linguistic Status of Biblical Hebrew Here I return to Lass’s first question: how far is the ancient corpus representative of the language in which it is written, and in what way(s)? The Semitist Edward Ullendorff famously asked nearly the same question of Biblical Hebrew almost 30 years ago: does Biblical Hebrew present us with adequate data to consider it a “language”? His conclusion to this question was negative: “In the sense in which I have been endeavoring to present the problem biblical

32

Chapter 2

Hebrew is clearly no more than a linguistic fragment” (Ullendorff 1977: 16). The validity of Ullendorff’s negative answer is entirely dependent upon whether one accepts his equation of “a language” and “a system of communication.” For Ullendorff, being representative of a language meant having access to a lexicon and grammatical system sufficient for typical human communication (e.g., “Pass the pepper and get me a spoon, please”). The Hebrew of the Bible lacks certain lexical items and grammatical forms, and thus Ullendorff deemed it insufficient as a “language.” There are two salient linguistic questions that spin off Ullendorff’s argument. First, does the linguistic system(s) represented in the Hebrew Bible sufficiently represent the language(s) that did exist in the biblical period (whatever we call it/them)? Second, even with a greater lexical inventory and a complete grammatical paradigm, is this entity we refer to as Biblical Hebrew a language in any technical sense of that word? On the issue of a representative linguistic system (which is also what I understand as the essential reference of Lass’s first question), Miller 2004 argues contra Ullendorff that the extant Hebrew data in fact do present us with a good representation of the linguistic system of ancient Hebrew. 12 While I do not doubt the validity of Miller’s points, which are firmly based in corpus linguistic principles, I do question her assumption about the nature of the data: should all the ancient Hebrew data be subsumed under a single linguistic system—one that would have had to span more than five centuries, a relatively large geographical area, and multiple social and religious classes? 13 There has been a consensus for over two centuries that the biblical and epigraphic evidence presents data from at least 3 historical stages of Hebrew, typically referred to as archaic, classical, and late Biblical Hebrew. Minimally, then, we must recognize three linguistic systems. If we add into the mix proposals for remnants of a northern dialect(s) (versus the dominant “southern” dialect of the biblical corpus), diglossia (with remnants of both the spoken lect and the dominant literary lect), we would need to recognize anywhere from 6 to 12 (if not more) linguistic systems. 14

12.  Putting aside the numerous lexical items that Ullendorff expects but does not find in the lexical inventory of Biblical Hebrew, more salient to the issue of a linguistic system are the supposed “grammatical gaps” that he adduces: the lack of many second-person-feminine forms, of certain Hophal forms, and of certain types of clauses, and a “dearth of genuine dialogue features, of anacoluthon, and especially of non-literary . . . sentence structure” (Ullendorff 1977: 14). See Miller 2004: 287 n. 24, and pp. 293–96 for a rebuttal. 13.  Caveats notwithstanding, it remains that teaching and research grammars marginalize data that might reflect alternate linguistic systems of ancient Hebrew. 14.  See, among many others, Driver 1970; Ullendorff 1977; Rendsburg 1990; 1991; 1992b, c, d; 1999; 2000; 2002; 2003; 2006; Young 1992; 1993; 1995; 1997; 1998; 1999; 2003b; 2005; 2008; 2009; Young, Rezetko, and Ehrensvärd 2008; Polak 2003; 2006a; 2006b; 2010.

Issues of Method and Theory

33

In light of all these issues, it is a significant linguistic concern whether it is responsible and accurate to speak of “Biblical” Hebrew as a single linguistic system. It is clear that we can no longer work with three broad biblical stages— archaic, classical, and late Biblical Hebrew—and then a sweeping postbiblical stage. 15 Instead, it seems increasingly likely that in order to make valid descriptive statements future linguistic studies will have to reckon with, in the least, archaic, archaizing, preexilic, exilic, Persian, Hellenistic, early Roman, and Tannaitic, along with Israelian, Judahite, faux Aramean, and oddly sectarian (i.e., certain Qumran texts such as the Temple Scroll) forms of Hebrew, some in chronological sequence, some coexistent (see Knauf 1990). The likelihood of multiple linguistic systems’ having representation in the ancient Hebrew data (and the need for separate descriptive grammars for each) brings us back to the initial prompt for this discussion: does the linguistic system(s) represented in the Biblical (and non-Biblical) Hebrew corpora sufficiently represent the language(s) that did exist in the ancient period? Let us set this question in context: Ullendorff assumed many fewer ancient Hebrew linguistic systems than I have here suggested are likely, and his conclusion was in the negative. With recent research pointing toward numerous linguistic systems’ proceeding from the ancient corpora, the odds have decreased even further that any one ancient Hebrew “language” is sufficiently represented. Of course, this conclusion is hardly illuminating, since it begs the question: what is a “language”? While in the last paragraph I recast Ullendorff’s study within the context of discerning the relationship of the biblical corpus to the actual linguistic system of ancient Hebrew, Ullendorff clearly set out to determine whether the ancient Hebrew data extant in the Hebrew Bible could suffice for human linguistic interaction (“a system of communication”), and since he deemed that it did not, ipso facto “Biblical Hebrew” was not a language. This, however, is but one of numerous ways to define a language, and so we come to the second question: is this entity we refer to as Biblical Hebrew a language in any technical sense of that word? Pateman 1983 conveniently considers five common ways that the concept of a language is defined and used referentially within linguistics: (64)  A language is . . . a. a natural kind (naturalism) b. an abstract object (platonism) c. a name given to a set of objects (nominalism) 15.  With varying details, the three-stage model of Archaic Biblical Hebrew, Classical (or Standard) Biblical Hebrew, and Late Biblical Hebrew has held sway for two centuries (for general overviews, see Kutscher 1982; Sáenz-Badillos 1993). For a recent challenge to this diachronic model, see Young, Rezetko, and Ehrensvärd 2008. For a variety of responses to the challenge, see the articles in Miller-Naudé and Zevit 2012. For an abstruse counterresponse, see Rezetko and Young 2014.

34

Chapter 2 d. a social fact, and that social fact is also a/the only linguistic fact (sociologism) e. a social fact, but that social fact is not a linguistic fact (dualism)

Ullendorff’s approach to language lies somewhere in the nexus of (64c) and (64d), and yet, Pateman argues that only two of the five classifications are cogent, that a language in some sense is a natural kind (64a) and that a language is a social but not a linguistic fact (64e). For Ullendorff (and most nongenerativists as well as philologists), the existence of a “Hebrew language” in the biblical period is indubitable. But if, to follow Pateman (and generativists in general), speaking of the “Hebrew language” or the “English language” is nothing more than a convenience from a linguistic perspective and only the idiolect has any status as a linguistic fact, 16 we cast into even sharper relief the theoretical and methodological problems I introduced above: how can we describe the linguistic features of “Biblical Hebrew” (or even northern, southern, literary, and colloquial “Hebrew”) if it is not a linguistic fact? It would seem that we cannot. However, the data from the Hebrew Bible and other ancient corpora do provide us with linguistic facts, so how do we classify these if “Hebrew” is not the proper linguistic category? Naudé summarizes this issue well and suggests the logical response:   The notion language like many concepts that have widespread pretheoretical distribution can be conceived in a variety of ways. Language as a sociopolitical notion is difficult to substantiate in linguistic research. The speakers of a language typically feel themselves bound to one another by way of their self-designated identity as, for example, the speakers of English. There are no empirical tests to assist in determining whether, for example, Cockney English and South African English are manifestations of the same entity, or whether a dialect spoken in Denmark and a comparable dialect spoken in Norway represent distinct entities (Danish and Norse). The standard attempts to develop tests designed to establish languagehood (such as mutual intelligibility, shared core 16.  I take it that the view that languages are social facts, but not linguistic facts, is expressed by Chomsky when he says . . . , ‘such notions as “the English language” are not linguistically definable, but are rather socio-political in nature’. This view arises not only from a theoretical preference for writing grammars for idiolects—or, more precisely, grammars for mentally represented objects—but from positive doubts about the linguistic definability of ‘German’, ‘Dutch’, etc. (Pateman 1982: 119)

It is important to recognize here the Chomskyan distinction between knowledge of language, what he refers to as “competence” and which he locates in the mental grammar and identifies as inaccessible to introspection, and a speaker’s beliefs about his/her own language. The latter concept is a sociopolitical one and may or may not (more likely the latter) have any direct connection to the formal features of the mental grammar. It is in the latter sense that Pateman, following Chomsky, can assert the sociological fact of, e.g, “English” while simultaneously denying any truth to English as a linguist fact (i.e., a feature within the mental grammar of the native speaker/listener).

Issues of Method and Theory

35

vocabulary, etc.) proved to be not very fruitful (Schütz 1972). The notion dialect is equally ill defined, in an empirical sense, just like the notion language, in that both notions suffer from the same defects.   The notion language must be qualified by reference to idiolects (the output of a single individual). Idiolect is a term that denotes something closer to the actual empirical object—the innate grammar—which generative linguists recognize as the proper object of study. One could argue that in purely linguistic terms entities such as the English language and the dialect of South African English do not exist as well-defined formal objects of scientific study, but individuals actually do exist and their output can be studied scientifically. However, it is not the total output of individuals which forms the basis of study but only the individual grammars present in the mind of the person in question. Such a concept of language enjoys widespread acceptance in contemporary synchronic linguistic theory and is equal to the description of language faculty and I-language of N. Chomsky (1995): it goes under the name grammar (rather than language), and it is generally recognized as representing the proper object of study of linguistics as a discipline. (Naudé 2003: 196–97; similarly, Naudé 2004: 97)

Thus, the task of formally describing the linguistic features of the ancient Hebrew data must focus on idiolect. The reader will immediately recognize the intimidating obstacles that result from taking this position. We do not, of course, have direct access to individuals from the ancient Hebrew world— only the products of their writing activity (this brings us back to the issue of how representative of the author’s idiolect any given ancient Hebrew text is). Moreover, the challenge of studying the texts as partial representations of the authors’ idiolects is further compounded by the complex compositional history of many of the texts: biblical scholars are in general agreement that texts such as the books of Isaiah and Ezekiel, for instance, exhibit multiple authors and editorial stages. In the case of texts that are the product of multiple hands over the course of decades or centuries (a possibility for the Pentateuch and the book of Psalms), the linguistic information will be mixed in accordance with the input of multiple idiolects. 17 17. Describing each idiolectal layer of ancient Hebrew for which we have access is clearly the ideal, and perhaps the next generation of scholarship will have the benefit of such a nuanced approach. At present, however, little exists beyond descriptions of a monolithic “Biblical Hebrew” or, at best, only slightly better descriptions within the three-stage model. As a first step, I have suggested elsewhere (Holmstedt 2006a) that the best—indeed, the only—way forward given the nature of the data is to start by giving each ancient Hebrew text its own descriptive treatment, considering each a distinct “bibliolect,” so to speak. With bibliolectal descriptions as a foundation, researchers could address the mixed lects of the compositionally rich texts, such as Isaiah, Ezekiel, and the Pentateuch. The result of this program would first of all be a view of the ancient Hebrew data unprejudiced by existing models of Hebrew diachrony and dialectal variation. At its best, this research program would produce a set of descriptive grammars as well as a final comparative synthesis leading to a

36

Chapter 2

Unfortunately, there exists no sufficiently nuanced study of the individual “grammars” represented by the ancient data with which the variations in the relative clause might be compared. However, features that exist in high enough numbers, such as spelling (Forbes 2012; Forbes and Andersen 2012) and the variation between ‫ אשׁר‬and ‫( שׁ‬see below, §7.4), provide us with statistically sound first steps in a wholesale diachrony of ancient Hebrew. 2.2.  The Linguistic Framework of This Study This work is grounded in two linguistic frameworks: typology and generative syntax. Typology is here taken in the first two of Croft’s definitions (2003: 1–2): typological classification and generalization. The typological classification of relative clause structures allows us to situate the features of the Hebrew relative clause within the observed oppositions salient for the Hebrew data, such as the order of relative head and relative clause and the nature of the relative elements used to introduce the relative clause. The typological generalizations, which have led to proposed absolute and implicational universals, suggest interconnectedness of features, such as the incompatibility of relative pronouns and resumptive pronouns, which we can test on the Hebrew data. So it is in this work that the features of the Hebrew relative clause are sorted using a cross-linguistic perspective. Croft’s third definition is not adopted in this work—that linguistic typology represents a functionalist theoretical approach to explain the specific linguistic structures in question and human language in general. Instead, I adhere to the view of language and phrase structure associated with Chomsky’s generative theory (Chomsky 1995; 2009; see also Adger 2003; Boeckx 2006; and Carnie 2006; 2010). Generative grammar provides a compelling view of human language and also proposes an abstract level (i.e., deep structure) of linguistic structures by which the diverse patterns of individual languages (“parameters”) observed in the cross-linguistic study can be related to each other by a small number of underlying “principles” (hence, Chomskyan generative theory is often referred to as the theory of “principles and parameters”). I see advantages in both typology and generative syntax, and my combination of the two in this work reflects a commitment to drawing on the descriptive, cross-linguistic strengths of typology along with the unifying abstraction of generative theory. 18 As Baker and McCloskey suggest, such a combination model of ancient Hebrew dialect rich on both the horizontal (dialect, diglossia) and vertical (diachrony) axes. 18.  For a discussion of typology and generative linguistics that has influenced my own views, see Newmeyer 1998: 297–369. On building productive relationships between typology and other frameworks, also see Bresnan 2007; and Nichols 2007.

Issues of Method and Theory

37

could thus form a kind of conduit for the best results of typology to flow into formal syntactic theory and vice versa. . . . This could provide an escape from those sterile debates between “functionalist” (typological) and “formal” approaches to language that have too long hindered dialogue and deepening of understanding. (Baker and McCloskey 2007: 295) 19

What this means for the current work in practical terms is that cross-linguistic (including comparative Semitic) relative clause data are used to help organize the salient features of the Hebrew relative clause (see especially chaps. 2–3). The identified salient structural features are then analyzed through a generative syntactic lens, which for the purposes of the Hebrew relative clause primarily refers to seeing phrase and clause structure as hierarchical and binary, allowing that constituents may “move” (by which variations YXZ, YZX, and ZXY may be derivations by movement of the single basic clause XYZ, rather than distinctly different, unrelated clauses), and recognizing the syntactic reality of phonologically null constituents. Although the overtly linguistic component of this monograph has been decreased in comparison with the doctoral thesis from which it has developed, it remains an example of how I think the philological and linguistic study of Biblical Hebrew can be fruitfully combined. There are few aspects of my analysis of the Biblical Hebrew relative clause that would differ in substance if my theoretical position were non-Chomskyan generative or even altogether nongenerative. My purpose for including the description below of the generative principles I identify as the most salient for understanding my presentation in this work is twofold. First, in the interests of scholarly transparency, it alerts the reader to the theoretical view of language. Second, it allows the reader to contextualize and so understand the language I use in my description, such as “overt,” “covert,” “null head,” “raising,” or “rightward movement.” 2.2.1.  Hierarchical and Binary-Branching Phrase Structure 20 Unlike a linear or “flat” analysis of the structure of a given clause, illustrated by the tree and bracket notation in (65), many forms of generative and nongenerative syntax have adopted a hierarchical approach. 19.  The debates have not always been “sterile” or friendly. To get a sense of the vitriol and animosity that has too often marred theoretical discourse in modern linguistics, see Harris 1993. 20.  Though minimalist phrase structure is explicitly hierarchical and binary and supports this position with data and an appeal to theoretical economy, there is certainly no consensus among linguists in general on whether the syntactic structure of human language reflects a flat or hierarchical structure, and if the latter, whether the hierarchy is fundamentally binary or “n-ary” (Culicover and Jackendoff 2005: 112–16; Berg 2009: 33–56, 325–28; Guevara 2007). Of course, empirical support and theoretical advantage are claimed by all who bother to make an explicit defense of their phrase structure(s).

38

Chapter 2 (65) 

[[Avigayil ] [gave ] [her ] [ sister] [ the ] [ teddy ] bear]]

Using the same clause (and simplifying a great deal), the hierarchical analysis discerns that there exists a closer relationship between the constituents her and sister than there is between gave and her even though the words in both pairs are adjacent. A hierarchical diagram might look like the one in (66). (66)  [[Avigayil ] [[gave [[her ] sister] [[ the ] [ teddy ] bear]]]. In the hierarchical approach to phrase structure, the syntactic elements relate to each other in terms of how they cluster together. For example, in the clause Avigayil gave her sister the teddy bear, we might suggest that Avigayil and gave relate to each other nonhierarchically as the two basic halves of the clause. But we would not put the rest of the clause on the same level: the words her sister, which seem to belong together, and the words the teddy bear, which also seem to form a group, both seem to form a group with the verb gave. These hierarchical relationships are represented in tree diagrams by the position and direction of the branches and in the bracketing system by the levels of brackets. Among the strengths of the hierarchical phrase structure is the ability to account for long-distance syntactic relationships, in which two syntactic elements that somehow depend on each other are separated by an arbitrary number of words. For example, in the first two examples in (67), the subject and verb are adjacent and so the subject-verb agreement is immediate, or “local.” But in the third example, the agreement is non-local, or long distant. (67)  The [ baby sg] [ cries sg]. The [ babies pl] [ cry pl]. The [ babies pl] in the nursery [cry pl]. The hierarchical phrase structure can account for long-distance dependencies by proposing that the intermediate branching junctures, or “nodes,” can relate to each other in the same way as linearly adjacent constituents. Thus, just as baby and cries can agree locally in (67a), so babies and cry in (67c) can agree locally at a higher level in the hierarchy of the phrase.

Issues of Method and Theory

39

(68)  [[[The ] babies pl [ in [[ the ] nursery ]]] [ cry pl]]. In this example, the element in the nursery is hierarchically dominated by the babies. The agreement between the plural the babies can occur between the higher node representing the babies and the now hierarchically adjacent plural verb cry. The structural representations in (66) and (68) both include an example of ternary branching, or a flat structure for two or more constituents. In (66), this concerns the teddy bear and in (68) in the nursery. Earlier forms of Chomskyan generative syntax allowed for “n-ary” branching (i.e., whatever number of branches appear to be required) and some non-Chomskyan generative frameworks maintain an n-ary principle of phrase structure. However, since the mid-1980s, the Government-and-Binding model, followed by minimalist syntax, adopted a strictly binary approach to constituent structure. Binarity in phrase structure is directly related to “Merge,” which is the primary operation in Chomsky’s bare phrase structure (Carnie 2010: 155–58; see also Boeckx 2008: 28–120). To simplify, Merge takes two constituents, for example, X and Y, from the lexicon and combines them into a set {X, Y}. This set includes both X and Y and has either the form ‘X, ‘X,Y’’ or ‘Y, ‘X,Y’’. The choice between these two options is based upon which item, X or Y, projects itself in order to give the new larger item its identity (see Adger 2003: 73–90; and Boeckx 2008: 91–98). In other words, the lexical item that projects shares its bundle of lexical features with the newly formed category: this lexical item that projects is called a head. 21 The structure of the newly merged constituent is represented by the diagram in (69). (69) 



X

X Y If we substitute the items ate and food for X and Y, the new object created after the operation Merge may have either the form ‘ate, ‘ate, food’’ or ‘food, ‘ate, food’’. The choice between these two depends on which one of these lexical items gives its identity to the larger object, a choice that is determined by the structure and lexicon of the particular language in question and how the 21.  Nouns are the heads of the phrase containing the noun and its modifiers, hence noun phrase (NP). Likewise verbs, determiners, adjectives, adverbs, and prepositions are the heads of their respective verb phrases (VPs), determiner phrases (DPs), adjective phrases (APs), adverb phrases (AdvP), and prepositional phrases (PPs).

40

Chapter 2

two items being merged can possibly relate to each other. In English, it is the verb ate that projects, producing the first of the two choices, ‘ate, ‘ate, food’’— what is typically referred to as a verb phrase (VP), as the tree in (70) illustrates. (70) 



ate

ate food Because one of the features of ate is its identity as a Verb, the new complex item shares the verbal identity of the projected head, resulting in the Verb Phrase identity of ate food. The examples so far have addressed how lexical items project to the phrasal level, but within generative syntax, there are two basic phrase categories: lower levels are fundamentally lexical in nature, and higher levels are functional (Fukui 2001). The lexical categories have substantive content since they line up with the lexical items that are inserted into an incipient clause (e.g., the nouns, verbs, adjectives, adverbs, etc.). These categories, which bear semantic features, play a critical role in the interpretation of the linguistic expressions, since it is within these categories that most of the lexical items interact with each other in terms of thematic roles (e.g., agent, patient, goal). The functional categories, in contrast, do not have substantive content and are not central to interpreting the linguistic expressions; rather, these categories are largely grammatical and computational, triggering agreement feature checking, setting the tense-aspect-mood, or, in the case of the outermost layer, establishing the clause type. The outermost functional layer is widely referred to the as the complementizer phrase, or CP for short. The CP establishes the clause’s illocutionary force—that is, what type of clause it is, such as main versus subordinate, or declarative versus interrogative (Haegeman and Guéron 1999: 520–38). The operation Merge applies to a phrase as many times as necessary in order to incorporate all the available constituents. Thus, in the Verb Phrase quickly ate food a first Merge would produce ate food, as in (70), and a second application of Merge would account for the adverb, quickly ate food, resulting in (71). (71) 



VP

quickly VP ate food

Issues of Method and Theory

41

Many languages (including Hebrew) exhibit this type of multiple projection in the CP, resulting in what is called the complex or articulated CP. The projected layers may contain function words that establish the illocutionary force, such as an interrogative word to establish an interrogative clause or a relative word to establish a relative clause, as well as two or more layers to host constituents that have been fronted, often for discourse pragmatic reasons such as Topic or Focus. I will illustrate what I think this looks like for the Hebrew clause below in §2.3.2. The point of this outline of minimalist phrase structure building is to highlight the fundamentally binary nature of the Merge operation: it combines one item, no matter how simple or complex, with another item, no matter how simple or complex, and results in a larger constituent in which one of the merged items is the head that projects its identity onto the whole. There is no room in this phrase structure for n-ary branching; it can only be binary. 2.2.2.  Constituent Movement Constituent movement is a hallmark of Chomskyan generative grammar as well as many non-Chomskyan derivatives (see Carnie 2006: 241–71). The concept of items moving around in the clause developed to address two questions. First, can variations of what appears to be the same basic clause be given a unified syntactic description? For example, is an interrogative clause derivationally related to a declarative clause, as in (72)? (72)  Constituent movement a. Noah has eaten his cereal. b. Has Noah eaten his cereal? In (72a) and (72b), the syntactic difference is the position of the finite auxiliary has—in the declarative clause, has follows the subject Noah, whereas in the question, has precedes the subject. If the two sentences, which are on the surface distinguished only by the one change in linear order and the semantic force (declarative versus interrogative) are structurally related, then one must be derived from the other. The solution is to argue that interrogative clauses reflect “subject–auxiliary inversion,” or more accurately, the movement of the finite auxiliary verb from its position in a declarative clause to a higher position, above the subject (Carnie 2006: 204–12). The second motivation for movement comes from “discontinuous constituents”—that is, where a sequence of items normally considered to create together a phrasal constituent are not adjacent in the linear syntax. In English, this primarily occurs with verb + preposition combinations (“phrasal verb”), as in (73a); interrogatives, as in (73b); and stylistic fronting, as in (73c).

42

Chapter 2 (73)  Constituent Discontinuity a. Verb + Particle (“phrasal verbs”) Benjamin woke up his sister. Benjamin woke his sister up. b. Interrogatives Miryam has finished playing. Has Miryam finished playing? c. Fronting Eliana liked the dress (but not the shoes). The dress Eliana liked (but not the shoes).

In each of the three types, the first, declarative clause exhibits the continuity of items in the verb phrase (woke up, has finished playing, and liked the dress). In the second clause, some component of the verb phrase is no longer adjacent to the item it relates to most closely. In (73a), the preposition of the phrasal verb appears after the complement; in (73b), the auxiliary appears before the subject; and in (73c), the complement appears at the front of the clause, separated from the verb by the subject. The basic idea is that the linear order of constituents in many actual clauses does not reflect the “original” order of those constituents. In Hebrew, discontinuity is extremely common, since many narrative clauses begin with the so-called wayyiqtol narrative “verb,” 22 switch to a subject, and then continue with the rest of the predicate. Consider the typical example in (74). (74)  ‫ֹלהים אֶת־הָאֹור‬ ִ ‫‏ וַ ּי ְַרא ֱא‬ and-saw God the-light (Gen 1:4) The clause in (74), which represents the narrative convention, shows the subject ‫ אלהים‬intervening between the verb ‫ ירא‬and its NP complement ‫האור‬ (which is preceded by the direct object marker ‫)את‬. A movement approach to discontinuity proposes that the components of a phrase are together at some point in the development of a clause. This allows for the hierarchical relationships to be satisfied. For those constituents that end up discontinuous, for various reasons one or more of the components moved to other positions in the linear structure of the clause. Constituent movement is integral to the final linear arrangement of constituents that, after every nec22.  I understand the complex wayyiqtol form to consist of wa+∅+yiqtol [preterite], where ∅ is an underspecified complementizer specific to narrative. For various explanations proposed for the history of the wayyiqtol form, see McFall 1982: 217–18; IBHS 544–45; Garr 1998: lxv–lxxiii; and most recently, Cook 2012: 256–59.

Issues of Method and Theory

43

essary Merge is completed, split at “Spell-Out” to produce, on the one hand, the “Phonological Form,” and, on the other hand (with possible further re­ arrangement), the “Logical Form” or semantic interpretation. 23 2.2.3.  Null Constituents The final salient theoretical concept that factors into my analysis of the relative clause is the existence of null or “covert” items, also known as empty categories. Null constituents are those that lack a phonological shape and yet are syntactically real. 24 The English phenomenon of the “wanna-contraction” (a colloquial contraction of want to) in (75) illustrates the syntactic reality of phonologically empty constituents (see Carnie 2006: 322; Featherston 2001). (75)  Wanna-contraction a. I want to read this novel vs. I wanna read this novel b. I want this novel to be considered for a prize vs. *I wanna this novel to be considered for a prize c. This novel, I want __ to be considered for a prize vs. *This novel, I wanna be considered for a prize Example (75a) illustrates how want and to are often contracted to wanna in colloquial English when they are immediately adjacent. Example (75b) shows that when a constituent intervenes between want and to, the two words cannot be contracted. Finally, example (75c) demonstrates that when the noun phrase this novel is moved from its position after want, the result is the adjacency of want and to. However, because the movement leaves a “trace” (the empty category marking the position from which a constituent has been moved), the two words want and to only appear to be adjacent; syntactically they are not; contraction of want and to is thereby prohibited. Hebrew, and many other Semitic languages, use null constituents in just about every syntactic position, including a null copula in so-called verbless clauses. 25 In the relative clause, null constituents are often used as the heads of relative clauses as well as the position that corresponds to the head’s syntactic role within the relative. Both types are illustrated in (76). 23. See Carnie 2006: 358–60 for an accessible entry to the minimalist model. For how different articulations of generative theory may motivate constituent movement, the interested reader should see Carnie 2006; and then Boeckx 2006. 24.  Null constituents are connected within generative theory to constituent movement, verbal valency, and argument structure. However, there have also been arguments that experiments suggest these empty categories to have psychological reality (see Akmajian et al. 2010: 459–62 for a summary of one such experiment). 25.  For an overview of null constituents in Hebrew, including discussion of Hebrew as a “pro-drop language,” see Creason 1991; and Holmstedt 2013a.

44

Chapter 2 (76)  ‫ית ּלֹו‬ ָ ‫ר־ע ִׂש‬ ָ ‫וְ ָׁש ַכח ֵאת ֲא ֶׁש‬ ‘and he will forget ___ that you did ___ to him’ (Gen 27:45)

In (76), the direct object marker ‫ את‬signals that the head of the relative clause is a noun phrase, which in this case is null. Based on the discourse context and the relative clause that modifies it, the null head is easily interpretable as the deception that Jacob perpetrated against his brother Esau. The null position within the relative marks the position of the complement of the verb ‫עשׂית‬, a position that corresponds to the syntactic role the null head would have if it were fully realized within the relative. Since Hebrew allows an overt constituent to exist in the position to “resume” the relative head, in this work I refer to examples similar to (76) as having “null resumption.” Example (76) serves to illustrate not just the presence of null constituents but that the concepts I have described in the last three subsections—phrase hierarchy, constituent movement, and null constituents—will figure prominently in my representation and analysis of the Hebrew relative clause data. Indeed, if my representation and analysis were to depart from what I have presented here, my argument would risk lacking conceptual coherence, a danger I warned against in the introduction to this chapter. 2.3.  A Brief Orientation to Biblical Hebrew Syntax As a specific type of clause, we expect the relative to exhibit grammatical features that distinguish it from other clause types, and it meets those expectations. Among other characteristics, the relative is typically introduced by a specific lexical item that marks it as a subordinate clause and, while still permitting some word order variation to signal pragmatically marked constituents, the relative shows a more constrained word order than is normal in Hebrew. Indeed, Hebrew is often described as a relatively “free” word-order language. This characterization is only superficially accurate and, in fact, to a great extent obscures the real nature of Hebrew syntax. To be sure, Hebrew appears to allow a greater degree of constituent movement than, say, English; however, this movement operates according to only a few principles that are driven by clear syntactic or pragmatic features. Moreover, constituent movement applies exclusively to the order of subject, verb, and verbal complements and modifiers. Or, to put it another way, the relatively fixed order within subclausal constituents, such as PPs, has been wholly ignored in the characterization of Hebrew word order. In the following section, I sketch an approach to Hebrew phrase structure based on published analyses of word order in the books of Genesis (Holmstedt 2002; 2011), Proverbs (Holmstedt 2005), and Ruth and Jonah (Holmstedt 2009), though my observations in reading throughout the Hebrew Bible confirm that my sketch is accurate on a larger scale (see Holmstedt 2013b).

Issues of Method and Theory

45

2.3.1.  A Typological Description of Hebrew Constituent Order The fundamental question involved in determining how a language patterns syntactically is whether a head (i.e., the constituent being modified) precedes or follows its modifier. For each basic type of head (noun, verb, and adposition), there are three basic types of modifiers (complement, 26 adjunct, 27 and specifier 28). Using these three syntactic categories, the basic oppositions are headcomplement versus complement-head, head-adjunct versus adjunct-head, and head-specifier versus specifier-head (see Dryer 1997; 2007; Alexiadou 2002). The first goal is to determine if a language exhibits strong tendencies for each grammatical category and each syntactic relationship. If so, these strong tendencies are taken as the “basic” syntactic pattern of the language for that construction. The subsequent goal is to determine linguistic motivation(s) for any deviation from the identified norm. In table 2.1 below, I illustrate these relationship, using English examples (Holmstedt 2005: 140). The examples in table 2.1 illustrate that English is strictly head-initial for the order of head and complement, strictly head-final for the order of head and specifier, and both head-initial and head-final for the order of head and adjunct, with greater weight given to the head-final examples because they occur in less-restricted environments. English exhibits no single order for all 26. David Crystal, in his Dictionary of Linguistics and Phonetics, defines complement as follows: complement (n.) ( comp) A term used in the analysis of grammatical function, to refer to a major constituent of sentence or clause structure, traditionally associated with ‘completing’ the action specified by the verb. In its broadest sense, complement therefore is a very general notion, subsuming all obligatory features of the predicate other than the verb, e.g., objects (e.g., She kicked the ball) and adverbials (e.g., She was in the garden). (Crystal 2008: 92)

27.  Crystal defines adjunct thus: adjunct (n.) A term used in grammatical theory to refer to an optional or secondary element in a construction: an adjunct may be removed without the structural identity of the rest of the construction being affected. The clearest examples at sentence level are adverbials, e.g., John kicked the ball yesterday instead of John kicked the ball, but not *John kicked yesterday, etc.; . . . In X-bar syntax, an adjunct is one of the major components of a phrasal category (the others being head, complement and specifier). (Crystal 2008: 12)

28.  Crystal’s definition of specifier is below, though articles (also referred to as determiners) are no longer considered specifiers in generative syntax; see my discussion of (82) and n. 33. specifier (n.) (spec, Spec) A relation in the X-bar theory of phrase structure. Specifiers are normally seen as combining with a single-bar category to form the related double-bar category. For example, in John is a student, a is the specifier of the noun, student, and in She is very happy, she is the specifier of the adjective, happy.

46

Chapter 2

Table 2.1.  Typological Illustration of English Word Order Heads

Complements Adjuncts (≈ obligatory modifiers) (≈ optional modifiers)

Specifiers

Nouns

destruction of the city

big cities cities in Africa

our/the(?)a city

Verbs

destroy the city

run quickly, quickly run

They destroyed cities

runs very quickly

Straight down the street

Adpositions in the city a.  See notes 28 and 33.

grammatical categories but is fairly consistent within each syntactic category; in this way, English is a typical SVO language by typological standards. In previous studies on the Hebrew word-order data in Proverbs (2005), Ruth and Jonah (2009), and Genesis (2002: 126–59; see also 2011), I have argued that Hebrew is primarily head-initial, as in (77)–(81). (77)  Preposition with Nominal Complement (= Preposition + Object): head-initial ‫ל־ּפנֵ י ְתהֹום‬ ְ ‫ַע‬ ‘upon the surface of the water’ (Gen 1:2) (78)  Noun with Noun Complement (= Bound Phrase): head-initial 29 ‫ֶמ ְמ ֶׁש ֶלת ַהּיֹום‬ ‘rule of the day’ (Gen 1:16) (79)  Noun with Adjunct (= Noun + Adjective): head-initial 30 ‫ַה ָּמאֹור ַהּגָ ד ֹל‬ ‘the big(ger) light’ (lit., ‘the light the big(ger)’ (Gen 1:16) (80)  Verb with Complement (= Verb + Object): head-initial ‫ָּב ָרא אֹתֹו‬ ‘(he) created him’ (Gen 1:27) 29.  Most types of the Hebrew “construct” do not fit the noun-complement category; rather, they are various types of genitives (IBHS §9.5.3k; cf. Kroeze 1993; 1997) and have a different structure (Ritter 1988). Even so, the order is strongly head-initial. 30.  If we follow current generative theory and take adjuncts to be left-adjoined, the postnominal ordering of an adjective phrase in Hebrew must reflect the raising of the noun to a functional phrase within the DP, above NP but below the Determiner (see Shlonsky 2004).

Issues of Method and Theory

47

(81)  Verb with Adjunct (= Verb + Adverb/Adverbial constituent): a. head-initial 31 ‫ת־א ְׁשּתֹו‬ ִ ‫וַ ּיֵ ַדע ָא ָדם עֹוד ֶא‬ ‘and Adam knew again his wife’ (Gen 4:25) b. head-final ‫יֹוסף‬ ֵ ‫ּכֹה ָא ַמר ִּבנְ ָך‬ ‘thus said your son, Joseph’ (Gen 45:9) Examples (77)–(78) and (80) demonstrate that the head-complement relationship in Hebrew is firmly head-initial while the head-adjunct relationship (79) and (81) shows some variation. Nominal heads precede their adjuncts, 32 illustrated in (79), but verbal heads, illustrated in (81), may precede (i.e., headinitial) or follow (i.e., head-final) their modifying adjuncts. The examples in (77)–(81) indicate that Hebrew is a strongly head-initial language, though the verb allows some variation with its noncomplement arguments. Many languages, however, exhibit some word order disharmonies, specific types of constructions whose features run counter to the dominant patterns. For example, Hebrew appears to exhibit a basic word-order disharmony with regard to the order of a head and its article, which may be taken as a type of specifier. While the basic pattern elsewhere in Hebrew is head-initial, the order article-noun appears to be head-final (82). (82)  Noun with Specifier: head-initial ‫ָה ָא ֶרץ‬ ‘the earth’ (Gen 1:1) But I stress “appear,” because the head-final order is only accurate if one takes the noun to be the syntactic head of a determined noun phrase. Since the late 1980s, the consensus that emerged in generative syntax is that cross-linguistic evidence pointed toward the determiner (D) being the head of the determiner phrase (DP), which is the functional layer, with the lexical noun and its projected phrase contained within the DP. If this is accurate, the Hebrew order article-noun is head-initial and not disharmonic with the dominant head-initial pattern throughout Hebrew. 33 31.  The Verb in Hebrew raises out of the VP to T, and, in the case of the past tense wayyiqtol, the Verb is further raised to C. 32.  Numerals can both precede and follow their nominal heads in Hebrew and are thus an exception to the strong head-adjunct order otherwise exhibited by nominal heads. The divergence of numerals from other types of nominal modifiers, however, is quite common in languages. 33.  For a recent discussion of the noun phrase and the “DP-hypothesis,” see Alexiadou, Haegeman, and Stavrou 2007 (esp. pp.  1–6 for a brief introduction). The function of the specifier became a question after the DP-hypothesis was accepted, and determiners were no

48

Chapter 2

The second possible disharmony in Hebrew concerns the order of subject (= specifier) and verb (= head). If Hebrew had a basic verb-subject order, as the consensus holds, this head-initial pattern would align with the rest of the headinitial patterns that predominate in Hebrew. And yet, I have argued in multiple contexts that Hebrew exhibits a disharmony on precisely this issue—by the biblical period, it had changed from the inherited Semitic verb-subject order to a basic subject-verb order, as in (83). 34 (83)  Verb with Specifier: head-final (subject-verb) = basic pattern ‫ת־חּוָ ה ִא ְׁשּתֹו‬ ַ ‫וְ ָה ָא ָדם יָ ַדע ֶא‬ ‘the man (S) knew (V) Eve, his wife’ (Gen 4:1) The subject-verb order in (83) is paralleled by the basic (and dominant) order in clauses with a null copula, both the variety with nominal complements (the so-called verbless clause), as in (84), and the variety with participial complements, as in (85). 35 (84)  Specifier with Null Copula (and Nominal Complement): head-final (subject-verb) 36 ‫ל־ּפנֵ י ְתהֹום‬ ְ ‫וְ ח ֶֹׁשְך ַע‬ ‘and darkness (S) (was [V]) upon the surface of the deep’ (Gen 1:2) (85)  Specifier with Null Copula (and Participial Complement): headfinal (subject-verb) 37 ‫ל־ּפנֵ י ַה ָּמיִ ם‬ ְ ‫ֹלהים ְמ ַר ֶח ֶפת ַע‬ ִ ‫רּוח ֱא‬ ַ ְ‫ו‬ ‘and the spirit of God (S) (was [V]) hovering over the surface of the waters’ (Gen 1:2) Since Hebrew allows significant overt constituent movement, it is not surprising that all three types of Hebrew predicate structures also exhibit verb-subject order. The key is that the variation occurs within well-defined contexts. In clauses with a finite verb, the verb-subject order is found when there is an initial function word, such the ‫ אשׁר‬in (86) or a constituent fronted for pragmatic longer considered specifiers. Carnies notes that, “around the same time, it was suggested that specifiers have a special role, serving as the identifiers of subjects of various kinds of phrase” (Carnies 2010: 128). 34.  For the data and analysis, see Holmstedt 2002: 126–59; 2005; 2009; 2011; for an initial investigation of the diachronic change, see Holmstedt 2013b. 35.  On the word order of null-copula clauses, see Buth 1999. I follow Cook (2008; 2012) in understanding participial clauses as a subtype of copular clauses, wherein the participle is an event adjective that is the complement of a null copula. This not only acknowledges the cross-linguistic evidence suggesting such an analysis but also allows a unified approach to “verbless” clauses and participial clauses. 36.  For the list of data, see Holmstedt 2002: 158–59. 37.  For the list of data, see Holmstedt 2002: 156–58.

Issues of Method and Theory

49

reasons, as with the Focus-fronted complement in (87). In very rare cases, the verb itself may be fronted for Focus, as in (88). (86)  Verb-Subject with Initial Function Word ‫ֹלהים ַל ֲעׂשֹות‬ ִ ‫ר־ּב ָרא ֱא‬ ָ ‫אכּתֹו ֲא ֶׁש‬ ְ ‫ְמ ַל‬ ‘all his work that God created by doing’ (lit., ‘that created [V] God [S] . . .’) (Gen 2:3) (87)  Verb-Subject with Fronted Complement ‫ֹלהים‬ ִ ‫ת־ענְ יִ י וְ ֶאת־יְ גִ ַיע ַּכ ַּפי ָר ָאה ֱא‬ ָ ‫ֶא‬ ‘my affliction and the labor of my hands God has seen’ (lit., ‘my affliction . . . [C] has seen [V] God [S]’) (Gen 31:42) (88)  Verb-Subject with Fronted Verb ‫ת־ח ְר ָּפ ִתי‬ ֶ ‫ֹלהים ֶא‬ ִ ‫אמר ָא ַסף ֱא‬ ֶ ֹ ‫וַ ּת‬ ‘and she said, “God has taken away my disgrace’ (lit., ‘taken away [V] God [S]’) (Gen 30:23) With null copulas, even with an initial function word or fronted modifier, the order of subject–(null copula)–complement is preserved in the strong majority of cases. This suggests that in the minority of cases in which we find the non-basic order, such as when a nominal complement (89) or participial complement (90) precedes the subject, it parallels the third type of environment in finite verbal clauses, i.e., the nominal or participial complement has been fronted for pragmatic reasons. (89)  Null Copula with Fronted Nominal Complement: head-final ‫ָל ָמה ָא ַמ ְר ָּת ֲאח ִֹתי ִהוא‬ ‘why did you say, “my sister she [is]”?’ (Gen 12:19) (90)  Null Copula with Fronted Participial Complement: head-final ‫ל־ּב ִלי ִהּגִ יד לֹו ִּכי ב ֵֹר ַח הּוא‬ ְ ‫ַע‬ ‘because he did not tell him that fleeing he (was)’ (Gen 31:20) The analysis I have suggested challenges the consensus that Hebrew is a basic verb-subject language. However, not only does my explanation account for the variation, it provides a unified account of basic word order in finite verbal and null-copula (“verbless” and participle) clauses. Within the consensus view, the asymmetry between finite verb-subject and null-copula subject-verb order has not been addressed. Whether Hebrew is concluded to be a subject-verb language (my analysis) or a verb-subject language (the consensus in Hebrew studies), the answer must be grounded in a typologically minded collection and analysis of the data as

50

Chapter 2

well as a theoretically aware explanation of the syntax in terms of phrase structure. To put it another way, typological categories and criteria are needed to formulate the syntactic question(s), which can then only be answered within an explicit syntactic theory. Additionally, the constituent fronting in (87)–(90) indicates that not just syntactic concerns are involved but also information structure issues (sometimes referred to as discourse-pragmatics). 2.3.2.  A Generative Description of Hebrew Phrase Structure Combining the generative phrase structure described above in §2.2 and the word order patterns presented in the previous section, we can see how Hebrew phrase structure operates. Take the combination of the definite article and a bare noun in (91), repeated from (82). (91)  ‫ָה ָא ֶרץ‬ ‘the earth’ (Gen 1:1) DP D N hā ʾāreṣ The noun is bare and so has no need to project a structure greater than simply itself. It is merged with the determiner, which projects a phrasal level to include both it and its complement, the noun. Similarly structured is the PP in (92). (92)  ‫ֵּבין ָהאֹור‬ ‘between the light’ (Gen 1:4) PP P DP bên hāʾôr The preposition ‫ בין‬requires a nominal complement, which is the DP ‫ האור‬in (92). Thus, the P merges with the DP complement to project the larger structure of the PP. Similarly, though the semantics of the bound construction in Hebrew, illustrated in (93), can be interpreted so that the clitic host is agent, possessor, or theme (hence, the semantics are contextually determined), the syntax is simply that of a head noun merged with a DP complement (see Ritter 1988; Shlonsky 2004: 1506). (93)  ‫ֹלהים‬ ִ ‫ֶצ ֶלם ֱא‬ ‘image of God’ (Gen 1:27)

Issues of Method and Theory

51

DP N DP ṣelem ʾĕlohîm In (93), the second noun, ‫אלהים‬, serves as the clitic host of the bound form of ‫צלם‬. The acontextual interpretation of the phrase leaves three options freely available: God is the author of the image (agent), owner of the image (possessor), or the image is about God (theme), regardless of who made it or owns it. Aside from the contextually felicitous interpretation, the syntax in each case reflects the merger of the head, which is the bound noun ‫צלם‬, and a nominal complement, which is the clitic host ‫אלהים‬. Beyond the relatively simple structures in (91)–(93), the projections and constituent movement within the growing phrase structure becomes increasingly complex. Consider the noun modified by an adjective in (94): (94)  ‫ַה ָּמאֹור ַהּגָ ד ֹל‬ ‘the big(ger) light’ (lit., ‘the light, the big(ger)’ (Gen 1:16) DP D ham

XP

māʾôr NP AP NP haggādôl tmāʾôr In (94), the noun has no complement to merge with and so projects only itself to the phrasal level. An adjective is then adjoined to the NP. If we follow current generative theory and take adjuncts to be left-adjoined, the postnominal ordering of an adjective in Hebrew must reflect the raising of the noun to a functional phrase XP, which is above NP but below the Determiner (see Shlonsky 2004). The functional phrase XP is then merged as the complement of the determiner, and the DP is projected. Moving to the clausal level understandably increases the complexity, though none of the basic issues differs from what occurs in the complex DP in (94). For example, in (95) the verbal Complement ‫ את אשׁתו‬is left in place below the verb, and the adverb ‫ עוד‬is left-adjoined above the original sites of the verb and subject. This is similar to the adjunction of the adjective in (94).

52

Chapter 2 (95)  ‫ת־א ְׁשּתֹו‬ ִ ‫וַ ּיֵ ַדע ָא ָדם עֹוד ֶא‬ ‘and knew again Adam his wife’ (Gen 4:25) ConjP wa- CP C CP ∅

yēdaʿ TP

Subj DP VP ʾādām

ʿôd VP

tʾādām VP tyēdaʿ ʾet ʾištô The additional complexity of the verbal clauses concerns the raising of the subject to the functional phrase TP, 38 and in examples such as (95), also the verb. For the subject, the raising occurs in every finite clause type, whether the predicate is headed by a null copula or a finite verb (clauses with infinitives and imperatives have their own complexities). 39 Though the precise language and theoretical articulation for such subject raising has changed in the last 30 years of generative syntax, the essence of the explanation has not: the subject raises so that its agreement features (person, gender, and number) may match the verb. A Hebrew clause whose development finishes with subject-raising (and has no further verb-raising) results in subject-verb order, which is the basic order in Hebrew and is seen in clauses like (96)—that is, clauses that do 38.  Current convention is to use the label “T(ense)” for the functional head associated with the tense/event structure of the clause. This phrasal level was previously referred to as the “IP,” standing for “inflectional phrase,” which betters suits Hebrew. 39.  In brief, infinitive verbs have no agreement features and so do not require their subjects to raise for agreement matching. Imperatives do have agreement features, but as fundamentally modal verbs, they are raised into the CP just as jussives and irreal perfects and imperfects are. On modality as a trigger for verb-raising to the CP in Hebrew, see Holmstedt 2002: 134–39; 2005: 150–51; and 2011: 18–20; see also Cook 2012: 233–37.

Issues of Method and Theory

53

not have some additional constituent at the front of the clause that motivates further movement of the verb. 40 (96)  ‫ַחּיָ ה ָר ָעה ֲא ָכ ָל ְתהּו‬ ‘a wild animal ate him’ (Gen 37:20) In the case of those clauses that do have an initial constituent, such as the narrative ∅ (resulting in the wayyiqtol form) in (95) or the ‫ כי‬in (97), the verb is raised into the domain of the CP to a phrasal position below the head C. 41 In (95), after the verb-raising, which must have become a conventional pattern in Hebrew narrative, the preterite yiqtol was fused with the narrative function word, reduced at this point to gemination of the following consonant (e.g., wa-y-yiqtol) or lengthening of the preceding vowel (e.g., wa-a-ʾeqtol), and the initial conjunction wa-. 42 Though the causal ‫ כי‬clause in (97) does not exhibit (97)  ‫ֹלהים‬ ִ ‫י־ל ַקח אֹתֹו ֱא‬ ָ ‫ִ‏‏‏ּכ‬ ‘because God took him’ (Gen 5:24) CP C CP kî lāqaḥ ʾotô TP Subj DP TP ʾĕlohîm tʾĕlohîm VP tlāqaḥ tʾotô 40. On subject-verb as the basic order in Hebrew, see above, example (83) and the sources listed in n. 52. 41.  The phenomenon whereby an initial constituent, such as ‫ כי‬or even a pragmatically fronted phrase (for Topic or Focus) effects the movement of a constituent within the clause, such as the verb in examples (95) and (97), is called “triggered inversion.” Similar triggered inversion as I claim for Biblical Hebrew has been observed in Modern Israeli Hebrew (Shlonsky and Doron 1992; Shlonsky 1997; Doron 2000a, b). Note that I do not claim any direct syntactic relationship between Ancient and Modern Hebrew; however, whatever the historical facts may be, typologically Ancient and Modern Hebrew share the feature of triggered inversion. 42.  What may have been vowel lengthening in Biblical Hebrew is better analyzed as vowel backing, from [a] to [ɔ] in Masoretic Hebrew.

54

Chapter 2

the fusion of the verb and function word, it does show verb-raising as well as an additional feature: the light complement ‫ אתו‬attaches to the verb to create a complex constituent before verb-raising, resulting in the raising of the complement as well. For the study of the relative clause, one final feature of Hebrew word order variation and phrase structure to note is “the recursive CP ”—that is, the ability of the CP to project multiple phrasal positions for pragmatically motivated raising and successive syntactically triggered verb-raising. Clauses such as (98) illustrate well the flexibility of the CP in Hebrew. (98)  ‫ִּכי ָּכל־א ֵֹכל ָח ֵמץ וְ נִ ְכ ְר ָתה ַהּנֶ ֶפׁש ַה ִהוא ִמּיִ ְׂש ָר ֵאל ִמּיֹום ָה ִראׁש ֹן ַעד־יֹום ַה ְּׁש ִב ִעי‬ ‘because anyone who eats leavened bread—that person shall be cut off from Israel from the first day to the seventh day’ (Exod 12:15) In the clause in (98), the initial ‫ כי‬establishes it as a subordinate clause (in this case, the context indicates that the multivalent ‫ כי‬is causal). The constituent following the ‫כי‬, the complex DP ‫כל אכל חמץ‬, is left-dislocated (i.e., a casus pendens or nominative absolute) and resumed within the core clause by the co-referential DP ‫הנפשׁ ההוא‬. The ‫ כי‬has also triggered verb-raising, and so the verb ‫ נכרתה‬has moved higher than its subject, ‫הנפשׁ ההוא‬. Both the leftdislocated phrase and the raised verb are positioned within the CP, which has projected levels to accommodate them both. (99)  Hebrew Phrase Structure Skeleton CP Extreme Top

CP C TopP* (kî, ʾăšer)

Topic-fronted FocP* phrase(s) Focus-fronted CP phrase(s)

triggered V

TP



core SV



clause

Issues of Method and Theory

55

Based on the variation illustrated by the clausal examples in (95)–(98), I proposed in Holmstedt 2014b that the phrasal skeleton in (99) accounts for the word order variation in Hebrew. 43 Whatever phrases are needed to account for clause type and constituent movement, whatever the reason, are projected within the simple constraints of the skeleton in (99). 2.3.3. Summary Within the view of Hebrew syntax that I have sketched, the relative clause is both mundane and important. It is important due to its near ubiquity in texts. It is mundane because, as a subordinate clause that is often introduced by an overt function word, both the basic and overwhelmingly dominant order in relatives with finite verbs is verb-subject, similar to the ‫ כי‬clause in (97). In fact, Hebrew relatives appear to be more constrained in word order than most clause types, allowing constituent fronting only in a small minority of cases. For example, out of 3,945 ‫ אשׁר‬clauses with a finite verb, only 29 have subjectverb order. 44 Similarly conservative, but from the opposite direction, out of 355 ‫ אשׁר‬relatives with a null copula and participial complement, only 2 have the participle fronted before the subject (Job 6:4 and Eccl 8:11). In addition to the small number of subject-verb finite relatives and verb-subject participial and null-copula relatives, there are also rare examples of nonverbal constituents out of their normal position, as in (100). (100) Verb-subject following a fronted (Focus) complement ‫ם־ּבנֵ י יִ ְׂש ָר ֵאל‬ ְ ‫ר־ׁשם ְּב ִרית יְ הוָ ה ֲא ֶׁשר ָּכ ַרת ִע‬ ָ ‫ָה ָארֹון ֲא ֶׁש‬ ‘the ark which there is the covenant of Yhwh that he cut with the children of Israel’ (2 Chr 6:11). In (100), the expected order in the null-copula clause is the subject ‫ ברית‬and then the copular complement ‫שׁם‬. Instead, this clause gives us the complement before the subject. There are two possible explanations, both of which are syntactically plausible and contextually felicitous. The adverb ‫ שׁם‬may be Focus-fronted to reinforce the importance of the location in question (the Temple) as the residence of the Ark of the covenant. Alternatively, the subject ‫ברית‬, which is modified by both the clitic host ‫ יהוה‬and a relative clause of its own, ‫אשׁר כרת עם בני ישׂראל‬, may reflect “Heavy Noun Phrase Shift” (HNPS), a cross-linguistically common phenomenon wherein “heavy” or complex constituents are moved to the right of “lighter” or less complex constituents, even 43.  For a more-detailed study of constituent placement and movement to the clausal edges, e.g., left-dislocation, fronting, right-dislocation, and extraposition, see Holmstedt 2014b. 44.  Lev 4:22; Deut 9:2; 2 Kgs 22:13; Isa 62:2; 66:13; Jer 1:17; 22:25; 31:32; 39:17; Mic 6:12; Zech 1:15; 11:2; Mal 2:14; Ps 1:3; 89:22; 104:17; 144:8, 11; Job 3:23; 5:5; 15:18; 19:27; Eccl 7:22; Esth 6:6, 7, 9 (2×), 11. This includes a nonrelative ‫ אשׁר‬clause in Jer 29:25.

56

Chapter 2

when the resulting word order reflects a disharmonic pattern (see Haegeman 1994: 418–23). The fact that examples such as (100) are rare, as are any divergent word orders within relative clauses, suggests that this particular clause type is simply not conducive for carrying the kind of extralinguistic discourse information that is signaled by Focus-fronting. And yet, while the internal syntax of the Hebrew relative may exhibit less variation than other clause types, the variation in the relative word, the relative head, the use of resumption, the position of the relative clause vis-à-vis its head, and the semantics of both the relative as a whole and the predicate within all testify to the Janus nature of this grammatical item.

Chapter 3

Relative Words 3.1.  Relative Word Typology The constituents that mark relative clauses are referred to variously (and often inaccurately) as relative pronouns, relative adverbs, and more generally, relative particles. The different types of relative elements exhibit distinct characteristics, either in their morphological shape or syntactic behavior. Figure 3.1 summarizes the types of relative elements exhibited cross-linguistically (de Vries 2002: 62). relative elements relative pronouns

relative particles

relative complementizers

relative markers

resumptive pronouns relative affixes

Figure 3.1.  Relative Elements. Relative and resumptive pronouns, as “pronouns,” carry agreement features (person, gender, number, or case 1) that match in some way the agreement features of the relativized noun in its role within the relative clause. Note that resumptive pronouns can be free or clitic. Relative markers are similar in that they show some characteristics of pronouns, e.g., by carrying agreement features; however, the agreement features on relative markers do not reflect the relative head’s syntactic role within the relative but, rather, its role within the matrix clause (relative markers provide what is essentially redundant 1.  I use case in this work to refer both to “morphological case,” which is the morphological variation that some languages use to signal syntactic relations, such as nominative, accusative, etc., in Latin; and to “abstract case” (often written as capital “Case”), which is the corresponding notion of case for languages lacking the relevant morphological variation—that is, such languages exhibit the same syntactic relationships between constituents as languages that show case morphologically. See Crystal 2008: 66–67, s.v. “case.”

57

58

Chapter 3

information, which explains why this is a very rare relative element type cross-linguistically). 2 In contrast to the pronouns and markers, relative complementizers and affixes do not carry agreement features. The complementizers and affixes are distinguished from each other by their morphosyntactic status: relative complementizers are separate words, whether free or bound, whereas relative affixes are verbal affixes indicating in some way that the clause in which the verb resides is a relative. English uses both relative pronoun and relative complementizer strategies for forming relative clauses. The relative pronouns are the wh-series, e.g., who, which, where, when (which are homophonous with the same series of interrogative pronouns), and the relative complementizer is that. These two types are illustrated in (101) and (102). (101) English Relative Pronoun The book which was published last year is already out of print. 3 (102) English Relative Complementizer The book that was published last year is already out of print. The relative complementizer that in (102) does not carry any agreement features associating it with the relative head, which is one reason why that is classified as a complementizer. It may seem that the relative pronoun which in (101) also lacks such features, but this is not accurate upon further consideration. The relative pronouns in English do not carry person, gender, or number features but instead agree with their heads in terms of the feature “human-ness” (103). Thus, the head dog, which has the semantic feature [-human], requires the relative pronoun which and disallows the pronoun who, but a head with the semantic feature [human], such as man, requires the pronoun who and dis­ allows the pronoun which.

2.  Deutscher 2001 describes the Old Akkadian relative marker šu, which was declined for case, number, and gender matching the relative head’s role in the matrix clause. He contextualizes its rarity by setting it in cross-linguistic relief: there are remnants in Arabic, with ʾallaḏi, a partial parallel in the “case attraction” construction of Greek and Latin, a possible parallel in Old Germanic (Icelandic), and another partial parallel with participial relatives in Tümpisa Shoshone. 3. Unlike that, the relative word which can introduce both restrictive and nonrestrictive relatives in English. However, without intonation (a pause would indicate a nonrestrictive relative) or commas (the orthographic convention for indicating a nonrestrictive relative), relative clauses introduced by which are semantically ambiguous—they can be restrictive or nonrestrictive, whereas relative clauses introduced by that are always restrictive).

Relative Words

59

(103) English [± human] Agreement with Relative Pronouns a. The man who(m)/*which I met last month has already left the company. b. Dog my which/*who I bought last month was a golden retriever. Note also that, in the literary/high register of English, the relative pronouns also carry case features reflecting the syntactic role of the head within the relative. For instance, in (103a) the role of the NP the man inside the relative is the accusative complement of the verb met; thus, the accusative pronoun whom is acceptable in this context within all registers and prescribed for high literary English. Relative pronouns marked for the case, gender, and number of the relative head are used in German, illustrated in (104). (104) German [case, gender, number] Agreement with Relative Pronouns Der Hund, den (ms acc) ich letzte Woche gesehen habe, war Jens *Der Hund, die (fs acc/cp acc) ich letzte Woche gesehen habe,   war Jens *Der Hund, das (ns acc) ich letzte Woche gesehen habe, war Jens ‘The dog that I saw last week was Jen’s’ The examples in (104) illustrate how in German the relative pronoun must agree in gender and number with its antecedent as well as the case of the relative gap within the relative. In (104), the relative head is co-indexed with the gap filling the accusative complement role within the relative; thus the relative pronoun must carry the accusative case or result in an unacceptable construction. English and German together illustrate the use of relative complementizers (English) and relative pronouns (English, German) but not relative markers, affixes, or resumptive pronouns. 4 Many languages also use a “zero” relativization strategy: instead of explicit marking by a relative complementizer, the complementizer is null and the head and relative clause appear simply to be juxtaposed. Such relatives are often referred to as “unmarked” or “bare” relatives. Consider the English pairs in (105)–(108), in which ∅ represents the position of the null-relative word (the “?” signals a grammatically marginal construction).

4.  Sells points out that English relatives sometimes appear to use resumptive pronouns, e.g., I’d like to meet the linguist that Mary couldn’t remember if she had met him before (Sells 1984: 11). But he shows how the properties of the pronouns in such cases in English differ from the properties of resumptive pronouns in, for example, Hebrew, Swedish, Irish, and Welsh. Sells calls the former, English, kind “intrusive pronouns” and the latter “resumptive pronouns.” See also McCloskey’s (2006) helpful overview of resumption.

60

Chapter 3 (105) The dog ∅ I bought ___ was a golden retriever cf. The dog that I bought ___ was a golden retriever (106) The dog ∅ ___ running down the street is rabid cf. The dog that ___ is running down the street is rabid (107) The dog ∅ ___ injured by the car was recovering well cf. The dog that ___ was injured by the car was recovering well (108) ?The dog ∅ ___ black with dirt was washed with a hose cf. The dog that ___ was black with dirt was washed with a hose

A relative-clause analysis is the only syntactically acceptable option for the clauses in (105)–(108), even though none includes an overt relative element. Take the relative in (105), for example: the head dog is the subject of the finite verbal predication was a golden retriever and therefore cannot also be the object of the verb bought apart from a context in which the head dog is coindexed with the object position (the gap in [105]) in an embedded structure, i.e, a relative clause. The same is true of the participial zero-relatives in (106) and (107) as well as the adjectival zero-relative in (108). The difference with the examples in (107)–(108) is that the subject within the relative clause is relativized. There are often language-specific constraints on zero subject-relatives. For instance, it is significant that the zero-relatives versus relatives introduced by a complementizer in (106)–(108) also differ in whether they contain a finite copula. Consider the pair in (109) and note the contrast in acceptability between the zero and marked relatives. (109) *The dog ∅ ___ bit the man was a golden retriever cf. The dog that ___ bit the man was a golden retriever The ill-formed relative in (109) accords with the Implicational Tendency in (4c), that a subject-relative clause is marked by a sentence-initial element. English does not allow zero relativization from the subject position within the relative, particularly when the relative contains a finite verb (see also the questionable relative in [108]). However, this appears to be a fairly weak tendency since many languages use zero relativization broadly, including relativization from the subject position (de Vries 2002: 58–60). Finally, it is helpful to note that a number of typological implicationals are associated with the combination of relative elements used in forming relative clauses. Table 3.1 visually presents all the logical combinations of relative elements, with representatives listed from the 172 languages that de Vries used in his study (de Vries 2002: 177, with the addition of Hebrew, Akkadian, and Ugaritic).

61

Relative Words

Table 3.1.  Combination of Relative Elements Rel Pro

Rel Comp

Rel Marker

Res Pro

(impossible)

+3 +17 (Arabic (Clas(Akan, Urhobo, Farsi, sical), Geez, Hebrew, . . .) Akkadian (Old), Ugaritic, . . .)

Rel Affix

+1 (Hurric)



Rel Marker





Rel Comp

+2* (Arabic (Tunisian), Hungarian [*plus some Germanic dialects]

Rel Affix +4 (Jacaltec, Kongo, Shona, Swahili, . . .)



Particularly relevant to Hebrew are two implicational universals, (110)–(111), repeated from (3g–h): (110) The use of a relative pronoun excludes a resumptive pronoun or clitic, and vice versa. (111)  The use of a relative complementizer excludes a relative particle, and vice versa. The implicational universals in (110) and (111) reflect the cross-linguistic fact that languages avoid double marking of a relative clause. Thus, very few languages combine, say, a relative pronoun and complementizer (*I know the man whom that you saw). The only context in which double marking a relative frequently occurs is when the subordinating element is not a pronoun, and there is a resumptive pronoun within the relative (e.g., a complementizer–resumptive combination). 3.2.  Hebrew Relative Elements Hebrew exhibits four types of relative elements—pronouns, markers, complementizers, and resumptive pronouns—and also uses the markers and all but one of the complementizers in combination with the resumptive pronouns. These are the patterns we will explore in this section. The data for the system of relative markers are incomplete, which together with the picture that emerges from the larger Semitic context (see appendix A), suggests that the marker system had been displaced, and the few extant examples are a remnant. The marker system is discussed below in §3.2.1. The majority of Hebrew

62

Chapter 3

relatives are either introduced by one of two complementizers, ‎‫אשׁר‬‎(§3.2.2) or ‫ה‬‎(§3.2.3), or are zero-marked (§3.2.5). Two other relative elements are used, but less commonly: the complementizer ‫שׁ‬‎(§3.2.2) and the relative pronouns that are homophonous with the interrogatives ‫ מי‬and ‫מה‬‎ (§3.2.4). Setting aside the rather small corpus of relatives representing the likely remnants of the declinable relative marker system (see below, §3.2.1), Hebrew relative elements share the morphosyntactic feature of being indeclinable. That is, unlike, for example, German relative words (104), which exhibit morphological agreement features with the head of the relative clause, the Hebrew relative elements ‫אשׁר‬‎ (112)–(115), ‫שׁ‬‎(116)–(118), and ‫ ה‬‎(119)–(122) do not decline and thus do not match their heads’ person, gender, number, or case. (112) ‫ת־ה ָא ָדם ֲא ֶׁשר יָ ָצר‬ ָ ‫ֶא‬ ‘(acc) the man(ms) who(m) he formed’ (Gen 2:8) (113) ‫‏ה ִא ָּׁשה ֲא ֶׁשר נָ ַת ָּתה ִע ָּמ ִדי‬ ָ ‘(nom) the woman(fs) who(m) you put with me’ (Gen 3:12) (114) ‫‏ה ְּמ ָל ִכים ֲא ֶׁשר ִאּתֹו‬ ַ ‘(nom) the kings(mp) who (were) with him’ (Gen 14:5) (115) ‫יֹוסף‬ ֵ ‫ר־ׁש ַלח‬ ָ ‫ת־ה ֲעגָ לֹות ֲא ֶׁש‬ ָ ‫‏א‬ ֶ ‘(acc) the carts(fp) that Joseph sent’ (Gen 45:27) (116) ‫‏ה ָא ָדם ֶׁשּיָבֹוא ַא ֲח ֵרי ַה ֶּמ ֶלְך‬ ָ ‘(nom) the man(ms) who will come after the king’ (Eccl 2:12) (117) ‫‏ּכ ְׁשגָ גָ ה ֶׁשּי ָֹצא ִמ ִּל ְפנֵ י ַה ַּׁש ִּליט‬ ִ ‘(acc) like an error(fs) that proceeds from the ruler’ (Eccl 10:5) (118)  5‫ל־ה ַּמ ֲע ִׂשים ֶׁשּנַ ֲעׂשּו ַּת ַחת ַה ָּׁש ֶמׁש‬ ַ ‫ת־ּכ‬ ָ ‫‏א‬ ֶ ‘(acc) all the events(mp) that have happened under the sun’   (Eccl 1:14) (119) ‫ל־רע‬ ָ ‫‏ה ַּמ ְל ָאְך ַהּג ֵֹאל א ִֹתי ִמ ָּכ‬ ַ ‘(nom) the messenger(ms) who (was) protecting me from every evil’   (Gen 48:16) (120) ‫‏ ָה ַע ְל ָמה ַהּי ֵֹצאת ִל ְׁשאֹב‬ ‘(nom) the maiden(fs) who (will) go out to draw water’ (Gen 24:43) 5.  There are only three cases where ‫ שׁ‬follows a feminine-plural noun: Song 4:1; 6:5, 6. In each case, the verb within the relative is a Qal perfect 3rd-common plural (‫ גָ ְלשּׁו‬in 4:1 and 6:5 and ‫ ָעלּו‬in 6:6). The verb, therefore, signals the plural features of the relative head but is ambiguous with regard to gender. Moreover, the feminine-plural noun preceding the ‫ שׁ‬in each case is the clitic host of the bound noun ‫‘ עדר‬flock’. It is probably the collective ‫ עדר‬that governs the plural verb (see IBHS 113, §7.2.1d for a brief discussion of collectives and their verbs).

Relative Words

63

(121) ‫‏הּכ ֲֹהנִ ים ַהּנִ ּגָ ִׁשים ֶאל־יְ הוָ ה‬ ַ ‘(nom) the priests(mp) who (are) approaching Yhwh’ (Exod 19:22) (122) ‫מֹועד‬ ֵ ‫ת־הּנָ ִׁשים ַהּצ ְֹבאֹות ֶּפ ַתח א ֶֹהל‬ ַ ‫‏א‬ ֶ ‘(ACC) the women(fp) who (are) serving (at) the entrance of the tent of meeting’ (1 Sam 2:22) The variation between masculine (112), (114), (116), (118), (119), and (121) and feminine (113), (115), (117), (120), and (122) and between singular (112), (113), (116), (117), (119), and (120) and plural (114), (115), (118), (121), and (122), without any corresponding variation in the morphology of the relative elements, leads to one conclusion: the Hebrew relative elements ‫אשׁר‬‎, ‫שׁ‬, and ‫ ה‬are not pronouns. Similarly, the lack of any case distinction supports this conclusion: the relative heads in (113), (114), (116), (119), (120), and (121) are in nominative syntactic roles, while those in (112), (115), (117), (118), and (122) are in accusative roles. Instead of comparing the Hebrew words to German relative pronouns, or even the English wh-series of pronouns, we are clearly dealing with items similar to English that and French que (in contrast to French qui). Thus, the Hebrew relative words should be syntactically categorized as complementizers. 6 As complementizers, ‫ אשׁר‬and ‫ שׁ‬function most commonly to establish that the following clause is a relative. But ‫ אשׁר‬and ‫ שׁ‬are similar to English that in having a second function: introducing a complement clause (123). (123)  He saw that he had struck the ball well In (123), the finite clause he had struck the ball well is the complement of the verb saw and is licensed for this role by the complementizer that. The Hebrew data in (124)–(125) illustrate that, like English that and French que, Hebrew ‫ אשׁר‬and ‫ שׁ‬also introduce verbal complements (for further discussion, see below, §7.2–3). (124) Verbal Complement Clause Introduced by  7‫אשׁר‬ ‫ֹלהים‬ ִ ‫ם־יֹוד ַע אָנִ י ֲא ֶשׁר יִ ְהיֶ ה־טֹּוב ְליִ ְר ֵאי ָה ֱא‬ ֵ ַ‫ִכּי גּ‬ ‘because I know that it will go well for those who fear God’   (Eccl 8:12)

6.  As I defined it above in §2.2.1, a complementizer heads the complementizer phrase. Its function is to establish what type of clause it is, such as main versus subordinate or declarative versus interrogative (Haegeman and Guéron 1999: 520–38). 7. For ‫ אשׁר‬complement clauses, see Exod 11:7; Lev 5:5; 26:40; Num 32:23; Deut 1:31; 3:24; Josh 4:7; 1 Sam 15:20; 18:15; 2 Sam 1:4; 14:15; 1 Kgs 22:16; Jer 28:9; Ezek 20:26; Pss 10:6; 89:52 (2×); Eccl 5:4, 17; 6:10; 7:18, 22, 29; 8:12, 14; 9:1; Esth 1:19; 2:10; 3:4; 4:11; 6:2; Dan 1:8; Ezra 2:63; Neh 2:10; 7:65; 8:14–15; 10:31; 13:1, 19, 22; 2 Chr 2:7; 18:15. For full context with translation, see appendix A.8.

64

Chapter 3 (125) Verbal Complement Clause Introduced by  8‫שׁ‬ ‫ת־כּ ָלּם‬ ֻ ‫וְ יָ ַד ְע ִתּי גַ ם־אָנִ י ֶשׁ ִמּ ְק ֶרה ֶא ָחד יִ ְק ֶרה ֶא‬ ‘and I know also that one fate befalls all of them’ (Eccl 2:14)

The term complementizer has spread to cover a larger class of function words, those that allow a clausal item to fill any nominal slot, whether as the external or internal arguments of a predicate (i.e., the subject and object, respectively) or as a nominal modifier. This latter function is precisely what the relative complementizers in Hebrew do: they take clausal constituents and allow them to modify a head in essentially the same fashion as adjectival modification does. This nominalizing role is the reason why I prefer the broader and descriptively more transparent term nominalizer for Hebrew ‫ אשׁר‬and ‫שׁ‬, since they both introduce relative clauses and complement clauses (fulfilling the valency of both verbs and prepositions; on the latter, see §4.3.2 and appendix A.8.b–c). 3.2.1.  ‫ז‬-Relatives Based on historical-comparative reconstruction (see Deutscher 2001; 2002; Huehnergard 2006; Lipiński 2001: 332–35) and a few suggestive examples in the Bible, it seems that Hebrew used to employ a relative marker related to the proximal demonstratives using the ‫ז‬-base. The demonstrative sequence in Hebrew is ‫( זֶ ה‬ms), ‫( זֹאת‬fs), and the etymologically unrelated ‫( ֵא ֶּלה‬cp). Of these, only the ms ‫ זֶ ה‬is used as a relative marker (126). 9 But there are also the rare fs variants ‫ זֹו‬and ‫( זֹה‬127) 10 as well as a form found only in poetry and appearing to be indeclinable: ‫זּו‬. 11 as

(126) ‫ְמקֹום זֶ ה יָ ַס ְד ָּת ָל ֶהם‬ ‘a place(ms)i-that you established ___i for them’ (Ps 104:8) (127) ‫יתי וְ ֵעד ִֹתי זֹו ֲא ַל ְּמ ֵדם‬ ִ ‫ְּב ִר‬ ‘my covenant and my laws(fp)i that I will teach them ___i’   (Ps 132:12) 8.  See Job 19:29; Song 5:8; Eccl 1:17; 2:13, 14, 15; 3:18; 7:10; 8:14; 9:5. 9.  See Exod 13:8; 2 Sam 14:2; Pss 74:2; 78:54; 104:8, 26; Job 19:19; Prov 23:22. Note that ‫ זה‬is used with a fs relative head in 2 Sam 14:2 and Ps 74:2, indicating that, early on, the agreement features of the relative marker were disappearing. Also, it is possible to read the first occurrence of ‫ זה‬in Isa 25:9 as a relative, though the poetic structure of the verse suggests that both cases of ‫ זה‬are deictic demonstratives, e.g., “Behold, our god is this one—we waited for him and he shall save us!” (see also Blenkinsopp 2000: 360; contra, e.g., Wildberger 1997: 442) 10.  There is only one case of the relative marker usage: the ‫ זֹו‬in Ps 132:12. All other examples of ‫ זֹו‬and ‫ זֶ ה‬are more simply analyzed as demonstratives: Judg 18:4; 2 Sam 11:25; 1 Kgs 14:5; 2 Kgs 6:19; Ezek 40:45; Hos 7:16; Eccl 2:2, 24; 5:15, 18; 7:23; 9:13. 11.  Exod 15:13, 16; Isa 42:24; 43:21; Hab 1:11; Pss 9:16; 10:2; 17:9; 31:5; 32:8; 62:12; 68:29; 142:4; 143:8. One other example of ‫זו‬, in Ps 12:8, appears to be a simple demonstrative, “from this generation.”

Relative Words

65

(128) ‫ית ְב ַח ְס ְּדָך ַעם־זּו ּגָ ָא ְל ָּת‬ ָ ‫‏נָ ִח‬ ‘you lead in your faithfulness a people(ms)i that you redeemed ___i’   (Exod 15:13) In (126) and (127), the demonstrative/relative marker agrees in gender with the relative head—masculine in (126) and feminine in (127). There is no indication of number agreement in the extant data nor any sign of case agreement. In (128), the relative ‫ זו‬follows ms ‫ עם‬as its head, but in other examples (e.g., fs ‫ רשׁת‬in Ps 9:16 and fp ‫ מזמות‬in Ps 10:2) the gender and number of the head differ. Though the evidence of declination in the extant Hebrew markers is limited, comparison with Ugaritic, in particular, but also Old Akkadian and Classical Arabic (see chap. 8) leads to the reasonable reconstruction of a fully functioning relative-marker system in pre-Biblical Hebrew. This system would have shown agreement features on the relative markers matching the gender and number of the relative head, and also probably the case of the relative head in its matrix clause syntactic role. Though a clear pattern of agreement is no longer available, the sparse marker data do witness a few other significant features. For instance, example (129) demonstrates that the relative marker could co-occur with a resumptive pronoun (the clitic pronoun ‫ ו‬in the PP ‫ ;בו‬note that in this example, ms ‫ זה‬does not agree in gender with fs ‫)הר ציון‬. (129) ‫ר־צּיֹון זֶ ה ָׁש ַכנְ ָּת ּבֹו‬ ִ ‫‏ה‬ ַ ‘Mount Zioni, whichi you dwelt in iti’ (Ps 74:2) And, though this moves us in the direction of relative heads (chap. 4), it is worth noting here that the ‫ז‬-relative markers also serve as clitic hosts for head NP in the bound form, as in (126) above, in which the head is ‫מקֹום‬, ְ with the vowel of the first syllable reduced to a šewa in accordance with the phonological pattern of cliticization. The ‫ז‬-relatives also modify null heads, as in (130), as well as proper nouns (and so can introduce nonrestrictive relatives; see below, §6.3), as in (131). (130) ‫וְ ָא ֵׁשם זּו כֹחֹו ֵלאֹלהֹו‬ ‘and (he)i who hisi might is his god became guilty’ (Hab 1:11) (131) ‫ֲהלֹוא יְ הוָ ה זּו ָח ָטאנּו לֹו‬ ‘Is it Yhwhi, who we sinned against himi?’ (Isa 42:24) Consider also a subtle feature of (132), in which I have included the Masoretic ‫טעמים‬. (132) ‫אתי ִמ ִמ ְצ ָ ֽריִ ם‬ ֖ ִ ‫ַב ֲע ֣בּור ֶ֗זה ָע ָ ֤שה יְ הוָ ֙ה ֔ ִלי ְב ֵצ‬ ‘on account of (the thing)i that Yhwh did ___i for me when I left Egypt’ (Exod 13:8)

66

Chapter 3

Example (132) shows a disjunctive ‫( טעם‬a ‫ )רביע‬on the relative marker ‫זה‬. Since the relative element introduces the subordinate clause, it is unexpected that it is prosodically separated from that clause by a disjunctive ‫טעם‬. And yet, the presence of the ‫ רביע‬on the ‫ זה‬is paralleled by six cases of a ‫ רביע‬on the relative complementizer ‫אשׁר‬‎(Ezek 6:11; 40:44; Ps 33:22; 56:7; 119:38, 85). Moreover, there are 251 cases of ‫ אשׁר‬with another significant disjunctive break, the ‫טפחא‬. 12 This suggests that whatever the pattern of the ‫ טעמים‬signals, the distribution depends on the prosodic features of the context before and especially after the word one is analyzing (see Dresher 1994 on the ‫)טעמים‬. Finally, we must examine two examples often cited as ‫ז‬-relatives: 13 (133)  14‫ֹלהי יִ ְש ָר ֵ ֽאל‬ ֥ ֵ ‫הו֖ה ֱא‬ ָ ְ‫הו֑ה ֶז֣ה ִס ַ֔יני ִמ ְפ ֵ֕ני י‬ ָ ְ‫ָה ִ ֥רים נָ זְ ֖לּו ִמ ְפ ֵנ֣י י‬ ‘the mountains flowed 15 before Yhwh; this one, Sinai, (flowed) before Yhwh, God of Israel’ (Judg 5:5) (134) ‫ֹלהי יִ ְש ָר ֵ ֽאל׃‬ ֥ ֵ ‫ֹלהים ֱא‬ ִ֗ ‫פּו ִמ ְפ ֵנ֪י ֱא ֹ֫ל ִ ֥הים ֶז֥ה ִס ַינ֑י ִמ ְפ ֵנ֥י ֱ֝א‬ ֮ ‫ף־ש ַ ֣מיִ ם נָ ְט‬ ָ ‫ֶ ֤א ֶרץ ָר ֨ ָע ָשה ׀ ַא‬ ‘the earth shook—even the heavens dripped—before God; this one, Sinai, (quaked) before God the God of Israel’ (Ps 68:9) 12.  Gen 1:7; 14:6; 18:17; 22:2, 17; 23:9, 17; 24:51; 25:9; 26:2; 27:14; 30:37; 34:22; 38:14; 39:22; 41:21, 54; 43:17, 19; 47:11; 50:11; Exod 5:11, 13; 7:10, 13, 22; 8:11, 15, 23; 11:5; 12:29; 16:24; 25:22, 26; 26:5; 29:23; 30:6; 33:7, 16; 36:12; 37:13; Lev 1:8, 12; 3:3, 4, 9, 10, 14, 15; 4:2, 7, 8, 9, 18; 5:17; 6:8; 7:4; 8:31; 9:7, 21; 10:5, 6, 15; 11:47; Num 3:16; 5:7; 12:3; 20:9, 27; 21:30; 22:36; 23:2, 30; 28:23; 30:2; 31:12, 48; 33:6; 34:29; Deut 1:11; 3:8, 25; 4:47; 6:14, 19; 7:6; 11:7, 25; 12:21; 14:2; 15:6; 17:10; 18:2; 21:2; 26:18; 28:9; 29:10, 12; 31:3; 32:49; 34:1; Josh 8:6; 9:10; 13:6, 14, 25, 33; 14:7, 12; 17:5, 7, 16; 19:11; 22:4, 10, 29; 23:10, 16; 24:26; Judg 1:16, 20; 8:21, 26; 9:33; 10:4; 11:7, 36; 15:10; 16:3; 18:16; 21:12, 13; 1 Sam 10:16; 14:1; 15:7; 17:31; 23:19; 24:5; 28:17; 30:29, 31; 2 Sam 2:32; 7:10; 14:14; 17:9; 19:17; 21:5, 18; 24:19; 1 Kgs 1:49; 5:9, 26; 7:17, 20, 41, 42, 48; 9:25; 11:7; 13:3, 32; 17:3, 5; 2 Kgs 5:4; 6:16; 11:10; 14:25; 18:26; 23:7, 10, 11, 17; 24:13; 25:10; Isa 7:18; 24:2; 28:14; 36:11; 56:5; 65:18; Jer 17:19; 20:2; 23:40; 25:22, 26; 27:5; 29:23; 32:2; 33:8; 38:14; 43:13; 44:24; 48:8; 52:14; Ezek 8:14; 9:6; 11:23; 20:26; 31:9; 38:20; 41:22; 42:3; 47:16; Mic 4:6; 5:14; Mal 3:18; 1 Chr 17:9; 19:16; 22:2, 11; 27:31; 2 Chr 1:6; 4:12, 13, 19; 11:13; 13:4; 17:10; 21:16; 23:9; 30:7, 14; 32:3; Ps 66:16; 79:12; 145:18; Job 9:5; 15:28; 22:15; 29:25; 36:24; Prov 6:7; 17:8; 21:1; 22:28; 25:7; Ruth 2:3; Eccl 3:9; 9:2; Esth 1:2, 9, 12; 4:6; 8:5; Dan 1:8, 20; 8:2; Neh 3:25; 5:12; 6:8; 8:1. I thank Gary Rendsburg for asking me about the disjunctive ‫ טעם‬in Exod 13:8 and so spurring my investigation of this issue. 13.  The following discussion of Judg 5:5 and Ps 68:9 is a slightly condensed version of Holmstedt 2014a. See there for more in-depth discussion of the verses, their poetic structures, and the relevant biblical scholarship. 14.  I have included the ‫ טעמים‬in both (133) and (134). It is worth noting that the Masoretic tradition divides the verse in Judg 5:5 (133) by the ‫ אתנח‬after ‫יהוה‬, thus placing ‫זה סיני‬ in the second half of the verse. Similarly, in Ps 68:9 (134) the ‫ עלי ויורד‬breaks the verse at ‫אלהים‬, leaving ‫ זה סיני‬once again in the second half. 15.  Often repointed as a Niphal from ‫זלל‬, so “quaked”; see the nrsv; DCH, s.v. ‫;נזל‬ HALOT, s.v. ‫נזל‬.

Relative Words

67

Both examples are from notoriously difficult poems, which are often considered to be among the earliest texts in the Bible. Judges 5 is the Song of Deborah about the Israelite victory over the army of Hazor, and Psalm 68 covers numerous themes, often dealing with Yhwh’s kingship, but exhibits little discernible structure. It has become common to analyze ‫ זה סיני‬in both examples as remnants of the Proto–West Semitic ḏ-series “determinative-relative” and to take the ‫ זה‬as a marker of the genitive, resulting in “Yhwh of Sinai” (Judg 5:5) and “God of Sinai” (Ps 68:9). 16 There are two reasons that the genitive “Yhwh of Sinai” analysis is very unlikely. First, the Semitic š/ḏ constructions—and so certainly the Hebrew ‫ז‬ examples—have been misunderstood (see chap. 8 for additional discussion). And second, a genitive analysis of ‫ יהוה זה סיני‬in (133) and ‫ אלהים זה סיני‬in (134) makes little sense in the poetic context. First, the grammar of ‫זה סיני‬: there is but one grammatically defensible way to analyze this sequence of words—as a deictic demonstrative followed by an appositive proper noun that clarifies the deictic reference, ‘this (mountain), Sinai’. Elsewhere in the Bible, the demonstrative ‫ זה‬functions as an adjunct to a noun that it follows or serves in the role of a noun (as, e.g., a subject), pointing backward or forward to a referential element in the discourse. 17 And it is never 16.  See, among others, Allegro 1955: 311; IBHS 337, §19.5; Lipiński 2001: 334; Huehnergard 2006: 111; Pat-El 2010: 44; 2012: 323; 2013a: 2 (note that in this latest article, ‫יהוה‬ ‫ זה סיני‬is mistakenly cited as Exod 15:13); Eskhult 2013: 48. 17.  Though the rare ‫ זה משׁה‬and ‫ זה ארבעים שׁנה‬constructions might be considered a suggestive pattern, this use of ‫ זה‬has been largely misunderstood. It is not the attributive use of the ‫ זה‬preceding its head (cf. IBHS 309–11, §17.4.1–2, where the explanation is confused—they inexplicably call this the “predicate” use; GBH 500, §143i), nor it is adverbial (GBH 498, §143a; Pat-El 2007). Rather, there are two syntactic patterns that explain all the examples: the NP following ‫ זה‬is in apposition to it or is the complement of a null-copula clause, of which ‫ זה‬is the subject. In the case of Exod 32:1 (‫ֹלהים‬ ִ ‫ה־לנּו ֱא‬ ָ ‫אמרּו ֵא ָליו קּום ֲע ֵׂש‬ ְ ֹ ‫וַ ּי‬ ‫ה־היָ ה לֹו‬ ָ ‫)א ֶׁשר יֵ ְלכּו ְל ָפנֵ ינּו ִּכי־זֶ ה מ ֶֹׁשה ָה ִאיׁש ֲא ֶׁשר ֶה ֱע ָלנּו ֵמ ֶא ֶרץ ִמ ְצ ַריִם לֹא יָ ַד ְענּו ֶמ‬, ֲ the ‫ זה‬cannot modify ‫משׁה‬, since that implies at least two men named Moses, “this one (that led us)” versus “that one (that did not lead us)”. Rather, the ‫ זה‬is the people’s way of dismissively referring to Moses, who has not met their expectations. The proper noun ‫ משׁה‬follows the ‫ זה‬to clarify it as an appositive, as does a second appositive, the more complex ‫ האישׁ‬with its relative clause. Thus, ‘because this one, Moses, the man who took us out of the land of Egypt—we don’t know what has happened to him’. In the second pattern, where ‫ זה‬is followed by a temporal phrase, the ‫ זה‬points backward or forward to an event that can be quantified, mostly in years. Consider for example, Deut 8:2,‫ֹלהיָך זֶ ה ַא ְר ָּב ִעים‬ ֶ ‫יכָך יְ הוָ ה ֱא‬ ֲ ‫ל־ה ֶּד ֶרְך ֲא ֶׁשר ה ִֹל‬ ַ ‫ת־ּכ‬ ָ ‫וְ זָ ַכ ְר ָּת ֶא‬ ‫ת־א ֶׁשר ִּב ְל ָב ְבָך‬ ֲ ‫ׁשנָ ה ַּב ִּמ ְד ָּבר ְל ַמ ַען ַעּנ ְֹתָך ְלנַ ּס ְֹתָך ָל ַד ַעת ֶא‬. ָ Rather than interpret ‫ זה ארבעים שׁנה‬as some sort of questionably formed adverbial phrase (pace Pat-El 2007), it is much simpler and avoids imaginative grammatical creations to follow the Masoretes in understanding a pause before the ‫ זה‬and so take the ‫ זה‬as the subject of a new clause: ‘And you shall remember the whole road that Yhwh your God led you on. This was forty years in the wilderness in order to humble you, to test you, to know what was in your heart.’ Examples like Gen 27:36 should be read similarly but with a zero-relative following the ‫ זה‬clause: ‫אמר ֲה ִכי ָק ָרא ְׁשמֹו יַ ֲעקֹב‬ ֶ ֹ ‫וַ ּי‬

68

Chapter 3

a clitic bound to the following word. All this suggests that the translations (and the analyses behind them) ‘Yhwh of Sinai’ or even ‘Yhwh, the one of Sinai’ do not reflect Hebrew grammatical structure. Second, based on the poetic structure, the ‫ זה סיני‬of both verses should be understood as parallel to ‫ הרים‬in (133) and ‫ ארץ‬in (134). The context of both passages is imagery and behavior associated with Yhwh’s Sinai theophany. As Webb says with regard to Judg 5:5 (but also applicable to Ps 68:9), This powerful poetry draws on the historical memory of the exodus, when Yahweh as the divine warrior manifested his presence by making wind and water change their normal behavior for the sake of his people, and Sinai, where his presence was manifested on the mountain in cloud and fire and earthquake. . . . Here in Judges 5 it gives powerful poetic expression to the belief that the storm which broke over Sisera and his army and threw them into panic and retreat was no merely natural event, or amazing stroke of luck; it was unleashed by Yahweh, as he had divided the Sea at the exodus and shaken Mount Sinai by descending on it. (Webb 2012: 208)

Thus, “this (mountain), Sinai” is experiencing the effects of Yhwh’s power, not defining who Yhwh is (as a genitive phrase would do). The genitive analysis thus fails by tests of grammar and literary sensitivity. 18 Instead, both examples should be understood as the Masoretes did, with ‫ זה סיני‬beginning the second stich (and thus the second clause) of both poetic verses. 3.2.2.  ‫אשׁר‬- and ‫שׁ‬-Relatives For unknown reasons, at an early stage of Biblical Hebrew the language shifted from using the common Semitic strategy of demonstrative/relative markers to using two other relative elements, ‫ אשׁר‬and ‫שׁ‬, illustrated in (135)–(136). (135) ‫ת־ה ֶּׁש ֶמׁש‬ ַ ‫ת־ה ַּמ ֲע ֶׂשה ֲא ֶׁשר נַ ֲע ָׂשה ַת ַח‬ ַ ‫‏ ֶא‬ ‘the deedi that ___i was done under the sun’ (Eccl 8:17) (136) ‫ל־ה ַּמ ֲע ִׂשים ֶׁשּנַ ֲעׂשּו ַּת ַחת ַה ָּׁש ֶמׁש‬ ַ ‫ת־ּכ‬ ָ ‫‏א‬ ֶ ‘all the deedsi that ___i were done under the sun’ (Eccl 1:14) Both are clitics, and ‫ שׁ‬is often orthographically prefixed to the first constituent of the relative. 19 The examples provided above in (112)–(118) illustrate the indeclinable nature of both ‫ אשׁר‬and ‫שׁ‬. And examples (124) and (125) show how ‫ת־ּבכ ָֹר ִתי ָל ָקח‬ ְ ‫‘ וַ ּיַ ְע ְק ֵבנִ י זֶ ה ַפ ֲע ַמיִם ֶא‬Was he really named Cheater? He cheated me—This is two times (that) my birthright he has taken!’ 18.  The issues surrounding the analysis of ‫ זה‬in Judg 5:5 and Ps 68:9 drive the point home that linguists, whether historical-comparative or theoretical, who work with ancient languages must not—indeed, cannot!—ignore the literary context of the data they use. 19.  However, four times in the same Qumran text (4QMMT) ‫ שׁ‬is also unattached to the relative and has a word-final epenthetic ‎‫א‬:‎ 4Q394 frgs. 3–7 i 5, 12, 19; and ii 14.

Relative Words

69

both are used to introduce other nominal, nonrelative clauses (also see §§4.3.2, 7.2–3, and appendix A). Finally, both relative words are used with resumptive pronouns (137)–(138) (see §5.2.4). (137) ‫וְ ִכ ֶּפר ַהּכ ֵֹהן ֲא ֶׁשר־יִ ְמ ַׁשח אֹתֹו‬ ‘and the priesti whoi one anoints himi . . . shall make atonement’   (Lev 16:32) (138) ‫ֹלהיו‬ ָ ‫ַא ְׁש ֵרי ָה ָעם ֶׁשיֲ הוָ ה ֱא‬ ‘happy is the peoplei whoi Yhwh is itsi god’ (Ps 144:15) As I argued above, these features undeniably indicate that these two function words are to be classified as relative complementizers, not pronouns. 3.2.3.  ‫ה‬-Relatives The second most common relative element after the complementizer ‫אשׁר‬ is the complementizer ‫ה‬. The relative ‫ ה‬is homophonous with the prefixed definite article ‫ה‬. Whereas the article ‫ ה‬is a prefix attached to nouns, the relative ‫ ה‬is a clitic that attaches to the first word of the relative, such as the verb ‫ ההרימו‬in (139). (139) ‫ֹלהינּו ַה ֵה ִרימּו ַה ֶּמ ֶלְך וְ י ֲֹע ָציו וְ ָׂש ָריו וְ ָכל־יִ ְׂש ָר ֵאל ַהּנִ ְמ ָצ ִאים‬ ֵ ‫ית־א‬ ֱ ‫רּומת ֵּב‬ ַ ‫ְּת‬ ‘the offeringi of the house of our God that the king and his counselors and his officials and all Israel who were present lifted ___i up’ (Ezra 8:25) In (139), the ‫ ה‬introduces a relative with a finite verb; however, most ‫ה‬-relatives are null-copula clauses, with the ‫ ה‬attached to the participial or adjectival complement of the copula. A participial ‫ה‬-relative is illustrated in (140). (140) ‫יהן‬ ֶ ‫‏ וְ ֶׁש ַבע ַה ָּפרֹות ָה ַרּקֹות וְ ָה ָרעֹת ָהעֹֹלת ַא ֲח ֵר‬ ‘and the seven lean and ugly cowsi that ___i come up after them’   (Gen 41:27) However, the rare examples of the finite verbal ‫ה‬-relative, such as (139), are instructive in that, when taken at face value, they clearly show the ‫ ה‬to be a relative complementizer and not a definite article. There are a number of constraints on Hebrew ‫ה‬-relatives: they do not allow the presence of copular pronouns, negation, topicalization, or wh-elements, and relativize only the subject of the relative clause. This last descriptive fact—that ‫ה‬-relatives relativize only the subjects of the relatives (which are also typically participial relatives)—accords with the general tendency in (4d): for most languages, there is an incomplete participial relativization strategy that is restricted to subjects. Example (139) is the lone exception to this in Hebrew. Finally, since the head is always co-indexed with the subject position within ‫ה‬-relatives, and the ‫ה‬-relative is attached to some constituent modifying the

70

Chapter 3

subject-head (e.g., participle, adjective), resumptive pronouns do not occur in Hebrew ‫ה‬-relatives, as the unacceptable (and unattested) example in (141) illustrates. (141) *‫יהן‬ ֶ ‫‏ וְ ֶׁש ַבע ַה ָּפרֹות ָה ַרּקֹות וְ ָה ָרעֹת ָה ֵהּנָ ה עֹֹלת ַא ֲח ֵר‬ ‘and the seven lean and ugly cowsi that (*theyi) come up after them’ (Gen 41:27, modified) Most reference grammars agree that there are some instances of the Biblical Hebrew definite article ‫ ה‬used as a relative word. 20 Eighteen clear examples of this relative construction exist in the Hebrew Bible; 21 17 are in the form of ‫ ה‬followed by a perfective verb, as in (139), and 1 exhibits ‫ ה‬followed by a verbless clause consisting of one member, a clitic preposition and its cliticized object pronoun (142). (142) ‫יה‬ ָ ‫ת־הּׁשֹוק וְ ֶה ָע ֶל‬ ַ ‫וַ ּיָ ֶרם ַה ַּט ָּבח ֶא‬ ‘and the cook lifted the leg and ∅(the thing)i that ___i (was) on it’   (1 Sam 9:24) Like the example in (139), many instances of the ‫ה‬-relative occur in later biblical texts. 22 But there are also a number of occurrences in arguably older texts, as (142) shows; 23 however, in such older texts the perfective verb forms to which the relative ‫ ה‬is attached are similar to the participial forms, differing only by a single vowel or the placement of stress. Thus, for an example such as (143), Waltke and O’Connor argue that “such forms should probably be read as participles . . . the article with the perfective is unlikely in early texts” (IBHS 339). 24 (143) ‫ּנֹולד־לֹו ֲא ֶׁשר־יָ ְל ָדה־ּלֹו ָׂש ָרה יִ ְצ ָחק‬ ַ ‫ם־ּבנֹו ַה‬ ְ ‫ת־ׁש‬ ֶ ‫וַ ּיִ ְק ָרא ַא ְב ָר ָהם ֶא‬ ‘and Abraham called the name of his soni that ___i was born to him, who(m) Sarah bore for him, Isaac’ (Gen 21:3) It is the form ‫ הנולד‬in (143) that is in question. It is vocalized in the Masoretic Text as a ‫ה‬-relative prefixed to a 3ms perfective, ‫נֹולד‬. ַ But this form can be changed to the form ‫ּנֹולד‬ ָ ‫ה‬, ַ a ms participle, by simply altering the final 20.  See GKC 447; Peretz 1967: 109–10; Lambert 1972: 135; IBHS 338–40; GBH 503–5. 21.  Gen 18:21; 21:3; 46:27; Josh 10:24; 1 Sam 9:24; Isa 51:10; Ezek 26:17; Job 2:11; Ruth 1:22; 2:6; 4:3; Ezra 8:25; 10:14, 17; 1 Chr 26:28; 29:17; 2 Chr 1:4; 29:36. Other possible examples are 1 Kgs 11:9; Isa 56:3; Jer 5:13; Dan 8:1; and 1 Chr 29:8. 22.  See, for example, Dan 8:1; Ezra 8:25; 10:14, 17; 1 Chr 26:28; 29:8, 17; 2 Chr 1:4; 29:36. 23.  Waltke and O’Connor cite Gen 21:3 and Ruth 1:22 as examples of these “older” occurrences (IBHS 340). By “older,” I mean that these texts are arguably preexilic or exilic, not postexilic. 24. Cf. Peretz 1967: 110; GBH 504–505; see also Kautzsch (GKC 447), who suggests that “no doubt the authors in all these cases intended participles” [emphasis added].

Relative Words

71

vowel from pataḥ to qameṣ. 25 Similarly, there are a number of forms that the Masoretic Text presents as words with penultimate stress, as in (144). These forms can be read as participles by a simple shift of the word stress to the final syllable, that is, ‫ ַה ָּׁש ָ֫בה‬for (144) instead of the penultimate stress marked by the Masoretic ‫מ ְר ָכא‬, ֶ viz. ‫ה ָּׁ֫ש ָבה‬. ַ  26 (144) ‫מֹואב‬ ֽ ָ ‫ֹוא ִביָ ֙ה ִ֔היא ַה ָ ֥ש ָבה ִ ֽעם־נָ ֳע ִ ֖מי ִמ ְש ֵ ֥דה‬ ֲ ‫נַ ֲָע ֤רה ֽמ‬ ‘She is a Moabite girli that ___i returned with Naomi from the fields of Moab’ (Ruth 2:6) Even if we accept the proposals for revocalization or reaccentuation of suspect forms, we are still left with nine examples, similar to example (139), that cannot be explained by altering the vowels or the word accent. 27 Clearly, the morpheme ‫ ה‬does serve, mostly in late texts but also in a few early ones, to introduce a relative clause. And the use of ‫ ה‬as a relative complementizer is indisputably grammatical, so there is no good reason for any of the vocalization patterns or word accents to be altered. Linguistic study of a similar phenomenon in Modern Hebrew, the use of ‫ ה‬as a relative complementizer, highlights the salient features of the biblical ‫ה‬-relative. In her study of the Modern Hebrew ‫ה‬-relative, Siloni points out the cross-linguistic fact that many languages use a single element as both the syntactic head of the DP (see above, p. 47 and n. 33) and a relative complementizer in restricted environments (Siloni 1995: 445). Based on examples (145)–(147), she argues that this relative complementizer is null in many languages, as in the French (145) and English (146) participial/reduced relatives, but overt in languages such as Modern Hebrew (where the relative complementizer is homophonous with the definite article ‫)ה‬, as in (147). Siloni labels this particular relative construction a semi-relative. (145) Un homme lisant un journal dans la rue est un espion. (146) A man reading a newspaper in the street is a spy. (147) ‫( אישׁ הקורא עיתון ברחוב הוא מרגל‬Siloni 1995: 446) ‘a man reading a newspaper in the street is a spy’ Siloni compares the synonymous pair in (148)–(149) to illustrate that ‫ ה‬can function syntactically just like the more common Modern Hebrew relative complementizer ‫שׁ‬. 25.  See also 1 Kgs 11:9; Isa 56:3; Dan 8:1. 26.  See also Gen 18:21; 46:27; Isa 51:10; Job 2:11; Ruth 1:22; 4:3. 27.  Josh 10:24 (‫ה ָה ְלכּוא‬, ֶ ‘who walked’); Ezek 26:17 (‫‘ ַה ֻה ָּל ָלה‬who was praised’); Ezra 8:25 (‫‘ ַה ֵה ִרימּו‬who lifted up’); 10:14 (‫‘ ַהה ִֹׁשיב‬who caused to dwell’), 10:17 (‫‘ ַהה ִֹׁשיבּו‬who caused to dwell’); 1 Chr 26:28 (‫‘ ַה ִה ְק ִּדיׁש‬who sanctified’); 29:17 (‫‘ ַהּנִ ְמ ְצאּו‬who were found’); 2 Chr 1:4 (‫‘ ַּב ֵה ִכין‬in [the place] where he established’); 29:36 (‫‘ ַה ֵה ִכין‬what he established’).

72

Chapter 3

Table 3.2.  Distinctions between ‫ שׁ‬and ‫ה‬/Semi-Relatives in Modern Hebrew ‫שׁ‬-Relatives Verbal form Copular pronouna Negation Topicalization wh-elements Relativized element

finite (+ tense) obligatory possible possible obligatory subject/object/ . . .

‫ה‬/Semi-Relatives participial (– tense) impossible impossible impossible impossible subject only

a. Modern Hebrew uses 3rd-person independent pronouns as the copula markers in restricted environments (Doron 1983; 1986; Rothstein 1995). For the use of a pronominal copula in Biblical Hebrew, see Holmstedt and Jones 2014.

(148) ‫הנה האישׁ שׁחושׁב רק על כסף‬ (149) ‫הנה האישׁ החושׁב רק על כסף‬ ‘here is the man who thinks only about money’ Siloni does, however, describe at least seven characteristics that distinguish ‫ ה‬semi-relatives from ‫ שׁ‬full relatives; the seven are listed in table 3.2 (Siloni 1995: 452). As the table indicates, Modern Hebrew semi-relatives only occur with participles and only a relative head that functions as the subject within the relative. Additionally, the Hebrew semi-relatives exclude an overt copula, negation, topicalization, and any other relative element. From examples (139)–(144), it is clear that all seven characteristics in table 3.2 are also true of Biblical Hebrew ‫ה‬-relatives, excepting only the rare examples with a finite verb. 28 One feature that table 3.2 does not address and that has been almost wholly neglected in Biblical Hebrew scholarship is the use of ‫ה‬-relatives with an indefinite relative head, as in (150) and (151). (150) ‫ל־ה ָא ֶרץ‬ ָ ‫ל־חּיָ ה ָהר ֶֹמ ֶׂשת ַע‬ ַ ‫ּוב ָכ‬ ְ ‘and over every creaturei that ___i creeps on the ground’   (Gen 1:28) 29 28.  In contrast, Biblical Hebrew’s full relatives differ from Siloni’s analysis of Modern Hebrew’s full relatives in that neither copular pronouns nor initial wh-elements in free relatives are obligatory in Biblical Hebrew. 29.  See Gen 1:28; 7:21; 49:17; Exod 26:12; 38:26; Lev 11:46; 16:16; Judg 21:19; 1 Sam 25:10; Isa 65:2; Jer 27:3; 46:16; 50:16; Ezek 2:3; 21:19; 28:16; 32:22, 24; 47:2; Ps 31:19; Prov 26:18; Song 4:5; Dan 9:26; Ezra 10:17; 1 Chr 26:28; 2 Chr 31:6.

Relative Words

73

(151) ‫ית־אל ִמזְ ְר ָחה ַה ֶּׁש ֶמׁש‬ ֵ ‫ימה ֲא ֶׁשר ִמ ְּצפֹונָ ה ְל ֵב‬ ָ ‫יָמ‬ ִ ‫ִהּנֵ ה ַחג־יְ הוָ ה ְּב ִׁשלֹו ִמּיָ ִמים‬ ‫ית־אל ְׁש ֶכ ָמה‬ ֵ ‫ִל ְמ ִס ָּלה ָהע ָֹלה ִמ ֵּב‬ ‘see—the festival of Yhwh (is) at Shiloh from year to year, which is north of Bethel, on the east of a highwayi that ___i goes up from Bethel to Shechem’ (Judg 21:19) In (150), the indefinite NP ‫ חיה‬is modified by a participial phrase introduced by ‫ה‬. In (151), the indefinite noun ‫‘ מסלה‬highway’ is modified by the definite participle ‫‘ העלה‬that-going up’. This construction is similar to those in (139)– (144) in that the morpheme ‫ ה‬is used to introduce a relative clause. What is significant about identifying examples such as (150) and (151) as relative clauses is that such constructions have most often been analyzed as “attributive participles.” But this cannot be, without modifying the consensus definition of attributive modification (GKC 408, §126u; IBHS 258, §14.3.1a; GBH 481, §138a), because these examples do not exhibit “agreement in definiteness.” 30 Participial ‫ה‬-relatives are not alone in occasionally modifying an indefinite NP. This ‫ ה‬asymmetry also occurs with adjectives, as in (152) and (153). (152) ‫רּוח ָה ָר ָעה‬ ַ ‫וְ ָס ָרה ֵמ ָע ָליו‬ ‘and a spiriti that ___i (was) evil would depart from him’ (≈ an evil spirit) (1 Sam 16:23) (153) ‫וַ ּיָבֹא ֵה ֶלְך ְל ִאיׁש ֶה ָע ִׁשיר‬ ‘a traveler came to a mani that ___i (was) rich’ (≈ a rich man)   (2 Sam 12:4) 31 Reference grammars often refer to the patterns in (152) and (153) as “exceptions” (Davidson 1901: 133; IBHS 621; GBH 481–82, §138b) or “an apparent discrepancy in definiteness” (IBHS 260). It is common to propose either that the Hebrew text is mistaken (in terms of the consonants or vocalization) or that the NPs are inherently definite. Rather than assuming that the Masoretic Text is corrupt or that the Masoretes who preserved and recorded the vocalization of the text were mistaken, clauses like those in (152) and (153) should be analyzed as ‫ה‬-relatives. A relative analysis avoids the problem of either massaging Hebrew grammatical patterns so that attributive adjectives do not have to agree with the modified noun in definiteness, or altering the text of the Hebrew Bible.

30.  Many grammarians do note examples in which the noun and its adjective do not agree in definiteness but, if they provide explanations, they are ad hoc. I have yet to find a unified explanation that includes these asymmetric examples. 31.  See Gen 41:26; Lev 11:10; Judg 16:27; 1  Sam 12:23; 16:23; 19:22; 2  Sam 12:4; 1 Kgs 7:8, 12; 2 Kgs 20:13; 24:4; Jer 6:16, 20; 17:2; 22:17; 38:14; Ezek 9:2; 21:19; 40:28, 31; 42:9; Zech 4:7; Ps 62:4; 104:18; Eccl 11:5; Neh 9:35; 2 Chr 4:10; 23:20.

74

Chapter 3

Analyzing the participial and adjectival modifiers in (139)–(144) and (150)–(153) as relatives paves the way for recognizing two related generalizations that have significant impact on our description of Hebrew grammar. First, the difference between simple adjectival modification of an NP and relativeclause modification is grounded in one criterion: how complex is the internal structure of the modifier? Second, when a participial or adjectival modifier has the complexity that requires projecting to a full clause, it typically does so as a null-copula relative clause that is either unmarked (i.e., a zero-relative; see below §3.2.5) or introduced by ‫ה‬. That ‫ ה‬relativization (along with zero relativization) is the highly preferred choice with participles is confirmed by the surprisingly few such relatives introduced by ‫ אשׁר‬or ‫שׁ‬. 32 There is a similarly small representation in the Qumran texts, Ben Sira, and the Mishnah. 33 What is critical about these examples is that not a single one consists of a bare participle. Instead, all the participles following ‫ אשׁר‬or ‫ שׁ‬have their own modifiers, which requires the projection of a larger phrase structure. The same is descriptively accurate regarding adjectives. For example, throughout the Bible, there are only six cases of ‫ אשׁר‬and ‫ שׁ‬followed by an adjectival relative (Gen 30:35; Judg 20:31; Ezek 42:13; Job 39:30; Eccl 2:26; 6:10). 34 Of these, only one case is a simple adjective, Job 39:30, and in that verse, the relative head is not the subject within the relative (which is a defining feature of ‫ה‬-relatives). Thus, none of these could have been ‫ה‬-relatives due to their internal structure. When we pull all the pieces of ‫ ה‬relativization together, the attractive picture that emerges is a unified explanation that spans the cline of internal complexity for NP modification (§1.1.3). On the internally simple end are bare modifiers, a category that includes adjectives (154) as well as participles (155) and stative verbs (156). (154) ‫נ ַֹח ִאיׁש ַצ ִּדיק‬ ‘Noah (was) a righteous man’ (≈ a mani that ___i was righteous)   (Gen 6:9) 32. For ‫אשׁר‬‎/‫ שׁ‬+ participial relatives in the Bible, see Gen 7:8; 39:22; Num 21:34; Deut 1:4; 3:2; 4:46; 1 Kgs 5:13; Isa 11:10; 49:7; Jer 38:16; Ezek 9:2; 13:3; 43:1; Zech 11:5; Pss 115:8; 135:18; Eccl 4:1; 8:12, 14; Esth 8:8; Neh 5:2, 3, 4; 2 Chr 34:10. 33.  From Qumran, see 1QS 11:6; 4Q274 frg. 1, col. i 8; 4Q410 frg. 1 3; 4Q419 frg. 1 9; 4Q504 frg. 1–2 (recto) vi 5. In Ben Sira, see 13:2 and 38:5. For the Mishnah, there are over 300 examples of ‫ שׁ‬+ participle. However, they are constructions in which ‫ ה‬either could not be used (since the head is not the subject within the relative, e.g., ‫‘ בזמן שׁ‬at the time that . . .’) or is never used for whatever reason (e.g., ‫‘ מי שׁ‬whoever . . .’). 34.  I found no ‫ אשׁר‬or ‫ שׁ‬+ adjectival relatives in the Qumran texts or Ben Sira. In the Mishnah, there are 31, but only 1 is a bare adjective (Kil. 5:1): see Kil. 4:1; 5:1; Maʿaś. 3:5; Šabb. 15:1; ʿErub. 6:3; 8:3; 10:15; Šeqal. 1:7; Beṣah 5:2; Moʿed Qaṭ. 3:9; Meg. 3:3; Yebam 4:13; Ned. 3:11; 8:1; B. Bat. 1:6; Sanh. 7:8; Šebu. 2:4; 3:7, 8; Hor. 3:7; Zebaḥ. 9:2; 13:7; Ḥul. 1:7; 10:3; Bek. 6:7; 9:3; Kelim 1:3; Neg. 12:2; Tehar. 7:4; Nid. 1:7; 7:1.

Relative Words

75

(155) ‫ּובנֵ י יִ ְׂש ָר ֵאל י ְֹצ ִאים ְּביָ ד ָר ָמה‬ ְ ‘and the sons of Israel (were) going out with a high hand (≈ a handi that ___i was high)’ (Exod 14:8) (156) ‫ׁש־לנּו ָאב זָ ֵקן‬ ָ ֶ‫י‬ ‘we have an old father’ (≈ a fatheri that ___i is old’) (Gen 44:20) What binds these three types of NP modifiers together is their fundamental adjectival status (see Cook 2008 for the typological justification; 2012: 223–33). When the modification is bare, as in (154)–(156), there is no reason to project a full clausal structure (even though it may sound better in translation, which is a target-language issue irrelevant to Hebrew grammatical analysis). Thus, I would clarify my primary translations for (152) and (153) with those in (157) and (158), which signal the nonclausal status of the modifiers. (157) ‫רּוח ָה ָר ָעה‬ ַ ‫וְ ָס ָרה ֵמ ָע ָליו‬ ‘and an evil spirit would depart from him’ (≈ a spiriti that ___i was evil) (1 Sam 16:23) (158) ‫וַ ּיָבֹא ֵה ֶלְך ְל ִאיׁש ֶה ָע ִׁשיר‬ ‘a traveler came to a rich man’ (≈ a mani that ___i was rich)   (2 Sam 12:4) In contrast, when the modification itself includes a modifier and so moves rightward on the cline of internal complexity, the motivation for projecting a clausal structure grows. When the internal structure includes verbal arguments, which often happens with participles (159), a clausal structure is obligatory. 35 (159) ‫ל־צ ִּדיק ָע ָתק ְּבגַ ֲאוָ ה וָ בּוז‬ ַ ‫‏ּת ָא ַל ְמנָ ה ִׂש ְפ ֵתי ָׁש ֶקר ַהּד ְֹברֹות ַע‬ ֵ ‘Let be stilled lips of deceiti that ___i (are) speaking against a righteous person arrogantly, with pride and contempt’ (Ps 31:19) Examples such as (159) are unambiguously relative clauses. And adjective modification with internal modifiers, as with the PP internal to the adjective phrase in (160) and (161), are also arguably relative clauses in their structure. (160) ‫ל־ּביתֹו ְּב ִמ ְכ ַסת נְ ָפשׁ ֹת‬ ֵ ‫ּוׁש ֵכנו ַה ָּקר ֹב ֶא‬ ְ ‫וְ ָל ַקח הּוא‬ ‘and he and his neighbori that ___i (is) close to his house shall take [a lamb] according to the number of people’ (Exod 12:4) (161) ‫ם־ה ֵא ֶלּה ֵהנָּ ה׃‬ ָ ִ‫א־מ ָע ֵרי ַהגֹּוי‬ ֵ ֹ ‫ל־ה ָע ִרים ָה ְרחֹקֹת ִמ ְמָּך ְמאֹד ֲא ֶשׁר ל‬ ֶ ‫ֵכּן ַתּ ֲע ֶשׂה ְל ָכ‬ ‘thus you shall do to all the citiesi that ___i are very far from you, which are not from the cities of these nations here’ (Deut 20:15) 35.  When a clausal structure is projected for adjectives and participles, it includes a null copula (see chap. 2, n. 35). With stative verbs, there is no null copula.

76

Chapter 3

The use of the ‫ ה‬is overlaid on this spectrum of NP modification, but the ‫ה‬ is not obligatorily present. If the ‫ ה‬is absent, the structure of the modification may run from the simple, bare adjectives, participles, and stative verbs, as in (154)–(156), to the highly complex zero-relatives in (162) and (163) (on zerorelatives, see below, §3.2.5). (162) ‫‏וַ יְ ִהי ֵע ָׂשו ִאיׁש י ֵֹד ַע ַציִ ד ִאיׁש ָׂש ֶדה וְ יַ ֲעקֹב ִאיׁש ָּתם י ֵֹׁשב א ָֹה ִלים‬ ‘and Esau was a mani (who) ___i (was) knowing hunting, a man of the field; and Jacob (was) a quiet mani (who) ___i (was) living (in) tents’ (Gen 25:27) (163) ‫אֹלהים י ְֹד ֵעי טֹוב וָ ָרע‬ ִ ‫יתם ֵּכ‬ ֶ ‫‏ וִ ְיִה‬ ‘and you shall become like godsi (who) ___i know good and bad’   (Gen 3:5) The data adduced in this section also illustrate the variety of modificational complexities introduced by the ‫ה‬, from the bare participle in (164) to the more complex structures above in (159) and (160). (164) ‫מֹול ֵדְּתנּו ִמ ְּפנֵ י ֶח ֶרב ַהּיֹונָ ה‬ ַ ‫ל־א ֶרץ‬ ֶ ‫ל־ע ֵּמנּו וְ ֶא‬ ַ ‫קּומה וְ נָ ֻׁש ָבה ֶא‬ ָ ‘get up, let us return to our people, to the land of our birth, away from a swordi that ___i is oppressing’ (Jer 46:16) It appears that, whereas ‫ אשׁר‬and ‫ שׁ‬force a full clausal structure and so reside on the far end of the complexity cline, ‫ ה‬interacts with NP modification in either subclausal or clausal form. Thus, the ‫ ה‬straddles the major part of the cline, with only the limitations listed in table 3.2. These data suggest that the “agreement in definiteness” that grammars cite as a feature of attributive modification has been misunderstood. When the ‫ ה‬is attached to an NP modification—whether an adjective, a participle, or a stative verb—it functions simply as a subordinator. 36 When the ‫ ה‬is attached to an NP itself, whether standing in an argument role (e.g., subject, complement) or in apposition to another NP, it is a marker of identifiability (Bekins 2013). The examples I have presented also make it clear that ‫ ה‬need not be present for all the same types of NP modification to be used. But since unmarked subordination is inherently more difficult to process than marked subordination, the dominant pattern in Hebrew is to include the ‫ה‬, especially when it creates a symmetry of ‫ ה‬usage—the ‫ ה‬on the head marking identifiability and the ‫ ה‬on 36.  The common semantic feature of adjectives, participles, and stative verbs as NP modifiers is that they make a predicational assertion, not an identificational one, as a referential NP in apposition would make. This semantic feature distinguishes these types of NP modification, which are all associated with the subordinating ‫ה‬, from apposition, in which both the head and the modifier are NPs. On predication versus identification, see Stassen 1997: 101–13.

Relative Words

77

the modifier signaling predicational modification. These two metalinguistic motives, ease of processing and aesthetics (i.e, symmetry), have resulted in the common misanalysis of the function and distribution of ‫ה‬. Finally, we may note a possible diachronic feature of the ‫ ה‬subordinator. In his study of the article in Biblical Hebrew, Barr mentions the use of ‫ ה‬as a relative marker (1989: 322–25) and proposes that the relative article has a main function other than that of normal determination; it is frequent in some poetic texts in which the usual article is rare; and it may possibly suggest a path which leads from an older state of the language, in which determination by the article was unusual, to the classical state, in which such determination was central. (Barr 1989: 325)

Recent study of the definite article in Hebrew and Phoenician suggests that Barr was correct, that the origin of the Hebrew article lies in the function of ‫ה‬ as a subordinator (Gzella 2006; cf. Pat-El 2009). In this vein, I suggest that the article in Central Semitic, which was a relatively late innovation in each of the languages, began as a subordination marker for NP modification, not identifiability (i.e., definiteness). Rather, since NPs that are highly modified are often identifiable, it is likely that the ‫ ה‬was reanalyzed so that it took on the function of marking identifiability and so attached also to nonsubordinate NPs. As for the relative function, the ‫ ה‬may originally have had no constraints, as hinted at by the few examples with a finite verbal clause (see example [139]). But competition with the already existing relative-‫ ז‬marker and then the advent of the ‫ אשׁר‬complementizer (and then also ‫ )שׁ‬eventually restricted the ‫ ה‬to adjectival and participial relatives and thus the constraints listed in table 3.2. 37 3.2.4.  ‫מ‬-Relatives The final set of relative elements that Hebrew employs are homophonous with the interrogative pronouns ‫ מי‬and ‫( מה‬see GKC §137c; IBHS 319–21, 325, §§18.2.c, e, 18.3e; GBH 502–3, §144fa–g). Typically, the ‫ מי‬and ‫ מה‬pronouns are associated with a +Q(uestion) feature in the CP that establishes the clause as an interrogative. However, as we will see below, there are a few clear cases in which ‫ מי‬and ‫ מה‬establish the following clause as a relative, not an interrogative. In this way, these two pronouns are very similar to English 37.  Eskhult (2013) presents a slightly different argument, which remains unclear to me at many points. He seems to argue that the “abberant” use of ‫ ה‬as a relative word with finite verbs is the result of a combination of extension (applying the common ‫ ה‬+ participle syntax to finite verbs that are homophonous with their participial counterparts) and Aramaic influence in the postexilic period (specifically, that the usage of Aramaic ‫ די‬corresponds to both the usage of relative ‫ ה‬and ‫אשׁר‬, so that, “by a confusion of expressions, Aramaic dī with the perfect was occasionally rendered by ha- with the perfect”; Eskhult 2013: 55). The diversity of the data, as I have covered in this section, is more complex than Eskhult suggests and cannot be, in my opinion, explained as Eskhult does.

78

Chapter 3

wh-words (e.g., who, what, which), which also function as both interrogatives and relative pronouns. That ‫ מי‬and ‫ מה‬should be classified as pronouns, instead of, say, complementizers is established by their ± animacy agreement, with ‫ מי‬used for animate entities (mostly human) (165) and ‫ מה‬used for inanimate entities (166). (165) ‫ִמי ִהּגִ יד ְלָך‬ ‘Who told you . . . ?’ (Gen 3:11) (166) ‫אתי‬ ִ ‫ה־ּפ ְׁש ִעי ַמה ַח ָּט‬ ִ ‫ַמ‬ ‘What is my crime? What is my sin?’ (Gen 31:36) So, too, in introducing relative clauses the ‫ מ‬pronouns are distributed according to animacy, as in (167) [+ animate] and (168) [– animate]. (167) ‫ת־ח ָדׁש וְ לֹא ֲחנָ כֹו‬ ָ ִ‫י־ה ִ֞איׁש ֲא ֶׁשר ָּבנָ ה ַבי‬ ָ ‫ִמ‬ ‘whoever is the man who has built a new house and not dedicated it’ (Deut 20:5) (168) ‫ה־ּלְך‬ ָ ‫אמר נַ ְפ ְׁשָך וְ ֶא ֱע ֶׂש‬ ַ ֹ ‫ַמה־ּת‬ ‘whatever your soul says, I will do for you’ (1 Sam 20:4) The feature ±animacy by itself does not tell us whether relative ‫ מי‬and ‫מה‬ are pronouns or markers (since that distinction depends on case-marking; see above, §3.1). However, it is suggestive that there is not a single case of relative ‫ מי‬or ‫ מה‬co-occurring with a resumptive pronoun (see below, §3.2.6). Relatives formed with ‫ מי‬and ‫ מה‬are rare, with only 37 examples (and some of those can be read as true interrogatives). 38 This rarity may be due to the fact that Hebrew already had sufficient relative elements and, as I will discuss below, the relative use of ‫ מי‬and ‫ מה‬often signals a specific semantic interpretation. Of the 37 likely examples in the Bible, there is a lone example of a headed ‫ מה‬relative, given in (169), while the remaining 36 are headed covertly—that is, by a null head (see §4.3), as in (170) and (171). (169) ‫ְּוד ַבר ַמה־ּיַ ְר ֵאנִ י וְ ִהּגַ ְד ִּתי ָלְך‬ ‘and the word-what he shows me I will tell you’ (≈ whatever he shows me . . .) (Num 23:3) (170) ‫יתי ַמ ֲהרּו ֲעׂשּו ָכמֹונִ י‬ ִ ‫יתם ָע ִׂש‬ ֶ ‫ָמה ְר ִא‬ ‘what you saw (that) I did—hurry, do like me’ (≈ whatever you saw me do . . .) (Judg 9:48) 38.  See Exod 24:14; 32:26, 33; Num 23:3; Deut 20:5, 6, 7; Josh 14:15; Judg 7:3; 9:48; 10:18; 1 Sam 20:4; 2 Sam 20:11 (2×); Jer 5:15; 9:11; Hos 14:10; Prov 9:4, 16; Eccl 1:9 (2×); 3:15, 22; 5:9; 6:10; 7:24; 8:7; 9:4; 10:14; Esth 5:3, 6 (2×); 7:2 (2×); 9:12 (2×); 2 Chr 36:23.

Relative Words

79

(171) ‫ה־ׁשּנַ ֲע ָׂשה הּוא ֶׁשּיֵ ָע ֶׂשה‬ ֶ ‫ּומ‬ ַ ‫ה־ּׁש ָהיָ ה הּוא ֶׁשּיִ ְהיֶ ה‬ ֶ ‫ַמ‬ ‘whatever has been—it is what will be, and whatever has happened—it is what will happen’ (Eccl 1:9) In (169)–(171), the function of the ‫ מה‬partially parallels that of the other Hebrew relative elements but also illustrates differences specific to ‫מה‬-relatives. In terms of similarity, note for example that in (169) the head NP ‫ דבר‬is bound to the relative ‫מה‬, like the head of the ‫ז‬-relative is in (126) above or the head of the ‫אשׁר‬-relative is in (172): (172) ‫ת־הע ָֹלה ִל ְפנֵ י יְ הוָ ה‬ ָ ‫‏ּב ְמקֹום ֲא ֶׁשר־יִ ְׁש ַחט ֶא‬ ִ ‘in the placei-that he slaughters the burnt offering ___i before Yhwh’ (Lev 4:24) A cliticized relative NP is not a common pattern (although it does signal particular relative semantics; see §6.3), though the fact that this is used even within a very small corpus of ‫מ‬-relatives suggests that in most features these relatives match the other Hebrew relative strategies. For example, ‫ מי‬and ‫מה‬relatives may correspond to both subject (173)–(174) and complement position (175)–(176). (173) ‫ִמי־יָ ֵרא וְ ָח ֵרד יָ שׁ ֹב‬ ‘Whoeveri ___i is afraid and trembling should return’ (Judg 7:3) (174) ‫יאּנּו ִל ְראֹות ְּב ֶמה ֶׁשּיִ ְהיֶ ה ַא ֲח ָרי‬ ֶ ‫ִמי ִיְב‬ ‘Who will bring him to see whati ___i will be after him’ (Eccl 3:22) (175) ‫ת־מי ַת ֲעבֹדּון‬ ִ ‫ַּב ֲחרּו ָל ֶכם ַהּיֹום ֶא‬ ‘choose for yourselves today who(m)i you will serve ___i’   (Josh 24:15) 39 (176) ‫ה־ּלְך‬ ָ ‫אמר נַ ְפ ְׁשָך וְ ֶא ֱע ֶׂש‬ ַ ֹ ‫ַמה־ּת‬ ‘whateveri your soul says ___i , I will do for you’ (1 Sam 20:4) Additionally, the discourse context of ‫מ‬-relatives often indicates that the null head of the ‫מה‬-relative has an identifiable, specific interpretation, which is provided by the content of the relative. Thus, in (170) Abimelech wants his men to do precisely what he did (which was narrated in the preceding three clauses). This unique, identity reading of the null head is similar to the semantics of all null-head relatives with ‫ אשׁר‬or ‫( שׁ‬see §4.3). But the differences that ‫מ‬-relatives show versus other Hebrew relative elements are significant. First, notice that the relative ‫ מה‬can be used with another relative element to double mark a relative clause, as with the combination of 39.  On the ‫ את‬marking the headless ‫מי‬-relative, see Bekins 2014: 98–102.

80

Chapter 3

‫ מה‬and ‫ שׁ‬in (171) (also see below, §3.2.6). In fact, the status of ‫ מי‬as a relative element in the pattern ‫ מי האישׁ אשׁר‬is questionable—the ‫ מי‬certainly contributes to the free choice construction (see next paragraph) but does so as part of the relative head rather than as a relative element. Second, only about half of the likely ‫מ‬-relatives in the Bible have an identity reading. The other half present a distinct subtype of relatives that only appear with the ‫מ‬-relative elements: free-choice relatives. 40 The null head of a free-choice ‫מ‬-relative has a nonspecific, generic reading, as in (171). The context of (171)—indeed, the context of most free-choice relatives in the Bible— is sapiential discourse. For (171), the sage Qoheleth is making his case about the fundamentally cyclical nature of the human experience. The ‫מ‬-relatives in (171) are thus generic, universal statements and the null-relative heads are indefinite. In free-choice relatives, the speaker does not know or does not care what the referent of ‫ מה‬is, which accounts for the indeterminacy and genericity of these relatives (Heller and Wolter 2011: 170). The semantics of Hebrew ‫מ‬-relatives are thus similar to English relatives with a wh-word and -ever—e.g., “whoever” and “whatever.” Like Hebrew ‫מ‬-relatives, English -ever free relatives establish for the relative head either an identity (unique) reading, as in (177), or free choice (generic) reading, as in (178) (Dayal 1997). (177) Whichever movie is now playing at the Avon is making a lot of money. (178)  Whichever movie plays at the Avon makes a lot of money. In (177), the phrase “whichever movie” refers to a unique referent: the movie now playing, even though the speaker does not know the precise identity of the movie. In contrast, the phrase “whichever movie” in (178) does not have a unique referent but is a generic or kind-referring NP and thus refers to any and every movie that plays at the Avon. According to Dayal, the interpretation of “whatever movie” in these two examples is often related to the tense-aspect 40.  I have judged these to be identity ‫מ‬-relatives: Exod 32:26; 32:33; Num 23:3; Deut 20:5, 6, 7; Josh 24:15; Judg 7:3; 9:48; 10:18; 2 Sam 20:11 (2×). 2 Chr 36:23 may also be another example. I interpret the following as free-choice ‫מ‬-relatives: Exod 24:14; 1 Sam 20:4; Jer 5:15; 9:11; Hos 14:10; Prov 9:4, 16; Eccl 1:9 (2×); 3:15, 22; 5:9; 6:10; 7:24; 8:7; 9:4; 10:14; Esth 5:3, 6 (2×); 7:2 (2×); 9:12 (2×). Although some grammars list Judg 21:5; 2 Sam 18:12; 21:4; Isa 50:8 (2×), I have determined that a relative analysis of the ‫ מ‬pronouns in each of these is unlikely (or, in the case of 2 Sam 18:12, the text is simply problematic). Finally, I must note that, even in the texts that I have judged to be identity ‫מ‬-relatives, a nonrelative analysis could be felicitous. For example, in the case of Exod 32:33, ‫ִמי ֲא ֶׁשר‬ ‫א־לי ֶא ְמ ֶחּנּו ִמ ִּס ְפ ִרי‬ ִ ‫ח ָט‬, ָ the ‫ מי‬might introduce a relative (‘Whoever has sinned against me—I will wipe him from my book’) but it could also be an interrogative (‘Who has sinned against me? I will wipe him from my book’). Determining which reading is contextually most likely is no simple matter.

Relative Words

81

difference, between the present progressive in (177) and the simple present in (178). Though no specific verbal form is associated with the identity versus freechoice interpretation in Hebrew ‫מ‬-relatives, I have used the context of the relatives in (170) and (171) to show how the ‫מ‬-relative pronouns are used for both identity and free-choice types. The null-relative head in (170) refers to a unique event, whereas the null-relative heads in (171) refer to generic entities or events. A final example of a ‫ מ‬free-choice relative, in (179), will suffice to illustrate the nature of these constructions. (179) ‫ִמי־בַַעל ְּד ָב ִרים יִ ּגַ ׁש ֲא ֵל ֶהם‬ ‘whoever is an owner of issues shall draw near to them’ (≈ whoever has an issue . . .) (Exod 24:14). In the context of (179), Moses is taking Joshua with him farther up the mountain of Yhwh to get the tablets of the law. He instructs the elders waiting below about what to do if any of the people has a complaint that requires adjudication: any complainant should take the case to Aaron or Hur, who also are remaining below. Thus, when Moses says ‫מי בעל דברים‬, he has no knowledge of the referent, nor does he particularly care who comes with a complaint. The null head of the ‫מי‬-relative is thus generic or kind-referring, and the relative as a whole has a free-choice interpretation. 3.2.5. Zero-Relatives In addition to the five overt types of relative elements (‫ז‬-series, ‫אשׁר‬‎, ‫שׁ‬‎, ‫ה‬, and ‫מ‬-series), Hebrew also allows a zero 41 relativization strategy: instead of explicit marking by a relative element, a null (“zero”) relative complementizer is used, and the head and relative clause appear to be simply juxtaposed. 42 Consider the following minimal pair (I use ∅ to represent the null complementizer). 41. Zero-relatives are typically referred to as “asyndetic” or “unmarked” in Hebrew studies; see, for example, Joüon 1923: 481–82; Brockelmann 1956: 143–45; IBHS 338, §19:6; GBH 558–59, §158a; cf. GKC 486–89. 42. The following are the 298 zero-relatives I have identified in the Hebrew Bible (see appendix A.5 for the Hebrew with translation): Gen 1:1; 15:13; 24:22 (2×); 26:10; 29:25; 39:4; 42:28; 49:27; Exod 4:13; 6:28; 14:11; 15:17 (2×); 18:20; Lev 7:35; 13:20, 39; 17:11; 22:4; Num 7:13 (2×); Deut 4:15, 38; 7:1; 9:1 (2×), 14; 11:23; 32:11 (5×), 17 (4×), 37; Josh 7:21; Judg 8:1; 15:11; 16:15; 1 Sam 6:9; 10:11; 14:21; 26:14; 2 Sam 20:21; 21:4; 22:44; 23:1, 9; 1 Kgs 13:12; 17:14; 22:24; 2 Kgs 3:8; Isa 1:4, 30; 6:6; 10:3; 14:16–17 (2×); 15:7; 23:13; 28:16; 29:1; 30:5, 9; 31:1, 8 (2×); 32:13; 40:20 (2×); 41:24; 42:1 (2×), 11, 16 (2×); 44:1, 2; 48:17; 50:9; 51:1 (2×), 2, 7, 12 (2×); 53:7 (2×); 54:1 (2×); 55:13; 56:2 (2×); 62:1; 63:19 (2×); 64:2; 65:1 (3×); 66:1; Jer 2:2, 6 (2×), 8, 11; 5:15, 21 (3×); 13:20; 15:14; 23:9, 29; 30:21; 31:9; 36:2; 48:36; 49:12; 51:43 (2×); Ezek 20:25 (2×); 22:24 (2×); 38:11;

82

Chapter 3 (180) ‫ת־ה ָּד ָבר ֲא ֶׁשר נַ ֲע ֶׂשה‬ ַ ‫ְך־ּבּה וְ ֶא‬ ָ ‫ת־ה ֶּד ֶרְך ֲא ֶׁשר נֵ ֶל‬ ַ ‫ֹלהיָך ֶא‬ ֶ ‫ד־לנּו יְ הוָ ה ֱא‬ ָ ֶ‫וְ יַ ּג‬ ‘and let Yhwh your God tell us the way that we should walk in it and the thing which we should do’ (Jer 42:3) (181) ‫ת־ה ַּמ ֲע ֶׂשה ֲא ֶׁשר יַ ֲעׂשּון‬ ַ ‫ת־ה ֶּד ֶרְך יֵ ְלכּו ָבּה וְ ֶא‬ ַ ‫הֹוד ְע ָּת ָל ֶהם ֶא‬ ַ ְ‫ו‬ ‘and you shall make known to them the way ∅(that) they will walk in it and the work which they will do’ (Exod 18:20)

The minimal pair in (180) and (181) illustrates that the null complementizer of Hebrew zero-relatives may appear in the same configurations as overt relative elements: the zero-relative clause in (181) modifies a head in the accusative and contains an element resuming the head, just as the ‫אשׁר‬-relative clause does in (180). Also like marked relatives, zero-relatives allow a gap as well as a resumptive pronoun: (182) ‫ֹלהים לֹא יְ ָדעּום‬ ִ ‫ֹלה ֱא‬ ַ ‫יִ זְ ְּבחּו ַל ֵּׁש ִדים לֹא ֱא‬ ‘they sacrifice to demonsi ∅(that)i ___i are not divine, godsj ∅(that)j they have not known themj’ (Deut 32:17) 43 (183) ‫וַ ּיִ ּפֹל ְּב ַׁש ַחת יִ ְפ ָעל‬ ‘he has fallen into a piti ∅(that)i he made ___i’ (Ps 7:16) 44

Hos 4:14; 6:3; Amos 2:13; 3:9; Jonah 1:10; Mic 5:2; Hab 1:5, 6, 8, 14; 3:16; Zeph 2:1, 2; 3:3, 17; Zech 6:12; Mal 2:16; Ps 4:8; 5:5; 7:9, 16; 9:16, 18; 12:6; 15:3–5 (8×); 16:4; 17:1, 9; 18:3, 44; 21:5; 25:12 (2×); 32:2 (2×); 33:12; 34:9; 35:8; 38:14; 44:2; 46:2; 51:10; 52:9; 56:4, 10; 58:5; 65:5 (3×); 66:9; 68:31; 71:18; 74:2 (2×); 78:6, 8 (2×), 39; 81:6; 83:15 (2×); 84:6 (2×); 90:5, 15 (2×); 103:5; 104:9, 32; 105:8; 107:40; 109:16; 118:22; 141:9; Job 3:3 (2×), 15, 16; 4:7 (3×); 6:10, 17; 7:2 (2×); 9:26; 11:16; 13:28; 17:3; 18:21; 19:3; 21:27; 27:3 (2×); 28:1, 7 (2×); 29:12, 16; 31:12 (2×); 32:19; 33:21; 34:7–8 (2×); 35:5; 36:27; 38:19, 24 (2×); Prov 6:16; 8:32; 9:5, 16; 22:29; 23:32 (2×); 26:17; 30:15 (2×), 17 (2×), 18 (2×), 24, 29 (2×); Eccl 1:13, 18; 2:1, 26; 5:12; 10:5; 12:11; Lam 1:1, 10, 14, 21; 3:1, 25, 41, 57; 4:14, 17; Ezra 1:5; Neh 8:10; 1 Chr 12:24; 15:12; 16:15; 29:3 (2×); 2 Chr 1:4; 15:11; 16:9; 18:23; 20:22; 24:11; 28:9; 29:21; 30:19; 31:19. 43.  See also Exod 18:20; Deut 32:17; Isa 42:1; 44:1, 2; 63:19; Jer 2:6, 8; 5:21; Ezek 22:24; Hab 1:14; Pss 18:3; 32:2; 34:9; Prov 26:17; Job 3:15; Neh 8:10. 44.  See also Gen 15:13; 24:22; 26:10; 29:25; 39:4; 42:28; 49:27; Exod 4:13; 9:4; 14:11; 15:17; Lev 7:35; Num 7:13; Deut 32:11, 35; Josh 7:21; Judg 8:1; 1 Sam 6:9; 26:14; 2 Sam 20:21; 22:44; 1 Kgs 13:2; Isa 1:30; 6:6; 15:7; 28:16; 30:9; 40:20; 41:2, 24; 42:16; 48:17; 51:1, 2, 7, 12; 53:7; 54:1; 55:13; 56:2; 61:10, 11; 62:1; 64:2; 65:1; 66:1; Jer 2:11; 13:20; 15:14; 23:9, 29; 36:2; 48:36; Hos 4:14; 6:3; Jonah 1:10; Mic 5:2; Hab 1:6, 8; Zeph 2:1; Zech 6:12; Mal 2:16; Pss 4:8; 5:5; 7:16; 8:9; 9:16, 18; 12:6; 16:4; 17:1; 18:44; 25:12; 33:12; 35:8; 38:14; 42:2; 49:14; 51:10; 56:4, 10; 58:5; 65:5; 68:31; 71:18; 74:2; 78:6; 81:6; 83:15; 90:5, 15; 103:5; 118:22; 119:136; 125:1; 129:6; 141:9; Prov 8:32; 30:17; Job 1:1; 3:3; 6:17; 7:2; 9:26; 11:16; 13:28; 18:21; 21:27; 28:1; 29:16; 31:12; 36:27; 38:19, 24; Eccl 10:5; Lam 1:10, 14, 21; 3:1; Ezra 1:5; 1 Chr 12:24; 15:12; 16:15; 29:3; 2 Chr 1:4; 15:11; 16:9; 20:22; 24:11; 28:9; 29:27; 30:19; 31:19.

Relative Words

83

The two relatives in (182) show both a subject position gap and a complement position resumption, and the relative in (183) shows a complement position gap. Thus, the syntactic and semantic features of zero-relatives parallel those of overtly marked relatives in Hebrew, with two qualifications. First, since zerorelatives have no overt signal of their status as subordinate NP modifiers, they are almost always immediately adjacent to the relative head; zero-relatives are only extremely rarely extraposed (see §5.2.5). Second, zero-relatives are exclusively restrictive in their semantics (see §6.3). 3.2.6.  Combination of Hebrew Relative Elements At the end of §3.1 above, I reintroduced two implicational universals, (110)–(111), which concern the patterns by which relative elements are combined with each other in a single relative clause. What has emerged from the cross-linguistic study of relatives is that relatives are rarely double-marked by overt relative elements, and when they are double-marked, it is almost exclusively with the combination of a nonpronominal element with a resumptive pronoun. The Hebrew data for ‫ז‬-relatives, ‫אשׁר‬- and ‫שׁ‬-relatives, and zero-relatives support these implicational universals: each type allows the use of an overt resumptive pronoun, as in (184)–(187), verifying that none of these relative elements should be classified as pronouns but as nonpronominal markers or complementizers. 45 (184) (= [129]) ‫ר־צּיֹון זֶ ה ָׁש ַכנְ ָּת ּבֹו‬ ִ ‫‏ה‬ ַ ‘Mount Zioni, whichi you dwelt in iti’ (Ps 74:2) (185) (= [137]) ‫וְ ִכ ֶּפר ַהּכ ֵֹהן ֲא ֶׁשר־יִ ְמ ַׁשח אֹתֹו‬ ‘and the priesti who(m)i one anoints himi . . . shall make atonement’    (Lev 16:32) (186) ‫ֹלהיו‬ ָ ‫ַא ְׁש ֵרי ָה ָעם ֶׁשיֲ הוָ ה ֱא‬ ‘happy is the peoplei whoi itsi god is Yhwh’ (Ps 144:15) (187) (= [182]) ‫ֹלהים לֹא יְ ָדעּום‬ ִ ‫ֹלה ֱא‬ ַ ‫יִ זְ ְּבחּו ַל ֵּׁש ִדים לֹא ֱא‬ ‘they sacrifice to demons ∅(that) are not divine, godsi ∅(that)i they have not known themi’ (Deut 32:17) The ‫מ‬-relative elements show some declination (for animacy) and, in contrast to the Hebrew markers and complementizers, do not allow resumptive pronouns but may co-occur with other elements, the complementizers ‫אשׁר‬ 45.  The relative ‫ ה‬is also a complementizer, but since it only relativizes from the subject position and never allows resumption (see example [140] above), it is only the fact that the ‫ה‬ does not decline as pronouns do that provides direct evidence of its classification.

84

Chapter 3

and ‫שׁ‬, as in (188) and (189). Both features suggest that ‫ מי‬and ‫ מה‬are the lone relative pronouns in Hebrew. (188) ‫יֹואב‬ ָ ‫ר־ל ָדוִ ד ַא ֲח ֵרי‬ ְ ‫ּומי ֲא ֶׁש‬ ִ ‫יֹואב‬ ָ ‫ִמי ֲא ֶׁשר ָח ֵפץ ְּב‬ ‘Who-that (= whoever) delights in Joab and who-that (= whoever) is for David, (follow) after Joab!’ (2 Sam 20:11) (189) (= [171]) ‫ה־ׁשּנַ ֲע ָשׂה הּוא ֶׁשּיֵ ָע ֶשׂה‬ ֶ ‫ּומ‬ ַ ‫ה־ּׁש ָהיָ ה הּוא ֶׁשּיִ ְהיֶ ה‬ ֶ ‫ַמ‬ ‘What-that (= whatever) has been—it is what will be, and what-that (= whatever) has happened—it is what will happen’ (Eccl 1:9) The portrait of the relative elements in Hebrew, given in this section as well as §3.1.1–5, allows us to refine a number of grammatical issues in Hebrew relativization, among which is the persistent but inaccurate classification of ‫זה‬‎, ‫אשׁר‬, or ‫ שׁ‬as a pronoun (see most recently Faist and Vita 2008; Israel 2011; and Pat-El 2012). 46 Moreover, this sketch provides analytical direction when someone is faced with unclear, if not strange, examples such as (190), in which we find the complex word ‫ה ֲא ֶשׁר‬. ַ (190) ‫ּוב ַק ְׁש ְּתָך ַא ָּתה ַמ ֶּכה‬ ְ ‫ית ְּב ַח ְר ְּבָך‬ ָ ‫‏ לֹא ַת ֶּכה ַה ֲא ֶׁשר ָׁש ִב‬ nrsv: ‘No! Did you capture with your sword and your bow those    whom you want to kill?’ (2 Kgs 6:22) niv: ‘Do not kill them. . . . Would you kill men you have captured    with your own sword or bow?’ (2 Kgs 6:22) If we consider ‫ ַה ֲא ֶשׁר‬without the Masoretic vocalization tradition, which is a later addition to the biblical text, the item ‫ האשׁר‬could be analyzed in two ways, which appear to be reflected in the two modern English translations I have provided in (190). It is not entirely clear how the translator(s) of the nrsv understood the ‫האשׁר‬. They have included the finite verb ‫ תכה‬in the question, which suggests that they either moved the -‫ ה‬on the front of ‫ האשׁר‬to the front of ‫ תכה‬by emendation, or they took the interrogative as unmarked and the item ‫ האשׁר‬as a complex relative element. Either way, it is a poor translation and reflects a confused analysis of the syntax. The niv translation, in contrast, takes the item ‫האשׁר‬ as the interrogative plus relative complementizer. This is also the analysis reflected by the vocalization of the Masoretic Text of B19a: ‫ ַה ֲא ֶשׁר‬would have been vocalized as ‫ ָה ֲא ֶשׁר‬if the Masoretes had taken it to be the article/relative item ‫ ה‬attached to the relative complementizer ‫אשׁר‬. That is, if the Masoretes had understood the ‫ ה‬to be an article, the vowel of relative ‫ ה‬would have been lowered to a qameṣ in front of the glottal stop ‫א‬. We can support the Masoretic interpretation and the niv translation with the Implicational Universal (see 46. See Schneider 1993: 256, §53.4.2.1 for the rare, accurate assessment that ‫“ אשׁר‬ist kein Relativpronomen.”

Relative Words

85

example [3h] on p. 4), repeated in (111), that one relative complementizer and another relative complementizer do not co-occur. 3.3.  The Etymologies of ‫ אשׁר‬and ‫שׁ‬ In the first half of this chapter, I described the Hebrew relative elements with regard to their typological characteristics, often previewing their syntactic and semantic features, which I discuss further in chaps. 4–6. In this final section, I will address the etymology of ‫ אשׁר‬and ‫שׁ‬. 47 The relative elements ‫ז‬‎, ‫ה‬‎, and ‫ מ‬are clearly related to the correspondingly homophonous ‫ ז‬demonstratives, ‫ ה‬article, and ‫ מ‬interrogatives. The ‫ ז‬elements and the ‫ מ‬elements have uncontroversial Semitic etymologies, while the origin of the ‫ ה‬is unknown. 48 But controversy has been stirred up with regard to the etymologies of the complementizers ‫ אשׁר‬and ‫שׁ‬. The etymologies of the Hebrew relative words were the object of considerable study in the nineteenth and first half of the twentieth centuries, as were the Akkadian ša and Phoenician ʾaš. 49 Bergsträsser’s 1909 study of Hebrew ‫שׁ‬‎ established a consensus for the origins of Hebrew ‫ שׁ‬and its relationship to ‫אשׁר‬. This consensus lasted for a century, with few dissenters (e.g., Brockelmann 1956: 145–46). The appearance in the last ten years of five articles addressing some aspect of the etymology of Hebrew ‫ אשׁר‬and ‫ שׁ‬suggests that, its overall cogency aside, Bergsträsser’s argument contained some nagging holes. 50 Specifically, questions remain about the apparent linguistic novelty of ‫אשׁר‬: why were Hebrew and a few Canaanite sister-dialects (Moabite and Edomite) innovative in using ‫ אשׁר‬as a relative word and not retaining any hint of nominal semantics? 51 Also, is it plausible that Hebrew has three relative words, ‫אשׁר‬‎, 47.  Much of the material for the following discussion was published as Holmstedt 2007; I have, however, updated the references to reflect subsequent studies and added discussion where necessary. 48. See Miller 2004: 301–4 for a brief but insightful discussion. I concur with the proposal that the article in NWS developed from the deictic *han “here!” (note the grammaticalization paths “here” > demonstrative and demonstrative > definite in Heine and Kuteva 2002: 109–10, 172, 330. Taking it as reasonable that the adverb ‘here’ can cover the deictic *han, I see the combination of these two paths as the likely origin of the Hebrew article and its cognates. Note also the path demonstrative > relative (Heine and Kuteva 2002: 113–15), which I suggest is in NWS a development before the grammaticalization of *han into an article; this accounts for the relative use of ‫ה‬, and especially the asymmetric examples (noun, ‫ה‬-adjective and ‫ה‬-noun, adjective), which I note above, in §3.2.3. On the origin of the Semitic article, see also Gzella 2006; Pat-El 2009. 49. See in particular Joüon 1913; Gaenssle 1915; Langdon 1915; Eitan 1928; Ravn 1941; Gevirtz 1957; Schuster 1965; Pennacchietti 1968; Aartun 1974, 1978; Dietrich and Loretz 1984; Garbini 1985; Niehr 1986; Deutscher 2001; 2002; Gai 2002; Johnson 2005; Israel 2003; 2011; Faist and Vita 2008. 50.  Israel 2003; 2011; Huehnergard 2006; Holmstedt 2007; Pat-El 2012. 51.  The oft-repeated suggestion (see, among many others, IBHS 332 n. 4) that Judg 5:27 might reflect the original nominal status of ‫ אשׁר‬is unwarranted; there is nothing in ‫באשׁר כרע‬

86

Chapter 3

‫שׁ‬, and the ‫ז‬-series (not to mention ‫ ה‬or the ‫מ‬-series), particularly when the typical derivation of ‫ שׁ‬and the ‫ ז‬elements is from the same Proto-Semitic determinative-relative *ḏū/ṯū? 52 It is to these questions, and so the origins and relationship of ‫ אשׁר‬and ‫שׁ‬, that I turn in the third section of this chapter. 3.3.1.  ‫ אשׁר‬and Its Semitic Relatives Beginning with the Hebrew data that provide the most linguistic information (in this case, the Tiberian Masoretic tradition provides the most phonological information, since the earlier nonbiblical texts lack vocalization), it is important to note that the word ‫ אשׁר‬never appears in free form in the Hebrew Bible. Rather, the medieval vocalization of the word—‫ ֲא ֶשׁר‬/ʾăšer/— suggests that it was a phonological clitic. 53 Based on comparison with other forms, this vocalization suggests that, if Hebrew ‫ אשׁר‬had once had a free form, it would have been either *ʾašir or *ʾašar, with the former much more common in Hebrew, while the latter is favored by the comparative Semitic evidence. The CaCaC underlying word pattern is typical as the base for nouns, although no such noun exists with the discontinuous root [ʾ-š-r]. Two Hebrew nouns of this root based on different nominal patterns are attested, however—*ʾāšur (CaCuC) and *ʾaššur (CaCCuC), 54 both meaning ‘step, footstep’—as are verbs with the meaning ‘to stride’. The connection between the function word ‫אשׁ‬ ‫ ר‬and the nominal and verbal derivations from the root [ʾ-š-r] might be coincidental (many examples of such differing, root-based homophony exist in Hebrew), but it is also quite plausible, given the likely etymology of ‫אשׁר‬. The Semitic cognates of a noun *‫ אשׁר‬are well attested. 55 In East Semitic, Akkadian provides us with a noun ašru(m) (clitic form: ašar) ‘place, site, ‫ שׁם נפל שׁדוד‬that would necessitate a nominal analysis for ‫אשׁר‬. Rather, this phrase presents a straightforward example of a null-headed relative, ‘in (the place) that he went down, there he fell, destroyed’. 52. See, in general, Pennacchietti 1968; also Moscati et al. 1980: 113–14; Lipiński 2001: 332–35. For the Semitic cognates of the relative *ḏū/ṯū, see Ugaritic d (Tropper 2000: 234–38); Byblian Phoenician z (Friedrich and Röllig 1999: 209; Krahmalkov 2001: 93–94); Edomite zy (Aḥituv 1999); Old Aramaic z and zy, Imperial Aramaic zy and dy, and Biblical Aramaic dî (Segert 1984: 177; Hug 1993: 135–36); Classical Arabic ʾallaḏī and ḏū (Wright 1898: 1.270–73; cf. Fischer 2002: 130–31); and Geʿez za (Dillmann 1907: 119; Lambdin 1978: 106; Tropper 2002: 47). 53.  By all six criteria listed in Zwicky and Pullum (1983: 502–13) for distinguishing clitics from affixes, the Hebrew word ‫ אשׁר‬is clearly a clitic, phonologically and prosodically, as are many other Hebrew function words, such as the article ‫ה‬, the basic conjunction ‫‘ ו‬and’, prepositions such as ‫‘ ב‬in’, ‫‘ כ‬as’, ‫‘ ל‬to’, and the nominalizer ‫שׁ‬. See Holmstedt and Dresher 2013. 54.  Neither noun is attested in free form; hence the *-marking. 55.  For general discussions, see GKC 444 n. 1; Gaenssle 1915: 25–29; Bauer and Leander 1922: 264; Cohen 1976: 37; Moscati et al. 1980: 113; Murtonen 1986: 103; IBHS 332 n. 2; GBH 109 n. 2; Lipiński 2001: 324–26, 522; HALOT 98, s.v. ‫;א ֶשׁר‬ ֲ Israel 2003; Huehnergard 2006.

Relative Words

87

region’. 56 In West Semitic, there are a number of languages that contain cognates: Ugaritic ʾaṯr used as a noun ‘place’ 57 and perhaps as a verb ‘to march’; 58 Old Aramaic ʾšr, Imperial Aramaic ʾtr, Biblical Aramaic and Syriac ʾătar—all as nouns denoting some variation of ‘place’; 59 Deir ʿAllā ʾšr ‘place’; 60 Arabic ʾiṯr, ʾaṯar ‘mark, footprint, track’, ʾṯr ‘to begin, choose; to follow’; 61 and Geʿez ʾašar ‘trace, footprint, path, track’ and ‘to follow, look for tracks’. 62 Based on the strength of the cognate attestation, it is now taken as fact that Hebrew ‫ אשׁר‬has no etymological relationship with the Proto-Semitic *ḏū/ṯū. The consensus is that ‫ אשׁר‬derives from a common Semitic verbal root [*ʾ-ṯ-r] ‘to stride, march’ and a noun *ʾaṯar with a semantic range of ‘step, trace, footprint’. In light of the wide geographic and temporal distribution of this Semitic word, what distinguishes the Hebrew reflex of Proto-Semitic *ʾaṯar is that none of the Hebrew attestations are nominal; i.e., Hebrew ‫ אשׁר‬is always used as a nominalizing function word in the extant data, primarily to introduce relative clauses. Thus, Hebrew is sometimes considered “innovative” in its use of *ʾaṯar, 63 and we must evaluate whether this is substantiated by the data. Consider first that, in addition to using ašru(m)/ašar as a noun ‘place’, Akkadian also at some point began to use the bound form ašar to introduce relative clauses, at first primarily with locative semantics, as in (191), and then with nonlocative semantics, as in (192). 64 56.  CAD A/2 456–60; von Soden 1995: 251, 282; 1965: 1.82–83; Black, George, and Postgate 2000: 29. Faist and Vita 2008, for Emar; Israel 2011, for Mari. 57.  Gordon 1965: 369; Segert 1984: 180; Sivan 1997: 84, 198; Tropper 2000: 798, 905. Bordreuil and Pardee 2009: 301; del Olmo Lete and Sanmartin 2003: 1.127. 58.  Gordon 1965: 369; Sivan 1997: 84; Tropper 2000: 547. 59.  Payne Smith 1903, s.v.; BDB, s.v.; DISO 27–28; Segert 1975: 528; DNSWI 125–29. 60.  See line 11 in Hoftijzer and van der Kooij (1976: 174) and comment on p. 285. Note that Hackett (1984: 25, 100) lists the word in question on line 9. Although the translations and comments in these works are ambiguous at best, the likeliest analysis of the item bʾšr is as a prepositional phrase with nominal ʾšr ‘place’ (cognate with Aramaic ʾtr) followed by a zero-relative, i.e., ‘On the site (where) the stick would lead the ewes, hares are eating grass’ (Lipiński 1994: 133; cf. Lemaire 1985: 26–39; Weippert 1991: 151–84; Djikstra 1995: 43–64; DNSWI 126). 61.  Lane 1863: 1.18–19. 62.  Dillmann 1867: cols. 739–40; Leslau 1987: 45. 63.  See, for example, Huehnergard 2006: 124. 64. While both of the examples cited above are null-head relative clauses (so also Huehnergard 2006: 107 n.  31), they illustrate the locative and nonlocative semantics of ašar nonetheless. The entry under ašar in CAD (A/2 413–16) as well as the entries under ašru(m) in von Soden (1965: 1.82–83) and Black, George, and Postgate (2000: 29) suggest that ašru(m) had added the relative function in the earliest attested stages, i.e., Old Akkadian. Additionally, the citations in the CAD entries for ašar and ašru(m) illustrate the very fine line between the locative noun ‘place’ followed by a zero-relative—for example, ‘the place (that/where)’—and its use as a locative relative, ‘where’. Despite standard assertions to the contrary (which go back at least as far as Kraetzschmar 1890: 296–302), a number of the

88

Chapter 3 (191) Old Assyrian ašar-Relative (Locative): BIN 4 5:26 (CAD A/2 413) ašar ṭuppū ibaššiūni tertaka lillikamma rel.loc tablets be(3mp dur) instruction-your go(2mp prec) ‘let your instruction go (as to) where [= the place that] the tablets are available’ (192) Old Assyrian ašar-Relative (Nonlocative): CCT 3 30:25 (CAD A/2   413) ašar damquni lu nīpuš rel good mod do(1cp pret) ‘let us verily do what [= the thing that] is good’

In light of the numerous glosses provided for ašar by CAD (‘as soon as’, ‘while’, ‘if’, ‘in case’, ‘with’, ‘before’, ‘in the presence of’, ‘from’, and ‘instead of’), 65 as well as the fact that some contexts unambiguously demonstrate that ašar is used following nonlocative heads, one wonders if many of the other locative-relative examples listed are a bit forced, thrust into the Procrustean bed defined by the nominal etymology of ašar. This seems particularly clear when we add the evidence of the relative use of ašar in the Akkadian of Emar and Mari (see Faist and Vita 2008 and Israel 2011, respectively). It is also relevant that Aramaic (193) and Ugaritic (194) attest the nonnominal use of *ʾaṯar, as the preposition ‘after, behind’. 66 (193) Biblical Aramaic *ʾaṯar PP: Dan 7:7 bāʾtar dənâ ḥāzê hăwêt bəḥezwê lêlyāʾ waʾărû in-place.of this see(ms ptcp) be(1cs perf) in-visions.of night   and-behold ḥêwâ rəbî ʿāyâ creature fourth ‘after this I saw in night visions—and behold—a fourth creature’ (194) Ugaritic *ʾaṯar PP: KTU2 1.5.6.23–25 bʿl mt my lim bn dgn my hmlt Baal.nom dead(3ms perf) wh people.nom son.nom Dagan.gen wh    multitudes.nom a ṯr bʿl ard barṣ after Baal.gen descend(1cs impf) in-earth entries could easily be analyzed as (null-head) relative clauses, without any locative semantics directly associated with the word ašru(m). 65. CAD A/2 413–16, s.v. ašar; A/2 456–60, s.v. ašru(m). 66.  DNWSI 127; Gordon 1965: §19.424 reads at least one case of Ugaritic aṯr as a relative. This has since been challenged by most scholars; see the discussions in Rainey 1971: 160–62; Aartun 1974: 29; and 1978: 81; Pardee 1981: 156; and Dietrich and Loretz 1984.

Relative Words

89

‘Baal is dead, what of the people? The Son of Dagan (is dead), what of the multitudes? After Baal I will descend into the earth’. Both Aramaic and Ugaritic reflect the use of this noun, ‘place’, as a preposition meaning ‘after’. 67 Together with the Akkadian data, this suggests that already in the second millennium the use of *ʾaṯar was a multifaceted function word. The semantic shifts witnessed in Akkadian (of which Hebrew ‫אשׁר‬ may be a result), Ugaritic, and Aramaic all appear to be exemplary cases of “grammaticalization.” 68 Specifically, the Akkadian change from a noun ‘place’ → locative relative → relative 69 is easy to conceptualize and is possibly paralleled by Greek pou, German wo, Persian kujā, and Chinese só. 70 Unfortunately, 67.  If there is a diachronic element to the relationship of *ʾaṯar ‘place’ and ‘after’ in Aramaic, the earliest Aramaic occurrences of the preposition b- ‘in, with’ followed by *ʾaṯar might provide the link relating to the reanalysis and semantic bleaching, e.g., mlk zy [ysq wymlk] bʾšrh ‘a king who will come up and rule in his place [→ after him]’ (Donner and Röllig 1966: 222 A2.2–3); cf. DNWSI 127. Similarly, with reference to the Ugaritic use of ʾaṯr Pardee states, “If these readings are correct, it becomes clear that ʾaṯr is not functioning as a relative pronoun, though the syntactic function of the word here is the very one that led to its becoming a relative pronoun (accusative of respect of a noun meaning ‘place’ = ‘in whatever place’ → ‘wherever’ → ‘which’)” (Pardee 1981: 156). 68.  For definition and discussion of grammaticalization, see n. 86 below. 69. See Heine and Kuteva 2002. Givón (1991: 257–310) presents cross-linguistic data from Lhasa Tibetan, Hewa (a Papua-New Guinean language), and Krio (an English-based Creole) to demonstrate that the development from a noun ‘step’ or ‘place’ to a relative word is attested elsewhere. Unfortunately, none of the data is as conclusive as he implies. Krio apparently uses a wh-word relativization strategy (which does not parallel Hebrew ‫ אשׁר‬at all). The Hewa data he supplies indicate that this language forms relatives by using a copy of the head noun; this implies that any item may function as the “relative word,” and therefore the locative example he cites is in no way analogous to Hebrew ‫אשׁר‬. Finally, the Lhasa Tibetan evidence, which suggests that this language has employed items that are also noun-class determinatives as relative words, is inconclusive. 70.  Kraetzschmar 1890: 298; Joüon 1913: 128–33; Israel 2003: 342–43. Note that unlike the other items, Persian kujā is a complex item, formed from the combination of the relative kū and the nominal gāh ‘place’ (see Nyberg 1974: 119; MacKenzie 1971: 52). In a recent study, Park (2015) argues that Hebrew ‫ אשׁר‬parallels Chinese zhe as a “light noun”—that is, a class of nouns that “tend to lose their referentiality and assume a more grammatical meaning” (Simone and Masini 2014: 63). Light nouns are obviously examples of grammaticalization, but Chinese zhe is hardly a suitable comparison for ‫אשׁר‬. First, Park herself points out that zhe can function as a noun in that it can be modified by quantifiers, demonstratives, and numerals. Yet this never happens for ‫אשׁר‬. Moreover, Simone and Masini’s typology of the “light noun” class indicates that this is a fairly complex noun class with testable features— tests that cannot be applied to ‫אשׁר‬, since it never behaves like a noun in Hebrew but had become a nominalizer before the stage witnessed in the first-millennium corpora (the Bible and the epigraphic texts). Finally, not only is the typological syntactic profile of Chinese (or Japanese or any of the other “non-Western” languages to which Park refers) extremely different from Hebrew (or any Semitic language), the specific syntax of zhe clauses is significantly different from ‫ אשׁר‬clauses (see Aldridge 2008 on zhe clauses). Park does not address any of

90

Chapter 3

the data do not allow us to reconstruct a sure diachronic path of change between the lexical and functional uses of *ʾaṯar in Akkadian, or in Ugaritic and Aramaic, and since the discernible phonetic changes involved (i.e., the use of the bound form for Akkadian ašar and the quiescence of the /ʾ/ in the Aramaic form) reflect regular sound changes in these languages, we cannot with confidence invoke the process of grammaticalization 71 as it is now commonly understood, either within the specific languages in question or in Semitic as a whole. If we should be tentative in describing what happens with *ʾaṯar in languages that exhibit both nominal and functional uses, how much more careful we should be for Hebrew, in which *ʾaṯar serves only as a function word. We simply lack the necessary data to complete the reconstruction, and if we were to have adequate second-millennium data from Akkadian, Aramaic, and Ugaritic to call the story of Semitic *ʾaṯar a case of grammaticalization, 72 we would still need to account for the Hebrew situation, in which ‫ אשׁר‬witnesses only the nominalizer use of *ʾaṯar. Thus, grammaticalization is an inappropriate description for the etymology of ‫ אשׁר‬regardless of the processes that occurred before Hebrew. This does not preclude, however, the grammaticalization of ‫ אשׁר‬within Hebrew. I shall return to this question below, the answer to which carries implications for the etymology of Hebrew ‫שׁ‬. First, let us consider the possible cognate evidence for ‫שׁ‬, assuming for the moment the current consensus that it has no relation to ‫אשׁר‬. these issues in her study but compares Hebrew with Chinese in her “broader typology” without the necessary critical reflection. Her other novel proposal, that ‫ אשׁר‬introduces “standalone nominalizations” suffers the same basic issue—Hebrew is assumed to by typologically similar enough to Chinese for the comparison to be legitimate. Note that my analysis of some of the data Park cites has changed in this work; also see appendix B for examples. 71.  Following the critique of grammaticalization in Joseph 2004: 45–71. 72.  The Canaanite of the el-Amarna tablets reflects a terminus a quo for the use of *ʾaṯar as a relative word. Rainey asserts that the function of ʾašar is “not a true relative” and that, however the Hebrew relative developed, “it was hardly under the influence of EA Akkadian!” (Rainey 1996: 3.70–71); this unsubstantiated conviction aside, the syntax of many of the examples mirrors relative-clause structure, and thus they should be analyzed as relatives. It is tempting to suggest that el-Amarna Canaanite is the missing “grammaticalization” link, showing us a transition point squarely between the early second-millennium situation represented by Old Babylonian and Old Assyrian, in which ašru(m)/ašar is used both as a noun and a relative word, and later Hebrew usage, in which ‫ אשׁר‬has lost its nominal lexical entry and retained only its relative function. But once again, without any demonstrable nominal use in Hebrew, we are left without a sure diachronic path; in other words, given the gap between the dual use of ‫ אשׁר‬in pre-Hebrew Semitic and the singular use of ‫ אשׁר‬in Hebrew, it is possible that Hebrew and Canaanite actually inherited the nominalizing use directly from Akkadian rather than as the end-product of a grammaticalization process that began in East Semitic and wound its way through Northwest Semitic.

Relative Words

91

3.3.2.  ‫ שׁ‬and Its Semitic Relatives Bergsträsser’s 1909 argument convinced the majority of the field, until recently, that the etymology of ‫ שׁ‬placed it in historical line with the Akkadian relative ša. Stepping back a bit farther, we note that in Old Akkadian relative clauses were introduced by a š-based relative marker that was declined for case, number, and gender (Huehnergard 2006: 115–16; see below, §8.2). As Deutscher (2001; 2002) noted, since the case that this relative element carried corresponded to the role of the relative head in the matrix clause and not the subordinate relative clause, it was not a pronoun (but, in the terminology of this work, a marker). Deutscher also observed that the use of relative markers is rare across languages. This is probably due to the redundancy of the case-marking in the matrix clause as well as the lack of information provided about the head’s syntactic role in the relative clause; thus, in terms of syntax and language processing, this form of relative element was “inefficient and dysfunctional” (Deutscher 2001: 408). Deutscher then noted that, while the case system in Akkadian remained functional until 1000 b.c.e., a functioning system of declined relative markers disappeared by 2100 b.c.e., leaving only the fossilized ša (the former ms acc form) as an indeclinable relative complementizer. Thus, East Semitic had an earlier system that did not survive into the second millennium. The full system of declined relative markers to parallel the East Semitic system is also typically reconstructed for West Semitic, though it is only completely attested fully, with case-marking, in Classical Arabic (see below, §8.8). In the remaining West Semitic languages, the system of relative markers has either simplified to gender and number agreement (e.g., Ugaritic, Epigraphic South Arabian), number agreement only (e.g., Geʿez), or a single frozen form (e.g., Aramaic), as happened in East Semitic. Both East and West relative-marker systems are reconstructed to be developments from a single Proto-Semitic paradigm of “determinative-relative” markers *ḏū/ṯū. 73 The East Semitic š- version and the West Semitic z-/ḏ- versions both represent simple phonological shifts within a tight range of dental to post-alveolar fricatives. The East Semitic š- is the –voiced post-alveolar fricative while the West Semitic ḏ- is the –voiced dental fricative, and the z- is the +voiced alveolar fricative. While West Semitic appears to have inherited a set of subordination markers based on *ḏ-, within Northwest Semitic, there is clear evidence that another relative element, based on š-, was adopted. An š- relative has been identified in the Late Bronze Age alphabetic cuneiform text from Tanaach, 74 possibly 73. See Faber 1984: 189–224; Fronzaroli 1987: 267–74; Huehnergard 2006. 74. KTU2 1995: 4.767. See Cross (1968: 44–45) for the view that the text contains a relative: kkbʾ lpʿm / kpr š yḥtk l /dw ‘Kôkaba to Puʿm / The fee fixed [lit., “which is/was

92

Chapter 3

in the Late Bronze Age Lachish ewer, 75 in an Ammonite amulet/seal ca. 600 and in a Philistine text. 77 Additionally, I take the š in Punic and a few late colonial Phoenician inscriptions 78 as well as the relative ʾš 79 in Phoenician (Standard though late Neo-Punic) 80 and Ammonite 81 to be cognate. 82 b.c.e., 76

set or fixed”; Cross’s n. 24] (has been) remitted to / him’. Cf. Hillers 1964: 45–50; Tropper 2000: 76–77; Huehnergard 2006: 106 n. 19. 75.  Krahmalkov (2001: 94) reads the text as mtn šyt[n . . . ]ty lʾlt ‘[This is] the gift that . . . tay presented to Elath’. This differs from Cross’s reading, however: mtn.ṯśy⸢l⸣[rb]ty ʾlt ‘Mattan. A tribute to my Lady ʾElat’ (Cross 1954: 20–21). 76.  Jackson 1983: 77–80; Aufrecht 1989: 145–48, no. 56. 77.  Cross 1996: 64–65. 78.  Friedrich and Röllig 1999: 72–73; contra Krahmalkov 2001: 93–95; Huehnergard 2006: 105–6 and nn. 24, 109. 79.  To connect ʾš with š, we must suppose that the initial /ʾ/ is prothetic. Garbini (1985: 185 n. 1; cf. Israel 2003: 335–40) suggests that the Phoenician ʾš derived from the noun ʾš ‘man’; I have looked through the grammaticalization literature for any cross-linguistic support for the development of ‘man’ → ‘who /what’ (i.e., a relative word) and found no success; it is typologically unattested and therefore highly unlikely (especially when other paths of development are available). Though the Latino-Punic evidence, which represents the cognate relative word as ys, may appear to support the proposed grammaticalization of the Phoenician relative ʾš < ʾš ‘man’, such an argument would reflect a misunderstanding of the vowel represented by the grapheme {y} in Latino-Punic. As Kerr convincingly demonstrates, the {y} in Latino-Punic and Greco-Punic (from El-Hofra) and the first Punic rendition in the Poenulus represents vowel reduction of any unaccented vowel or accented vowel in a closed syllable to šĕwa /ə/ (Kerr 2010: 49–60, 102–5). Kerr also notes the use of {y} as a Hiatustilger, which would represent in Latino-Punic the same phonological phenomenon as the prothetic ʾalep on the earlier ʾš (2010: 59). Gevirtz argues that the addition of initial ʾalep occurred in Phoenician only when needed to aid the pronunciation of an initial two-consonant cluster (Gevirtz 1957: 125; cf. Garr 1985: 47), but his argument is weakened by the counterevidence mentioned in his n. 12; moreover, the use of prothetic ʾalep to allow a clitic to stand as a free form both is logical and finds support in the use of an epenthetic ʾalep four times with the Qumran relative ‫שׁ‬, resulting in the form ‫( שׁא‬see above, n. 19) as well as the Latino-Punic evidence cited above. Finally, Gevirtz’s proposal that relative ʾš derived from the copular ʾiṯ (Ugaritic), ‫אשׁ‬/‫שׁ‬ ֵ ֵ‫( י‬Hebrew), and ‫( איתי‬Aramaic), fails by the same typological lack of support as Garbini’s derivation from ‘man’ does. 80.  DISO 285–86; DNSWI 1089–94; Krahmalkov 2000: 77–80; and 2001: 93–95. 81.  Jackson 1983: 51–52; Aufrecht 1989: 214–19, no. 80. One occurrence of ʾš also exists in the Khirbet el-Mudeiyineh inscription discussed in Rainey 2002; Rainey argues that the text is actually written in “Israelite/Phoenician” (p. 82), but the context as well as the language suggest that the language is either Moabite or Ammonite. Also it is possible that the Deir ʿAllā ‘Balaam’ text, whatever its linguistic identity, contains an example of relative ʾš (Hackett 1984: 31, 101; Huehnergard 2006: 3), but for a rebuttal of the relative reading for the ʾš in this text, see Garr 1985: 85. 82.  Though I found no cogent argument against an etymological connection between š and ʾš in Northwest Semitic languages, there has often been a strange reluctance to support it; see Harris 1936: 55; Gevirtz 1957; Schuster 1965; Friedrich and Röllig 1999: 73; Krahmalkov 2001: 94. Krahmalkov is a good example of the linguistically muddled thinking about these function words that is rife in Comparative Semitics: he asserts that the occurrences of the Phoenician item š- are not as a relative word but a as a “determinative pronoun,

Relative Words

93

The obvious question is whether the š-based relative elements in Northwest Semitic are related to Akkadian ša. Numerous Hebraists and Semitists in the late nineteenth century attempted to avoid the Akkadian origin for Hebrew (and Northwest Semitic) ‫ שׁ‬by deriving it from ‫אשׁר‬, or vice versa—an explanation that Huehnergard has reinvigorated (2006; followed by Pat-El 2010; 2012). Bergsträsser and the majority of scholars in the twentieth century took the Akkadian etymology as the simplest explanation. So, assuming the data lend themselves to an informed conclusion, which is it—Akkadian ša or Hebrew ‫?אשׁר‬ 3.3.3. Are ‫ אשׁר‬and ‫ שׁ‬Related? The nineteenth century produced a number of creative proposals for the etymologies of ‫ אשׁר‬and ‫שׁ‬, and the excellent summaries and critiques in Gaenssle 1915; Israel 2003; and Huehnergard 2006 allow me here to summarize even more briefly. The proposals fall into three basic camps: (1) those that derive one item from the other; (2) those that derive both items from a shared proto-form; and (3) those that conjecture no etymological connection between the two. Most nineteenth-century scholarship falls into the first camp, which assumes that these two function words are related and so focuses on discerning which one is earlier and the nature of the derivational process. There are two directions for deriving -‫ שׁ‬and ‫ אשׁר‬in this camp, and both had adherents. Many followed the view of Gesenius that ‫ אשׁר‬is the earlier form and ‫ שׁ‬a later reduced form; the processes involved were identified as the aphaeresis of the initial /ʾ/ and the assimilation of the final /r/, thus accounting for the gemination of the first consonant when ‫ שׁ‬is prefixed (Gesenius 1817: 224; 1835: 3.1345–46; Olshausen 1861: 439). Alternatively, the opposite derivation was proposed, perhaps first by Sperling, that ‫ שׁ‬was the earlier form and that ‫אשׁר‬ was a lengthened form by means of a prothetic /ʾ/ and the dissimilation of the final gemination first into /l/, which later “hardened” to /r/ (Sperling 1876: 15–22; Brown 1885: 249–50; 1886: 117–18; see Eitan 1928 for a variation on this proposal). The second camp derives both ‫ אשׁר‬and ‫ שׁ‬from a common reconstructed or hypothetical source, either *ʾšl or *šl. The proto-form *ʾšl was promoted by Ewald; he suggests that this form began as the combination of three demonstrative particles, *ʾ, *š, and *l, by analogy with Arabic allaḏī (Ewald 1844: 384). The developmental path from *ʾšl includes similar sound changes as the previously noted proposals: *ʾšl becomes ‫ אשׁר‬by the “hardening” of the /l/ to /r/, and *ʾšl becomes ‫ שׁ‬by the assimilation of the /l/ and aphaeresis of the /ʾ/. Böttcher suggests what he must have considered a simpler hypothesis than Ewald’s: he begins with the proto-form *šl, on analogy with the reconstruction serving primarily to express an indirect genitive relationship” (2001: 94). See below, §8.10; and Holmstedt 2014a for my assessment of the misguided “genitive” analysis.

94

Chapter 3

of the Hebrew article haC- as *hal (Böttcher 1866–68: 2.78–82). The addition of a prothetic /ʾ/ and the “hardening” of the /l/ produced ‫אשׁר‬, and the assimilation of the /l/ to the initial consonant of the host word produced the clitic ‫שׁ‬. The problems with both of these first two approaches were well known even in the nineteenth century (see Hommel 1878: 708–15; Kraetzschmar 1890). Both Gesenius’s and Sperling’s proposals begin and end with the existing lexemes, although they also both resort to sound changes that are either not attested at all in Hebrew or West Semitic (the assimilation of /r/) or not well attested in Semitic until the late first millennium b.c.e. (the prothesis of /ʾ/ and especially the aphaeresis of /ʾ/). 83 The objections to Ewald’s solution are similar: the “hardening” of /l/ to /r/ as well as the assimilation of /l/ are both ad hoc, unattested sound changes, and the aphaeresis of /ʾ/ arguably does not occur in Semitic until late in the first millennium b.c.e. Böttcher’s proposal combines all of the difficulties of the previous proposals and thus suffers all of the already-mentioned weaknesses of unattested and ad hoc sound changes. Thus, the third camp regarding ‫ אשׁר‬and ‫ שׁ‬that emerged toward the end of the nineteenth century built on the perceived weaknesses of the previous two basic camps as well as the presence of a potential Akkadian cognate, ša, and became the consensus in the twentieth century. This story is one of distinct origins for the two Hebrew nominalizers ‫ אשׁר‬and ‫שׁ‬. In accordance with the cognate evidence provided above in §3.3.1, ‫ אשׁר‬is connected to the Akkadian noun-cum-subordinator ašru(m)/ašar, and again in accordance with the apparent cognates listed above in §3.3.2, ‫ שׁ‬is connected to the Akkadian relative ša (Hommel 1878; Stade 1879: 133; Delitzsch 1886: 44; Kraetzschmar 1890; Gaenssle 1915). 84 This is the position advocated by Bergsträsser, whose argument seems to have effected the closure of the vigorous, century-old debate, at least until the appearance of Huehnergard’s recent challenge. 85 83.  On the evidence related to the prothesis of /ʾ/ and the aphaeresis of /ʾ/, see Garr 1985: 47–48, 50–52, respectively. For further discussion of the aphaeresis of /ʾ/, see n. 79 above. 84.  Gaenssle promotes this view independently of Bergsträsser, of which he does not appear to have been aware. Note that Gaenssle’s dissertation was published in two versions: a University of Chicago monograph in 1915 and a two-part submission in AJSL 31 [1914] 3–66; and 31 [1915] 93–159. For discussion of the Akkadian relative clause and specifically the relative word ša, see Langdon 1915; Ravn 1941; von Soden 1995: 237–39; Deutscher 2001; 2002; Gai 2002; Huehnergard 2005a: 185–88; Johnson 2005. 85. While Kautzsch (GKC 444 n. 1) considered the etymology “still a matter of dispute” in 1910, undoubtedly writing before Bergsträsser’s article had appeared, by the time that Joüon discussed the etymology in 1913: 128–39, only the theory of the distinct etymologies of the two relative words was presented; the other theories were ignored. See also the unambiguous presentation in Bauer and Leander’s historical grammar (1922: 264). The “separate etymology” view has rarely been questioned since 1909, with the notable exceptions of Eitan 1928: 178–84; Brockelmann 1956: 145–46; and Huehnergard 2006; 2013a.

Relative Words

95

In his 2006 study, Huehnergard presented a novel approach to an old solution: using grammaticalization theory, 86 he argued for the derivation of ‫ שׁ‬as a reduced form of ‫אשׁר‬. His main points are as follows: (1) that Hebrew ‫ אשׁר‬is derived from the common Semitic substantive ‘place’; (2) that it has undergone “grammaticalization” into a relative word; and (3) that ‫ שׁ‬is simply the reduced, clitic form of ‫אשׁר‬, formed by the aphaeresis of the initial /ʾ/-syllable and assimilation of the final /r/ to the initial segment of the clitic’s host. Huehnergard suggested that the reanalysis and phonetic reduction of Hebrew ‫ אשׁר‬must have taken place by the twelfth century b.c.e., in order to account for the presence of Hebrew ‫ שׁ‬in Judg 5:7 as well as the early first-millennium appearance of Phoenician ʾš and Ammonite š. Thus, the process had to have occurred in the Canaanite ancestor that produced Hebrew, Phoenician, Moabite, Ammonite, and Edomite. 87 Huehnergard frames his account of ‫’שׁ‬s origins in terms of three core processes of grammaticalization theory: (1) categorial reanalysis (‫ אשׁר‬as a substantive to ‫ אשׁר‬as a function word), (2) semantic change (the lexical meaning ‘step’ or ‘place’ was fully bleached so that the item became a function word), and (3) phonetic reduction (from a free form *ʾašar to a clitic with the reduced shape of ‫)שׁ‬. In his opinion, taken together, these explain ‫ אשׁר‬and ‫שׁ‬ so well that he calls the history of ‫“ אשׁר‬a parade example of the process 86.  Grammaticalization, at its simplest, is a label for discussing a subset of diachronic changes that occur in language. In particular, it is the convergence of reanalysis (i.e., the categorial reassignment of a lexeme), semantic change (typically from more concrete to more abstract meaning), and phonetic reduction. For introductions to grammaticalization as a theory as well as the history of the concept, see Hopper and Traugott 2003. For discussions of whether grammaticalization is legitimately a distinct phenomenon or instead an epiphenomenon, see Newmeyer 1998; and the responses in Heine and Kuteva 2002: 2–5; and Hopper and Traugott 2003: 132ff. For a critique of grammaticalization from the perspective of traditional historical linguistics, see Joseph 2001; 2004. For instance, particularly relevant to the current study is Joseph’s claim that grammaticalization theory, it seems to me, perhaps inadvertently, often takes stances that are quite at odds with constructs and notions about language and language change that have long been held and upheld within traditional historical linguistic frameworks; for those schooled traditionally, therefore, grammaticalization comes across as just flat out wrong. For instance, just to give a taste of what is to come, certain ways in which phonetic reduction is invoked in discussions of grammaticalization fly in the face of what is known about the regularity of sound change and the sorts of conditioning that can hold on sound changes. (Joseph 2004: 47)

87. The recent argument that Hebrew should be distinguished as a Transjordanian dialect, along with Moabite and Aramaic, over against the Canaanite languages, such as Phoenician, on the basis of one isogloss (the lexeme used as the copular verb) is wholly unconvincing (see Rainey and Notley 2006: 112; Rainey 2007a, b). See Hackett and Pat-El 2010 for a rejoinder presenting the current consensus, that Hebrew should remain grouped with Phoenician and Moabite as a Canaanite dialect.

96

Chapter 3

of grammaticalization.” Furthermore, he cites cross-linguistic data as support for the idea that the grammaticalization of a word is not a sudden and final event, but that “earlier forms may coexist with later ones,” in our case, the coexistence of the long form ‫ אשׁר‬and its reduced form ‫שׁ‬‎ (Huehnergard 2006: 120–21). The occasional addition of another ‫ אשׁר‬or ‫ שׁ‬on a newly found ostracon notwithstanding, it is unlikely that the etymological discussion about ‫ אשׁר‬and ‫ שׁ‬will ever hinge on new data. Rather, the issues are methodological: is it linguistically reasonable that the two items were related? If reasonable, then is it likely? Or is it reasonable that ‫ אשׁר‬and ‫ שׁ‬had separate origins? If so, what were they? My response in Holmstedt 2007 had the goal of evaluating the data and the implications of the proposals, especially that of Huehnergard 2006, rather than strongly advocating my own synthesis. Thus, while I concluded that separate origins were more likely, my concern was that any reconstruction follows the accepted principles of grammatical analysis and reconstruction and does not ignore the history of Hebrew and its general historical context in favor of simplifying the overall Semitic reconstruction. 88 I will here present my own synthesis more fully and so will only briefly summarize my criticism of Huehnergard’s creative but ultimately misguided return to the nineteenthcentury position. My historical linguistic arguments against Huehnergard’s proposal were (and remain) twofold. First, it is not clear that Hebrew was innovative in using a form of ʾšr as a subordinating function word, the repeated assertions of innovation notwithstanding. There is evidence that ʾšr was used as a subordinating function word in Akkadian by the first half of the second millennium (Emar and Mari; see n. 56). Second, the consonant /r/ never assimilates in Hebrew. My comparative linguistic arguments against Huehnergard are also twofold. First, creating symmetry in the larger Semitic picture by proposing that ‫אשׁר‬ was shortened to ‫ שׁ‬by loss of the initial /ʾ/ (an attested sound change) and assimilation of the final /r/ to the next word (an unattested change) is simply special pleading. Second, telling this story for Hebrew fails to account for the use 88.  The methodological nuance of my argument appears not to be understood by Pat-El, whose 2012 summary of my 2007 argument is inaccurate. When I suggested that the absence of a full diachronic spectrum of data prohibits us from reconstructing a path of grammaticalization for ‫אשׁר‬‎ (Holmstedt 2007: 181), I meant that, if the word ‫ אשׁר‬has subordinating functions as far back as Akkadian (cf. its dialectal forms in Emar and Mari), it is not logical to argue that the word underwent grammaticalization in Hebrew. Indeed, it may never have experienced grammaticalization in Semitic, but as far back as we can reconstruct, the word always had a twofold lexical entry, the nominal half of which fell into early disuse in Hebrew. In no way did I reject “the etymology of ʾăšer . . . because not all languages show the change ʾaṯar > ʾăšer and ʾăšer > šeC” (Pat-El 2012: 320).

Relative Words

97

of both a z-relative and an ʾš/š-relative in Phoenician, where the ʾš/š can hardly be explained away as a shortening of ʾšr, which never occurs in Phoenician. Pat-El (2012) has added her voice to the discussion in support of Hueh­ nergard’s position by examining some syntactic features of clauses introduced with ‫ז‬‎, ‫אשׁר‬‎, and ‫שׁ‬. She argues that ‫ שׁ‬differs in its basic syntax from ‫ אשׁר‬and ‫זה‬, and (counterintuitively) this difference suggests that ‫ שׁ‬is a reduced form of ‫אשׁר‬. In particular, she argues that, while all three introduce relatives with verbal clauses, with null-copula clauses and prepositional complements, only ‫ זה‬and ‫ אשׁר‬introduce clauses consisting only of a “noun attribute,” for which Pat-El cites Judg 5:5, ‫יהוה זה סיני‬, and 1 Sam 13:8, ‫למועד אשׁר שׁמואל‬. There are two fundamental problems with this argument. First, Pat-El misinterprets Judg 5:5 (see above, §3.2.1); ‫ יהוה‬is not in the same clause as ‫זה סיני‬. Since this is her lone example of this construction with ‫זה‬, her argument fails due to lack of evidence. 89 The second point of failure is statistical and evidential. On the one hand, she draws a conclusion from the absence of any example of ‫ שׁ‬+ “noun attribute,” when there are only 136 occurrences of ‫ שׁ‬in the entire Hebrew Bible. Given the rarity of what she calls the “genitive” construction in the first place, it is hardly to be expected that even 1 example would exist out of 136 cases of ‫שׁ‬. On the other hand, the 12 examples of ‫ אשׁר‬+ “noun attribute” adduced by Pat-El to distinguish the syntax of ‫ אשׁר‬from ‫ שׁ‬cannot bear the weight that she places on them. Twelve examples out of 5,500 ‫ אשׁר‬clauses in the Hebrew Bible is even less frequent (.002) than 1 ‫ שׁ‬clause out of 136 would have been (.007). 90 And once these 12 are considered carefully, only 2 (1 Sam 13:8 and 1 Kgs 11:25) fit the pattern that Pat-El describes; the other 10 have much simpler explanations grounded in well-attested grammatical

89. In Pat-El 2010, she cites Mic 5:4 (‫)וְ ָהיָ ה זֶ ה ָׁשלֹום‬, Hab 1:11 (‫)וְ ָא ֵׁשם זּו כֹחֹו ֵלאֹלהֹו‬, and Ps 34:7 (‫)זֶ ה ָענִ י ָק ָרא‬, and other examples. All have more likely interpretations than Pat-El’s “genitive marker” analysis. In Mic 5:4, the ‫ זה‬is the subject of the verb ‫ היה‬and the noun ‫שׁלום‬ is the complement, ‘this one will be peace’; in Hab 1:11, the ‫ זו‬is the relative marker of a nullhead relative, and the relative clause itself is a null-copula clause with ‫ כחו‬the subject and ‫ לאלהו‬the copular complement, ‘. . . and he shall be guilty, (he) who his strength is his god’; and in Ps 34:7, while difficult (I suggest that ‫ זה‬is the subject, by which “David” is pointing at himself, followed by ‫ עני‬as a substantival adjective in apposition, ‘this one, a poor fellow’), ‫ זה‬is certainly not a “genitive marker.” 90.  1 Sam 13:8 (‫מּואל‬ ֵ ‫ּמֹועד ֲא ֶׁשר ְׁש‬ ֵ ‫יָמים ַל‬ ִ ‫יחל ׀ ִׁש ְב ַעת‬ ֶ ִ‫וַ ּי‬‎); 1 Kgs 7:48 (‫וַ ּיַ ַעׂש ְׁשֹלמֹה ֵ ֚את ָּכל־‬ ‫ה ֵּכ ִלים ֲא ֶׁשר ֵּבית יְ הוָ ה‬‎ַ ); 11:25 (‫ת־ה ָר ָעה ֲא ֶׁשר ֲה ָדד‬ ָ ‫ל־יְמי ְׁשֹלמֹה וְ ֶא‬ ֵ ‫וַ יְ ִהי ָׂש ָטן ְליִ ְׂש ָר ֵאל ָּכ‬‎); 2 Kgs 2:3 (֒‫יׁשע‬ ָ ‫ל־א ִל‬ ֱ ‫ית־אל ֶא‬ ֵ ‫ר־ּב‬ ֵ ‫יאים ֲא ֶׁש‬ ִ ‫י־הּנְ ִב‬ ַ ֵ‫וַ ּיֵ ְצאּו ְבנ‬‎); 10:29 (‫ית־אל‬ ֵ ‫יהם ֶעגְ ֵלי ַהּזָ ָהב ֲא ֶׁשר ֵּב‬ ֶ ‫א־סר יֵ הּוא ֵמ ַא ֲח ֵר‬ ָ ֹ‫ל‬ ‫וַ ֲא ֶׁשר ְּב ָדן‬‎); 25:15 (‫ב־ט ָּב ִחים‬ ַ ‫ר־ּכ ֶסף ָּכ ֶסף ָל ַקח ַר‬ ֶ ‫א ֶׁשר זָ ָהב זָ ָהב וַ ֲא ֶׁש‬‎ ֲ ); Jer 32:2 (‫וְ יִ ְר ְמיָ הּו ַהּנָ ִביא ָהיָ ה‬ ‫הּודה‬ ָ ְ‫ית־מ ֶלְך י‬ ֶ ‫כלּוא ַּב ֲח ַצר ַה ַּמ ָּט ָרה ֲא ֶׁשר ֵּב‬‎ָ ); 52:19 (‫ב־ט ָּב ִחים‬ ַ ‫ר־ּכ ֶסף ָּכ ֶסף ָל ַקח ַר‬ ֶ ‫א ֶׁשר זָ ָהב זָ ָהב וַ ֲא ֶׁש‬‎ֲ ); Ezek 42:11 (‫ה ְּל ָׁשכֹות ֲא ֶׁשר ֶּד ֶרְך ַה ָּצפֹון‬‎ ַ ), 12 (‫ה ְּל ָׁשכֹות ֲא ֶׁשר ֶּד ֶרְך ַה ָּדרֹום‬‎ ַ ); Hos 12:9 (‫אּו־לי ָעֹון‬ ִ ‫יִמ ְצ‬ ְ ‫לֹא‬ ‫ר־ח ְטא‬ ֵ ‫א ֶׁש‬‎ ֲ ); 2 Chr 23:9 (‫ת־ה ְּׁש ָל ִטים ֲא ֶׁשר‬ ַ ‫ת־ה ָּמגִ ּנֹות וְ ֶא‬ ַ ‫יתים וְ ֶא‬ ִ ִ‫ת־ה ֲחנ‬ ַ ‫וַ ּיִ ֵּתן יְ הֹויָ ָדע ַהּכ ֵֹהן ְל ָׂש ֵרי ַה ֵּמאֹות ֶא‬ ‫ֹלהים‬ ִ ‫)ל ֶּמ ֶלְך ָּדוִ יד ֲא ֶׁשר ֵּבית ָה ֱא‬. ַ

98

Chapter 3

patterns. 91 Moreover, for 1 Sam 13:8, the context suggests that the head of the ‫אשׁר‬-relative is not ‫ מועד‬but ‫ שׁבעת ימים‬and that the one-word relative clause is elliptical for ‫אשׁר אמר שׁמואל‬, i.e., ‘and he waited the seven days for the meeting, which Samuel had said’ (this harks back to 10:8, where Samuel told Saul ‫)שׁבעת ימים תוחל עד בואי אליך‬. This leaves only 1 Kgs 11:25, which has obvious textual problems apart from the ‫ אשׁר‬clause, since ‫ את הרעה אשׁר הדד‬has no clear syntactic role in the verse. So, once the extremely few data adduced for Pat-El’s methodologically suspect case are considered more carefully, the evidence disappears, and consequently the argument fails. In sum, the proposed phonological changes for relating ‫ אשׁר‬and ‫ שׁ‬do not occur in Hebrew, and the conjectured syntactic differences between ‫ אשׁר‬and ‫ שׁ‬that support their etymological connection do not exist. There is therefore no cogent argument for a shared etymology for ‫ אשׁר‬and ‫שׁ‬. This moves us back to the argument that ‫ אשׁר‬and ‫ שׁ‬are etymologically unrelated, leaving us with three relative elements, ‫זה‬‎, ‫שׁ‬‎, and ‫( אשׁר‬not to mention ‫ ה‬and ‫)מ‬. What is a historically and linguistically plausible explanation for this apparent abundance of relative redundancy? First, we may note that ‫ זה‬is used rarely and only in poetic texts, which are likely reflections of an older stage of the language; thus, ‫ זה‬does not overlap with the other relative elements throughout the corpus (see §3.2.1). Second, the functions of ‫ ה‬and ‫ מ‬are constrained syntactically and semantically, and so neither of these overlaps fully with the other relative elements (see §3.2.3–4). This leaves the overlap between ‫אשׁר‬ and ‫ שׁ‬at the heart of the messy redundancy—except that these two words do not fully overlap either. As I demonstrate in §7.4, there is a statistically clear diachronic change between ‫ אשׁר‬in earlier texts and ‫ שׁ‬in later texts of the Second Temple period. The question then returns to where ‫ שׁ‬came from. And what about the few examples of ‫ שׁ‬that are in early texts (Judg 5:7 [2×]; 6:17; 7:12; 8:26; 2 Kgs 6:11; and also perhaps Gen 6:3; Jonah 1:7, 12; 4:10)? These data provide a critical clue in reconstructing the history of ‫ שׁ‬in Hebrew. The Judges examples are used exclusively in discourses set in the northern tribal areas of Ephraim, Naphtali, and Manasseh. The 2 Kings 6 occurrence is set in the mouth of an 91.  See 1 Kgs 7:48; 2 Kgs 2:3; 10:29; 25:15; Jer 32:2; 52:19; Hos 12:9; 2 Chr 23:9. For 1 Kgs 7:48; 2 Kgs 2:3; 10:29; Jer 32:2; 2 Chr 23:9, the ‫ בית‬should be understood to function as either a locative adverbial or an abbreviation of ‫ ;בבית‬see GBH 428–29, §126h; in Hos 12:9, the NP is a complement of a null copula, “iniquity which was sin”; in 2 Kgs 25:15 and Jer 52:19, the repetitive NPs are the copular complements in null-head relatives, “(the items) that were really gold and (the items) that were really silver.” Ezek 42:11, 12 reflect the adverbial (locative-directional) use of ‫דרך‬, giving the ‫ דרך‬+ noun (cardinal direction) collocation prepositional semantics (i.e., ‘at/to the way of the north/south’ > ‘on the north/south’), which is particularly prevalent at the end of Ezekiel.

Relative Words

99

Aramean king, which cannot be historically accurate since ‫ שׁ‬was never used at any stage of Aramaic. Thus, Young proposes, We do not know of this word in our Aramaic sources at all. We must therefore raise the possibility that beside genuine foreign and dialectal forms, the Hebrew author could also draw on a body of clichéd “non-standard” forms. To draw some modern analogies, while parodies of foreign or dialectal speech will utilize certain language features which are felt to be absolutely characteristic of the target of the parody . . . , other accent features used will be from the general category of “funny speech”, which is built from a mishmash of many different varieties of “non-standard” language. (Young 1995: 65–66; see also Rendsburg 1990: 123 n. 29)

The use of ‫ שׁ‬in this verse is thus a technique by which to characterize the king of Aram as “foreign.” Similarly, the three examples of ‫ שׁ‬in Jonah can be explained as intentional literary devices. In 1:7, ‫ שׁ‬is placed in the mouth of the non-Israelite sailors when they speak among themselves (whereas when they address the Israelite Jonah, they use ‫ אשׁר‬in v. 8). Similarly, when Jonah addresses the sailors in v. 12, he matches their statement in v. 8 by using ‫שׁ‬, even though he uses ‫ אשׁר‬elsewhere. Finally, in 4:10 both ‫ שׁ‬and ‫ אשׁר‬are placed in the mouth of ‫יהוה‬, suggesting that the literary device of “us” versus “foreigners” was now being turned into a theological point: 92 Certainly Yhwh was not perceived as a foreigner to the book’s Israelite audience, and, in any case the presence of ‫ אׁשר‬attenuates any suggested foreign characterization. Why, then, does the author switch the relative words within a single utterance of Yhwh’s? Again, it fits the book’s overall rhetorical purpose, which builds strongly in the final chapter: Yhwh is not just the deity of the Israelites, his domain and care extends well beyond the borders of Israel. Thus, the author 92.  Muraoka (2012) has recently speculated that the Jonah variation between ‫ שׁ‬and ‫אשׁר‬ can be explained in terms of diglossia, with ‫ שׁ‬marking the vernacular and ‫ אשׁר‬the higher register. While this is a creative suggestion, it is unfortunate that Muraoka conflates the distinct phenomena of diglossia and social registers. While Muraoka describes the variation in Jonah in terms of education, diplomacy, and friendliness, diglossia is something quite different. Ferguson defined diglossia in his seminal 1959 article (and affirmed in his 1991 follow-up study) as follows: A relatively stable language situation in which, in addition to the primary dialects of the language (which may include a standard or regional standards), there is a very divergent, highly codified (often grammatically more complex) superposed variety, the vehicle of a large and respected body of written literature, either of an earlier period or in another speech community, which is learned largely by formal education and is used for most written or formal spoken purposes, but is not used by any sector of the community for ordinary conversation. (Ferguson 1959: 336; 1991: 218)

It is implausible that the conversation between Jonah includes diglossic variation. For additional discussion, see Holmstedt and Kirk forthcoming.

100

Chapter 3

has used a subtle shift in style, viz., the switch in relative words from ‫ אׁשר‬to ‫שׁ‬, to reinforce his point linguistically. (Holmstedt 2006a: 18)

Both the 2 Kings and Jonah usages imply that ‫ שׁ‬was recognizable but not native to the grammar of the Hebrew audience of those two books. To summarize, then, we have a marginal, northern use in an arguably very early text—monarchic, if not premonarchic period—and then two examples of literary “foreignness” in exilic (2 Kings) and postexilic (Jonah) works. And, as I demonstrate in §7.4, it is not until the Hellenistic period and the books of Ecclesiastes and Ben Sira that we see the change from ‫ אשׁר‬to ‫ שׁ‬and the latter’s diffusion in Hebrew grammar. The distribution of ‫ שׁ‬suggests to me that the following sketch is not only historically plausible but likely in its outlines. The early use of ‫ שׁ‬reflects either Phoenician or Assyrian influence on Northern Israel, either during the premonarchic period or the divided monarchy. The possibilities are numerous for the specific context wherein language contact and borrowing resulted in  ‫’שׁ‬s entering into a northern dialect. Perhaps some northern tribes were exploring connections with their Phoenician neighbors in eleventh century b.c.e. Or ‫שׁ‬ may reflect direct Neo-Assyrian linguistic influence on the Northern Kingdom: it may have had the twofold effect of representing a prestige item and distinguishing the Israelian Hebrew from Judahite Hebrew and its use of ‫אשׁר‬. The Phoenicians themselves (as well as some other Canaanites, taking the Lachish ewer and Tanaach texts into account) may have adopted š- from Assyrian during the Middle Assyrian period, perhaps especially during Tiglath-pileser I’s (1114–1076 b.c.e.) aggressive attention to northern Syria–Palestine (Rainey and Notley 2006: 107; Van de Mieroop 2007: 182). If so, then š- in TyroSidonian may reflect a different policy of appeasement toward the Assyrians and Tiglath-pileser I than their northern neighbor, Byblos, had, in which the West Semitic z- was retained. For the later Hebrew change and diffusion from ‫ אשׁר‬to ‫שׁ‬, language contact is again the likely source. At this stage, however, the historical picture shifts to the exilic communities in Babylon, known from the Murashu and Al-Yahudu archives (see Pearce 2006; Zadok 2009; Abraham 2011; Pearce and Wunsch 2014). Given the degree of late Babylonian linguistic influence exhibited in the emerging archival texts from these Judean communities, it may have been that, among other items, one of the long-lasting effects of this language contact situation was the change from ‫ אשׁר‬to ‫שׁ‬. Moreover, the superficial similarity of the two relative complementizers would have contributed to the ease of borrowing as well as a folk etymology that connected the two—a folk etymology that has continued into the modern period and misled both comparative Semitists (e.g., Huehnergard 2006; 2013a; Pat-El 2012) and linguists (see, e.g, Givón 1991; Hendery 2010).

Relative Words

101

This historical sketch accounts for the asymmetry in the larger Semitic picture by providing a historically grounded account of language contact, borrowing, and change-and-diffusion with regard to ‫ שׁ‬in Hebrew. As for ‫אשׁר‬, the reasons why Hebrew moved this marginally used Semitic subordinator, used as far back as early second-millennium Akkadian, to a central grammatical role is unclear. But Hebrew was not alone, since even within the sparse extant Moabite and Edomite evidence, there are occurrences of ‫אשׁר‬, indicating that those languages patterned with Hebrew. It is possible that this southern Canaanite dialect group was innovative with ‫אשׁר‬, not in its grammaticalization, but in its replacement of the inherited West Semitic ‫ז‬-relative. 3.4. Summary The concern of this chapter was a description of the grammatical elements that Hebrew uses to introduce relative clauses. I presented and explained data for five distinct overt relative elements (‫ז‬‎, ‫אשׁר‬‎, ‫שׁ‬‎, ‫ה‬‎, and ‫ )מ‬as well as a nullrelative element used in zero relativization. In addition, I addressed the sticky wicket of the etymologies of ‫ אשׁר‬and ‫שׁ‬. Besides briefly describing the history of research on this issue, I particularly addressed my previously published criticism of the resurrected old proposal in Huehnergard 2006 and then added careful consideration of even-more-recent arguments advanced in Pat-El 2010; 2012. At the end of the matter, I agreed with my 2007 argument, that the separate etymology analysis that was the consensus in the twentieth century remains the most coherent and compelling among the possibilities: ‫ אשׁר‬is derived from a common Semitic noun ‘place’, and ‫ שׁ‬is cognate with the East Semitic (and Phoenician and Ammonite) š-relative. To address the perplexing question about how Hebrew ended up with two reflexes, ‫ ז‬and ‫שׁ‬, of the Proto-Semitic *ḏū/ṯū, I again concurred with my statement, that “two centuries of historical and comparative linguistic investigation should have taught us by now that our desire for symmetry within language families and dialect geography must often take a back seat to the messy and asymmetry-producing realities of language contact, competing dialects, social registers, and the use of dialectal variation for reasons of style and rhetoric in literary compositions” (Holmstedt 2007: 191). It is with these factors in mind that I concluded by offering a plausible sketch that satisfies historical and linguistic principles of reconstruction.

Chapter 4

Relative Heads In considering relative heads, we are faced with the “problem of the pivot” (de Vries 2002: 1): unlike other subordinate clauses, the relative clause is linked to the main clause by a constituent, the relative head, that has both a role within the subordinate clause and a role within the matrix clause. This dual syntactic role distinguishes relative clauses from other clause types that turn out to be only superficially similar. For English, the contrast is illustrated well by the two types of that clause (see above, [9] and [10]); for Hebrew, this is paralleled by the two types of ‫ אשׁר‬clauses, such as (195) and (196) (repeated from [11] and [12]): (195) (= [11]) ‫אתיָך ֵמ ֶא ֶרץ ִמ ְצ ַריִ ם ִמ ֵּבית ֲע ָב ִדים‬ ִ ‫הֹוצ‬ ֵ ‫להיָך ֲא ֶׁשר‬ ֶ ‫ָאנ ִֹכי יְ הוָ ה ֱא‬ ‘I am Yhwhi, your god, whoi ___i brought you out from the land of Egypt, from a house of slaves’ (Deut 5:6) (196) (= [12]) ‫ֹלהיָך‬ ֶ ‫ית ֲא ֶׁשר נְ ָׂש ֲאָך יְ הוָ ה ֱא‬ ָ ‫ָר ִא‬ ‘you saw that Yhwh, your god, carried you’ (Deut 1:31) In the relative in (195), the head ‫ יהוה‬plays two roles, one within the matrix clause (as the complement of the null copula, the subject of which is ‫ )אנכי‬and one as the subject of the verb ‫ הוצאתיך‬within the relative clause. In contrast, the ‫ אשׁר‬clause in (196) has no head, and the NP ‫ יהוה‬only plays one syntactic role, as the subject of the verb ‫ נשׂאך‬within the subordinate clause; the clause as a whole is not a relative but is the complement of the matrix verb ‫ראית‬. In relative clauses such as (195), the head NP faces both the matrix clause and the subordinate relative clause. The Janus-like character of the relative head, as the syntactic pivot, aids not only in the proper classification of a clause such as (195) as a relative (versus, for example, a complement clause, as with [196]), but also in the precise identification of the head when there are multiple NP candidates available, as in (197) and (198): (197) ‫א־ת ָע ֶׂשינָ ה‬ ֵ ֹ ‫ל־מ ְצֹות יְ הוָ ה ֲא ֶׁשר ל‬ ִ ‫‏ וְ ָעׂשּו ַא ַחת ִמ ָּכ‬ ‘(and if) they do any of the commandmentsi of Yhwh which ___i should not be done’ (Lev 4:13) 102

Relative Heads

103

(198) ‫אתֹת ֲא ֶׁשר ִצּוָ הּו׃‬ ֹ ‫ל־ה‬ ָ ‫ל־ּד ְב ֵרי יְ הוָ ה ֲא ֶׁשר ְׁש ָלחֹו וְ ֵאת ָּכ‬ ִ ‫וַ ּיַ ּגֵ ד מ ֶֹׁשה ְל ַא ֲהר ֹן ֵאת ָּכ‬ ‘and Moses told Aaron all the words of Yhwhi who ___i sent him and all the signsi that he commanded ___i him’ (Exod 4:28) The head for the ‫אשׁר‬-relative (197) could legitimately be either ‫ מצות‬or ‫יהוה‬. The 3fp passive verb ‫ תעשׂינה‬within the relative, and no additional indication of a syntactic role for a ms constituent combine to signal ‫ מצות‬unambiguously as being the head. Whereas the verbal agreement features in (197) make the decision between the two available potential heads fairly simple, the case of (198) is much more difficult. For the ‫אשׁר‬-relative in (198), many commentators (e.g., Keil 1878: 389; Clements 1972: 28; Childs 1974: 91; Durham 1987: 52–53; Propp 1999: 183) and translations (nrsv, niv, njps) take ‫‘ דברי‬words of’ to be the relative head, and therefore the relative clause to be about what Moses was sent with or sent to say, not by whom he was sent. However, the 3ms verb ‫ שׁלח‬points to ‫יהוה‬ as the agentive subject, and the 3ms clitic pronoun as the verbal complement suggests a reference to Moses; the result points to a relative modifying ‫ יהוה‬as the one who sent Moses, the superficial parallel with the next relative clause notwithstanding (so Murphy 1866: 38; nkjv). The initial ambiguity concerning the heads in (197) and (198) simply illustrates that determining the head of a relative clause is not straightforward in every single case. Sometimes information within the relative clause, such as the verbal agreement features in both (197) and (198), provide the disambiguating clues. For other examples, such as (199), there is no grammatical solution, leaving only the literary context to provide clues (and even they are not always determinative). (199) ‫ת־ה ַחּיִ ים‬ ַ ‫א־עזַ ב ַח ְסּדֹו ֶא‬ ָ ֹ ‫אמר נָ ֳע ִמי ְל ַכ ָּל ָתּה ָּברּוְך הּוא ַליהוָ ה ֲא ֶׁשר ל‬ ֶ ֹ ‫‏וַ ּת‬ ‫ת־ה ֵּמ ִתים‬ ַ ‫וְ ֶא‬ ‘and Naomi said to her daughter-in-law: He is blessed to Yhwh who ___ has not abandoned his steadfast love for the living and the dead! (Ruth 2:20) What is the head in (199)? Is it the closest potential candidate, ‫יהוה‬, which is embedded within a ‫ ל‬PP, or is it the pronoun ‫הוא‬, which refers to Boaz? In my grammatical commentary on the book of Ruth, I summarized the choices and their implications: The choice is significant, since if Noʿomi recognizes Yhwh’s faithfulness here, the ultimate resolution of the book begins here. Many commentators take Boaz as the head of the relative and thus the null subject within the relative as well as the antecedent of the pronoun in ‫חסדו‬. This makes discourse sense in that Boaz is the topic of the blessing and there is no explicit switch of topic. However, it is an overwhelming tendency in the syntax of relative clauses that, unless explicitly

104

Chapter 4

identified (e.g., by the use ‫ בעז‬within the relative) or a clear result of syntactic movement (e.g., the relative head fronted and thus moved away from its relative clause), the nearest grammatically acceptable antecedent is the relative head. This strongly suggests that ‫ יהוה‬is the head as well as the verbal subject of ‫עזב‬ and the antecedent for the pronoun in ‫חסדו‬. The syntactic ambiguity is probably intentional: the narrator uses this clause to signal that Noʿomi’s redemption has begun. (Holmstedt 2010: 141–42)

Thus, only a strong preference for the head to be immediately adjacent to its modifying relative clause provides a weak syntactic clue. However, this is balanced by the discourse context, which leans in favor of the further constituent as the head. Thankfully, the interpretive opacity presented by (199), or even (197) and (198), is not common, and identifying the certain, or at least very likely, relative head is simple (if this were not the case, the discourse-processing load of relatives would be so high as to preclude the frequent usage). In the remainder of this chapter, I will survey the various typologies of relative heads and their salient features, and then present the Hebrew patterns. 4.1.  Relative Head Typology As the NP modified by the relative clause, the head is the constituent about which the predication within the relative makes an assertion. Thus, the head is not only the syntactic anchor for the relative clause but also its semantic anchor. Cross-linguistic study shows that a broad selection of constituents may serve as a relative head, and the head may occur in a number of syntactic positions (inside or outside the relative clause, before or after the relative clause). In the following four sections, I sketch a typology of relative-clause heads— whether the head is overt or null, where the head is positioned in the phrasal hierarchy of the relative clause, what the linear position of the head is vis-à-vis the relative clause, and how heads, determiners, and case-markers interact. 4.1.1.  Overt and Null Heads In the most easily identifiable relative clauses, the heads are obvious—they are typically full NPs and are positioned immediately before the relative word, as in (200). However, rich discourse information often allows a relative clause to appear to lack a head, as in (201): (200) The park where we camped had wonderful trails. (201)  Where we camped had wonderful trails. In (200), the head of the relative clause is clear: it is the NP the park. In contrast, in (201) it is not immediately transparent that where is the relative head; the problem is that the wh-word where is the relative pronoun in (200), and so how can it also be the relative head in (201)? Simply, the relative pronoun

Relative Heads

105

where is not the head; it remains the relative pronoun and nothing more. The syntactic head of the relative clause in (201) is null (see above, §2.2.3, on the syntactic reality of null, or covert, constituents). This type of relative has variously been called ‘free’, ‘independent’, and ‘headless’, although since they are not in fact syntactically free, independent, or lacking a head, I will employ the more accurate label null-head relative clause. The interpretation of a null relative head is first constrained by the semantic domain established within the relative clause; thus, in example (201), the wh-word where can only refer to an entity that legitimately belongs to the membership set established by the phrase we camped—i.e., a place related to camping (the +locative feature of the head is, in English, confirmed by the +locative features associated with the relative word where, as opposed to who or which). The head is secondarily constrained by the predicate, in (201) the phrase had wonderful trails, narrowing the membership set to {camping places with good trails}, which in all likelihood restricts the referent to a nonurban, recreationally oriented location. One typological statement relates to the presence of an overt or null head in a relative clause: it is a General Tendency that most languages use nullhead relatives (4a). Clearly Hebrew supports this typological statement; see below, §4.3. 4.1.2.  Hierarchical Position of the Head In addition to the basic binary opposition between overt head and null head, there is also a basic binary opposition concerning the hierarchical position of the relative head: it resides either inside (internally-headed) or outside (externally-headed) the relative clause. English uses externally-headed relatives, as all of the English examples used in this chapter illustrate. 1 However, some languages, such as Bambara (202) and Ancash Quechua (203), typically position the relative head inside the relative clause itself. (In both examples, the head is indicated by an underline and bold typeface, and the relative clause is enclosed in brackets in the interlinear glossing.) (202) Bambara Internally-Headed RC ne ye tyὲ mìn ye, ò bè f ìnì f ère (Lehmann   1986: 665) [I compl mani reli saw] dem.3i ipfv cloth.def sell ‘the man that I saw (, he) sells the cloth’ 1.  Mallinson and Blake (1981) assert that internally-headed relatives do occur in English, but rarely: for example, What jewelry the thief left behind wasn’t worth worrying about (1981: 355). Further research on relatives would now identify this as a null-head relative with a free-choice interpretation and a complex wh-phrase. In English, it is more typical to see the use of -ever phrases in these relatives, such as Whatever jewelry the thief left behind. See Dayal 1997 and Grosu 2002 for similar examples and discussion.

106

Chapter 4

(203) Ancash Quechua Internally-Headed RC nuna bestya-ta ranti-shaq-n alli bestya-m ka-rqo-n   (Cole 1987: 277) [man horse-acc buy-perf-3] good horse-evid be-past-3 ‘the horse that the man bought was a good horse’ In both (202) and (203), the NP that is semantically modified by the relative clause is a constituent within the syntactic boundaries of the relative. Compare the accusative case-marking on the head in (203) with a similar but externallyheaded relative in Ancash Quechua (204). (204) Ancash Quechua Externally-Headed RC nun ranti-shaq-n bestya alli bestya-m ka-rqo-n   (Cole 1987: 279) [man buy-perf-3] horse.nom good horse-evid be-past-3 ‘the horse that the man bought was a good horse’ The external relative head in (204) is marked for nominative case. Whereas the internal relative head bestyata in (203) receives accusative case from the verb rantishaqn ‘bought’ within the relative clause, the external head in (204) received nominative case as the subject of the predicate alli bestyam karqon ‘was a good horse’ in the matrix clause. The word order also distinguishes (203) and (204): the NP bestya cannot be a part of the relative clause in (204) since subordinate clauses, like relative clauses, in Quechua are strictly verb-final, which effectively bars bestya from being an argument of the verb rantishaqn ‘bought’ in the relative. In (203), the NP bestyata precedes the embedded verb and can thus be one of its arguments. The hierarchical position of the head relates closely to the linear order of head and relative. Thus, externally-headed relatives are either prenominal or postnominal, and internally-headed relatives are circumnominal. Given the close relationship of these features, I will postpone a discussion of typological universal and tendencies until the end of the next section. 4.1.3.  Linear Order of the Head and Relative Clause The internally-headed relatives in examples (202) and (203) introduce us to another parameter of relative heads: the placement of the head vis-à-vis the relative clause. There are four main structural types, schematized in table 4.1 (from de Vries 2002: 20, with slight modification). Both the English and Hebrew examples given in this work illustrate that both languages form relatives based on the structural strategy in (a): the relative head always precedes the relative clause. 2 Compare the Hebrew and its 2.  In fact, this is true of Semitic in general: no Semitic language uses circumnominal (internally-headed) relatives, or the few that use prenominal relatives (Amharic, Tigré) have

Relative Heads

107

Table 4.1.  Linear Order of Head and Relative Clause a.  postnominal RC b.  prenominal RC c.  circumnominal RC d.  correlative RC

[CP-matrix . . . [N RC] . . .] [CP-matrix . . . [RC N] . . .] [CP-matrix . . . [[RC . . . N . . .]] . . .] [CP-matrix . . . [RC (. . .) N . . .] [CP-matrix . . . (Dem) . . .]

English translation in (195) with the prenominal relative—type (c)—of the Mandarin Chinese example in (205): (205) Mandarin Chinese Prenominal RC (Lehmann 1984: 64) wŏ bă nĭ gĕi wŏ de shū diūdiào-le I acc [you give I nmlz] book lose-perf ‘I lost the book (that) you gave me’ Similarly, in the externally-headed Ancash Quechua relative, given above in (204) and repeated here as (206), the relative head bestya ‘horse.nom’ follows the relative clause. (206) Ancash Quechua Prenominal RC (Cole 1987: 279) nun ranti-shaq-n bestya alli bestya-m ka-rqo-n [man buy-perf-3] horse.nom good horse-evid be-past-3 ‘the horse that the man bought was a good horse’ In contrast, the internally-headed Ancash Quechua relative, (203), repeated here as (207), is circumnominal (type [c]): constituents within the relative clause come both before and after the head bestyata ‘horse-acc’. (207) Ancash Quechua Circumnominal RC (Cole 1987: 277) nuna bestya-ta ranti-shqa-n alli bestya-m ka-rqo-n [man horse-acc buy-perf-3] good horse-evid be-past-3 ‘the horse that the man bought was a good horse’ Finally, three relatives from Hindi are given in (208)–(209) to illustrate the nature of correlatives (Type [d]). In each example, the relative clause is underlined and in boldface (Srivastav 1991: 639–40; translations slightly modified). For more on the structure of correlatives, see §5.1. (208) jo laRkii khaRii hai vo ambii hai rel girl standing is dem tall is lit., ‘that girl is standing, that/she is tall’ arguably been influenced by the syntax of a non-Semitic substratum. J. Cale Johnson (2005) has argued that Akkadian used an internally-headed relative strategy; for my response to this unlikelihood, see §8.2.

108

Chapter 4

(209) vo laRkii lambii hai jo khaRii hai dem girl tall is rel standing is lit., ‘that girl is tall, who is standing’ 3 Two typological statements concerning the position of relative heads and relative clauses are relevant to Hebrew. First, (3f) states that, if a language has postnominal relative clauses, relative complementizers are clause-initial (IU). All of the Hebrew examples I present throughout this work show (3f) to be accurate for Hebrew, since all the relative elements are located between the head and the relative clause—i.e, they are clause-initial. Second, (4b) predicts that, in general, languages with postnominal relative clauses are also verb-object (VO)—i.e., head-initial within the VP. Hebraists are unanimous that Hebrew is a VO language (it is the position of the subject—that is, SVO or VSO—that is debatable; see §2.3.1). 4.1.4.  Position of Determiner and Case-Marker The positions of the determiner and case morphemes with respect to the relative head are also typologically distinguishing features. Example (210) reflects the fact that in English the determiner always precedes the noun, whether it is the definite article or a pronominal phrase (e.g., her man, the company’s man). In contrast to the “initial” determiner in English, the Swedish example in (211) and the Indonesian example in (212) illustrate “middle” and “final” placement of the determiner in relativization. (210) Initial Position of the Determiner with Respect to NP Head and RC I spoke with the man who knows you (211) Middle Position of the Determiner with Respect to NP Head and RC Jag talade med mann-en vilken känner dig (de Vries 2002: 19) I spoke with man-the rel knows you (212) Final Position of the Determiner with Respect to NP Head and RC Orang yang datang itu Ahmed (Lehmann 1984: 95) person rel come det Ahmed ‘the man who came was Ahmed’ Languages that have overt case-marking fall into two groups with regard to the position of the case morpheme and the relative head and/or the relative clause. German represents languages that allow the head to be case-marked (213), whereas Shoshone represents those that mark both the head and the relative clause (214). 3. The more literal English gloss, ‘the girl is tall who is standing’ (i.e., without the comma), is not a correlative but an extraposed relative; both the semantics and the syntactic structure of correlative relative clauses are different from extraposed relative clauses. See Srivastav 1991; de Vries 2002: 145–50.

Relative Heads

109

(213) Case-Marked on the Head Ich fürchte den Herr-n der eine Pistole trägt   (de Vries 2002: 19) I fear the gentleman-acc who a gun carries (214) Case-Marked on the Head and the RC Nɨ u pui-“ka-ha-ntɨ tu”ku-i un ti“ka-”pɨh-a (Lehmann 1984: 79) I it see-res-ref-ptcp meat-acc poss.3 eat-ptcp-perf-acc ‘I see the meat that he ate’ The pattern of the determiner and relative head accords with the general pattern in each language for the position of determiners. However, the issues involved in the syntax of determiners are sufficiently complicated that no clear universals or tendencies with particular regard to relative clauses are discernible. The same is true of case-marking: it is worth noting in a given language, but no insightful typological statements have emerged with regard to relative clauses. I have now surveyed the salient features of relative heads in cross-linguistic perspective, noting whether typologists have been able to determine universals (absolute or implicational) or tendencies (general or implicational) arising from a comparison of the data. We are now in a position to turn to an informed study of relative heads in Hebrew relative clauses. 4.2.  Overt Relative Heads in Hebrew Hebrew exhibits a fair amount of diversity in the type of nominal heads modified by relative clauses. Indeed, though prepositional and clausal heads are rare, the fact that they appear to be grammatical (even if marginal) suggests that some speakers, at least, allowed the definition of nominal to be stretched. The examples given in (215)–(222) represent the full range of constituents modified by relative clauses that are found in the Hebrew Bible. Note that my examples are primarily ‫אשׁר‬‎-relatives, since this relative element is the most common in Hebrew; however, when possible I provide ‫שׁ‬‎, ‫ה‬, ‎‫ז‬,‎and zerorelative data, which suggest that none of these relative words shows any further restrictions on the type of relative head it may modify. (215) Bare Indefinite NP a. ‫יתם‬ ִ ִ‫ֶׁש ֶקר ֲא ֶׁשר לֹוא ִצּו‬ ‘a falsehood that I did not command them’ (Jer 29:23) 4 4.  For examples with ‫ אשׁר‬in the Pentateuch alone, see Gen 1:12, 29; 20:9; 27:27; 31:13; 34:14; 38:25; 39:20; 40:3; 41:38; 44:8, 15; 49:28; Exod 12:30; 21:13; 22:15; 30:33, 38; 32:1, 23; 34:18; Lev 4:24, 33; 5:2; 6:18; 7:2; 14:13, 32; 15:5, 18, 33; 17:13; 20:10, 11, 12, 13, 14, 15, 16, 17, 18, 20, 21; 21:19; 22:4, 5, 6; 27:9, 26; Num 5:10, 30; 9:6, 17; 12:11; 14:8; 19:20;

110

Chapter 4 b. ‫ּול ָא ָדם ֶׁשּלֹא ָע ַמל־ּבֹו‬ ְ ‘and to a man that has not toiled over it’ (Eccl 2:21) c. ‫ַעם־זּו ּגָ ָא ְל ָּת‬ ‘a people that you have redeemed’ (Exod 15:13) d. ‫ֶח ֶרב ַהּיֹונָ ה‬ ‘a sword that oppresses’ (Jer 46:16) 5 e. ‫ן־א ָדם ָח ִציר יִ ּנָ ֵתן‬ ָ ‫ּומ ֶּב‬ ִ ‫וַ ִּת ְיר ִאי ֵמ ֱאנֹוׁש יָ מּות‬ ‘You have feared a man (that) dies and a human (that) is made (as) grass’ (Isa 51:12) 6

(216) Modified Indefinite NP a. ‫ֹלהינּו‬ ֵ ‫ֹלהים ְקר ִֹבים ֵא ָליו ַּכיהוָ ה ֱא‬ ִ ‫גֹוי גָ דֹול ֲא ֶׁשר־לֹו ֱא‬ ‘a great nation that has gods near to it like Yhwh, our god, whenever we call to him’ (Deut 4:7) b. ‫יִ ְהיּו ַּכ ֲח ִציר ּגַ ּגֹות ֶׁש ַּק ְד ַמת ָׁש ַלף ֵיָבׁש‬ ‘they will be like grass of rooftops, which before it has shot up it withers’ (Ps 129:6) c. ‫ל־צ ִּדיק ָע ָתק ְּבגַ ֲאוָ ה וָ בּוז‬ ַ ‫‏ׂש ְפ ֵתי ָׁש ֶקר ַהּד ְֹברֹות ַע‬ ִ ‘lips of deceit that (are) speaking against a righteous person arrogantly, with pride and contempt’ (Ps 31:19) d. ‫וַ ּיִ ַּקח ָה ִאיׁש נֶ זֶ ם זָ ָהב ֶּב ַקע ִמ ְׁש ָקלֹו‬ ‘and the man took a (nose) ring of gold (that) its weight was a half-shekel’ (Gen 24:22) (217) Determined NP a. ‫ן־ה ָא ָדם‬ ָ ‫ר־ל ַקח ִמ‬ ָ ‫ת־ה ֵּצ ָלע ֲא ֶׁש‬ ַ ‫ֶא‬ ‘the rib that he took from the man’ (Gen 2:22) b. ‫ַה ַּמ ֲע ֶׂשה ֶׁשּנַ ֲע ָׂשה ַּת ַחת ַה ָּׁש ֶמׁש‬ ‘the work that was done under the sun’ (Eccl 2:17) c. ‫יתי וְ ֵעד ִֹתי זֹו ֲא ַל ְּמ ֵדם‬ ִ ‫ְּב ִר‬ ‘my covenant and testimony that I will teach them’ (Ps 132:12) 27:18; 34:13; Deut 2:36; 3:4; 4:10; 6:11; 8:9; 11:12; 12:5, 21; 14:23, 24; 16:2, 6, 11, 22; 17:15; 20:20; 22:28; 23:9, 11, 24; 26:2; 28:33, 36, 49; 29:25. 5.  See Exod 26:12; Judg 21:19; 1 Sam 25:10; Jer 27:3; 50:16; Ezek 2:3; 47:2; Nah 3:17; Ps 62:4; Dan 9:26; 2 Chr 31:6. 6.  See Gen 15:13; Lev 7:35; 2 Sam 22:44; Isa 48:17; 51:12; Pss 5:5; 18:44; 25:12.

Relative Heads

111

d. ‫ל־ּביתֹו ְּב ִמ ְכ ַסת‬ ֵ ‫ּוׁש ֵכנֹו ַה ָּקר ֹב ֶא‬ ְ ‘his neighbor that (is) close to his house’ (Exod 12:4) e. ‫ית ָּלנּו‬ ָ ‫ה־ה ָּד ָבר ַהּזֶ ה ָע ִׂש‬ ַ ‫ָמ‬ ‘What is this thing (that) you have done to us?’ (Judg 8:1) 7 (218) Proper Noun a. ‫ר־ה ְת ַה ַּל ְכ ִּתי ְל ָפנָ יו יִ ְׁש ַלח ַמ ְל ָאכֹו ִא ָּתְך‬ ִ ‫יְ הוָ ה ֲא ֶׁש‬ ‘Yhwh, who(m) I walked before him, will send his angel with you’ (Gen 24:40) b. ‫לּולי יְ הוָ ה ֶׁש ָהיָ ה ָלנּו‬ ֵ ‘If it had not been Yhwh, who was for us, . . .’ (Ps 124:1) c. ‫ֲהלֹוא יְ הוָ ה זּו ָח ָטאנּו לֹו‬ ‘was it not Yhwh, who we sinned against him’ (Isa 42:24) d. ‫יה ַא ָּתה ֵאל ֳר ִאי‬ ָ ‫‏וַ ִּת ְק ָרא ֵׁשם־יְ הוָ ה ַהּד ֵֹבר ֵא ֶל‬ ‘and she named Yhwh, who was speaking to her: “You are ElRoi”’ (Gen 16:13) e. ‫ִּכי יְ הוָ ה ַקּנָ א ְׁשמֹו ֵאל ַקּנָ א הּוא‬ ‘because Yhwh, (who) his name is Jealous—he is a jealous god’ (Exod 34:14) (219) NP Bound to Relative Element 8 a. ‫סּורים‬ ִ ‫סּורי ַה ֶּמ ֶלְך ֲא‬ ֵ ‫ר־א‬ ֲ ‫ל־ּבית ַהּס ַֹהר ְמקֹום ֲא ֶׁש‬ ֵ ‫וַ ּיִ ְּתנֵ הּו ֶא‬ 7.  See Judg 8:1; Isa 53:7; Zeph 2:1; Pss 8:9; 25:12; 33:12. 8.  The list below includes all the examples in which ‫ אשׁר‬acts as the clitic host for a bound element. The majority are with the quantifier ‫כל‬, though many are also with the noun ‫מקום‬. The list also includes examples of bound-form nouns that have been grammaticalized as subordinate conjunctions (‫“ מפני‬because”) or prepositions (‫“ מלבד‬apart from”). Though the latter result in a different ‫ אשׁר‬clause (e.g., the ‫ אשׁר‬clause is often a nonrelative nominalized clause), such examples illustrate the general principle that the ‫ אשׁר‬clause can serve as a clitic host. For non-‫ כל‬examples, see: (42×) Gen 39:20; 40:3; Exod 19:18; Lev 4:24, 33; 6:18; 7:2; 13:46; 14:13, 32; 27:8; Num 6:21; 9:17–18; Deut 20:20; 22:24; 23:5; 1 Sam 3:13; 2 Sam 13:22; 15:21; 1 Kgs 21:19; Jer 22:12, 25; 44:23; Ezek 6:13; 21:35; 23:28; 48:22; Hos 2:1; Mal 2:9; Esth 4:3; 8:17; Neh 4:14. With ‫ כל‬as the head (or quantifier of a null head), see: (174×) Gen 1:31; 12:20, 13:1; 14:23; 20:7; 24:2, 36; 25:5; 31:1, 12, 21; 34:29; 35:2; 39:5, 6, 22; 41:56; 45:10, 11, 13; 46:1, 32; 47:1; Exod 6:29; 7:2; 9:19, 25; 10:12; 18:1, 8, 14; 20:11; 25:22; 31:6; 34:32; 35:10; 38:22; 40:9; Lev 8:10; 14:36; 27:8, 28; Num 15:23; 16:26, 30, 33; 18:13; 19:14; 22:2; Deut 3:21; 5:27, 28; 10:14; 12:11; 13:16; 18:18; 29:1, 8; Josh 1:16; 2:13; 6:17, 21, 25; 7:15, 24; 9:9, 10; 22:2; 23:3; Judg 3:1; 7:18; 9:25, 44; 11:24; 1 Sam 2:22; 3:12; 14:7; 15:3; 19:18; 25:21, 22; 30:18, 19; 2 Sam 3:19, 25; 6:12; 7:3; 9:9; 11:22; 14:20; 16:21; 1 Kgs 2:3; 10:2; 11:38, 41; 14:29; 15:7, 23, 31; 16:14; 19:1; 20:4; 22:39; 2 Kgs 8:6,

112

Chapter 4 ‘and he put him in the round house, the place-that the king’s prisoners (were) confined’ (Gen 39:20 Kethiv) b. ‫ְמקֹום ֶׁשּיִ ּפֹול ָה ֵעץ‬ ‘the place-that the tree falls’ (Eccl 11:3) c. ‫ְמקֹום זֶ ה יָ ַס ְד ָּת ָל ֶהם‬ ‘the place-that you established for them’ (Ps 104:8) d. ‫אכה‬ ָ ‫ל־ה ְּמ ָל‬ ַ ‫‏ ְל ַבד ִמ ָּׂש ֵרי ַהּנִ ָּצ ִבים ִל ְׁשֹלמֹה ֲא ֶׁשר ַע‬ ‘apart from the officers-who are supervising belonging to Solomon, who were over the work’ (1 Kgs 5:30) e. ‫ל־ּכן יִ ְת ַרת ָע ָׂשה ָא ָבדּו‬ ֵ ‫ַע‬ ‘therefore the abundance-(that) it made they lost’ (Jer 48:36) 9

(220) Clitic Pronoun 10 a. ‫יׁשי‬ ִ ‫ֹלהים ְׂש ָכ ִרי ֲא ֶׁשר־נָ ַת ִּתי ִׁש ְפ ָח ִתי ְל ִא‬ ִ ‫אמר ֵל ָאה נָ ַתן ֱא‬ ֶ ֹ ‫וַ ּת‬ ‘and Leah said: God has provided my reward, who (I) gave my maid to my husband’ (Gen 30:18) b. ‫ל־ּתִראּונִ י ֶׁש ֲאנִ י ְׁש ַח ְרח ֶֹרת‬ ְ ‫ַא‬ ‘do not look at me who (I) am dark’ (Song 1:6) c. ‫ל־עֹונֵ ִכי‬ ֲ ‫מּוליו ַהּס ֵֹל ַח ְל ָכ‬ ָ ְ‫ל־ּת ְׁש ְּכ ִחי ָּכל־ּג‬ ִ ‫‏ ָּב ֲר ִכי נַ ְפ ִׁשי ֶאת־יְ הוָ ה ַוְא‬ ‘Bless Yhwh, O my soul, and do not forget all his benefits who (he) forgives all your iniquity, . . . [through v. 5]’ (Ps 103:2–5) 11 23; 10:34; 12:20; 13:8, 12; 14:28; 15:6, 16, 21, 26, 31; 18:12; 20:13, 15, 17; 21:17; 23:28; 24:5; Isa 39:2, 4, 6; Jer 1:7, 17; 26:8; 31:37; 32:23; 38:9; Ezek 12:14; 14:22, 23; 16:37, 63; 40:4; 44:5; 47:9; Pss 96:12; 109:11; 119:63; 146:6; Job 1:10, 12; 42:10; Prov 17:8; 21:1; Ruth 3:16; 4:9; Eccl 1:13, 16; 3:14; 8:3; Esth 2:13; 3:12; 4:1, 7; 5:11; 6:13; 8:9; Neh 9:6; 1 Chr 10:11; 13:14; 16:32; 2 Chr 9:1; 33:8. For ‫שׁ‬, there are five examples: Eccl 1:7; 5:15; 7:14; 11:3, 8. For ‫ז‬-relatives, there is only Ps 104:8. 9.  See Gen 1:1; 39:4; Exod 4:14; Lev 7:35; Deut 4:15; 32:37; 1 Kgs 17:14; Isa 29:1; 48:17; 51:1; Jer 2:8; 36:2; 48:36; Amos 3:9; Hos 1:2; Mic 5:2; Hab 3:16; Ps 4:8; 7:16; 9:16; 18:44; 25:12b; 44:2; 56:2, 10; 71:18; 81:6; 90:15; 118:22; 141:9; Job 3:3a; 6:17; 18:21; Lam 1:14, 21; 3:25, 57; 1 Chr 29:3b; 2 Chr 20:22; 24:11; 29:27; 30:19; 31:16, 19. 10.  In each of these, the pronoun is the nearest antecedent (a linear feature that is often lost in translation, since the Hebrew possessive pronoun is enclitic, whereas English, for example, has prenominal possessive pronouns). Moreover, the person-gender-number features within the relative also point to the pronoun as the head. See Gen 11:7; 30:18; Exod 29:33; Deut 4:19; 1 Kgs 3:12, 13; 15:13; Isa 47:15; Ezek 16:59; 20:21; 47:14; Ps 31:8; 103:2–5; Eccl 5:15; 10:16, 17; Song 1:6; 5:2; 11.  See also Pss 33:15; 103:2–5; 104:3; 113:5–6; 144:10; 147:8; Job 9:4–7.

Relative Heads

113

(221) Prepositional Phrase ‫ל־ׁשֹלמֹה‬ ְ ‫ל־ּדוִ ד וְ ֶא‬ ָ ‫ת־ּפ ֶסל ָה ֲא ֵׁש ָרה ֲא ֶׁשר ָע ָׂשה ַּב ַּביִ ת ֲא ֶׁשר ָא ַמר יְ הוָ ה ֶא‬ ֶ ‫וַ ּיָ ֶׂשם ֶא‬ ‫ְבנֹו‬ ‘and he set the image of the Asherah that he made in the Temple, where Yhwh had said to David and to Solomon his son: . . .’   (2 Kgs 21:7) (222) Entire Clause 12 ‫א־ראּו ֲאב ֶֹתיָך וַ ֲאבֹות‬ ָ ֹ ‫ל־מ ְצ ַריִ ם ֲא ֶׁשר ל‬ ִ ‫ּוב ֵּתי ָכ‬ ָ ‫ל־ע ָב ֶדיָך‬ ֲ ‫ּוב ֵּתי ָכ‬ ָ ‫ּומ ְלאּו ָב ֶּתיָך‬ ָ ‫ל־ה ֲא ָד ָמה ַעד ַהּיֹום ַהּזֶ ה‬ ָ ‫יֹותם ַע‬ ָ ‫ֲאב ֶֹתיָך ִמּיֹום ֱה‬ ‘and your houses and the houses of all of your servants and the houses of all of Egypt shall be filled [with locusts, v. 4]—which your fathers and your ancestors have never seen from the day they came to exist upon the land until this day’ (Exod 10:6) The data presented in (215)–(222) provide no discernible pattern of divergence among the Hebrew relative words. While I did not identify any examples of a ‫ז‬-relative following a modified indefinite NP or a ‫ ז‬or zero-relative that modifies a clitic pronoun, or anything other than an ‫אשׁר‬-relative that modifies a PP or CP (a clause), I take these lacunae to be statistical coincidences. The ‫ז‬-relatives are used infrequently in the Hebrew Bible, so it is not at all surprising to have some constructions unattested. Similarly, zero-relatives are dispreferred outside poetic contexts, a fact that constrains the data. Of the type of heads provided, only the clitic pronouns, PP, and CP options carry any controversy. This has more to do with the infrequent use of these relative heads and the subsequent lack of investigation—they have simply gone unnoticed (excepting only Gaenssle 1915), rather than there being any principled objection to these grammatical options. 4.3.  Null Relative Heads in Hebrew As with Hebrew’s selection of allowable relative heads, there is nothing surprising or controversial about its use of null relative heads, except for perhaps the extent to which they are employed (and the general lack of clarity with which they have been described). Though overtly-headed relative clauses constitute the great majority of Hebrew ‫ אשׁר‬clauses, there are just over 1,000 12.  Although relative clauses with an entire clause as the antecedent are not uncommon cross-linguistically (e.g., English: Adam fell down the stairs—which wasn’t a good thing), they are rare in ancient Hebrew (and in written English as well). In regard to ‫אשׁר‬, Gaenssle (1915: 58) compares this function to Syriac d and Akkadian ša as well as Latin quale and lists the following verses as examples in the Hebrew Bible: Exod 10:6; Jer 7:31; 32:35; Esth 4:16 (see also GBH 564). I have identified the following as additional examples: Deut 17:3; Josh 4:23; 2 Sam 4:10; Jer 19:5; 48:8; 2 Chr 3:1. Ps 139:15 may also be an example.

114

Chapter 4

examples of null-head relatives in the Hebrew Bible. The minimal pair in (223) (an overt head) and (224) (a null head, indicated by ∅) sufficiently illustrates that most null heads are easily identifiable and are used in the same contexts as overt heads. 13 (223) Hebrew Overt-Head Relative Clause ‫ל־ה ְּד ָב ִרים ֲא ֶׁשר ָע ָׂשה‬ ַ ‫וַ יְ ַס ֵּפר ָה ֶע ֶבד ְליִ ְצ ָחק ֵאת ָּכ‬ ‘and the servant related to Isaac all the things that he had done’   (Gen 24:66) (224) Hebrew Null-Head Relative Clause ‫ית ּלֹו‬ ָ ‫ר־ע ִׂש‬ ָ ‫ף־א ִחיָך ִמ ְּמָך וְ ָׁש ַכח ֵאת ֲא ֶׁש‬ ָ ‫ַעד־ׁשּוב ַא‬ ‘until the anger of your brother turns from you and he forgets ∅(the thing) that you did to him’ (Gen 27:45) In (223), the head of the relative is overt and easy to identify: it is, and can only be, the DP ‫הדברים‬. In contrast, in (224) the syntactic head of the relative is null—there is a direct object marker ‫ את‬followed by an ‫ אשׁר‬clause but no NP between to receive the object-marking of the ‫ את‬or to serve as the overt head of the relative. 14 Many null-head relatives in Hebrew are like the one in (224) in that the direct object marker ‫ את‬precedes the relative clause. However, it is significant that ‫ את‬does not mark clauses as complements, which means that ‫ אשׁר עשׂית לו‬in (224) cannot be the complement of the verb ‫שׁכח‬. Moreover, ‫ אשׁר‬itself cannot be the verbal complement since it is a relative element—a function word that classifies a clause or phrase rather than fulfilling any verbal valency requirements (valency refers to the number and type of arguments required by a verb to produce a well-formed clause; Crystal 2008: 507). 15 Therefore, we are left with one option: a null NP must be present. 16 This is also what the contrast between the minimal pair in (223) and (224) suggests—that there is a nominal syntactic head to the relative (224) that simply cannot be seen (i.e., it is phonologically underspecified but syntactically real). Using the context as the guide, in (224) the head must be interpretable as the complement of both the matrix verb ‫ שׁכח‬and the embedded verb within the relative, ‫עׂשית‬. While ‫ שׁכח‬would allow a “what,” “who,” “how,” or “when” as its complement, the verb ‫ עשׂה‬is more constrained. Thus, the null head could not be interpreted as “who,” since ‫ עשׂה‬does not select a human complement 13.  For similar minimal pairs, see Exod 29:1 // Gen 9:24 and 2 Sam 11:27 // Gen 38:10. 14. See Bekins 2014 for a recent insightful and linguistically-informed study of Hebrew ‫את‬. 15.  For an introduction to verbal valency in Hebrew, including its advantages over voice and transitivity, see Cook 2014; forthcoming. 16.  Throughout this work, I use ∅ for a null item, though in most generative discussions, these sorts of null elements are represented as pro (“little pro”).

Relative Heads

115

(unless the agent is divine). As for distinguishing whether the null head of (224) is a “what,” “how,” or “when,” the identity of the null head is bound within the discourse, and thus its interpretation must be contextually appropriate, such as ‘something that Esau will be able to forget’ and ‘something that Jacob did to Esau’. I have suggested in my translation the broad (and bland) ‘the thing’, but it would be contextually appropriate to fill in ‘the deceit’ or ‘the trick’. The interpretation of the head in a null-head relative is constrained by the syntactic and semantic domain established in the relative clause. Two factors determine the interpretation: the position of the gap (or trace) or resumptive within the relative clause and the discourse context (see Haegeman and Guéron 1999: 604–9). For instance, when we consider the English headless relative what I saw was excellent out of context, the referent of the relative wh-word is ambiguous. The what could refer to anything from A to Z. However, if we supply the contextual data, I went to a horse auction last night, and what I saw was excellent, then it is reasonable to reconstruct the referent of the null head as some variation of horse(s). The interpretion of a null-relative head is in many cases at once intuitive and complicated. For example, in the relative in (225), the intuitive reading of the null is similar to English “what,” as in “he knew what his son had done.” (225) ‫ר־ע ָׂשה־לֹו ְּבנֹו ַה ָּק ָטן‬ ָ ‫וַ ּיֵ ַדע ֵאת ֲא ֶׁש‬ ‘and he knew ∅(the action)i whi that his youngest son had done ___i to him’ (Gen 9:24) The position of the gap in the relative indicates that the referent of the null head (∅) is the complement of the verb ‫עשׂה‬, a bivalent verb that selects an NP complement. This part is fairly straightforward. But when we press further and ask what “what” refers to, the identification of the null head in the discourse context becomes slightly more involved. The prior context suggests that the referent of the null head is the deed or action within the event of Noah’s youngest son, Ham, “seeing his father’s nakedness” (Gen 9:22). Thus, the ∅ (“what”) performed by Ham is not an NP with a concrete referent but an NP that refers to an event. To recognize this requires moving beyond the opacity of the typical English translation “what.” The null-head relatives in (226)–(227) further illustrate how the reader/listener must use discourse cues to reconstruct the content of a null-relative head. (226) ‫ר־ּת ַב ְּׁשלּו ַּב ֵּׁשלּו‬ ְ ‫ֵאת ֲא ֶׁשר־ּתֹאפּו ֵאפּו וְ ֵאת ֲא ֶׁש‬ ‘∅(the manna) that you want to bake, bake! And ∅/(the quail) that you want to boil, boil!’ (Exod 16:23)

116

Chapter 4

(227) ‫מֹואב ֵמ ֵע ֶבר ְליַ ְר ֵּדן יְ ִריחֹו ִמזְ ָר ָחה‬ ָ ‫ֵא ֶּלה ֲא ֶׁשר־נִ ַחל מ ֶֹׁשה ְּב ַע ְרבֹות‬ ‘these (are) ∅(the inheritances) that Moses divided in the plains of Moab, beyond the Jordan (to) Jericho, to the east’ (Josh 13:32) The context of (226) (looking back to Exod 16:13) suggests that the referent of ‫ תאפו‬and ‫ תבשׁלו‬is the bread-like manna and the quail, respectively, that Yhwh provided during the wilderness episode. The context of (227) (i.e., Joshua 13) suggests a referent such as “inheritances” or “possessions,” since the chapter narrates part of the allotment of Canaan for the Israelite tribes. Like ‫אשׁר‬, the ‫שׁ‬- and ‫ז‬-relative elements may modify null heads, as in (228)–(230). (228) ‫יכה ִת ְר ֶעה‬ ָ ‫ַהּגִ ָידה ִּלי ֶׁש ָא ֲה ָבה נַ ְפ ִׁשי ֵא‬ ‘tell me, ∅(O man) who(m) my soul loves: where do you pasture (your flock)?’ (Song 1:7) 17 (229) ‫כּו־בי‬ ִ ‫ה־א ַה ְב ִּתי נֶ ְה ְּפ‬ ָ ֶ‫וְ ז‬ ‘and ∅(the friend) who(m) I loved turned against me’ (Job 19:19) (230) ‫ל־מ ְׁש ָּכ ִבי ַּב ֵּלילֹות ִּב ַּק ְׁש ִּתי ֵאת ֶׁש ָא ֲה ָבה נַ ְפ ִׁשי‬ ִ ‫ַע‬ ‘upon my bed during night I sought ∅(the man) who(m) my soul loves’ (Song 3:1) Null heads also occur in ‫ה‬-relatives, as (231) demonstrates. (231) ‫לּו־ׁש ָּמה‬ ָ ‫ְך־סד ֹם וַ ֲעמ ָֹרה וַ ּיִ ְּפ‬ ְ ‫וְ ֵע ֶמק ַה ִׂש ִּדים ֶּב ֱאר ֹת ֶּב ֱאר ֹת ֵח ָמר וַ ּיָ נֻ סּו ֶמ ֶל‬ ‫וְ ַהּנִ ְׁש ָא ִרים ֶה ָרה ּנָ סּו‬ ‘and the valley of Siddim had many tar pits, and the kings of Sodom and Gomorah fled, and they fell there, and ∅(the kings) who remained fled to the hills’ (Gen 14:10) In (231), the content of the relative, ‫(‘ הנשׁארים‬those) who remained’, makes clear that the referent of the relative cannot be the kings of Sodom and Gomorah; ‘those who remained’ can only refer to the other three kings mentioned in Gen 14:8: the kings of Admah, Zeboiim, and Bela. I mention this to establish that the ‫ה‬-relative in (231) cannot be extraposed, modifying some constituent earlier in the verse (on relative extraposition, see below §5.2.4). Instead, the null head that the ‫ה‬-relative modifies must be co-referential with the other three kings (from 14:8) who did not succumb to the tar pits. Finally, there are 47 cases of null-head zero-relatives, as in (232).

17.  Song 1:7 provides the only example of a relative modifying a vocative (null) head in the Hebrew Bible.

Relative Heads

117

(232) ‫ד־תּ ְשׁ ָלח‬ ִ ַ‫ִבּי ֲאד ֹנָ י ְשׁ ַלח־נָ א ְבּי‬ ‘by me, O Lord, send [the message] by the hand of ∅(the man)i whi (that) you shall send ___i’ (Exod 4:13) 18 The doubly null nature of such clauses—no overt head and no overt relative word—makes them understandably more difficult to process. This may explain why most of the existing examples are in poetic texts. Although null-head relatives are not uncommon in the Hebrew Bible, they are often misinterpreted (see below, §7.4; Holmstedt 2001). For instance, Exod 14:13 has been analyzed and translated in various ways, sometimes as a relative clause (with the antecedent inside the relative!; see the nrsv, niv, nas95, and njps; Durham 1987: 188), other times as the “modal” use of ‫‘ אשׁר‬just as’ (equivalent to ‫ ;כאשׁר‬Childs 1974: 216; Clements 1972: 83). Consider the verse given with its context in (233). (233) 10And Pharaoh drew near, and the Israelites lifted their eyes, and behold, the Egyptians were coming after them. And the Israelites feared greatly and cried out to Yhwh. 11And they said to Moses, “Was it because there were no graves in Egypt that you have taken us away to die in the wilderness? What is this that you have done to us by bringing us out of Egypt? 12Is this not the thing that we told you in Egypt, ‘Let us alone and let us serve the Egyptians’? Because serving the Egyptians is better than dying in the wilderness.” 13And Moses said to the people, “Do not be afraid; stand firm, and see the deliverance that Yhwh will accomplish for you today; [because you who see the Egyptians today—you shall never see them again]. 14Yhwh will fight for you, and you shall keep silent.” ‫ד־עֹולם‬ ָ ‫ת־מ ְצ ַריִ ם ַהּיֹום לֹא ת ִֹסיפּו ִל ְרא ָֹתם עֹוד ַע‬ ִ ‫יתם ֶא‬ ֶ ‫ִּכי ֲא ֶׁשר ְר ִא‬ ‘because ∅(you)i whi who ___i see the Egyptians today—you shall never see them again’ (Exod 14:13) The first option I described above, that the second ‫ אשׁר‬clause in Exod 14:13 is an internally-headed relative, yields the translation “for the Egyptians whom you see today you will never see again” (njps). Clearly this reflects a misunderstanding of Hebrew grammar: Hebrew only uses an externally-headed relative clause formation (see below, §4.4). Furthermore, the second option, that the ‫ אשׁר‬clause in Exod 14:13 is analogous to a ‫ כאשׁר‬clause (see below, 18.  See Exod 4:13; 1 Kgs 22:24; Isa 31:1; 41:24; 54:1; 63:19 (2×); 65:1 (2×); Jer 2:8, 11; Mal 2:16; Ps 15:3–5 (8×); 16:4; 65:5 (3×); 66:9; 141:9; Job 18:21; 29:12, 16; 38:19, 24 (2×); Prov 8:32; 9:16; 30:15, 18, 29; Eccl 1:18; 2:1, 26; Lam 1:14; 4:14; Neh 8:10; 1 Chr 15:12; 2 Chr 1:4; 16:9; 31:16.

118

Chapter 4

§4.3.2) ‘( just) as/when’, avoids directly addressing the structure and function of the simple ‫ אשׁר‬clause. On the basis of similar observations, Vervenne examines the function of the second ‫ אשׁר‬clause in Exod 14:13 and proposes a third option: the ‫ אשׁר‬clause in (233) is an “initial superordinate clause” (Vervenne 1995: 197), by which I understand him to mean a null-head relative. This conclusion is the simplest analysis, since it avoids treating the ‫ אשׁר‬clause as a more complex ‫ כאשׁר‬clause, and a grammatically accurate analysis, since it avoids the implicit (and typologically unlikely) claim that Hebrew employs internally-headed relatives. 4.3.1.  Excursus on Hebraist Terminology for Null-Head Relatives All reference grammars that I consulted at least mention the phenomenon of null-head relatives, though they do not use that language, nor do they describe the issues as I have done. Since there is more diversity on this topic than on most concerning Hebrew relativization, the following survey may be of some use to readers who are also consulting the typical reference works. Ewald 1891 labels the two types of relative clauses independent relatives and dependent relatives, respectively. Ewald’s distinction, however, is not as clear as it may seem since he defines both types based on the function of the relative word rather than the function of the relative clause as a whole. Thus, an independent relative sentence is one that proceeds “from a word [e.g., ‫]אשׁר‬ which indicates a person or a thing (qui, or, with less indication of life, quod . . .)” (Ewald 1891: 207). In contrast, a dependent relative clause is one in which the relative word “merely serves to gather up a thought and show the relation in which it stands, i.e., a conjunction, as, gaudeo quod vales” (Ewald 1891: 207). It is difficult to ascertain precisely how Ewald can view the indeclinable BH ‫ אשׁר‬as both a conjunction and a referential word indicating “a person or thing” (Ewald 1891: 207). Unlike Ewald, Kautzsch distinguishes between “complete” and “incomplete” relative clauses based on whether the relative clause is dependent on a noun or not, respectively (GKC 485). In Kautzsch’s view, the ‫ אשׁר‬preceding incomplete relative clauses (i.e., those not dependent on a governing substantive) is “in reality . . . still a demonstrative belonging to the construction of the main clause as subject or object” (GKC 445; note Kautzsch’s bracketed qualification of this statement on p. 446, however; it is not quite clear how Kautzsch understands the syntax of ‫—אשׁר‬as a demonstrative or as a relative pronoun). Likewise, Waltke and O’Connor distinguish two types of relative clause: dependent (or attributive) relative clauses and independent relative clauses (IBHS 331). However, they qualify the label “independent relative” as a con-

Relative Heads

119

tradiction in terms; in their view, such relative clauses “are not relative to anything” (IBHS 331). Rather, this type of “relative” clause “functions as a principal part of the main verbal clause.” Finally, rather than providing neat categories or labels, Joüon and Muraoka merely present the ‫אשׁר‬-relative clauses that are used after nouns (GBH 558–62) in contrast to the ‫אשׁר‬-relative clauses that are “substantival” (GBH 562–63). According to Joüon and Muraoka, “In all these cases ‫ אׁשר‬has the effect of a relative pronoun used absolutely like Lat. qui, quem, quod etc.; the one who, that which etc.” Though the style of presentation differs among all of the ancient Hebrew grammars, there does appear to be general agreement regarding relative clauses with overt relative heads and those without heads (or, more accurately, with null heads). 4.3.2.  Preposition/Conjunction + Relative Word Combinations Before leaving behind the use of null-head relatives in Hebrew, we must examine one final type of construction: the comparatively frequent use of a preposition or subordinating (clitic) conjunction attached directly to a relative element, such as ‫( כאשׁר‬234) or ‫( יען אשׁר‬235). 19 (234) ‫יתי ֵּכן יֵ ָע ֶׂשה ָל ֶהם‬ ִ ‫ַּכ ֲא ֶׁשר ָע ִׂש‬ ‘Like-that I did, thus will be done to them’ (Ezek 12:11) (235) ‫יַ ַען ֲא ֶׁשר ִמ ֵּלא ַא ֲח ֵרי יְ הוָ ה‬ ‘because-that he wholly followed after Yhwh’ (Deut 1:36) Since Hebrew does not allow pied-piping (see §5.2.2), the preposition and conjunction in each example is a constituent of the matrix clause, not of the relative clause. The question that remains is how to analyze the ‫ כאשׁר‬and ‫יען אשׁר‬ and numerous other cases with similar prepositions and conjunctions. In light of the syntax of ‫ אשׁר‬that I have so far described, as well as the phenomenon of null heads, we have two clear syntactic options. Either the preposition + ‫ אשׁר‬collocations as in (234) are null-head relatives in which the null head is the complement of the preposition or the ‫ אשׁר‬functions not as a relative item but as a more general nominalizer, and the ‫ אשׁר‬clause is itself the complement of the preposition. The same basic two options exist for conjunction + ‫אשׁר‬ collocations as in (235): either there is a null head that is the clitic host of the conjunction, or the nominalized ‫ אשׁר‬clause is the clitic host. 19.  Note that all the prepositions and conjunctions are cliticized to the following ‫אשׁר‬. I signal this in my translation by the hyphen (-) joining the two items.

120

Chapter 4

The only relevant test for determining whether the relative or nominalized clause analysis is more likely is to examine the subordinate clause itself. If it is complete without assuming the role of a raised relative head, then the nominalized clause analysis is simpler. However, if the subordinate clause does not seem syntactically or semantically complete without the syntactic presence of a noun that has been raised as the (null) relative head, then the relative analysis is obligatory. Using this criterion, let us examine the two examples in (234) and (235). For the ‫ כאשׁר‬example in (234), we must ask whether the clause that the ‫ אשׁר‬introduces syntactically requires any additional constituent that a co-referential trace left by a raised relative head would represent. To put it more simply, is ‫ עשׂיתי‬a complete clause by itself? If so, then the ‫ אשׁר‬nominalizes the clause to fill the role of the complement of ‫כ‬. This analysis results in a syntactically similar construction as the preposition attached to an infinitive, ‫ּכ ֲעשֹׂות‬. ַ But ‫עשׂ‬ ‫ ה‬is not complete by itself: this verb is almost always bivalent and thus requires a nominal complement. Therefore, the valency of ‫ עשׂה‬strongly suggests that a null nominal complement, here the trace of a relativized noun, is syntactically present: ‘I did X’ > ‘Xi that I did ti’. A more accurate English glossing, then, would be ‘Like (the thing) that I did, thus will be done to them’. For the conjunction+‫ אשׁר‬example in (235), the question is the same: is there a null noun that hosts the clitic conjunction, or is the ‫ אשׁר‬clause itself the nominalized host? The clause following the ‫ אשׁר‬provides the test: is the verb ‫ מלא‬complete without an additional constituent—a relativized noun—playing a syntactic role? Unlike the ‫ כאשׁר‬example in (234), the answer in this case is yes: the ‫ מלא‬has its valency filled without any additional complement (though ‫ מלא‬is typically bivalent with a nominal complement, in this example we see the idiomatic phrase ‫מלא אחרי‬, which leaves no room for another complement of the verb). The conclusion for ‫ יען אשׁר‬in (235) is that the ‫ אשׁר‬does not introduce a null-head relative; rather, the ‫ אשׁר‬nominalizes the following clause, and the whole functions as the clitic host of the clitic conjunction ‫יען‬. We can apply the same criterion—whether or not the ‫ אשׁר‬introduces a null-head relative or nominalizes the clause as the prepositional complement depends on the valency of the verb within the relative—to the full range of preposition+‫ אשׁר‬and conjunction+‫ אשׁר‬examples. Let us first consider the three clitic prepositions, ‫ב‬‎, ‫כ‬, and ‫ל‬, beginning with ‫ ב‬in (236)–(237). Semantically, the preposition -‫ ב‬most often indicates either spatial or temporal inclusion (hence the common English glosses ‘in’ or ‘at’), though there are a number of derived nuances. 20 20.  Though the dominant sense of ‫ ב‬is spatial or temporal, Waltke and O’Connor note the significant diversity of derivative nuances, including accompaniment (‘with’), instrumental (‘with, by’), exchange (‘in exchange for’), specification (‘with regard to’), norm (‘in the manner of’), capacity (‘as’), and causal (‘because’) (IBHS §11.2.5; also GBH §133c;

Relative Heads

121

(236)  21‫ּוב ֲא ֶׁשר ָּת ִלינִ י ָא ִלין‬ ַ ‘and in-that you lodge I will lodge’ (Ruth 1:16) (237) ‫ל־ה ָא ָדם‬ ָ ‫ל־ּבית ִמ ְׁש ֶּתה ַּב ֲא ֶׁשר הּוא סֹוף ָּכ‬ ֵ ‫ית־א ֶבל ִמ ֶּל ֶכת ֶא‬ ֵ ‫ל־ּב‬ ֵ ‫‏ טֹוב ָל ֶל ֶכת ֶא‬ ‘to walk to a house of mourning is better than walking to a house of feasting because-that it is the end of every man’ (Eccl 7:2) 22 In (236), the ‫ באשׁר‬clause precedes the main verb, which is ‫אלין‬. This verb, ‘to lodge’ or ‘spend the night’ typically takes a locative complement with a ‫ב‬-PP. This suggests a locative semantic nuance for the ‫ב‬, which in turn strongly suggests that there is a null locative head for the ‫ אשׁר‬clause. This is supported by the verb within the ‫ אשׁר‬clause, which is also ‫—לין‬it requires a locative complement, though none is overt. Therefore, there must be a null locative relative head that satisfies both the valency of the preposition ‫ ב‬and the valency of the verb ‫ תליני‬within the ‫ אשׁר‬clause. Thus, ‘in the place where you lodge, I will lodge’. In contrast to (236), the ‫ באשׁר‬in (237) does not require a null head. The null-copula clause within the ‫ אשׁר‬clause, ‫הוא סוף כל אדם‬, is semantically complete—it is not lacking any necessary syntactic roles. There is therefore no need for a co-referential relative head to have a role within the clause. The larger context of the ‫ באשׁר‬clause in (237) suggests that the ‫ ב‬most likely has a causal nuance. Since there is nothing in the semantics of causal ‫ ב‬that requires a nominal complement (unlike a spatial ‫)ב‬, these observations taken together indicate that the best analysis for example (237), and others like it, is that the ‫ אשׁר‬clause is a nominalized clausal complement of the ‫ב‬, not a null-head relative. Applying this test to the 19 ‫ באשׁר‬examples in the Hebrew Bible (see n. 21) results in 14 null-head relatives (Gen 21:17; 23 Judg 5:27; 17:8, 9; 1 Sam 23:13; 2 Kgs 8:1; Isa 47:12; 56:4; 65:12; 66:4; Job 39:30; Ruth 1:16, 17; Eccl 3:9) and 5 ‫ אשׁר‬clause complements of the preposition (Gen 39:9, 23; Jonah 1:8; 24 Eccl 7:2; 8:4; all represent the causal use of ‫)ב‬. GKC §119h–q). Besides the variety of prepositional uses, ‫ ב‬can also be selected by a verb as its complement. 21.  See Gen 21:17; 39:9, 23; Judg 5:27; 17:8, 9; 1 Sam 23:13; 2 Kgs 8:1; Isa 47:12; 56:4; 65:12; 66:4; Jonah 1:8; Job 39:30; Ruth 1:16, 17; Eccl 3:9; 7:2; 8:4. 22.  Pat-El 2013b limits the meaning of ‫ ב‬to ‘in, on’ and explicitly denies any causal meaning, though she does not cite any support. On the causal use of ‫ב‬, see the references cited above in n. 20. 23.  In Gen 21:17, there is no additional syntactic position for the null head of the relative; however, this is due to the overt resumption of the locative relative head by the locative adverb ‫שׁם‬. 24.  Jonah 1:8 (‫י־ה ָר ָעה ַהּזֹאת ָלנּו‬ ָ ‫‏הּגִ ָידה־ּנָ א ָלנּו ַּב ֲא ֶׁשר ְל ִמ‬ ַ ) is difficult to analyze without ignoring a maqqep connecting ‫ באשׁר למי‬prosodically (and thus probably phrasally) to ‫הרעה‬ ‫הזאת לנו‬. There simply must be a full clause following the ‫אשׁר‬, which means a null copula. And the only syntactic analysis that makes any sense is to take the ‫ הרעה הזאת‬as the subject

122

Chapter 4

Unlike ‫ב‬, the preposition ‫ כ‬does not carry a basic temporal or spatial sense; rather, Waltke and O’Connor suggest that it has three basic uses: (1) agreement in quantity or measure (including approximation); (2) agreement in kind (i.e., comparison); and (3) correspondence (IBHS 202–5). 25 In the 511 occurrences of ‫ כ‬attached to ‫אשׁר‬, the preposition appears to use two of these nuances: comparison or approximation/correspondence in time. The example in (238), repeated from (234), illustrates the comparative use while the example in (239) illustrates the temporal use. (238) (= [234]) ‫יתי ֵּכן יֵ ָע ֶׂשה ָל ֶהם‬ ִ ‫ַּכ ֲא ֶׁשר ָע ִׂש‬ ‘like-(the thing) that I did, thus will be done to them’ (Ezek 12:11) (239) ‫ר־תּמּוּ ָכּל־אַנְ ֵשׁי ַה ִמּ ְל ָח ָמה ָלמוּת ִמ ֶקּ ֶרב ָה ָעם׃‬ ַ ‫וַ יְ ִהי ַכ ֲא ֶשׁ‬ ‘and it was, when-that all the men of war had completely died from the midst of the people, (Yhwh spoke to me)’ (Deut 2:16) The null-head relative analysis for (238) was discussed above for (234), and, indeed, most examples of comparative ‫ כ‬combined with an ‫ אשׁר‬clause can be compellingly analyzed as a null-head relative. 26 More often than not, and the ‫ לנו‬as the copular complement (see Trible 1994: 138; many translations and commentators fill in the null copula with some form of “come,” e.g., Tucker 2006: 26; or opt for some more periphrastic rendering, e.g., nrsv, njps; Allen 1976: 206; Stuart 1987: 454). The initial ‫ ב‬is causal, and its complement is the ‫ אשׁר‬nominalized clause, which is a one-part null-copula clause, ‘it is for whom?’. In this analysis, which accounts for all the necessary syntactic components, the question the sailors ask Jonah is not “why” or “because of whom” has the storm happened, but more specifically “for whom?” The question betrays their sensitivity to divine wrath—they know that someone on the boat is being chased by divine wrath. 25.  Waltke and O’Connor propose that what is sometimes labeled the temporal use of ‫כ‬ is in fact related to either approximation (‘about that time’) or correspondence (‘at the [same] time’; IBHS 205). 26.  See Gen 7:9, 16; 8:21; 12:4, 11; 17:23; 18:5; 21:1 (2×), 4; 24:51; 27:4, 9, 14, 19; 34:12, 22; 40:22; 41:13, 21, 54; 43:17; 44:1; 47:11; 50:6, 12; Exod 1:12, 17; 2:14; 5:13; 7:6, 10, 13, 20, 22; 8:11, 15, 23; 9:12, 35; 12:25, 28, 32, 50; 13:11; 16:24, 34; 17:10; 21:22; 23:15; 27:8; 34:4; 39:1, 5, 7, 21, 26, 29, 31, 43; 40:19, 21, 23, 25, 27, 29, 32; Lev 4:20; 8:4, 9, 13, 17, 21, 29, 31, 34; 9:7, 10, 21; 10:5, 15, 18; 16:15, 34; 24:19, 23; Num 1:19; 2:33; 3:16, 42, 51; 5:4; 8:3, 22; 14:17, 28; 15:14, 36; 17:5, 12, 26; 20:9, 27; 21:34; 22:8; 23:2, 30; 26:4; 27:11, 22, 23; 31:7, 31, 41, 47; 32:25, 27; 33:56; 36:10; Deut 1:11, 19, 21; 2:1, 12, 14, 22, 29; 3:2, 6; 4:5, 33; 5:12, 16, 32; 6:3, 19, 25; 9:3; 10:5, 9; 11:25; 12:20; 13:18; 15:6; 18:2; 19:8, 19; 20:17; 22:26; 23:24; 24:8; 26:15, 18, 19; 27:3; 28:9; 29:12 (2×); 31:3, 4; 34:9; Josh 1:3; 4:8 (2×), 12, 23; 6:22; 8:2, 5, 6, 31, 33; 9:21; 10:1, 28, 30, 39, 40; 11:9, 12, 15, 20; 13:6, 8, 14, 33; 14:2, 5, 7, 10, 12; 21:8; 22:4; 23:5, 8, 10, 15; 24:5; Judg 1:7, 20; 2:15 (2×), 22; 6:27a, 36, 37; 7:17; 9:33; 11:36; 15:10, 11; 1 Sam 2:16, 35; 17:20; 24:5; 28:17; 2 Sam 3:9; 5:25; 7:10, 15, 25; 13:29; 15:26; 16:23; 24:19; 1 Kgs 1:30; 2:31, 38; 3:6; 5:19, 26; 8:20; 9:5; 11:38; 12:12; 20:34; 21:11 (2×); 2 Kgs 7:7, 10, 17; 8:18, 19; 15:9; 17:23, 26, 41; 21:3, 13, 20; 24:13; Isa 10:11; 11:16; 14:24 (2×); 24:2; 25:11; 51:13; Jer 7:14; 13:5; 17:22; 18:4; 26:11; 27:13; 39:12; 40:3; 44:17; 50:15; Ezek 2:2; 9:11; 12:7, 11; 15:6; 16:48; 24:18; 37:7, 10; 41:25; 46:7, 12; Joel 3:5; Amos 5:14; 9:9; Obad 15; Jonah 1:14; Mic 3:3, 4; Hag 1:12; Zech 1:6; 10:6; Pss 48:9; 56:7; Job 4:8; 29:25; 42:9; Prov 24:29; Ruth 1:8; Eccl 8:7; 9:2 (2×); Esth 2:20a;

Relative Heads

123

the valency of the verb within the ‫ אשׁר‬clause, such as ‫‘ צוה‬to command’, ‫דבר‬ ‘to speak’, and ‫‘ עשׂה‬to do’, requires the (mostly null) resumption of a null head to fulfill the verbal semantics. One more example will suffice to demonstrate the analytical process for the ‫ כ‬+ null-head ‫אשׁר‬-relative: (240) ‫יוּכלוּן‬ ְ ‫ת־אַמ ְתּחֹת ָה ֲאנָ ִשׁים א ֶֹכל ַכּ ֲא ֶשׁר‬ ְ ‫ל־בּיתֹו ֵלאמֹר ַמ ֵלּא ֶא‬ ֵ ‫ת־א ֶשׁר ַע‬ ֲ ‫‏וַ יְ ַצו ֶא‬ ‫ְשׂ ֵאת‬ ‘and he commanded (the servant) that was over his house, “Fill the sack of the men (with) food like-that they are able to carry’   (Gen 44:1) In (240), the ‫ אשׁר‬clause includes both the finite verb ‫ יוכלון‬and its infinitival complement ‫שׂאת‬. While it might be tempting to understand this as a case of simple ‫ אשׁר‬nominalization, e.g., ‘fill their sacks with food as they are able to carry’, this analysis ignores the valency requirement of the infinitive from the verb ‫נשׂא‬, which is bivalent and thus requires an NP complement (see BDB, HALOT, DCH). The only analysis of ‫ כאשׁר‬that fulfills the verbal semantics within the ‫ אשׁר‬clause is a null-head relative analysis, ‘like (the amount) that they can carry (it)’. The cases in which an ‫ אשׁר‬clause following a ‫ כ‬is better understood as a nominalized clause are those in which the comparison is not about a thing (e.g., a command or specific act) but about the event as a whole, as in (241). (241) ‫‏‏וְ ִה ָכּה יְ הוָ ה ֶאת־יִ ְשׂ ָר ֵאל ַכּ ֲא ֶשׁר יָ נוּד ַה ָקּנֶ ה ַבּ ַמּיִ ם‬ ‘and Yhwh shall smite Israel, like-that a reed sways in the water’   (1 Kgs 14:15) In (241), the verb ‫ ינוד‬is monovalent—that is, it requires a subject, here ‫הקנה‬, but no complement. There is thus no syntactic necessity for the null resumption that would correspond to a null-relative head. Moreover, the comparison is between the manner in which Yhwh shall smite Israel and the manner in which a reed sways. One could propose that the null head of the ‫אשׁר‬-relative is a manner phrase, such as ‘(the way/manner) that a reed sways’ (this is the approach I advocated in Holmstedt 2002: 76–79). However, taking the ‫ אשׁר‬clause as the nominalized clausal complement of the preposition is both simpler and more economical than the null-head analysis, even in cases where a null ‘manner’ head would also make good sense. Consider another examples of this in (242): 27 6:10; 9:31 (2×); Dan 1:13; 9:12, 13; Ezra 4:3; Neh 5:12; 1 Chr 14:16; 15:15; 17:9, 23; 22:11; 24:19; 2 Chr 2:2; 6:10; 7:18; 10:12; 21:6, 7; 23:3; 29:8; 30:7; 33:22; 27.  See Gen 26:29 (2×); Exod 33:11; 40:15; Lev 4:21, 31, 35; 18:28; 24:20; 27:14; Num 2:17; 11:12; 14:19; 27:13, 14; Deut 1:31, 44; 6:16; 8:5; 12:22; 16:10; 28:29, 49, 63; 30:9; 32:50; Josh 1:5, 17; 4:14; 14:11; Judg 6:27b; 7:5; 8:8; 16:9, 22; 1 Sam 4:9; 6:6; 15:33; 20:13; 23:11; 24:14; 26:20, 24; 2 Sam 10:2; 16:19; 19:4; 1 Kgs 1:37; 3:14; 8:25, 57; 9:2, 4; 14:10, 15; 2 Kgs 2:19; 5:26; 10:15; 23:27; Isa 9:2; 10:10; 20:3; 31:4; 52:14; 55:10; 65:8; 66:20, 22; Jer 2:36; 5:19; 7:15; 12:16; 13:11; 19:11; 23:27; 31:28; 32:42; 42:2, 18; 43:12; 44:13, 30;

124

Chapter 4

(242) ‫ל־ר ֵעהוּ‬ ֵ ‫ל־פּנִ ים ַכּ ֲא ֶשׁר יְ ַד ֵבּר ִאישׁ ֶא‬ ָ ‫וְ ִד ֶבּר יְ הוָ ה ֶאל־מ ֶֹשׁה ָפּנִ ים ֶא‬ ‘and Yhwh would speak to Moses, face to face, like-that a man speaks to his friend’ (Exod 33:11) Though the null-head analysis ‘like (the way/manner) that a man speaks to his friend’ makes sense, it is also grammatically felicitous, and linguistically more economical, to take the event of companions speaking as the object of comparison; hence the ‫ אשׁר‬clause as the complement of the ‫כ‬. When the preposition ‫ כ‬has a temporal sense, as in (239), the nominalization analysis (versus the null-head relative analysis, i.e., ‘[the time] that . . .’), is always simpler and thus preferred. Since verbs rarely (if ever) require a temporal complement, the much simpler analysis of the temporal ‫כ‬‎ + ‫אשׁר‬ clause examples is that the ‫ אשׁר‬nominalizes the following clause to function as the complement of the preposition. 28 Out of the 511 occurrences of ‫כאשׁר‬ in the Hebrew Bible, 329 are best analyzed as null-head relatives (see above, n. 26), and 182 are more simply taken as nominalized complements of the ‫כ‬ (see above, nn. 27–28). The third clitic preposition that attaches to the relative word is the preposition ‫ל‬. Hebrew ‫ ל‬is often glossed with English ‘to’ or ‘for’, and the semantic range of ‫ ל‬is about as diverse as its English counterparts. Unlike ‫ כ‬and ‫ב‬, however, there are no clear examples of ‫ לאשׁר‬among the 39 occurrences in the Hebrew Bible that are nonrelative. Instead, all these cases of ‫ ל‬select a non-event nominal complement—i.e., a noun. If the ‫ ל‬carries a purpose or result meaning and selects an event complement, it is attached to an infinitive complement, never a full (nominalized) clausal complement. Example (243) illustrates ‫ ל‬preceding ‫ אשׁר‬and provides a contextually appropriate rendering of the null-relative head. (243) ‫ְל ַמ ֲע ֵׂשה יָ ָדיו יִ ְׁש ַּת ֲחוּו ַל ֲא ֶׁשר ָעׂשּו ֶא ְצ ְּבע ָֹתיו‬ ‘to the work of (their) hands they bow down, to-∅/(the work/idols) that (their) fingers have made’ (Isa 2:8) 29 48:13; 50:18; Ezek 1:16; 10:10; 16:50, 59; 20:36; 23:18; 24:22; 43:22; 48:11; Amos 2:13; 3:12; 5:19; Obad 16; Zech 7:3; 8:13, 14; 14:5; Mal 3:17; Job 10:19; Eccl 4:17; 5:14; 11:5; Lam 1:22; Esth 2:20b; Neh 3:33; 4:1, 6, 9; 5:6; 6:1, 3, 16; 7:1; 13:19; 1 Chr 17:1, 13; 2 Chr 6:16; 7:17. 28.  See Gen 18:33; 20:13; 24:22, 52; 27:30, 40; 29:10; 30:25; 32:3, 32; 37:23; 40:14; 43:2, 14; Exod 10:10; 17:11 (2×); 32:19; Lev 4:10; Deut 2:16; Josh 2:7; 3:7; 4:1, 11; 5:8; Judg 3:18; 8:33; 11:5, 7; 1 Sam 1:24; 8:1, 6; 12:8; 24:2; 28:18; 2 Sam 12:21; 16:16; 17:12; 20:12, 13; 1 Kgs 2:24; 2 Kgs 14:5; Isa 23:5; 26:9; 29:8 (2×); Jer 17:22; 38:28; 39:4; Ezek 35:11; 37:18; Hos 7:12; 9:13; Zech 7:13; Pss 33:22; 51:2; Job 29:4; Eccl 5:3; 8:16; Esth 4:16; 2 Chr 25:3. 29.  See Gen 27:8; 43:16; 44:4; 47:24; Exod 16:16; Lev 5:24; 27:24 (2×); Num 5:7; Josh 17:16 (2×); Judg 21:5; 1 Sam 30:27 (3×), 28 (3×), 29 (3×), 30 (3×), 31; 2 Kgs 10:22; Isa

Relative Heads

125

The preposition ‫ מן‬may also be attached to ‫( אשׁר‬in which case, it becomes a proclitic ‫)מ‬. As with the other clitic prepositions, ‫ מן‬exhibits diverse nuances, from spatial and temporal to cause and comparison (IBHS §11.2.11; 212–14). In the 17 examples with ‫אשׁר‬, we find 4 meanings used: spatial (‘from’), as in (244); partitive (‘one/some of’), as in (245); comparative, as in (246); and causal, as in (247). (244) ‫ל־ה ָּכבֹד ַהּזֶ ה‬ ַ ‫ּומ ֲא ֶׁשר ְל ָא ִבינּו ָע ָׂשה ֵאת ָּכ‬ ֵ ‫ל־א ֶׁשר ְל ָא ִבינּו‬ ֲ ‫ָל ַקח יַ ֲעקֹב ֵאת ָּכ‬ ‘Jacob has taken all that belonged to our father, and from-∅(the property) that (it) belonged to our father he has made all of this wealth’ (Gen 31:1) 30 (245) ‫ן־ּבנֵ י ַהּיֹונָ ה ֵמ ֲא ֶׁשר ַּת ִּׂשיג יָ דֹו‬ ְ ‫ן־הּת ִֹרים אֹו ִמ‬ ַ ‫ת־ה ֶא ָחד ִמ‬ ָ ‫ֶא‬ ‘one of the turtledoves or of the doves (out of-∅[those] that he can afford)’ (Lev 14:30) 31 (246) ‫יתי ִּכי ֵאין טֹוב ֵמ ֲא ֶׁשר יִ ְׂש ַמח ָה ָא ָדם ְּב ַמ ֲע ָׂשיו‬ ִ ‫וְ ָר ִא‬ ‘and I saw that nothing is better than-that man rejoices in his deeds’   (Eccl 3:22) 32 (247) ‫ל־הּנָ ֶפׁש‬ ַ ‫ֵמ ֲא ֶׁשר ָח ָטא ַע‬ ‘because-that he sinned concerning the dead’ (Num 6:11; see also   Isa 43:4) In both (244) and (245), the internal structure of the ‫ אשׁר‬clause (i.e., the need for a subject in [244] and the valency of the verb ‫ תשׂיג‬in [245]) as well as the context suggest that both are null-head relatives. In contrast, lack of any clear syntactic role of a null head within the ‫ אשׁר‬clauses in (246) and (247) suggest that these are nonrelative ‫ אשׁר‬nominalized clauses. There are a variety of other prepositions and conjunctions that occur with ‫ אשׁר‬clauses throughout the Hebrew Bible, though none are used frequently. For prepositions that typically take a locative NP complement, such as ‫ אל‬in (248) and ‫ תחת‬in (249), the null-head relative analysis is the most sensible. 33

2:8; 8:23; 31:6; 49:9; Jer 27:5; 38:20; 50:20; Ezek 23:40; Amos 6:10; Mal 3:18; Job 12:6; Eccl 9:2. 30.  See Gen 31:1; Exod 5:11; 29:27 (2×); Isa 47:13; Ruth 2:9; Esth 4:11; 1 Chr 17:13. 31.  See Lev 14:30; Jer 40:7. 32. The ‫ מאשׁר‬in Josh 10:11, Judg 16:30, 2 Sam 18:8, and 2 Kgs 6:16 also includes a comparative ‫מן‬, but the internal structure of the ‫ אשׁר‬clause of each example suggests a null-head relative analysis, e.g., ‘many more died by hailstones than-∅ (those) who(m) the Israelites killed by the sword’ (Josh 10:11). 33.  On the one example of ‫עם אשׁר‬, Gen 31:32, see below, §5.2.2, n. 27.

126

Chapter 4

(248) ‫ר־ּד ַּב ְר ִּתי ָלְך‬ ִ ‫ת־ה ָעם ֶאל ֲא ֶׁש‬ ָ ‫‏וְ ַע ָּתה ֵלְך נְ ֵחה ֶא‬ ‘and so, go, lead the people to-∅(the place) that I spoke to you’   (Exod 32:34) 34 (249) ‫צּומים יְ ַח ֵּלק ָׁש ָלל ַּת ַחת ֲא ֶׁשר ֶה ֱע ָרה ַל ָּמוֶ ת נַ ְפׁשֹו‬ ִ ‫ת־ע‬ ֲ ‫ָל ֵכן ֲא ַח ֶּלק־לֹו ָב ַר ִּבים וְ ֶא‬ ‘therefore I will give him a portion among the many and with the mighty he will apportion plunder, in return ∅/(for the fact) that he emptied his life to death’ (Isa 53:12) 35 For those prepositions whose complements are less semantically restricted, the nominalized complement clause analysis for the ‫ אשׁר‬clause is the simplest and thus most elegant analysis. Such examples include ‫ אשׁר‬used with ‫‘ על‬upon, on’ > ‘because’ (250), ‫‘ עד‬until, while’ (251)–(252), ‫‘ אחרי‬after, behind’ (253), and the complex prepositions ‫‘ מפני‬on account of, because’ (254). (250) ‫ת־אּיֹוב‬ ִ ‫א־מ ְצאּו ַמ ֲענֶ ה וַ ּיַ ְר ִׁשיעּו ֶא‬ ָ ֹ ‫ֹלׁשת ֵר ָעיו ָח ָרה ַאּפֹו ַעל ֲא ֶׁשר ל‬ ֶ ‫ּוב ְׁש‬ ִ ‘and his anger burned at his three friends because-that they could not find an answer but condemned Job’ (Job 32:3) 36 (251) ‫יְמי ָה ָר ָעה‬ ֵ ‫ימי ְּבחּור ֶֹתיָך ַעד ֲא ֶׁשר לֹא־יָ בֹאּו‬ ֵ ‫ת־ּבֹור ֶאיָך ִּב‬ ְ ‫ּוזְ כֹר ֶא‬ ‘and remember your creator in the days of your youth, while-that the days of calamity have not come’ (Eccl 12:1) 37 (252) ‫אתי וַ ִּת ְר ֶאינָ ה ֵעינַ י‬ ִ ‫ר־ּב‬ ָ ‫יהם ַעד ֲא ֶׁש‬ ֶ ‫א־ה ֱא ַמנְ ִּתי ְל ִד ְב ֵר‬ ֶ ֹ ‫וְ ל‬ ‘and I did not believe their words until-that I came and my eyes saw’ (2 Chr 9:6) 38 34.  See Exod 32:34; Num 33:54; Ezek 1:12; 42:14; Ruth 1:16. 35. For ‫ תחת‬with the nuance of ‘in return for’ or ‘as recompense for’ as an extension of the core ‘in place of’, see Gen 30:15; 44:4 (HALOT s.v., 3b). For ‫תחת אשׁר‬, see Num 25:13; Deut 21:14; 22:29; 28:47, 62; 1 Sam 26:21; 2 Kgs 22:17; Isa 53:12; Jer 29:19; 50:7; Ezek 36:34; 2 Chr 21:12; 34:25. 36.  See IBHS 218 for a discussion of the preposition ‫ ַעל‬to introduce a cause. For ‫על‬ ‫ אשׁר‬that should be anlayzed as ‫ על‬+ ∅ head + ‫אשׁר‬-relative, see Gen 47:6; Exod 16:5; 1 Sam 30:14; 2 Sam 15:20; 1 Kgs 18:12; Isa 29:12; Jer 15:4; Ezek 1:20; Hag 1:11; Ps 119:49. For ‫ על אשׁר‬that are a preposition+complement sequence, see Exod 32:35; Num 20:24; Deut 29:24; 32:51 (2×); 1 Sam 24:6; 30:14; 2 Sam 3:30; 6:8; 8:10; 12:6; 21:1; 1 Kgs 9.9; 16:7; 2 Kgs 18:12; 22:13; Jer 16:11; 22:9; Ezek 23:30; 35:15; 39:23; Job 32:3; Esth 1:15; 8:7; 1 Chr 13:10; 18:10; 2 Chr 7:22; 34:21. For ‫על ְּד ַבר ֲא ֶשׁר‬, ַ lit., ‘because of the fact that’, see Deut 22:24 (2×); 23:5; 2 Sam 13:22. See appendix A for a full list of the complementizer examples, with translation. 37.  See IBHS 215 for a discussion of the preposition ‫“ ַעד‬to mark the time before which an event takes place.” The collocation of ‫ עד‬and the negative ‫לא‬, i.e., ‘while/during not X event/ action’ in (251) results in the sense of ‘before X event/action’. See also Eccl 12:1, 2, 6; Neh 4:5. 38.  See Gen 27:44; 28:15; 29:8; 33:14; Exod 23:30; 24:14; 32:20; Lev 22:4; Num 11:20; 20:17; 21:22; 32:17; Deut 2:14, 29; 3:20; 9:21; Josh 1:15; 3:17; 8:26; 17:14; Judg 4:24; 1 Sam 22:3; 30:4; 2 Sam 17:13; 1 Kgs 10:7; 17:17; 2 Kgs 17:20, 23; 21:16; Isa 6:11; Ezek 34:21; Hos 5:15; Jonah 4:5; Mic 7:9; Ps 112:8; Ruth 1:13; 3:18; Eccl 2:3; Neh 2:7; 1 Chr 19:5; 2 Chr 9:6.

Relative Heads

127

(253) ‫ר־ׁש ְּל ָחּה ָלׁשּוב ְל ַק ְח ָּתּה ִל ְהיֹות לֹו ְל ִא ָּׁשה ַא ֲח ֵרי‬ ִ ‫א־יּוכל ַּב ְע ָלּה ָה ִראׁשֹון ֲא ֶׁש‬ ַ ֹ‫ל‬ ‫ֲא ֶׁשר ֻה ַּט ָּמ ָאה‬ ‘her first husband who cast her out shall not be able to return to take her to be his wife after-that she has been defiled’ (Deut 24:4) 39 (254) ‫וְ הַר ִסינַ י ָע ַׁשן ֻּכּלֹו ִמ ְּפנֵ י ֲא ֶׁשר יָ ַרד ָע ָליו יְ הוָ ה ָּב ֵאׁש‬ ‘and Mount Sinai—all of it was smoking because-that Yhwh had descended upon it in fire’ (Exod 19:18) 40 As in example (235), when ‫ אשׁר‬is preceded by the conjunctions ‫‘ יען‬because’ (255), ‫‘ למען‬for the purpose of’ (256), and ‫‘ עקב‬consequence’ (257), the ‫ אשׁר‬marks the nominalized complement, or if the valency of these particles is null, then the ‫ אשׁר‬clause is simply the nominalized clausal clitic host. (255) ‫ה־ח ְברֹון ְל ָכ ֵלב ֶּבן־יְ ֻפּנֶ ה ַה ְּקנִ ּזִ י ְלנַ ֲח ָלה ַעד ַהּיֹום ַהּזֶ ה יַ ַען ֲא ֶׁשר ִמ ֵּלא‬ ֶ ‫ל־ּכן ָהיְ ָת‬ ֵ ‫ַע‬ ‫ֹלהי יִ ְׂש ָר ֵאל‬ ֵ ‫ַא ֲח ֵרי יְ הוָ ה ֱא‬ ‘because of this, Hebron became an inheritance for Caleb, son of Jephunneh the Kenizzite, until this very day because-that he was fully after Yhwh, the God of Israel’ (Josh 14:14) 41 (256) ‫ל־ּפ ֶטר ָר ַחם ְל ַמ ַען ֲא ִׁש ֵּמם ְל ַמ ַען ֲא ֶׁשר יֵ ְדעּו‬ ֶ ‫נֹותם ְּב ַה ֲע ִביר ָּכ‬ ָ ‫אֹותם ְּב ַמ ְּת‬ ָ ‫וָ ֲא ַט ֵּמא‬ ‫ֲא ֶׁשר ֲאנִ י יְ הוָ ה‬ ‘and I defiled them through their gifts when (they) devoted every firstborn in order to horrify them for the purpose that they would know that I am Yhwh’ (Ezek 20:26) 42 (257) ‫ת־ה ָּד ָבר ַהּזֶ ה וְ ַעל ֲא ֶׁשר‬ ַ ‫ת־ה ִּכ ְב ָׂשה יְ ַׁש ֵּלם ַא ְר ַּב ְע ָּתיִ ם ֵע ֶקב ֲא ֶׁשר ָע ָׂשה ֶא‬ ַ ‫וְ ֶא‬ ‫א־ח ָמל‬ ָ ֹ‫ל‬ ‘and the lamb he shall restore fourfold as a consequence that he did this thing and on account that he did not show pity’ (2 Sam 12:6) 43 Additionally, ‫ שׁ‬clauses may be preceded by PPs, as in the two representative examples, (258) and (259). 44

39.  See Deut 24:4; Josh 7:8; 9:16; 23:1; 24:20; Judg 11:36; 19:23; 2 Sam 19:31. Also with ‫אחר‬, see Ezek 40:1; Ruth 2:2. 40.  See Exod 19:18; Jer 44:23. 41.  See Gen 22:16; Deut 1:36; Josh 14:14; Judg 2:20; 1 Sam 30:22; 1 Kgs 3:11; 8:18; 11:11, 33; 14:7, 15; 16:2; 20:28, 36; 2 Kgs 1:16; 10:30; 21:11, 15; Jer 19:4; 25:8; 29:23, 25, 31; 35:18; Ezek 12:12; 16:43; 21:9; 26:2; 31:10; 44:12; Ps 109:16; 2 Chr 1:11; 6:8. 42.  See Gen 18:19; Lev 17:5; Num 17:5; Deut 20:18; 27:3; Josh 3:4; 2 Sam 13:5; Jer 42:6; Ezek 20:26; 31:14; 36:30; 46:18. 43.  See Gen 22:18; 26:5; 2 Sam 12:6. 44. For ‫בשׁ‬, see Gen 6:3; Jonah 1:7, 12; Eccl 2:16; 8:17; for ‫כשׁ‬, see Eccl 5:14; 9:12; 10:3; 12:7; for ‫משׁ‬, see 2 Kgs 6:11; Eccl 5:4; for ‫עד שׁ‬, see Judg 5:7; Ps 123:2; Song 1:12; 2:7, 17; 3:4, 5; 4:6; 8:4; for ‫על שׁ‬, see 1 Chr 27:27; for ‫עם שׁ‬, see Eccl 1:11; 6:10.

128

Chapter 4

(258) ‫ל־ה ָא ֶרץ ְּכ ֶׁש ָהיָ ה‬ ָ ‫וְ יָ ׁש ֹב ֶה ָע ָפר ַע‬ ‘and the dust returns to the earth like-∅/(the condition/state) that it had been (Eccl 12:7) (259) ‫ָח ְדלּו ְפ ָרזֹון ְּביִ ְׂש ָר ֵאל ָח ֵדּלּו ַעד ַׁש ַּק ְמ ִּתי‬ ‘peasantry in Israel ceased; it ceased until-that I arose’ (Judg 5:7) The same criterion applies to these clauses that I applied above to sort out the null-head relatives from the nominalized ‫ אשׁר‬clauses. There are no examples of this phenomenon with ‫ז‬-relatives. 4.4.  Hierarchy, Linear Order, Determiners, and Case-Marking in Hebrew Relatives Typologically, the hierarchical relationship of the relative head and relative clause, the linear position of the relative head vis-à-vis the relative clause, the position of the determiner with respect to the head, and the position of casemarkers are all distinguishing features. The usage in Hebrew with regard to these cross-linguistic categories is straightforward. Example (260), repeated from (223), illustrates that Hebrew relatives are externally-headed. There is no example of an internally-headed Hebrew relative. (260) (= [223]) Hebrew Externally-Headed Relative Clause ‫ל־ה ְּד ָב ִרים ֲא ֶׁשר ָע ָׂשה‬ ַ ‫וַ יְ ַס ֵּפר ָה ֶע ֶבד ְליִ ְצ ָחק ֵאת ָּכ‬ ‘and the servant related to Isaac all the things that he had done’   (Gen 24:66) Hebrew prohibits the final position for the determiner but allows either the initial (261) and middle (262) determiner strategies since nouns in general may be determined by a prefixed article or an enclitic possessive pronoun. (261) Initial Position of the Determiner in Hebrew Relative Clauses ‫אכה ֲא ֶׁשר ָע ָׂשה ַה ֶּמ ֶלְך ְׁשֹלמֹה‬ ָ ‫ל־ה ְּמ ָל‬ ַ ‫וַ ִּת ְׁש ַלם ָּכ‬ ‘and all the work that the king Solomon had done was complete’   (1 Kgs 7:51) (262) Middle Position of the Determiner in Hebrew Relative Clauses ‫אכּתֹו ֲא ֶׁשר ָע ָׂשה‬ ְ ‫יעי ְמ ַל‬ ִ ‫ֹלהים ַּבּיֹום ַה ְּׁש ִב‬ ִ ‫וַ יְ ַכל ֱא‬ ‘and God finished on the seventh day his work that he had done’   (Gen 2:2) Finally, Hebrew does not exhibit morphological case-marking on the relative word, the relative head or the relative clause. Moreover, these features of relative clause structure appear not to point toward any insightful typological universals or tendencies.

Chapter 5

Relative Syntax 5.1.  Relative Syntax 5.1.1.  Typological Features The syntactic diversity of relative clause structures attested cross-linguistically is typically sorted by two primary criteria: embeddedness and resumption (see §1.1 and table 1.1). The first criterion concerns whether the relative clause is embedded within the phrasal structure headed by the noun it modifies or is directly adjoined to the larger clause. Those that are embedded may precede, follow, or surround the relative head (see §4.1.3 and table 4.1). English relatives, as in (263), are of the embedded variety, with the relative clause contained and right-adjoined within the larger (DP) phrase structure of the noun it modifies. (I have included brackets to clarify the embedded nature of the English relative.) (263)  I saw [DP the [NP dog [RC that was black]]] In (263), the English determiner the marks the beginning of the DP, with the relative clause that was black adjoined to the right of the relative head dog. Relative clauses that are not embedded are known as correlatives. In this structure, the relative is left-adjoined and separated from its correlate in the matrix clause (de Vries 2002: 17). The correlate in the matrix clause is a demonstrative or a full copy of the modified noun, and the modified noun inside the relative clause is accompanied by a specific correlative marker (Nikolaeva 2006: 503). The Hindi examples in (264) and (265) illustrate correlative relatives, and the scheme in (266) clarifies the structure. 1 1.  Though another type of Hindi relative in (a) is also sometimes called a correlative, Srivastav distinguishes the type in (264) and (265) from those in (a): (a) vo laRkii lambii hai [jo khaRii hai]    dem girl tall is rel standing is    lit., ‘that girl is tall, who is standing’

She argues that the syntax of (264) and (265) is left-adjoined, i.e., the relative clause does not originate within the matrix clause, or specifically, within the phrase structure of the head NP/DP. The relative in the type illustrated in (a), in contrast, does originate within the matrix clause and is subsequently moved rightward, i.e., extraposed. Supporting this differentiation

129

130

Chapter 5

(264) [jo laRkii khaRii hai] vo ambii hai rel girl standing is dem tall is lit., ‘that girl is standing, that/she is tall’ (Srivastav 1991: 639;    translation slightly modified). (265) [jis aadmi ka kutta bemaar hai] us aadmi ko mai ne dekha rel man gen dog sick is dem man obj I erg saw lit., ‘which man’s god is sick, that man I saw’   (Nikolaeva 2006: 503). (266) [matrix [CP-correl [NP-rel wh NP]i . . . ti . . .] [matrix . . . Dem . . .] While there are implicationals associated with the use of correlatives, there are none for the nonuse of correlatives. 2 Hebrew uses the embedded strategy for RCs, not the correlative strategy. The second criterion for distinguishing relative-clause structure concerns resumption, which is related to the recoverability of the grammatical function of the relative head within the relative clause (Nikolaeva 2006: 503–4). Resumption is one of four strategies for indicating the modified noun, along with nonreduction, relative pronouns, and a gap. The Udihe example in (267) illustrates nonreduction, which occurs primarily in internally-headed / circumnominal relative clauses. (267) si anda-i ŋenee-ni [bi agʾa-i xoton-du badgii-tigi-ni] you friend-2sg went-3sg I brother-1sg city-loc living-lative-3sg ‘your friend went to the city where my brother lives’   (Nikolaeva 2006: 503) In (267), the relative head is within the boundary of the relative clause and may itself fill the syntactic roles within both the relative and the matrix clause. In contrast, English disallows the relative head to fill more than one syntactic role and uses a gap (268) with either a relative complementizer or a relative pronoun (269), which carries (partial) agreement features indicating the role of the relative head within the relative (see §3.1 on relative pronouns). (268) The friend that Eliana called ____ on the phone. (269) The friend who(m) Eliana called ___ on the phone. Whereas in both English examples, with or without a (partially) inflected pronoun, there is a gap within the relative clause, the Irish example in (270) between the two is the fact that the left-adjoined correlatives allow the head to be internal to the relative, whereas the other type does not. 2. De Vries (2002: 38) lists and subsequently discusses three implication universals (correlatives have internal heads, no relative affixes, and maximalizing semantics) and two implicational tendencies (correlatives use relative pronouns and not complementizers).

Relative Syntax

131

illustrates resumption by an anaphoric pronoun (note my use of subscript coindexation in the example to signal the co-referential relationship between the relative head and the resumptive pronoun). (270) an ghirseachi ar ghoid na síogaí íi the girl rel stole the fairies her ‘the girl who(m) the fairies stole’ (McCloskey 2006: 95) For the languages that use resumption by an anaphoric pronoun or coreferential adverb (e.g., locative there) or epithet, there is a related pattern of usage that corresponds to the “accessibility” of the syntactic position. This pattern follows what Comrie and Keenan called the “Noun Phrase Accessibility Hierarchy” (NPAH), given in (271) (Keenan and Comrie 1977: 66). (271) The Noun Phrase Accessibility Hierarchy Subject > Direct Object > Indirect Object > Oblique > Genitive > Object of Comparison The NPAH primarily addresses the positions from which languages can access a noun for relativization; Keenan and Comrie summarize the hierarchy as follows: On the basis of data from about fifty languages, we argue that languages vary with respect to which NP positions can be relativized, and that the variation is not random. Rather, the relativizability of certain positions is dependent on that of others, and these dependencies are, we claim, universal. The Accessibility Hierarchy (AH) . . . expresses the relative accessibility to relativization of NP positions in simplex main clauses. Here, “>” means ‘is more accessible than’. . . . The positions on the AH are to be understood as specifying a set of possible grammatical distinctions that a language may make. We are not claiming that any given language necessarily distinguishes all these categories, either in terms of RC formation or in terms of other syntactic processes. For example, some languages (e.g., Hindi) treat objects of comparison like ordinary objects of prepositions or postpositions. In such cases we treat these NPs as ordinary Obl[iques], and the O[bject of] Comp[arison] position on the AH is unrealized. (Keenan and Comrie 1977: 66)

The hierarchy also strongly predicts the positions in which a language using a resumptive strategy may use the resumptive pronouns. Keenan and Comrie note that, if a language uses resumption, it will use it first and more often in the less accessible positions “down” the NPAH, i.e., rightwards in (271) (Keenan and Comrie 1977: 92). For instance, if a given language exhibits relative resumption in the genitive and object of comparison position, it may or may not also use resumption in higher positions on the NPAH. However, if a language allows or requires resumption at the direct object position, it will not only allow but probably require resumption in each of the less accessible positions down (along the right of) the hierarchy.

132

Chapter 5

5.1.2.  Phrase Structure Features Within generative syntax, the discussion goes beyond the typological criteria I have so far described and focuses on the implications of these features for the phrase structure of relative clauses. Though the precise phrase structure of the relative clause has been and will continue to be controversial in generative studies, 3 many of the issues, both data-driven and theory-internal, are simply not relevant for ancient Hebrew (for which we have no access to the subtle semantic features by which many of the syntactic distinctions are discerned). In Holmstedt 2002, I reviewed the proposals for the structure of restrictive relatives that were current at that time (I take up nonrestrictive phrase structure in §6.1.2.1). I presented them within the configuration of relativization that was then dominant (Holmstedt 2002: 240–66). As is common in modern linguistics, shortly after I completed that work, the momentum shifted away from the analysis dominant at that time—the revised standard analysis—to another analysis that had been gaining ground through the 1990s—the raising analysis (also sometimes referred to as the promotion theory). The raising/promotion analysis has continued to be the preferred option until the time of this writing. In a linguistic nutshell, the primary difference between the revised standard analysis that I used in Holmstedt 2002 and the raising analysis is the strategy by which the relative head and the relative clause are related. In the revised standard approach, a wh-word or null operator (Op) exists in the relative clause and is fronted from a base position in the clause to the initial position; this relative operator is related to the relative head (which is generated outside the relative clause) by means of an interpretive relation (e.g., predication; Safir 1986; Fabb 1990; Haegeman 1994: 408–9). The scheme in (272) illustrates this approach to relativization, where the co-indexation between the relative head and the wh-word/(null) operator signals the interpretive relationship that is created (de Vries 2002: 73). (272) [dp [d′ D [np [n′ Ni [cp Op/whi . . . ti . . . ]]]]] In contrast, the raising/promotion analysis dispenses with the null operator and the need to connect the head and relative by an interpretive relationship; instead, the relative noun (DP) combined with a wh-word is generated within the relative and raised to the front of the relative; the wh-word remains at the front of the clause (Spec, CP), while the noun (DP) itself is further raised to a higher projection of the DP, which then functions as a constituent within the matrix clause. A simplified scheme representing this is given in (273) (for more details, see de Vries 2002: 85–86). 3.  For overviews, with critical analysis, see Alexiadou et al. 2000; Bianchi 2001; 2002; de Vries 2002: 70–86; and Hulsey and Sauerland 2006.

Relative Syntax

133

(273) [dp [d′ D [np NPi [cp whi . . . wh-DPi . . . ]]]]



Though the differences between the revised standard analysis and the promotion analysis are not critical to the analysis of the Hebrew relative clause, I must highlight one aspect of the raising analysis, that the relative clause is a complement of the determiner (D) rather than an adjunct. This may appear counterintuitive, since relative clauses, like adjectives, do not appear to be obligatory (which is the nature of a complement) and are thus often taken as adjuncts by a variety of grammarians as well as theoretical linguists (see, e.g., McCawley 1998: 381–90). However, the raising and D-complement analysis does not imply that the relative clause from which the NP was raised to become the complement of the D is obligatory. All that is implied is that the presence of a D requires an NP complement, however the NP is manifested. In the case of a relative clause, even if it is an optional adjunct, when it is inserted into the phrase, the NP is raised to a position at which it may serve as the complement to the D. The simplified relative clause scheme in (273) illustrates how this analysis coordinates with the “DP-hypothesis,” in which the NP is considered to be the complement of the determiner. Since the crosslinguistic evidence regarding the position of the determiner indicates that it must be positioned outside the head and relative, it only follows that a raised (i.e., relativized) noun along with the following clausal structure must be the determiner’s complement. 4 Whatever one thinks of the DP-hypothesis or even generative argumentation in general, it is undeniably coherent and reflects a careful consideration of cross-linguistic variation. Given the “project-as-required” phrase structure of minimalism (see §2.2.1), it is not surprising that the resulting structure of a complex constituent such as a relative clause is significantly different from a simpler constituent, such as an adjective. What is required is a study of Hebrew noun modification strategies, to discern both how the structures differ and how a unified analysis may still be provided (to confirm the intuition of most grammarians that these strategies share something in common). Until this kind of study is completed, such that implications of reclassifying, for example, the relative clause and adjective modification as complementation—or one as a complement and then the other as an adjunct—are fully spelled out, I have taken an agnostic position on the syntactic status of the relative clause. That said, in the remainder of this chapter, both out of convenience (i.e., assuming much from Holmstedt 2002) and for lack of a more compelling alternative, I position the relative as a

4.  For example, de Vries cites the circumnominal relative strategy in Mohave as the “most clear-cut piece of evidence for the D-complement hypothesis” (de Vries 2002: 76).

134

Chapter 5

left-adjoined structure within a DP, as in (274) with a Hebrew parallel in (275). Note the obvious parallels to the adjective structure in (276). (274) The man that I saw DP D NP the mani CP

whi that I saw twh-man

(275) ‫יתי‬ ִ ‫יאל ֲא ֶשׁר ָר ִא‬ ֵ ‫‏ה ִאישׁ גַּ ְב ִר‬ ָ ‘the man, Gabriel, that I saw’ (Dan 9:21) DP D NP hā ʾîši CP

whi ʾăšer raʾîtî twh-ʾîš

(276) ‫ַה ָּמאֹור ַהּגָ ד ֹל‬ ‘the big(ger) light’ (lit., ‘the light, the big[ger])’ (Gen 1:16) DP D XP ham

māʾôr NP

AP NP haɡɡādôl tmāʾôr

Relative Syntax

135

In the relatives in (274) and (275), the NPs man and ‫אישׁ‬, which are associated with a wh-feature, are raised first to the highest position of their respective clauses (the Spec,CP), where they leave the wh-feature behind before again being raised to an NP projection within the DP. If we compare this to the adjective structure, the NP projections of the relative clauses in (274) and (275) are situated in the same position as the XP in which the raised NP of the adjectival structure in (276) resides. The English relative in (274) could also appear with an overt wh-word, who or whom, in place of that, though the non-wh complementizer that is similar in status and function to the Hebrew ‫ אשׁר‬in (275). Both English that and Hebrew ‫ אשׁר‬are not initially associated with the relativized noun. Rather, both that and ‫( אשׁר‬as well as all the other Hebrew nonpronominal relative elements, ‫שׁ‬‎, ‫ה‬‎, and the ‫ז‬-relatives) are inserted into the C position of the relative CP after the relative raising, to classify the clause type overtly—though, since zero-relatives are common, this is clearly not always an obligatory insertion (and, when optional, is probably motivated by either the style or interpretive/ processing needs perceived by the author-speaker). 5.1.2.1. Resumption The linguistic literature addressing the structure and function of relative clause resumption is voluminous, in part because it relates to more-general pronoun anaphora but also because an analysis that captures the full crosslinguistic variety has been elusive. The proposals fall into three categories that emphasize a different approach: those that describe the difference primarily in syntactic terms (e.g., Shlonsky 1992; Doron and Heycock 1999; Boeckx 2003; see McCloskey 2006 for a full summary); those that suggest a dominant semantic solution (e.g., Doron 1982; Sells 1984); and finally, those that find a pragmatics answer (e.g., Prince 1990; 1997; Suñer 1998). Shlonsky’s (1992) study of resumption in Modern Hebrew and Palestinian Arabic is an example of a syntax dominant solution (more recently, see, e.g., Alexopolou 2006). He suggests that the use and nonuse of resumption in the Hebrew language is not simply about the optional insertion of a pronoun with features matching the relative head. Rather, resumption and nonresumption in Hebrew relative clauses reflect different syntactic structures associated with distinct (though homophonous) relative complementizers (both manifested phonologically as ‫)שׁ‬. Therefore, though the ‫ שׁ‬complementizers in the minimal pairs in (277)–(278) (resumption or gap at the accusative complement position) and (279)–(280) (resumption or gap at the embedded subject position) appear identical, Shlonsky argues that they carry different features that are ultimately related to resumption (Shlonsky 1992: 444). (277)  ‫האישׁ שׁראיתי אותו‬

136

Chapter 5

(278) ___ ‫האישׁ שׁראיתי‬ ‘the man that I saw him/___’ (279) ‫האישׁ שׁחשׁבת שׁהוא מלמד אנגלית‬ (280) ‫האישׁ שׁחשׁבת שׁ__מלמד אנגלית‬ ‘the man that you thought that he/___ teaches English’ Often clauses such as (277)–(280) are used to illustrate the supposed free variation of resumption in Modern Hebrew (see Hayon 1973; Glinert 1989; Borer 1984), but Shlonsky argues that resumption is obligatory in these environments in the same way that resumption is obligatory in NP-internal (281) and “oblique” (i.e., prepositional complement) (282) positions (Shlonsky 1992: 445). (281) ‫*אשׁת‬/‫האישׁ שׁראיתי את אשׁתו‬ ‘the man who I saw his wife’ (282) ‫*על‬/‫האישׁ שׁחשׁבת עליו‬ ‘the man that I thought about him’ In (281), the resumptive constituent is a cliticized possessive pronoun, which means it resides within the structure of the NP that hosts it (i.e., it is “NPinternal”). When a relative head corresponds to an NP-internal position, the absence of resumption results in an ungrammatical clause (see Glinert 1989: 364; also Sells 1984: 65). Similarly, the example in (282) demonstrates that, when the relative head corresponds to a prepositional complement within the relative, i.e., an “oblique” position within the relative clause, it is required for grammaticality (see Glinert 1989: 362; also Cole 1976; Borer 1984: 220 n. 1). The case of (282) also relates to the prohibition of preposition stranding in Hebrew relatives (cf. English any object that he sits on; see below, §5.2.2). Shlonsky’s “last-resort” proposal assumes the revised standard analysis of relativization (see above). 5 His solution to the variation in resumption is that two relativization strategies exist in Modern Hebrew: one strategy that includes wh-movement and one in which wh-movement is blocked. The movement strategy leaves a gap (i.e., no resumptive pronoun) in the relative when the movement is from accusative complement, embedded subject, or embedded accusative complement positions. The other strategy disallows wh-movement, requiring an anaphoric pronoun to be inserted in the embedded position corre5.  Within a raising analysis of relativization, Shlonksy’s last-resort proposal for resumption takes the form of a stranding analysis, which is precisely what Boeckx (2003) argues. In raising-and-stranding terms, some part of the raised relative head (the D of the relativized DP) is left behind in the raising and then manifested as a co-referential pronoun. See also McCloskey 2006: 109–11.

Relative Syntax

137

sponding to the relative head. This pronoun insertion is a “last resort” strategy to overcome what would otherwise be a failure to meet grammaticality (see also Chomsky 1995: 28, 257, 261, 280). Both strategies are associated with the relative complementizer ‫שׁ‬, which Shlonsky argues exists in two forms, one that has features associating it with the movement relative and a second that has features restricting it to the nonmovement relative (see Holmstedt 2002: 273–77 for a more detailed summary). Though Shlonsky’s last-resort proposal (or its more recent “stranding” counterpart; see n. 5) is an economical solution to the syntax of the resumptionversus-nonresumption variation, he admits that the choice between the two is simply “free.” The lack of any insight into the motivation for a speaker to choose one strategy over the other, when both are available, is what renders such syntactic analysis unsatisfying to many, especially those of us with philological leanings. The syntactic description may be accurate, but it seems to be only part of the full grammatical story. And this is why semantic and pragmatic analyses have been offered: to address the perceived inadequacy of a syntaxonly solution. Doron (1982) offers a semantic assessment of the difference between resumption and the lack of resumption in Modern Hebrew relatives (see also Sells 1984). She argues that the difference fundamentally relates to an issue of scope: a gap (283), which is necessarily bound to and within the scope of the relative head, results in a single semantic interpretation; in contrast, a resumptive pronoun (284) allows for ambiguity in terms of scope (which of two available pronouns is bound by the relative head) and results in two interpretations. (283) ‫האשׁה שׁאהרון הראה לה‬ ‘the womani that Aaron showed ti [= the woman] to her   [= woman B]’ (284) ‫האשׁה שׁאהרון הראה לה אותה‬ ‘the womani that Aaron showed heri to her’ (same meaning as   [283]) or ‘the womani that Aaron showed her [= woman B] to heri’ [= the   woman] Doron argues that the differences in scope also relate to the differences between de dicto (nonspecific) and de re (specific/individual) readings. Modern Hebrew relatives with a gap, like (283), may have either a de dicto reading or a de re reading. Thus, (283) could be interpreted as the type of woman (de dicto) that Aaron showed someone else or a specific woman (de re). The presence of overt resumption, as in (284), forces a de re reading. Unfortunately, it is impossible to apply Doron’s proposal to Biblical Hebrew, since this would require the type of semantic judgments only native

138

Chapter 5

speakers can provide. In any case, the proposed semantic differences between resumption and nonresumption in Modern Hebrew may not be accurate (see Prince 1990 for a brief discussion), and Doron herself has subsequently proposed a different analysis for resumption (Doron and Heycock 1999). Therefore, while a semantic solution for the variation between resumption and a gap in relative clauses may exist, a compelling one has yet to be given and, even if one were submitted, we could not test it on an ancient language. Whereas semantic judgments regarding the subtleties of syntactic variation are largely unavailable to the ancient languages researcher, the existence of large texts makes available a rich discourse context. Hence, it is possible that pragmatics may offer a fuller solution to the resumption problem. Prince (1990; 1997), among others (see also Suñer 1998; Sharvit 1999), offers such a discourse-pragmatics solution. Based on her observation that nonrestrictive and indefinite restrictive relatives exhibit a much higher frequency of resumptive elements than do definite restrictive relatives, she suggests that the relationship among the definiteness of the head, the restrictiveness of the relative, and resumption is the result of the information status of the head of the relative within the “hearer’s knowledge-store” (Prince 1990: 491). Prince’s basic approach to the information status of restrictive relative clauses can be briefly summarized as follows. Definite nouns represent entities that are already known to the hearer (i.e., they have already been invoked in the discourse, or they are implicit within it); thus, when definite nouns are used again in a discourse, the hearer is instructed to activate the referent of the definite noun. When definite nouns are modified by restrictive relatives, the relative is not merely additional information; the relative presents information necessary for the proper identification of the definite noun’s discourse referent. Thus, the information within the relative must already by known to the hearer as well, since the definite noun and its relative are considered to work together in order to refer to their referent (Prince 1990: 491). Prince describes the information status of resumption in relatives in terms of Heim’s metaphor of a filing cabinet and file cards on which information is stored. Consider Prince’s English examples in (285). (The English examples are illustrative in terms of the types of heads and relatives, but they do not illustrate the issue of resumption since English does not employ the same type of relative clause resumption as appears in Hebrew, Standard Arabic, Yiddish, Irish, Spanish, etc.; see Sells 1984) (285) a. He bought a house in Society Hill and a house down the shore. b. The house that’s in Society Hill is a colonial. c. The First Lady who introduced ice cream in America lived in it for three years. d. But the part of the roof that’s over the dormer is shot.

Relative Syntax

139

In the typical case, if the NP is indefinite, it represents a ‘Brand-new’ entity, as in [(a)], and the hearer must add that entity, or construct a new file card. . . . And typically, if the NP is definite, it represents something already invoked in the discourse model, as in [(b)], or something assumed to be present in the hearer’s knowledge-store, as in [(c)], in which case the hearer must activate the appropriate existing file card, or else it represents an entity which the hearer is assumed to be able to infer on the basis of prior knowledge s/he is assume to have, as in [(d)], in which case the hearer must construct a file card out of existing material. (Prince 1990: 491)

In contrast to definite nouns with restrictive relatives, indefinite nouns with restrictive relatives represent brand-new discourse entities, as in (286). Although both the head and the information within the relative are discourse-new, according to Prince, they are given separate entries in the hearer’s knowledgestore even though they are associated with each other (Prince 1990: 492). (286) a. He bought a house which he’ll move into in June. b. A realtor that I had recommended found it for him. Regarding both the relative clauses in (286), Prince argues that “the hearer has to add a new file card, as signaled by the indefiniteness of the [head] NPs. And, in each case, the file card to be added need represent only the entity described by the head, the information in the relative clause simply being an additional property of that entity to be noted on the independently constructed file card” (Prince 1990: 492). Similarly, nonrestrictive relatives, as in (287), provide discourse-new information, even if the referential heads they modify are not discourse-new. (287) a. He bought a house, which, by the way, I had found, and a car. b. The house, which is in Society Hill, is a colonial. c. Dolley Madison, who he always admired, lived in it for three years. d. But the roof, which is slate, is shot. For the nonrestrictive relatives in (287), Prince argues that file cards must be constructed based solely upon the head and that the nonrestrictive clauses represent information “that presumably does not yet exist on those file cards and which must be added to them” (Prince 1990: 491). So how does the definiteness of the head and the restrictiveness of the relative clause relate to the use of resumption? Prince suggests that resumptive pronouns may be used in English and Yiddish relative clauses when “the entity evoked by the whole NP is in fact evoked by the head, the relative clause serving simply to predicate some property of that entity” (Prince 1990: 492), as in (288). (288)  He bought a house which he’ll move into it in June.

140

Chapter 5

Prince also claims that her file card account “means that resumptive pronouns work exactly like ordinary discourse pronouns” and also explains “why one finds other anaphoric expressions in the place of resumptive pronouns, e.g., demonstratives, as in [(289)a], co-referential full NPs, as in [(289)b], and even referentially relative but non-co-referential pronouns, as in [(289)c], and full NPs, as in [(289)d]” (Prince 1990: 492). (289) a. b. c. d.

I had a handout and notes from her talk that that was lost too. He’s got this lifelong friend who he takes money from the parish to give to this lifelong friend. I have a manager, Joe Scandolo, who we’ve been together over twenty years. You assigned me to a paper which I don’t know anything about the subject.

Prince’s analysis of English and Yiddish suggests that resumption may have a pragmatic cause and function, even if the specific cause and the specific function differ from language to language. 6 Even so, the precise function for resumption that Prince has identified with regard to restrictiveness and the definiteness of the head does not adequately explain the Biblical Hebrew data. Simply, we have clear nonrestrictive relatives with resumption (290) and without (291). (290) ‫ֹותי וְ ֻחקּ ָֹתי‬ ַ ‫ְל ַמ ַען ָדּוִ ד ַע ְב ִדּי ֲא ֶשׁר ָבּ ַח ְר ִתּי אֹתֹו ֲא ֶשׁר ָשׁ ַמר ִמ ְצ‬ ‘for the sake of David, my servanti, whoi I chose himi, who has kept my commandments and my statutes’ (1 Kgs 11:34) (291) ‫רוּשׁ ַלםִ ֲא ֶשׁר ָבּ ָח ְר ִתּי‬ ָ ְ‫וּל ַמ ַען י‬ ְ ‫ֵשׁ ֶבט ֶא ָחד ֶא ֵתּן ִל ְבנֶ ָך ְל ַמ ַען ָדּוִ ד ַע ְב ִדּי‬ ‘one tribe I will give to your son, for the sake of David my servant and for the sake of Jerusalemi, whichi I chose ___i’ (1 Kgs 11:13) 7 6. Interestingly, Prince admits that her discourse/pragmatic analysis of resumptive pronouns in relative clauses does not explain the Modern Hebrew or Arabic data, since Semitic resumptive pronouns seem to exhibit different pragmatic functions from those in English and Yiddish (Prince 1997: 231–34). 7.  The verb ‫‘ בחר‬to choose’ exhibits resumption with ‫ ב‬and with ‫ את‬as well as the lack of resumption. Since resumption with this verb cannot be obligatory, given that it takes two complements and yet can stand the lack of resumption (in any case, there is no discernible semantic difference between the two complements), I suggest that the presence of a resumptive constituent is needed to disambiguate the head (e.g., Num 16:5; 1 Sam 10:24; 1 Kgs 11:32 [cf. 1 Kgs 11:36]; 1 Kgs 11:13, 34) or is pragmatically motivated (e.g., Judg 10:14; Isa 41:8 [cf. Isa 43:10]; Jer 33:24). With that said, it must be admitted that with this particular verb, there are 23 specific verses—when a relative clause with ‫ בחר‬modifies the noun ‫מקום‬ ‘place’ and the head refers to the ‘designated place of sacrifice’ (see Deut 12:5, 11, 14, 18, 21, 26; 14:23, 24, 25; 15:20; 16:2, 6, 7, 11, 15, 16; 17:8; 18:6; 23:17; 26:2; 31:11; Josh 9:27; Neh 1:9)—where it is difficult to determine a rhyme or reason for the presence or absence of

Relative Syntax

141

5.1.2.2. Extraposition Extraposition describes the placement of a constituent near the end of a clause (Ouhalla 1999: 87; Crystal 1997: 146; cf. Haegeman 1994: 60–63). 8 Extraposition often occurs with relative clauses but is a general phenomenon that is also used with a broad range of constituents, from appositive NPs to complement NPs and assorted PPs (see de Vries 2002: 235–38; Baltin 2006). The minimal pairs below illustrate extraposition of relative clauses (292) and PPs (293) in English (the extraposed clause in the second clause of each pair is in brackets, and its “normal” position is marked by co-indexed gap). (292) Some friends whom we hadn’t seen for years visited. Some friends ___i visited [whom we hadn’t seen for years]i. (293) Some children with Halloween costumes on came by. Some children ___i came by [with Halloween costumes on]i. In the first clause in (292), the relative clause modifies the DP some friends and, in the “normal” order, it follows immediately after its head. However, in the example with extraposition, there is discontinuity between its head some friends and the relative whom we hadn’t seen for years: the VP visited intervenes. Similarly, in the first example in (293) the PP with Halloween costumes on follows immediately after the NP it modifies, some children. In contrast, the second, extraposition example, shows the PP at the end of the clause, with the VP came by intervening. There are three basic analyses of the syntax of extraposition (see de Vries 2002: 239–68; and Baltin 2006 for detailed discussion and relevant literature): (1)  as rightward movement (e.g., Baltin 1984; Büring and Hartmann 1997) or base-generation at the right (e.g., Culicover and Rochemont 1990); (2) as stranding produced by the leftward movement of the non-extraposition resumption. The pragmatic motivation for resumption in these cases may be rather subtle and possibly is theologically nuanced. 8.  Within Hebrew studies, there has been variation in the definition of the term extraposition. For example, Khan 1988; Sivan and Yona 1994; Zewi 1996; 1999; 2001; and Shimasaki 2002: 245–49 use the term to refer to constituents at the front of a clause, whereas I (Holmstedt 2001; 2014b) use it for movement toward the end of the clause. Though the equation with fronting or even casus pendens (= left dislocation) seems to have been more common in the early and mid-twentieth century (see, for example, Jespersen, whose definition and subsequent examples indicate he used extraposition for constituent at the left or front of the clause [Jespersen 1969: 35–38; cf. 1964: 95; 1928–49: 3.72, 357; 7.223]), for at least the last three decades, using the term in reference to constituents at the end of the clause has been increasingly accepted (see, for example, Guéron 1980; Mallinson 1986; Baltin 2006). In this book, I have followed the definition current in general linguistics; for an introductory source, see Crystal 2008: 182, “Extraposition”: “A term used in grammatical analysis to refer to the process or result of moving (or extraposing) an element from its normal position to a position at or near the end of the sentence” (emphasis added).

142

Chapter 5

constituent (e.g., Kayne 1994); and (3) as specifying coordination (de Vries 2002: 268–78; 2006). Though de Vries’s argument for specifying coordination is attractive, since the structure of extraposition has no discernible bearing on the identification or determination of function for Hebrew relative extraposition, I remain agnostic in this book and will simply note the expected (adjacent) position and the resulting position by subscript co-indexation, as in the two examples in (292)–(293). I will take up the discussion of the pragmatic function of relative clause extraposition in the context of the Hebrew examples in §5.2.4. With the typological and phrase structure features I have now described, we are now adequately prepared to explore the syntactic features of the Hebrew relative. 5.2.  Relative Syntax in Hebrew With regard to the syntactic features that distinguish relative clauses crosslinguistically, a number of features may quickly be dispensed with. First, as the prototypical example in (294) illustrates, the Hebrew relative clause is embedded, not correlative; the relative head is not inside the relative, nor is there is a full copy or demonstrative “correlate” in the matrix clause. Second, all Hebrew relatives are externally-headed and the relative is postnominal (i.e., follows the relative head and relative element). (294) ‫אתי‬ ִ ‫ר־ּב ָר‬ ָ ‫ָה ָא ָדם ֲא ֶׁש‬ ‘the mani whi that I created ___i’ (Gen 6:7) In terms of the extraction site for relativized words, Hebrew exhibits significant variety concerning the syntactic role that the relative head corresponds to (i.e., has been promoted from) within the relative clause. There are numerous examples of the head corresponding to the subject position (295), the complement position (296), and even some adjunct positions representing a variety of semantic roles, such as the temporal head in (297) or the locative head in (298). (295) Promotion from the Subject Position ‫ִסיחֹון ֶמ ֶלְך ָה ֱאמ ִֹרי ֲא ֶשׁר ָמ ַלְך ְבּ ֶח ְשׁבֹּון‬ ‘Sihon, king of the Amoritesi, whi that ___i reigned in Heshbon’ (Josh 13:21) (296) Promotion from the Complement Position ‫ַעם ֲא ֶשׁר לֹא־יָ ָד ְע ָתּ‬ ‘a peoplei whi that you do not know ___i’ (Deut 28:33) (297) Promotion from an Adjunct Position (Temporal) ‫ָכּל־יְ ֵמי ֲא ֶשׁר ַהנֶּ גַ ע בֹּו‬ ‘all the daysi whi that the disease (is) in him ___i’ (Lev 13:46)

Relative Syntax

143

(298) Promotion from an Adjunct Position (Locative) ‫ר־דּ ֶבּר ִאתֹּו‬ ִ ‫וַ יַּ ֵצּב יַ ֲעקֹב ַמ ֵצּ ָבה ַבּ ָמּקֹום ֲא ֶשׁ‬ ‘and Jacob set up a pillar in the placei whi that (God) spoke with him ___i’ (Gen 35:14) Although the corpus of marked relative clauses not introduced by ‫ אשׁר‬is significantly smaller than the ‫ אשׁר‬corpus, enough data exist to make some descriptive statements about the extraction sites. For instance, ‫שׁ‬‎ (299)–(301), ‫ז‬-series (302)–(304), and ‫מ‬-pronoun (305)–(306) relatives appear to allow the same range of syntactic diversity regarding the site from which the relative head is promoted: (299) ‫שׁ‬-Relative: Head Promoted from Subject ‫‏ה ָא ָדם ֶׁשּיָבֹוא ַא ֲח ֵרי ַה ֶּמ ֶלְך‬ ָ ‘the mani whi that ___i will come after the king’ (Eccl 2:12) (300) ‫שׁ‬-Relative: Head Promoted from Complement ‫ל־מ ֲע ַשׂי ֶשׁ ָעשׂוּ יָ ַדי‬ ַ ‫‏כ‬ ָ ‘all my worksi whi that my hands made ___i’ (301) ‫שׁ‬-Relative: Head Promoted from Adjunct ‫ל־מקֹום ֶשׁ ַהנְּ ָח ִלים הֹלְ ִכים ָשׁם‬ ְ ‫‏א‬ ֶ ‘to the placei whi that the rivers flow therei’ (Eccl 1:7) (302) ‫ז‬-Relative: Head Promoted from Subject ‫ִמ ְפּנֵ י ְ ֭ר ָשׁ ִעים זוּ ַשׁדּוּנִ י‬ ‘because of wicked peoplei whi that ___ i destroy me’ (Ps 17:9) (303) ‫ז‬-Relative: Head Promoted from Complement ‫ְמקֹום זֶ ה יָ ַס ְד ָּת ָל ֶהם‬ ‘a placei whi that you established ___i for them’ (Ps 104:8) (304) ‫ז‬-Relative: Head Promoted from Adjunct ‫אֹורָך ְבּ ֶד ֶרְך־זוּ ֵת ֵלְך‬ ְ ְ‫ו‬ ‘in the wayi whi that you should walk ___i’ (Ps 32:8) (305) ‫מי‬-Relative: Head Promoted from Subject ‫ִמי־יָ ֵרא וְ ָח ֵרד יָ שׁ ֹב‬ ‘Whoeveri whi ___i is afraid and trembling’ (Judg 7:3) (306) ‫מה‬-Relative: Head Promoted from Complement 9 ‫ה־ּלְך‬ ָ ‫אמר נַ ְפ ְׁשָך וְ ֶא ֱע ֶׂש‬ ַ ֹ ‫ַמה־ּת‬ ‘whateveri whi your soul says ___i, I will do for you’ (1 Sam 20:4) 9.  There are no examples of a ‫מ‬-pronoun relative in which the head has been promoted from an adjunct position within the relative.

144

Chapter 5

In contrast to the variety of extraction sites allowed by other relative words in Hebrew, ‫ה‬-relatives are restricted to relativization from the subject position (see §3.2.3), as in (307). (307) ‫ית־אל ְשׁ ֶכ ָמה‬ ֵ ‫ִל ְמ ִס ָלּה ָהע ָֹלה ִמ ֵבּ‬ ‘of a highwayi whi that ___i goes up from Bethel to Shechem’   (Judg 21:19) Moreover, as I mentioned in §3.2.3, since the relative ‫ ה‬is cliticized to the first element of the relative clause, which is overwhelmingly a participle or adjective that agrees with the promoted subject, there is no room for a resumptive element, as the unacceptable (and unattested) example in (308) illustrates. (308) ‫יהן‬ ֶ ‫*ה ֵהּנָ ה עֹֹלת ַא ֲח ֵר‬ ָ ‫‏ וְ ֶׁש ַבע ַה ָּפרֹות ָה ַרּקֹות וְ ָה ָרעֹת‬ ‘and the seven lean and ugly cowsi whi that (*theyi) come up after them’ (Gen 41:27, modified) Like marked relatives (excepting ‫ה‬-relatives), zero-relatives may modify heads that are co-referential with a variety of syntactic roles within the relative clause: the relative head in (309) corresponds to the role of the relative verbal complement, 10 the relative head in (310) to the role of a PP complement, 11 and the relative head in (311) to an adjunct position. 12 (309) ‫ית ֶּק ֶדם ּגָ ַא ְל ָּת ֵׁש ֶבט נַ ֲח ָל ֶתָך‬ ָ ִ‫זְ כֹר ֲע ָד ְתָך ָקנ‬ ‘remember your congregationi whi (that) you acquired ___i long ago, whi (that) you redeemed ___i (as) a tribe of your inheritance’   (Ps 74:2) (310) ‫צּורי ֶא ֱח ֶסה־ּבֹו‬ ִ ‫ֵא ִלי‬ ‘my god (is) my rocki whi (that) I take refuge in iti’ (Ps 18:3) (311) ‫ּומ ְׁש ַחת ָּבנָ יו ֵמ ִא ֵּׁשי יְ הוָ ה ְּביֹום ִה ְק ִריב א ָֹתם ְל ַכ ֵהן ַליהוָ ה‬ ִ ‫זֹאת ִמ ְׁש ַחת ַא ֲהר ֹן‬ ‘this is Aaron’s portion and his sons’ portion from Yhwh’s offerings by fire on the dayi whi (that) he brought them near ___i to be priests for Yhwh’ (Lev 7:35) Furthermore, unlike the syntax of zero-relatives in English and contrary to the Implicational Tendency (4c), zero Hebrew relatives may exist even when the gap within the relative is the subject position. The English example in (312) indicates the ungrammaticality of an unmarked relative clause when the head 10.  See Gen 26:10; 29:25; 42:28; Exod 14:11; 15:17 (2×); Deut 32:17; Judg 8:1; 2 Sam 22:44; Isa 6:6; 42:1; Ps 74:2. 11.  See Isa 44:1, 2; Jer 2:6; Hab 1:14; Ps 18:3. 12.  See Exod 18:20; Lev 7:35; Deut 32:35; Jer 2:6; 36:2; Mic 5:2; Pss 4:8; 56:4, 10; Job 3:3; 6:17; Lam 1:21; 2 Chr 20:22; 24:11; 29:27.

Relative Syntax

145

corresponds to the subject position and the relative verb is finite. In contrast, zero subject-relatives in Hebrew appear with null-copula clauses—whether the copular complement is a participle in lines a, c, and d of (313), an adjective in line b of (313), or a PP (314)—and finite verbs (315)–(316). (312) I knew the person who /that/*∅ talked to me 13 (313) ‫יתים‬ ִ ‫‏ּגֹוי ח ֵֹטא ַעם ֶּכ ֶבד ָעֹון זֶ ַרע ְמ ֵר ִעים ָּבנִ ים ַמ ְׁש ִח‬ ‘a nationi whi (that) ___i (is) sinning, a peoplej whj (that) ___j (is) heavy (with) iniquity, a seedk whk (that) ___k (is) doing evil, sonsl whl (that) ___l (are) acting corruptly’ (Isa 1:4) (314) ‫‏‏ ּב ְֶא ֶרץ לֹא ָל ֶהם‬ ‘in a landi whi (that) ___i (is) not to them’ (i.e., not theirs)   (Gen 15:13) (315) ‫א־יָבין יִ ָּל ֵבט‬ ִ ֹ ‫‏וְ ָעם ל‬ ‘and a peoplei whi (that) ___i does not understand will come to ruin’   (Hos 4:14) (316) ‫חֹול ְל ֶכם‬ ֶ ‫ל־ׂש ָרה ְּת‬ ָ ‫יכם וְ ֶא‬ ֶ ‫ל־א ְב ָר ָהם ֲא ִב‬ ַ ‫ַה ִּביטּו ֶא‬ ‘look to Abraham your father and to Sarahi whi (that) ___i bore you’   (Isa 51:2) 14 Whereas Hebrew relatives (excepting the ‫ה‬-relative variety) exhibit significant diversity in the position from which the relativized noun is raised, similar diversity does not characterize word order within relatives. Although Hebrew normally allows a significant amount of word-order variation (see §2.3), within 13. See McCawley 1998: 433. If we removed the auxiliary verb was from (312), the result appears to be a grammatical relative clause without an overt subject constituent within the relative, e.g., the person talking to me. However, depending on its function in a larger clause (e.g., I heard the person talking to me), this may be an example of a construction called a “small clause,” which, while similar in some ways to relatives, is technically not a relative clause (see Haegeman and Guéron 1999: 108–12). 14.  In the zero-relative data in appendix A.5 (which excludes participial relatives, since they are too numerous to list), there are 142 examples of zero subject relatives: Gen 15:13; 39:4; 49:27; Lev 13:20; 17:11; 22:4; Deut 4:38; 7:1; 9:1 (2×), 14; 11:23; 32:11 (5×), 17 (2×); 1 Sam 6:9; 10:11; 14:21; 26:14; 2 Sam 23:1, 9; Isa 1:4; 10:3; 23:13; 28:16; 30:5, 9; 31:1, 8 (2×); 32:13; 40:20 (2×); 50:9; 51:2; 53:7 (2×); 54:1 (2×); 55:13; 56:2 (2×); 62:1; 65:1 (3×); 66:1; Jer 2:2, 8, 11; 13:20; 23:29; 30:21; 49:12; Ezek 20:25; Hos 4:14; 6:3; Amos 3:9; Hab 1:6, 8; 3:16; Zeph 2:1; 2:2; 3:3, 17; Mal 2:16; Pss 5:5; 7:9; 9:18; 15:3–5 (8×); 16:4; 17:1, 9; 25:12; 34:9; 38:14; 46:2; 52:9; 58:5; 65:5; 66:9; 68:31; 71:18; 78:6, 39; 83:15 (2×); 90:5; 104:32; 107:40; 109:16; Job 3:16; 4:7 (3×); 6:10; 7:2 (2×); 9:26; 17:3; 18:21; 31:12 (2×); 32:19; 33:21; 35:5; Prov 8:32; 9:16; 22:29; 23:32 (2×); 26:17; 30:15 (2×), 17, 18, 24; Eccl 1:18; 2:1; 2:26; 12:11; Lam 1:1, 10; 3:25, 41; 4:17; 1 Chr 12:24; 2 Chr 28:9; 30:19; 31:16, 19.

146

Chapter 5

the embedded context of relative clauses there is much less variation than in matrix clauses. The reason is transparent: the pragmatic motivation that drives constituent movement operates primarily in matrix or weakly sub­ordinate (non­embedded) clauses. Embedded clauses are less salient since they mostly convey background information within the structure of the discourse. It is not surprising, then, that the majority of relative clauses with null-copula (i.e., “verbless”) clauses exhibit subject-(copula)-complement order (317), almost to the exclusion of any other constituent order, as in (318). (317) Relative Clause: Subject – (Null) Copula – Complement ‫ר־א ָּתה ָׁשם‬ ַ ‫ן־ה ָּמקֹום ֲא ֶׁש‬ ַ ‫ְּור ֵאה ִמ‬ ‘and look from the placei whi that you are therei’ (Gen 13:14) (318) Relative Clause: Adverb – Subject – (Null) Copula 15 ‫ֹלהים׃‬ ִ ‫ר־ׁשם ָה ֱא‬ ָ ‫ל־ה ֲע ָר ֶפל ֲא ֶׁש‬ ָ ‫‏ וַ ּיַ ֲעמֹד ָה ָעם ֵמ ָרחֹק ּומ ֶֹׁשה נִ ּגַ ׁש ֶא‬ ‘and the people stood at a distance while Moses drew near to the darknessi whi that therei was God’ (Exod 20:21) 16 Similarly, almost all Hebrew relatives with finite verbs exhibit verb-subjectcomplement order (319) and, accordingly, relatives with subject-verb order (320) are extremely rare. 17 (319) Relative Clause: Verb-Subject-Complement ‫וְ ִא ָשּׁה ֲא ֶשׁר יִ ְשׁ ַכּב ִאישׁ א ָֹתהּ ִשׁ ְכ ַבת־זָ ַרע‬ ‘(as for) a womani whi that a man lies with heri (for) copulation’   (Lev 15:18) (320) Relative Clause: Subject-Verb-Complement ‫וְ ק ָֹרא ָלְך ֵשׁם ָח ָדשׁ ֲא ֶשׁר ִפּי יְ הוָ ה יִ ֳקּ ֶבנּוּ‬ ‘and to you shall be called a new namei whi that the mouth of Yhwh shall designate iti’ (Isa 62:2) The phrase structure of Hebrew relatives must allow not only for word order variation within the relative but also for the use of relative complementizers in 15.  It is possible that the null copula raises to intervene between the raised adverb and the subject, because it is the pattern with pronominal copulas (reference); however, since these copulas are null, there is obviously no way to be certain. 16.  See Gen 1:29, 30; 2:11; 6:17; 7:2, 15; 17:12; 34:14; 38:30; Exod 9:26; 20:21; Lev 11:9, 21, 23; 13:45, 54, 57; 14:35, 40; 21:18, 21; 25:30; 27:24; Num 17:5; Deut 4:7, 8; 14:9; 17:15; 19:17; 20:15; Judg 19:12; 1 Sam 3:3; 9:10; 22:2; Ezek 8:3; 9:6; 17:3; 21:32; Ps 16:3; Job 4:19; 6:4; 12:10; Eccl 4:1. 17.  Out of the almost 1,200 Hebrew ‫אשׁר‬-relative clauses with an overt subject and a finite verb, an overwhelming 98% exhibit verb-subject word order. For the 2% that exhibit subject-verb word order, see Gen 13:16; Lev 4:22; Deut 9:2; 1 Sam 20:31; 2 Kgs 22:13; Isa 62:2; 66:13; Jer 1:17; 22:25; 29:25; 31:32; 39:17; Mic 6:12; Zech 1:15; 11:2; Mal 2:14; Ps 89:22; 104:17; 144:8, 11; Job 3:23; 5:5; 15:18; 19:27; Eccl 6:12; 7:22; Esth 6:6, 9, 11.

Relative Syntax

147

Hebrew (i.e., ‫אשׁר‬‎, ‫שׁ‬, and the ‫ז‬-series are similar to that in English, as opposed to the English relative pronouns, e.g., who, which). This combination requires that two complementizer (C) positions exist in the phrase structure: one to accommodate the relative complementizer and one to accommodate the raised verb (see §2.3.2 for discussion of the recursive CP), as in (321). 18 (321) (= [319]) ‫וְ ִא ָשּׁה ֲא ֶשׁר יִ ְשׁ ַכּב ִאישׁ א ָֹתהּ ִשׁ ְכ ַבת־זָ ַרע‬ ‘(as for) a womani whi that a man lies with heri (for) copulation . . .’   (Lev 15:18) DP D NP ∅ ʾiššâi CP

whi C′

C2 CP ʾăšer C1 TP yiškab

ʾîš T′ tyiškab VP tʾîš V′ V′ AdvP šikbat zāraʿ V PP tyiškaḇ ʾōṯāhi

18.  Note that, while in almost all of the following examples the nominal head of the relative clauses projects to a DP within the DP-Hypothesis, for the sake of space (literally, space on the page), I often abbreviate the structure to an NP.

148

Chapter 5

The arrows in (321) illustrate the triggered verb raising of ‫ ישׁכב‬to a position C within the CP. This raising-to-C analysis of the verb in relatives is borne out by the data: as I noted above in n. 17, out of the almost 1,200 Hebrew ‫א‬ ‫שׁר‬-relative clauses with an overt subject in a finite verbal clause, only 2% exhibit subject-verb word order. In other words, an overwhelming 98% of Hebrew ‫אשׁר‬-relatives clauses with a finite verb exhibit verb-subject word order (see §2.3 for a discussion of Hebrew constituent order in general). The basic recursive structure of the CP within Hebrew relative clauses extends beyond the two layers needed for the relative complementizer and the raised verb. Although the typical order of constituents in Hebrew relatives is complementizer-verb-subject, the small minority—the 2% like (322)—that show word-order movement beyond the raising of the verb to C necessitates a very flexible CP layer (for a discussion of the flexible Hebrew CP with regard to a wide variety of “edge” constituents, see Holmstedt 2014b): (322) (= [320]) ‫וְ ק ָֹרא ָלְך ֵשׁם ָח ָדשׁ ֲא ֶשׁר ִפּי יְ הוָ ה יִ ֳקּ ֶבנּוּ‬ ‘and to you shall be called a new namei whi that the mouth of Yhwh shall designate iti’ (Isa 62:2) DP D NP ∅ šēm ḥādāši CP

whi C′

C2 CP ʾăšer

pî yhwh C1

C′ TP
 tpî yhwh

yiqqŏbennûi T′

tyiqqŏbennû VP

Relative Syntax

149

In (322), after the syntactically triggered raising of the verb to C, the subject ‫ פי יהוה‬is further raised for Focus. By raising ‫ פי יהוה‬for Focus in (322), the prophet was isolating this agent out of all likely agents and asserting that it was God that actually accomplished the activity (see, e.g., Blenkinsopp 2003: 232, who translates it “that Yhvh himself will bestow”) There are also a few relative clauses, as in (323), that exhibit multiple embedding; this also necessitates a recursive CP structure. In (323), the relative clause contains within it another level of embedding: an interrogative clause headed by a wh-word, ‫למה‬. (323) . . . ‫יכם ז ֲֹע ִפים‬ ֶ ֵ‫ת־פּנ‬ ְ ‫ ֲא ֶשׁר ָל ָמּה יִ ְר ֶאה ֶא‬. . . ‫ת־אד ֹנִ י ַה ֶמּ ֶלְך‬ ֲ ‫יָ ֵרא ֲאנִ י ֶא‬ ‘I am afraid of my lord the kingi . . . whi who why should ___i (he) see your faces dejected . . . ?’ (Dan 1:10) DP ʾădōnî CP hammelek

whi C′

C2 CP ʾăšer

lāmmâ C′

C1 TP yirʾê

ti T′   tyirʾê VP tʾîš VP

V′ t lāmmâ

V DP   tyirʾê ʾeṯ pənêkem . . .

150

Chapter 5

Though word-order variation such as in (322) and (323) is common in main clauses in the Hebrew Bible, it is rare within relatives. And for those relative clauses that do exhibit atypical constituent order, it is often due to a combination of both syntactically triggered and pragmatically triggered movement. For example, in (324), the ‫ אשׁר‬triggers verb-raising of ‫( יוצק‬a syntactic trigger), followed by the extraposition of the “heavier” subject NP ‫( שׁמן המשׁחה‬a discourse-pragmatic trigger). (324) ‫ר־יוּצק ַעל־רֹאשֹׁו ֶשׁ ֶמן ַה ִמּ ְשׁ ָחה‬ ַ ‫וְ ַהכּ ֵֹהן ַהגָּ דֹול ֵמ ֶא ָחיו ֲא ֶשׁ‬ ‘and the priesti greater than his brothers, whi that upon hisi head is poured the oil of anointing’ (Lev 21:10) It is tempting to read the PP ‫ על ראשׁו‬with Focus in (324), so that the Focus allows the 3ms clitic pronoun to be isolated and contrasted with other potential entities: it is on the head of this particular priest that the oil is poured. Against this analysis, however, is the position of the verb ‫‘ יוצק‬is poured’ higher than the PP ‫ ;על ראשׁו‬without an explicit Focus marker, such as ‫גם‬, and lacking the original prosody of the verse, it is hazardous to assert that the PP ‫על ראשׁו‬ carries in situ Focus-marking. Instead, an analysis that accords with known patterns is that the PP is in its normal place after the verb and that the subject NP ‫ שׁמן המשׁחה‬has been extraposed in a case of Heavy Noun Phrase Shift (HNPS), in which a more complex constituent is extraposed for ease of processing (see §§2.3.3; 5.2.4; Holmstedt 2014b). Like the relative in (322), the relatives in (325) and (326) illustrate the Focus-induced raising of a constituent above the verb to position within the CP following the complementizer ‫( אשׁר‬see also Peretz 1967: 97). (325) ‫ֹלהים וַ ֲאנָ ִשׁים‬ ִ ‫ר־בּי יְ ַכ ְבּדוּ ֱא‬ ִ ‫ת־דּ ְשׁנִ י ֲא ֶשׁ‬ ִ ‫אמר ָל ֶהם ַהזַּ יִ ת ֶה ֳח ַד ְל ִתּי ֶא‬ ֶ ֹ ‫וַ יּ‬ ‘and the olive tree said to them: Should I cease (producing) myi fatness whi that by mei gods and men they honor’ (Judg 9:9) (326) ‫ֹלהיו‬ ָ ‫וּמ ְשׁ ַפּט ֱא‬ ִ ‫ר־צ ָד ָקה ָע ָשׂה‬ ְ ‫אֹותי יֹום יֹום יִ ְדר ֹשׁוּן וְ ַד ַעת ְדּ ָר ַכי יֶ ְח ָפּצוּן ְכּגֹוי ֲא ֶשׁ‬ ִ ְ‫ו‬ ‫לֹא ָעזָ ב‬ ‘me they seek day by day and the knowledge of my ways they desire, like a nationi whi that ___i righteousness has done and the justice of its god it has not abandoned’ (Isa 58:2) In (325), the fronted PP ‫‘ בי‬by me’ functions as a Focus, highlighting the olive tree’s claim that it is only by it that gods and men are honored. The context of (326) is the prophetic assertion that the people, who ironically think they are pious and faithful, continue to fail to do what God really considers righteous—eliminating oppression and suffering (see esp. Isa 58:6–7; see Watts 2005: 845; contra Blenkinsopp 2003: 176). Thus, in (326) the complement ‫‘ צדקה‬righteousness’ is Focus-fronted to contrast what the people think of

Relative Syntax

151

themselves (i.e., righteous) with the contrary assessment that God, through the prophet, provides. Due to the comparative paucity of non-‫ אשׁר‬relatives in Hebrew, the full extent of word-order variation allowed in these varieties of relative clauses is likely underrepresented by the data in the Hebrew Bible. Yet, two firm descriptive statements can be made. First, ‫ה‬-relatives, as in (327), show a highly restricted word order since the subject is always resumed covertly (null) and the predicate is almost always a null copula with participial or adjectival complement, which is the clitic host for the relative ‫ ;ה‬therefore, little room is left for word-order variation. (327) ‫ל־ה ָא ֶרץ‬ ָ ‫ל־חּיָ ה ָהר ֶֹמ ֶׂשת ַע‬ ַ ‫ּוב ָכ‬ ְ ‘and over every creaturei whi that-___i-creeps on the ground’   (Gen 1:28) Second, though the examples are sparse, the few examples of nor-‫אשׁר‬relatives with overt subject or finite verbs that do exist suggest both that these relative clauses witness the triggered raising of the verb over the subject for typical verb-subject order within the relatives and that more data would probably provide a range of pragmatically-influenced word-order diversity similar to what ‫אשׁר‬-relatives provide. Consider the following minimal examples with a finite verb and either an overt subject or a fronted complement or adjunct for ‫שׁ‬‎(328)–(330), the ‫ז‬-series (332), and the ‫מ‬-pronoun (333). (328) ‫שׁ‬-relative: verb-subject 19 ‫ל־מ ֲע ַשׂי ֶשׁ ָעשׂוּ יָ ַדי‬ ַ ‫יתי ֲאנִ י ְבּ ָכ‬ ִ ִ‫וּפנ‬ ָ ‘and I faced, I, all my worksi whi that my hands had done ___i’   (Eccl 2:11a) (329) ‫שׁ‬-relative: verb-complement-subject 20 ‫קֹוצר וְ ִח ְצנֹו ְמ ַע ֵמּר׃‬ ֵ ‫ֶשׁלֹּא ִמ ֵלּא ַכפֹּו‬ ‘(grassi) whi that a reaper does not fill his hand (with) ___i and a gatherer [fill] his cloak (with) ___i’ (Ps 129:7) (330) ‫שׁ‬-relative: subject-verb 21 ‫י־ח ֶרב ֵמ ַח ְל ֵלי ָר ָעב ֶשׁ ֵהם יָ זוּבוּ ְמ ֻד ָקּ ִרים ִמ ְתּנוּבֹת ָשׂ ָדי׃‬ ֶ ‫טֹובים ָהיוּ ַח ְל ֵל‬ ִ ‘‘better were the slain by the sword than the slaini by hunger, whi that theyi flow out, being pierced without the produce of the field’   (Lam 4:9)

19.  See Song 1:6b, 7a; 3:1, 2, 3, 4c; Eccl 2:11a, 11:3; 12:3; Ezra 8:20. 20.  See Ps 129:7; Song 3:11; Eccl 10:16 (verb-adjunct-subject). 21.  See Song 1:6a; 5:2; 6:5a; Lam 4:9.

152

Chapter 5

(331) ‫שׁ‬-relative: complement/adjunct-verb 22 ‫וּבן־‬ ִ ‫ן־ליְ ָלה ָהיָ ה‬ ַ ‫א־ע ַמ ְל ָתּ בֹּו וְ לֹא גִ ַדּ ְלתֹּו ֶשׁ ִבּ‬ ָ ֹ ‫ל־ה ִקּ ָיקיֹון ֲא ֶשׁר ל‬ ַ ‫אַתּה ַח ְס ָתּ ַע‬ ָ ‫אָבד׃‬ ָ ‫ַליְ ָלה‬ ‘you are concerned about the qiqayon-planti that you did not labor and that you did not grow, whi that ___i was the product of a night and perished in a night?’ (Jonah 4:10) (332) ‫ ז‬relative: verb-subject 23 ‫יְמינֹו׃‬ ִ ‫ַהר־זֶ ה ָקנְ ָתה‬ ‘(the) mounti whi that his right hand had acquired ___i’ (Ps 78:54) (333) ‫ מ‬relative: verb-subject 24 ‫ה־ּלְך‬ ָ ‫אמר נַ ְפ ְׁשָך וְ ֶא ֱע ֶׂש‬ ַ ֹ ‫ַמה־ּת‬ ‘whateveri whi your soul says ___i, I will do for you’ (1 Sam 20:4) Unsurprisingly, zero-relatives exhibit the same basic patterns of constituent order. Of the 38 examples in which the subject is overt, all but three (Isa 51:12, Mic 5:2, and Ps 4:8) reflect verb-raising to produce verb-subject order. 25 Of the three with subject-verb order, the meaning of Isa 51:12 is opaque, and both Mic 5:2 and Ps 4:8 are cases in which the head of the relative is the temporal NP ‫עת‬. I know of no good reason why this might make a difference, and it very well may be coincidental, but I also cannot see any clear discourse reason for Topic or Focus fronting of the subject. Among the verb-subject zero-relatives, when the clause contains more than simply an overt subject and verb, the dominant patterns are the same as those in marked relatives (as well as general, triggered verb-raising clauses): verbsubject-complement (334), verb-subject-PP (335), and verb-subject-adverb (336), except when the complement is a cliticized pronoun attached to the verb (337) or the PP is light, and thus raised with the verb (338). (334) ‫ירוּשׁ ָלםִ׃‬ ָ ‫ת־בּית יְ הוָ ה ֲא ֶשׁר ִבּ‬ ֵ ‫ֹלהים ֶאת־רוּחֹו ַל ֲעלֹות ִל ְבנֹות ֶא‬ ִ ‫ְלכֹל ֵה ִעיר ָה ֱא‬ ‘with regard to everyonei whi (that) God had stirred hisi spirit to go up and rebuild the House of Yhwh that was in Jerusalem’ (Ezra 1:5; see also 1 Kgs 17:14) 22.  See Jonah 4:10; Ps 122:4; 129:6; 136:23; Song 1:6c, 7b; 5:9; Eccl 4:2. 23.  See Exod 13:8; Ps 78:54. There are no examples of a ‫ז‬-relative containing an overt subject with the order subject–(finite) verb; there are also no examples of the order complement/adjunct–(finite) verb. 24.  There are no examples of a ‫מ‬-relative’s containing an overt subject with the order subject–(finite) verb; there are also no examples of the order complement/adjunct–(finite) verb. 25.  Gen 1:1; 42:28; Exod 6:28; 15:17; Deut 4:15; 32:17, 37; 1  Kgs 17:14; 22:24; Isa 29:1; 42:1, 11; 51:12; 63:19; Jer 2:6 (2×); 23:9; 51:43 (2×); Mic 5:2; Ps 4:8; 32:2; 78:8; 103:5; 118:22; Job 13:28; 18:21; 28:7 (2×); 38:19, 24; Prov 6:16; Eccl 1:13; Ezra 1:5; 2 Chr 1:4; 18:23; 28:9; 29:21.

Relative Syntax

153

(335) ‫וַ יְ ִהי ְבּיֹום ִדּ ֶבּר יְ הוָ ה ֶאל־מ ֶֹשׁה ְבּ ֶא ֶרץ ִמ ְצ ָריִ ם‬ ‘and it was, on the dayi whi (that) Yhwh spoke to Moses in the land of Egypt ___i’ (Exod 6:28; see also 1 Kgs 22:24; Eccl 1:3;   2 Chr 18:23) (336) ‫אָדם ָשׁם׃‬ ָ ‫א־ע ַבר ָבּהּ ִאישׁ וְ לֹא־יָ ַשׁב‬ ָ ֹ ‫ְבּ ֶא ֶרץ ל‬ ‘in a landi (that) no one passes through it and whi (that) no one lives therei?’ (Jer 2:6) (337) ‫ֹלהים לֹא יְ ָדעוּם ֲח ָד ִשׁים ִמ ָקּר ֹב ָבּאוּ לֹא ְשׂ ָערוּם‬ ִ ‫ֹלה ֱא‬ ַ ‫יִ זְ ְבּחוּ ַל ֵשּׁ ִדים לֹא ֱא‬ ‫יכם׃‬ ֶ ‫ֲאב ֵֹת‬ ‘They sacrificed to demonsi whi (that) ___i are not Gods, deitiesj whj (that) they had never known themh, new onesk whk (that) ___k recently arrived, whk (that) your ancestors had not feared ___k’   (Deut 32:17; see also Jer 23:9; Job 13:28; 28:7 [2×]) (338) ‫ל־אישׁ וְ לֹא־יַ ֲעבֹר ָבּ ֵהן‬ ִ ‫יה ְל ַשׁ ָמּה ֶא ֶרץ ִציָּ ה וַ ֲע ָר ָבה ֶא ֶרץ לֹא־יֵ ֵשׁב ָבּ ֵהן ָכּ‬ ָ ‫ָהיוּ ָע ֶר‬ ‫ן־אָדם׃‬ ָ ‫ֶבּ‬ ‘her cities have become an object of horror, a land of drought and a desert, a landi whi (that) no one lives in themi, and whi (that) no human passes through themi’ (Jer 51:43 [2×]; see also 2 Chr 1:4) And yet, within the verb-subject clauses, there is also evidence of discoursepragmatic influences, such as Topic or Focus-marking, which results in further constituent movement. For example, in (339) a non-light PP is positioned ahead of the subject; in (340), a PP is raised in front of the verb, and in (341) a light PP is left in situ below the subject. (339) ‫עוּריְ ִכי׃‬ ָ ְ‫ַה ַמּ ְשׂ ִבּיַ ע ַבּטֹּוב ֶע ְדיֵ ְך ִתּ ְת ַח ֵדּשׁ ַכּנֶּ ֶשׁר נ‬ ‘[he (v. 2)] who satisfies with goodness youri ornamentation, whi (that) like the eagle your youth is renewed’ (Ps 103:5) (340) ‫אמר ָל ֶהם‬ ֶ ֹ ‫וְ ָשׁם ָהיָ ה נָ ִביא ַליהוָ ה ע ֵֹדד ְשׁמֹו וַ יֵּ ֵצא ִל ְפנֵ י ַה ָצּ ָבא ַה ָבּא ְלשׁ ְֹמרֹון וַ יּ‬ ‫גוּ־בם ְבזַ ַעף ַעד‬ ָ ‫הוּדה נְ ָתנָ ם ְבּיֶ ְד ֶכם וַ ַתּ ַה ְר‬ ָ ְ‫יכם ַעל־י‬ ֶ ‫בֹות‬ ֵ ‫י־א‬ ֲ ‫ֹלה‬ ֵ ‫ִהנֵּ ה ַבּ ֲח ַמת יְ הוָ ה ֱא‬ ‫ַל ָשּׁ ַמיִ ם ִהגִּ ַיע׃‬ ‘behold, in the anger of Yhwh, the God of your ancestors, against Judah, he gave them into your hand, and you killed them in a ragei whi (that) up to heaven ___i reached’ (2 Chr 28:9) (341) ‫אַמ ַתּ ְח ִתּי וַ יֵּ ֵצא ִל ָבּם וַ יֶּ ֶח ְרדוּ ִאישׁ ֶאל־‬ ְ ‫הוּשׁב ַכּ ְס ִפּי וְ גַ ם ִהנֵּ ה ְב‬ ַ ‫ל־א ָחיו‬ ֶ ‫אמר ֶא‬ ֶ ֹ ‫וַ יּ‬ ‫ֹלהים ָלנוּ׃‬ ִ ‫אָחיו ֵלאמֹר ַמה־זֹּאת ָע ָשׂה ֱא‬ ִ ‘and he said to his brothers, “My silver has been returned and moreover—look!—it is in my sack!”; and their heart left (them) and

154

Chapter 5 they trembled toward each other, saying “What is thisi whi (that) God has done ___i to us!?”’ (Gen 42:28; see also Ps 32:2)

For the fronted PP in (340), Focus-raising makes contextual sense, such that the Focus-marking communicates the extreme sort of action/event that the prophet Oded then uses to chastise the Israelites for going too far in their treatment of the Judeans. For both (339) and (341), the larger patterns of Hebrew constituent order suggest that the subject NP ‫ נעוריכי‬in (339) and the PP ‫לנו‬ in (341) are positioned at the right edge of their clauses for right-edge Focus (see Holmstedt 2014b). Similar discourse-pragmatic influences also explain the variation in those examples lacking an overt subject, such as the PP–verb order in Exod 15:17, Ps 46:2, Job 17:3, 21:27, 31:12 [2×], and 2 Chr 28:9, and the complement-verb-PP order in Ps 141:9 and Prov 8:32. In this section, I described the basic syntactic structure within Hebrew relative clauses. This included discussions of the syntactic position from which the relative head is promoted (i.e., the extraction site) and word-order variation, including an introduction to the basic phrase structure of Hebrew relatives. In the next section, I consider the properties of relative clauses that have null (that is, non-overt) relative heads. 5.2.1.  Syntax of Null-Head Relative Clauses The structure of null-head (∅) Hebrew relatives is the same as that of the more common overtly-headed relatives. Null-head Hebrew relatives contain a null-relative pronoun (wh) and, typically, an overt relative complementizer, such as the ‫אשׁר‬-relative in (342), the ‫שׁ‬-relative in (343), the ‫זה‬-relative in (344), and the ‫ה‬-relative in (345). (342) ‫ית ּלֹו‬ ָ ‫ר־ע ִׂש‬ ָ ‫ף־א ִחיָך ִמ ְּמָך וְ ָׁש ַכח ֵאת ֲא ֶׁש‬ ָ ‫ַעד־ׁשּוב ַא‬ ‘until the anger of your brother turns from you and he forgets ∅(the thing)i whi that you did ___i to him’ (Gen 27:45) (343) ‫אתיו׃‬ ִ ‫אָה ָבה נַ ְפ ִשׁי ִבּקְַּשׁ ִתּיו וְ לֹא ְמ ָצ‬ ֲ ‫ל־מ ְשׁ ָכּ ִבי ַבּ ֵלּילֹות ִבּ ַקּ ְשׁ ִתּי ֵאת ֶשׁ‬ ִ ‫ַע‬ ‘upon my bed at night I sought ∅(the one)i whi that my soul loves ___i; I sought him, but found him not’ (Song 3:1) (344) ‫כוּ־בי׃‬ ִ ‫ה־אָה ְב ִתּי נֶ ְה ְפּ‬ ַ ֶ‫וְ ז‬ ‘and ∅(the people)i whi that I loved ___i have turned (their back) on me’ (345) ‫יאל ַהגֵּ ְר ֻשׁנִּ י׃‬ ֵ ‫אֹוצר ֵבּית־יְ הוָ ה ַעל יַ ד־יְ ִח‬ ַ ‫וְ ַהנִּ ְמ ָצא ִאתֹּו ֲא ָבנִ ים נָ ְתנוּ ְל‬ ‘and ∅(the person)i whi that with himi were found precious stones gave them to the treasure of the House of Yhwh, by the hand of Jehiel the Gershonite’ (1 Chr 29:8)

Relative Syntax

155

Though the variety with an overt relative word is much more common—the lack of both an overt head and an overt relative word understandably makes a relative clause challenging to identify and process—null-head zero-relatives, though rare, are attested (346). (346) (= [232]) ‫ד־תּ ְשׁ ָלח׃‬ ִ ַ‫ִבּי ֲאד ֹנָ י ְשׁ ַלח־נָ א ְבּי‬ ‘by me, O Lord, send [the message] by the hand of ∅(the man)i whi (that) you shall send ___i’ (Exod 4:13) 26 Like overtly-headed relatives, the null head of null-head relatives may function syntactically in its matrix clause as a subject (344) and (345), a complement (342) and (343), or even an adjunct—e.g., the clitic host in the bound relationship in (346). Finally, constituent order in null-head relatives reflects the same patterns and pragmatic motivations discussed above for overtly-headed relatives. 5.2.2.  Preposition Stranding, Pied-Piping, and Stacking Hebrew relatives have thus far exhibited syntax-related constraints on the placement of the relative head and relative word (externally-headed and postnominal; see §§4.1.2–3) and constituent order (the wh relative pronoun triggers verb-raising, resulting in a highly dominant verb-subject-complement order), but relative freedom concerning the syntactic role of the relative head (e.g., subject, complement, or adjunct). Three additional constraints on Hebrew relative clause formation concern preposition-stranding, pied-piping, and stacking. First, unlike English (e.g., the car that I rode in), Hebrew does not allow preposition stranding. Preposition stranding is the phenomenon wherein the preposition remains behind after the object/complement of the preposition has been moved up in the clause, as in the English example that’s the pen that I write with (see Haegeman 1994: 375). Instead, in Hebrew, objects of prepositions are either resumed, as in (347) (see below, §5.2.3, on resumption), or the preposition is deleted along with the relativized NP, as in (348). (347) ‫כוּ־בהּ‬ ָ ‫טֹּובה ֲא ֶשׁר יֵ ְל‬ ָ ‫ת־ה ֶדּ ֶרְך ַה‬ ַ ‫ֶא‬ ‘the good wayi whi that they shall walk in iti’ (1 Kgs 8:36) (348) ‫ַבּ ֶדּ ֶרְך ֲא ֶשׁר ָה ָל ְכ ִתּי‬ ‘on the wayi whi that I walked ___i’ (Gen 35:3) Second, Hebrew also disallows pied-piping (e.g., English the car in whichi I rode ___i). Pied-piping is the process whereby the noun within 26.  See above, §4.3, n. 18.

156

Chapter 5

a PP is relativized and the preposition is kept with the relativized noun, i.e., I rode in the car becomes the car in whichi I rode ti. In English, piedpiping involving phrases such as in which, with which, etc., is always legitimate (Haegeman 1994: 375). The opposite is true in Hebrew: it is never legitimate. 27 When a preposition precedes the relative word (e.g., ‫באשׁר‬‎, ‫)כאשׁר‬, the preposition belongs to the matrix clause, not to the following relative clause, as in (349). (349) ‫אָלין‬ ִ ‫וּב ֲא ֶשׁר ָתּ ִלינִ י‬ ַ ‘and in ∅i (the place) whi that you lodge ___i I will lodge’   (Ruth 1:16) Finally, Hebrew relatives may be stacked; that is, a string of juxtaposed relatives may modify the same initial head (the presence of the coordinating ‫ ו‬between relatives is irrelevant). Similarly, English allows the stacking of restrictive relatives: The article (that John wrote) (that was published last month), where the successive relatives are marked with parentheses. Compare (350), in which the two Hebrew relative clauses ‘that is in their land’ and ‘that they bring to Yhwh’ modify the head ‘the firstborn of everything’. (350) ‫ר־יָביאוּ ַליהוָ ה ְלָך יִ ְהיֶ ה‬ ִ ‫אַר ָצם ֲא ֶשׁ‬ ְ ‫ל־א ֶשׁר ְבּ‬ ֲ ‫כּוּרי ָכּ‬ ֵ ‫ִבּ‬ ‘the firstborn of everythingi (whi that is in their land), (whi that they bring to Yhwh) shall be yours’ (Num 18:13) 28 Comparison of the English example in (351) and the Hebrew examples in (352) illustrates that Hebrew (unlike English) allows the stacking of non­ restrictive relatives. 27.  In the entire Hebrew Bible, out of over 600 examples of a preposition followed by ‫אשׁר‬, there are only 4 examples for which it has been suggested that the preposition appears to belong better with the relative clause than with the matrix clause: Gen 31:32; Isa 47:12; Ezek 23:40; and Ps 119:49. However, the parallelism in Isa 47:12 and the verb within the relative in Ezek 23:40 suggest that the prepositions preceding the ‫ אשׁר‬in these two examples are part of the matrix clause and that the ‫ אשׁר‬clause is a null-relative. See Parunak (1996: 109), who suggests a possible pragmatic solution to Gen 31:32. And the ‫ על‬PP in Ps 119:49 appears to me to function as an appositive to the initial complement, ‫דבר‬, thus: “remember the word to your servant, (that is) concerning (the way) that you gave me hope’. 28.  See also Lev 3:4; 3:10; 3:15; 4:9; 4:18; 7:4; 20:10; 25:45; Num 18:13; 19:2; 35:34; Deut 4:46; 9:26; 11:10; 18:20, 21–22; 19:4; 21:3; 32:46, 49; Josh 13:21; 24:15; Judg 6:11; 18:22; 1 Sam 6:15; 12:13; 1 Kgs 2:44; 9:21; 11:34; 12:8; 21:25; 2 Kgs 11:10; 19:6; 20:18; 23:7, 12–13, 19; Isa 30:9–10; 39:7; 41:8, 9; Jer 7:14; 13:4; 21:4; 32:2, 3; 46:2; Ezek 20:9; 26:17; 32:24; 37:25; 44:10; 48:11; Amos 5:26; Mic 1:1; Nah 3:8; Pss 58:5–6; 78:42–43; 95:3–5; 104:16–17; Job 36:27–28; Ruth 4:15; Eccl 8:15; Dan 1:10; 2 Chr 2:11; 6:14–15; 8:8; 23:9. These data contradict O’Connor’s claim that “Hebrew only stacks relative clauses in verse [i.e., poetry]” (O’Connor 1997: 127).

Relative Syntax

157

(351) *I saw John, who was yelling, who(m) you dislike. 29 vs. I saw the boy that was yelling that you dislike. (352) ‫ֹותי וְ ֻחקּ ָֹתי‬ ַ ‫ְל ַמ ַען ָדּוִ ד ַע ְב ִדּי ֲא ֶשׁר ָבּ ַח ְר ִתּי אֹתֹו ֲא ֶשׁר ָשׁ ַמר ִמ ְצ‬ ‘for the sake of Davidi, my servant, (whi that I chose himi), (whi that ___i has kept my commandments and my statutes)’ (1 Kgs 11:34) In (352), the two Hebrew relative clauses are clearly nonrestrictive since they modify the proper noun ‫( דוד‬on restrictiveness, see below, §§6.1.2; 6.3). Like the stacked ‫אשׁר‬-relatives in (350) and (352), both ‫שׁ‬-relatives (353) and ‫ה‬-relatives (354) also appear in stacked examples (there are no stacked ‫ז‬-series relatives, though this is no doubt due to the paucity of data). (353) ‫אַל־תּ ְראוּנִ י ֶשׁ ֲאנִ י ְשׁ ַח ְרח ֶֹרת ֶשׁ ֱשּׁזָ ַפ ְתנִ י ַה ָשּׁ ֶמשׁ‬ ִ ‘do not gaze at mei (whi that Ii am dark), (whi that the sun has browned mei)’ 30 (354) ‫‏ּב ֲר ִכי נַ ְפ ִׁשי ֶאת־יְ הוָ ה‬ ָ ‫מּוליו‬ ָ ְ‫ל־ּת ְׁש ְּכ ִחי ָּכל־ּג‬ ִ ‫וְ ַא‬ ‫ל־עֹונֵ ִכי‬ ֲ ‫ַהּס ֵֹל ַח ְל ָכ‬ ‫ל־ּת ֲח ֻל ָאיְ ִכי‬ ַ ‫ָהר ֵֹפא ְל ָכ‬ ‫ּגֹואל ִמ ַּׁש ַחת ַחּיָ יְ ִכי‬ ֵ ‫ַה‬ ‫ַה ְמ ַע ְּט ֵר ִכי ֶח ֶסד וְ ַר ֲח ִמים‬ ‫‏ה ַּמ ְׂש ִּביַ ע ַּבּטֹוב ֶע ְדיֵ ְך‬ ַ ‘Bless Yhwh, O my soul, and do not forget all hisi benefits— whi that ____i forgives all your iniquity, whi that ____i heals all your diseases, whi that ____i redeems your life from the pit, whi that ____i crowns you (with) kindness and compassion, whi that ____i satiates your ornaments with goodness’   (Ps 103:2–5) 31 29.  My use of commas in the English examples is simply to signal a nonrestrictive (with commas) versus restrictive (without commas) reading. Thus, in (351), the example with commas represents the grammatically unacceptable stacking of nonrestrictives relatives. 30.  See Song 1:6, 7; 4:2; 6:6; Eccl 2:18, 19; 31.  See also Pss 33:15; 103:2–5; 104:3; 113:5–6; 144:10; 147:8, 14–16; Job 9:4–7. Note that these are all ‫ה‬-participial relatives; among the small data set of ‫ה‬-relatives with finite verbs, there are no cases of stacking. There are also cases of unmarked participial relative stacking, primarily in the Psalms (in some cases, the ‫ה‬-participial and unmarked participial relatives alternate, without any discernible difference). See, for example, Pss 15:2; 18:33–35, 47–48; 104:2–4, 13–14, 32 (note the alternation between the ‫ה‬-participial relative and the unmarked-yiqtol relative); 106:3; 113:5–7, 9; 146:6–7; 147:2–4, 6, 8–9; 14–17, 19; Job 9:8–9.

158

Chapter 5

In (353), the successive ‫שׁ‬-relatives modify the 1cs clitic pronoun ‫ ני‬attached to the verb ‫תראו‬. Both ‫שׁ‬-relatives provide nonrestrictive information about the speaker, the first-person referent. In (354), which I have organized by the poetic stichometry, the 3ms clitic pronoun in ‫ גמוליו‬is the relative head (the 3ms pronoun refers back to ‫ יהוה‬in the first line). Each of the ‫ה‬-participial relatives in the 3rd–7th lines describes an additional activity of Yhwh, all of which together serve as the motivation for the psalmist’s call to bless Yhwh. On the rare but attested relative modification of pronominal heads (as in both examples above), see §§3.2.1 and 3.3. Finally, zero-relatives can be stacked, as in (355). 32 (355) ‫ית ֶּק ֶדם ּגָ ַא ְל ָּת ֵׁש ֶבט נַ ֲח ָל ֶתָך‬ ָ ִ‫זְ כֹר ֲע ָד ְתָך ָקנ‬ ‘remember your congregationi whi ∅(that) you acquired ___i long ago, whi ∅(that) you redeemed ___i (as) a tribe of your inheritance’ (Ps 74:2) 5.2.3. Resumption In §3.1 I introduced the basic typological features of resumptive pronouns, in §3.2 I described how resumptive pronouns may co-occur with the other Hebrew relative words, and in §4.3.2 the issue of resumption (or lack thereof) figures into the analytical criteria for distinguishing null-head relatives from nominalized complements of prepositions and conjunctions. In this section, I will resume the analysis of Hebrew relative resumption by considering two fundamental questions: (1) How does the syntax of relatives with overt resumption relate to relatives without overt resumption? And (2) what motivates the choice between overt resumption and the omission of resumption (or covert resumption): syntax, semantics, or pragmatics? 33 5.2.3.1.  Previous Studies on Hebrew Relative Resumption The issue of resumption figures prominently in many reference works on Hebrew, and most provide adequate if not linguistically insightful descriptions of where and when resumption does and does not occur in relative clauses. What has yet to be determined, however, is whether we can formulate an insightful explanatory account. In other words, Hebraists have long been able to discern the “rhyme” of resumption but not the reason. The following comments are a brief summary of what scholarship on the topic has so far provided for us. Ewald writes that the presence of a resumptive element is the result of the “dismemberment” of the relative word ‫( ֲא ֶשׁר‬i.e., that the “relative pronoun” 32.  This is the opposite of English syntax. English zero-relatives may not be stacked, as in *the book ∅ I bought ∅ Ann had recommended was boring (compare the book that I bought that Ann had recommended was boring); see McCawley 1998: 433. 33.  Among myriad general linguistics studies on resumption, see especially Sells 1984; Prince 1990; Demirdache 1991; Shlonsky 1992; Prince 1997; Suñer 1998; Sharvit 1999a.

Relative Syntax

159

has “lost” its inflection. This implies that resumptive pronouns should be present in all Hebrew relative clauses (since ‫ ֲא ֶשׁר‬had certainly lost its inflection by the biblical stage of the Hebrew language). Ewald accounts for the lack of resumption in many clauses as a means of brevity, as in null-copula clauses such as ‫‘ ָה ֲאנָ ִשׁים ֲא ֶשׁר __ ִאּתֹו‬the-men who (they) (were) with-him’ (2 Sam 1:11; the ___ marks the missing resumptive constituent). Ewald also proposes (1891: 210–11) that the omission of a resumptive element is obligatory before finite verbs (which carry person, gender, and number morphological-agreement features), as in ‫‘ ֲא ֶשׁר __ ָא ַמר‬who (he) said’ (Gen 22:3). 34 He stipulates one case in which the resumptive element may never be omitted: “when the relative-word points to an idea which is to be closely subordinated” (Ewald 1891: 211). 35 Additionally, he observes that, when the relative clause is removed from its head noun by a significant distance, the resumptive element within the relative clause is almost always present and is often a full copy of the head constituent rather than a pronoun. Adding to the observations of Ewald, Kautzsch indicates that the “suppression of the retrospective pronoun takes places especially when it . . . would represent an accusative of the object, or when it would be a separate pronoun representing a nominative of the subject in a noun-clause” (GKC 445). He states further that omission of resumption is “noticeable” with verbs of speaking (verbum dicendi) and also that, with antecedents of place (e.g., ‫)מקֹום‬, ָ resumption may take the form of the adverb ‫‘ ָשׁם‬there’ (Kautzsch 1910: 445). Finally, Waltke and O’Connor add that resumption does not depend on the definiteness of the head of the relative since both definite and indefinite NPs are resumed within relatives (IBHS 333). Kropat suggests that the “younger language” (i.e., Late Biblical Hebrew) has a tendency to leave out resumption if it is in the subject or complement position within the relative (Kropat 1909: 67; cf. Polzin 1976). Kropat’s definition of resumption is strangely restrictive: he only considers resumption of the relative head by a subject pronoun, which is extremely rare throughout the Hebrew corpora, or a pronominal complement in the relative. He does not include examples of resumption as clitic pronouns within PPs or attached to NPs as possessive pronouns. Furthermore, Kropat dismisses the two examples of subject resumption (2 Chr 6:32 [1 Kgs 8:41]; 8:7 [1 Kgs 9:20]) and the three examples of complement resumption (1  Chr 6:50 [Josh 21:9]; 2  Chr 8:8 [1  Kgs 9:21]; 22:7) that do 34.  Contrary to Ewald’s statement in this matter, the following verses contain relative clauses in which both the subject pronoun and the finite verb are present: Deut 9:2; 2 Kgs 22:13; Jer 1:17; 29:25; 31:32; Zech 1:15; Mal 2:14; Job 19:27. 35.  This vague qualification suggests that Ewald connected the issue of resumption to the status of relative clauses vis-à-vis restrictiveness (see below, §6.3, on restrictiveness).

160

Chapter 5

appear in Chronicles as entlehnten (Kropat 1909: 67)—even though one of the examples, 2 Chr 22:7, does not correspond to earlier material at all. Peretz allows a broader definition of resumption that includes resumption within PPs or as possessive pronouns attached to NPs. He notes that the resumptive pronoun (‫ )הכיניו המוסב‬may occur in marked or unmarked (i.e., zero-) relatives, that it may occur in the subject (‫ )הנושׂא‬position, though mostly in the predicate (‫)המשׁלים‬, whether as the complement NP (‫המושׂא‬, or “object”), an adverbial constituent (‫)תיאור הפועל‬, or some other constituent to which the resumptive is attached (Peretz 1967: 86–87). Peretz provides examples of relatives with resumptives on PPs as well as relatives in which both the resumptive and preposition are omitted; he also suggests that subject resumption is obligatory when the relative clause is a null-copula clause, whereas in verbal clauses the subject is resumed by the agreement features of a finite verb: ‫הכינוי‬ ‫ הנשוא כלול בפועל‬,‫ כשמשפט הזיקה הוא שמני; אולם כשהוא פועלי‬,‫ אמור‬. . . ‫המוסב‬‎ (1967: 87, though see above, n. 34). Additionally, Peretz notes that resumption in the complement (‫ )מושׂא‬position may be almost freely omitted, except in cases where it is obligatory to avoid interpretive error (‫ למניעת טעות‬. . . ‫)נחוץ‬, when the relative is too far from the antecedent, when the resumptive is attached to an NP within a relative as a possessive pronoun, and when the resumptive is a pronominal clitic within a PP that is required by the verb (1967: 90). Finally, Peretz observes that resumption is often omitted following verbs of speech, when the antecedent specifies time or place, and to avoid having two prepositions side by side (1967: 91–92). There have been three short studies of resumption in Hebrew (Joosten 1993; Tsujita 1991; and Parunak 1996) that have moved the discussion in valuable directions. Joosten (1993) addresses agreement (“congruence”) and lack of agreement (“discongruence”) between a resumptive pronoun and its antecedent. Joosten argues that relative clauses with an overt relative word (e.g., ‫)אשׁר‬ are unequivocal regarding agreement. Contrast (356) and (357). (356) ‫יהוּדה‬ ָ ‫את ִמ‬ ָ ‫ר־בּ‬ ָ ‫ֹלהים ֲא ֶשׁ‬ ִ ‫ישׁ־ה ֱא‬ ָ ‫אַתּה ִא‬ ָ ‫ַה‬ ‘are youi the man of God whi that ____i (you) have come from Judah?’ (1 Kgs 13:14) (357) ‫ר־בּא יֹומֹו ְבּ ֵעת ֲעֹון ֵקץ‬ ָ ‫אַתּה ָח ָלל ָר ָשׁע נְ ִשׂיא יִ ְשׂ ָר ֵאל ֲא ֶשׁ‬ ָ ְ‫ו‬ ‘and youi, defiled wicked one, prince of Israel, whi that hisi day has come in the time of final punishment’ (Ezek 21:30) Joosten argues that examples such as (356)–(357) illustrate that, when an overt relative word is present, the clause is unambiguously marked as a relative clause; in other words, it does not matter syntactically that in (356) the 2ms head noun is resumed by a 2ms verb within the relative clause, whereas in (357) the 2ms head noun is resumed by a 3ms suffix. Based on these data, Joosten con-

Relative Syntax

161

cludes that the agreement features of the resumptive pronoun inside a marked relative clause have no syntactic significance: the clause is clearly marked as a relative apart from the agreement features of the resumptive element. In contrast, Joosten argues that agreement between the head noun and the resumptive element is prohibited in zero (unmarked)-relative clauses, as in (358). (358) ‫וְ ַע ָתּה ְשׁ ַמע יַ ֲעקֹב ַע ְב ִדּי וְ יִ ְשׂ ָר ֵאל ָבּ ַח ְר ִתּי בֹו‬ ‘and now, listen, Jacob my servant, and Israeli whi ∅ I have chosen himi’ (Isa 44:1) Joosten uses examples such as (358) to argue for the obligatory lack of agreement between the resumptive pronoun and the head of the relative clause in Hebrew zero-relatives (Joosten 1993: 279). He argues that, if the resumptive pronoun in (358) were 2ms and thus agreed with the (vocative case) head noun, i.e., ‫( ישׂראל בחרתי בך‬unattested), the result would be an ambiguous construction with two possible parsings: simple juxtaposition of a vocative and a verbal clause (i.e., ‘Israel, I have chosen you’), or a vocative modified by a relative clause (i.e., ‘Israel [whom] I have chosen’). In contrast, Joosten proposes that the lack of agreement in zero-relatives prohibits a nonrelative interpretation. Joosten’s argument faces a number of challenges. The first objection concerns his treatment of the vocative as second person. He concludes that the vocative should not be treated as third person (Joosten 1993: 279), a conclusion that is possible only after “excepting” all of the marked and zero-relative clauses that modify vocative heads and resuming them in the third person. 36 If, contrary to Joosten’s assertion, the vocative is indeed to be treated as third person for the purposes of grammatical agreement (if the vocative can be assigned person features at all), much of Joosten’s argument fails: there would no longer be a clear pattern of discongruence in zero-relatives between the head and the resumptive constituent within the relative. Clearly, the grammatical status of the vocative needs to be investigated in order to determine whether Joosten’s claim that vocative should be treated as second person is valid (for a brief discussion of the vocative, see IBHS 77; also see Miller 2010). The second objection to Joosten’s analysis concerns many of his grammatical assessments. For example, he presents example (359) as a zero-relative clause in which discongruence “signals subordination” (Joosten 1993: 278).

36.  The marked relative clauses that Joosten’s lists as “exceptions” are Ezek 21:30 and 26:17. The zero-relative clause that Joosten lists as an “exception” is 1 Sam 26:14. In addition, he arbitrarily discounts many of the relative clauses in hoy oracles (Joosten 1993: 279 n. 19). See Hillers 1983 on the syntax of hoy oracles.

162

Chapter 5

(359) ‫ח־פּיו‬ ִ ‫וּכ ִא ֵלּם לֹא יִ ְפ ַתּ‬ ְ ‫וַ ֲאנִ י ְכ ֵח ֵרשׁ לֹא ֶא ְשׁ ָמע‬ ‘and I am like a deaf man—I do not hear, and like a dumb mani whi ∅ does not open hisi mouth’ (Ps 38:14) In order to draw the conclusion that the 3ms verb within the relative as well as the resumptive 3ms possessive pronoun ‘his’ in the phrase ‫‘ פיו‬his mouth’ is discongruent with the head ‫‘ אלם‬a dumb man’, he must analyze the head NP ‘a dumb man’ as 1cs! Another questionable linguistic judgment concerns the identification of the head of the relative clause in (360). (360) ‫ֲאנִ י ַהגֶּ ֶבר ָראָה ֳענִ י‬ ‘I am the mani whi ∅/(that) ____i has seen affliction’ (Lam 3:1) In order to regard the relative clause in (360) as an example of discongruence between the head and the agreement features, Joosten must either view the head of the relative as the 1cs pronoun ‫( אני‬rather than the nearer NP ‫)הגבר‬, or view the NP ‫ הגבר‬as something other than 3ms. Neither option is acceptable. The fact that “the antecedent is the predicate to a first person subject” (Joosten 1993: 278) does not warrant the grammatical analysis of the predicate as first person. With regard to the agreement features of verbs within Hebrew relative clauses, as in (356)–(360), it is much simpler to argue that the agreement features merely serve to identify the precise antecedent of the relative clause. This proposal provides a consistent grammatical analysis for both the marked and zero-relative clause data. Thus, for example, in (361), the agreement features of the verb within the relative identify the distant 2ms pronoun ‘you’ as the antecedent (rather than the nearer vocative NP ‫)יעקב‬, whereas here, (362), agreement features of the verb within the relative identify either the 3ms NP ‫ חלל רשׁע‬or the 3ms NP ‫ נשׂיא ישׂראל‬as the head, both of which are nearer to the relative clause than the distant 2ms pronoun ‫אתה‬. (361) ‫אַתּה יִ ְשׂ ָר ֵאל ַע ְב ִדּי יַ ֲעקֹב ֲא ֶשׁר ְבּ ַח ְר ִתּיָך‬ ָ ְ‫ו‬ ‘and youi Israel, my servant, Jacob, whi that I have chosen youi’   (Isa 41:8) 37 (362) ‫ר־בּא יֹומֹו ְבּ ֵעת ֲעֹון ֵקץ‬ ָ ‫אַתּה ָח ָלל ָר ָשׁע נְ ִשׂיא יִ ְשׂ ָר ֵאל ֲא ֶשׁ‬ ָ ְ‫ו‬ ‘and you, defiled wicked one, prince of Israeli, whi that hisi day has come in the time of final punishment’ (Ezek 21:30) 37.  Although this does not directly pertain to the relative clause in (361), the initial string of constituents could represent an equative verbless clause—i.e., “you are Israel.” However, Isa 41:8–9 does not appear to present any main-line predicate but, rather, provides a rather long stretch of background information that identifies more specifically the addressee of the 41:10, which contains the main predicate in the form of a negative command, “Do not fear!”

Relative Syntax

163

With regard to the identification of zero-relative clauses, as in (358), repeated in (363), the solution boils down to a process of elimination rather than any one linguistic feature. (363) ‫ַע ָתּה ְשׁ ַמע יַ ֲעקֹב ַע ְב ִדּי וְ יִ ְשׂ ָר ֵאל ָבּ ַח ְר ִתּי בֹו‬ ‘and now, listen, Jacob my servant, and Israeli whi ∅ I have chosen himi’ (Isa 44:1) In (363), the string of constituents ‫ ישׂראל בחרתי בו‬has no other grammatical interpretation besides that of a zero-relative. The vocative NP ‫ ישׂראל‬cannot be the subject of the 1cs verb that follows it; nor can it be the complement, since the PP ‫ בו‬fulfills that role. Even if the resumption were in the 2ms (i.e., the unattested ‫ישׂראל בחרתי בך‬, a construction that would result in ambiguity, according to Joosten), there is only one logical way that a nonrelative analysis could be possible: if the phrase ‫ ישׂראל בחרתי בך‬was the content of the message. In other words, we would have to translate the verse as follows: ‘and now listen, Jacob, my servant: Israel, I have chosen you’. 38 However, the fact that the content of the message is clearly introduced in the following verse by the phrase ‫כה אמר יהוה‬‎ (44:2), which is then followed by second-person commands, indicates that the larger grammatical context prohibits such an interpretation. 39 In summary, Joosten’s proposal—that a difference in the agreement between the head and a resumptive constituent within a relative distinguishes overtly marked from zero-relative clauses and that the lack of agreement is necessary in order to signal the presence of zero-relative clauses—is an intriguing solution to a complex issue. However, it faces serious problems: not only are person features of vocative NPs uncertain (a significant component in Joosten’s analysis), but many of Joosten’s grammatical assessments are questionable. A simpler solution, which applies to both marked and zero-relative clauses in Hebrew, is to view the type of agreement features that Joosten has noticed as a way of identifying the precise head of the relative when multiple choices are available. Tsujita (1991) briefly addresses resumptive elements standing in the direct object position within the Hebrew relative clause (he lists 15 occurrences from the books of Genesis and Deuteronomy). Using these data, he argues against the position that the presence of a resumptive pronoun within the relative clause is entirely optional or stylistic. Rather, arguing on the basis of syntax, 38.  For the presence and function of the conjunction ‫ ו‬at the beginning of direct discourse, see Miller 1999. 39.  One interesting feature of resumptive constituents in zero-relatives that can be identified from Joosten’s article (although he does not highlight this) is that zero-relatives seem to modify the nearest antecedent only. In contrast, overtly marked relatives (probably because they are unambiguously marked as relatives) can modify antecedents that are not immediately preceding the relative word.

164

Chapter 5

information structure, and semantics, Tsujita concludes that there are three possible motivations for the presence of resumptive pronouns. First, he asserts that resumptive pronouns in the object position may serve to avoid ambiguity regarding the syntactic function of the antecedent, as in (364); compare the “ambiguous” clause in (365). (364) ‫ֹלהים ֲא ֶשׁר לֹא־יְ ָדעוּם‬ ִ ‫ֹלהים ֲא ֵח ִרים וַ יִּ ְשׁ ַתּ ֲחוּוּ ָל ֶהם ֱא‬ ִ ‫וַ יַּ ַע ְבדוּ ֱא‬ ‘and they served other gods, and they bowed down to them, godsi whi that they did not know themi’ (Deut 29:25) (365) ‫ת־הגֹּויִ ם ֲא ֶשׁר לֹא ָשׁ ֵמעוּ‬ ַ ‫ֶא‬ ‘the nationsi whi that ___i do not hear’ or ‘the nationsi whi that they did not hear ___i’ (Mic 5:14) 40 According to Tsujita, in (364) the object pronoun suffixed to the verb signals that the head of the relative corresponds to the object position within the relative, rather than to the subject position. In (365), there is no object pronoun, and hence, the relative head ‫ גוים‬could be the subject or object of the verb within the relative. Thus, according to Tsujita, some resumptive pronouns are used to signal clearly how the head of the relative functions within the relative clause. The second part of Tsujita’s argument is that resumptive pronouns serve to highlight the rheme 41 within the information structure of the discourse, as in (366). In this example, the object pronoun within the relative clause agrees with the rheme of the matrix clause, ‫אני‬, rather than with the nearer theme, 42 ‫יוסף אחיכם‬. (366) ‫ר־מ ַכ ְר ֶתּם א ִֹתי ִמ ְצ ָריְ ָמה׃‬ ְ ‫יכם ֲא ֶשׁ‬ ֶ ‫יֹוסף ֲא ִח‬ ֵ ‫ֲאנִ י‬ ‘Ii (am) Joseph, your brother, whi that you sold mei to Egypt’   (Gen 45:4) 40.  Strictly speaking, this clause is syntactically ambiguous; however, contextually the second reading (with the head as the object of the verb inside the relative) is highly unlikely. This raises the issue of whether syntactic ambiguity is a valid diagnostic for the analysis of ancient Hebrew, since we are unable to produce or elicit examples judged by native speakers as being truly ambiguous. In this, we are confronted with the difficulties of analyzing an ancient, unspoken language. 41.  Tsujita does not define his use of rheme. Crystal provides the following basic definition: “The rheme is defined as the part of a sentence which adds the most to the advancing process of communication . . . ; in other words, it expresses the largest amount of extra meaning, in addition to what has already been communicated” (Crystal 2008: 416). 42.  According to Crystal (2008: 484), the theme stands in opposition to the rheme as the part of the sentence that “adds least to the advancing process of communication . . . ; in other words, it expresses relatively little (or no) extra meaning, in addition to what has already been communicated.”

Relative Syntax

165

Tsujita argues that, in agreeing with a further antecedent, the resumptive pronoun highlights or emphasizes that constituent (i.e., ‫ אני‬in [366]). The weakness of this proposal is twofold. First, the identification of rheme and theme in null-copula clauses, such as (366), is difficult: it is not entirely clear why Tsujita identifies ‫ אני‬as the rheme and ‫ יוסף אחיכם‬as the theme. Since Joseph is revealing his identify, one could easily argue that ‘Joseph, your brother, who(m) you sold me into Egypt’ is the new or rhematic information. Second, as I have suggested above, it is simpler to argue that the resumptive pronoun may merely serve to identify the precise antecedent; in the case of (366), the resumptive pronoun identifies the more distant 1cs pronoun ‫ אני‬as the antecedent rather than the NP ‫יוסף אחיכם‬. Finally, Tsujita draws a connection between the presence of resumptive pronouns and “the determination of the antecedent.” Such cases, illustrated in (367), “make the antecedent more salient (prominent) and . . . lend greater specificity to the determination of the antecedent” (Tsujita 1991: 1581). (367) ‫ל־פּנִ ים׃‬ ָ ‫א־קם נָ ִביא עֹוד ְבּיִ ְשׂ ָר ֵאל ְכּמ ֶֹשׁה ֲא ֶשׁר יְ ָדעֹו יְ הוָ ה ָפּנִ ים ֶא‬ ָ ֹ ‫וְ ל‬ ‘and a prophet has not yet risen in Israel like Mosesi whi that Yhwh knew himi personally’ (Deut 34:10) Unfortunately, Tsujita does not address precisely how the resumptive constituent in (367) makes the subject ‫ משׁה‬more specific in the verse. In summary, while many of Tsujita’s proposals are suggestive, and in fact, point in the direction of potentially valid descriptions of resumption in Hebrew relatives, as they stand they are inadequate. Parunak (1996) also examines the feature of resumption in Hebrew relative clauses based on the book of Genesis. Since he draws a number of careful conclusions in his study, it is worthwhile to analyze his argument by the categories that he uses: resumption in the “nominative slot,” resumption in the “genitive slot,” and resumption in the “accusative slot.” Parunak first examines the presence or absence of resumptive constituents in the subject/nominative position within the relative clause. Based on the Genesis data, he makes the following observations. First, when the relative clause is a finite verbal clause, there is never resumption at the nominative slot. Second, in null-copula clauses, there are only eight occurrences of resumption: five are when a negative occurs inside the relative; three are when “in the absence of the resumption, there would be no noun to serve as the predicate of the ‫ אשׁר‬clause, and thus no clause at all” (Parunak 1996: 106). Third, five of six nominative slots are resumed in relative clauses that contain a participle/adjective as the copular complement; however, four of these also contain negatives, and the remaining example is “needed for reasons of balance” since “Genesis

166

Chapter 5

has no one-word ‫ אשׁר‬clauses” (1996: 107). 43 Finally, of 107 relative clauses with a PP as the copular complement, the single example that contains nominative resumption also contains a negative. On the basis of these observations, Parunak concludes that “[r]esumption of the pivot [i.e., the head of the relative] into a nominative slot is extremely rare. Most cases are examples of the negatives or existential transform, and are consistent with the hypothesis that ‫ אשׁר‬indeed resumes the pivot except when some other element must be clause-initial” (Parunak 1996: 108). Thus, Parunak suggests that resumption that does occur in the nominative position is motivated purely by syntactic concerns. If we look outside the book of Genesis, we see that resumption in the nominative position is indeed rare (there are only 40 examples in the entire Hebrew Bible); 44 however, the additional data suggest that Parunak’s conclusions for resumption at the nominative position are not wholly accurate. First, there are a few examples of nominative resumption in relatives that contain a verbal clause; since none occurs in Genesis, Parunak does not consider this type. Second, the proposal that negation requires resumption only explains 11 of the 40 examples (Gen 7:2, 8; 17:12; 30:33; Num 17:5; Deut 17:15; 20:15; 29:14; 43.  Parunak has a lengthy paragraph discussing the use of participial predicates in ‫אשׁר‬ relative clauses. He begins with the statement that “compared with prepositional predicates, participial predicates (with or without resumption) are rare in ‫ אשׁר‬clauses” (Parunak 1996: 107). While it is true that null copulas with PP complements (his “prepositional predicates”) are the most common type of clause within ‫אשׁר‬-relatives besides a finite verbal clause, it is not quite accurate to describe “participial predicates” (i.e., a null-copula clause with a participial phrase as the copular complement) as “rare,” since there are nearly 350 occurrences in the Hebrew Bible. What is more interesting, however, is the hypothesis that he supplies for the reason why an ‫ אשׁר‬relative clause with a participle was chosen over what he calls “an attributive participle” (and what I analyze as a ‫ה‬-relative; see above, §3.2.3); he suggests that the ‫אשׁר‬-relative “puts the modifier in focus” (1996: 107). I propose a very different explanation, one that is based in the constraints of ‫ה‬-relatives versus the greater flexibility of ‫אשׁר‬-relatives. Simply, ‫ה‬-relatives (as well as unmarked participial relatives) are required to relativize the subject position (i.e., the relative head is promoted from the subject position within the relative clause). In contrast, ‫אשׁר‬-relatives allow promotion of the head from any syntactic position, regardless of the type of predicate. See, for example, ‫אַתּה שׁ ֵֹכב‬ ָ ‫אָרץ ֲא ֶשׁר‬ ֶ ‫ָה‬ ‫יה‬ ָ ‫‘ ָע ֶל‬the landi whi that you are lying upon iti’ (Gen 28:13). Such modification is not possible with ‫ה‬-participial relatives (see §3.2.3). 44.  The 40 examples break down as follows. There are 30 cases of nominative (i.e., subject) resumption in null-copula clauses within a relative (Gen 7:2, 8; 9:3; 17:12; 30:33; Lev 11:29, 39; Num 9:13; 17:5; 35:31; Deut 17:15; 20:15; 29:14; 1 Kgs 8:41; 9:20; 2 Kgs 25:19; Jer 40:7; Ezek 12:10; 20:9; 43:19; Hag 1:9; Ps 16:3; Song 1:6; Ruth 4:15; Eccl 4:2; 7:26; Neh 2:13, 18; 2 Chr 6:32; 8:7). There are only 5 examples of subject resumption in finite verbal clauses (1 Kgs 8:38; 2 Kgs 22:13; Ezek 14:5; Ruth 4:11; Eccl 9:9); only 1 of these is resumption by a subject pronoun (2 Kgs 22:13), whereas the remainder exhibit resumption by a co-referential nonpronominal NP. Finally, there are only 5 examples of subject resumption in participial clauses within relatives: Exod 5:8; Num 14:8; Deut 20:20, 1 Sam 10:19; Neh 2:13.

Relative Syntax

167

1 Kgs 9:20; 2 Chr 6:32; 8:7), and the existence of relative clauses that contain negation but not resumption (see Lev 27:22; Ezek 18:18; 36:31; Ps 119:85; Esth 4:16) suggests that negation cannot be connected to the phenomenon of resumption. Third, Parunak’s explanation for the few cases of resumption in Genesis that do not co-occur with negation—that resumption is necessary to create a grammatical clause 45 or that resumption is needed for “balance” (Parunak 1996: 107)—is not adequate. Although it is not precisely clear what Parunak means by the statement that “Genesis has no one-word ‫ אשׁר‬clauses” (1996: 107), if by this he means that there are no relative clauses with a one-constituent null-copula clause, then his assessment is inaccurate. There are clear examples with just a PP (see Gen 1:7, 29; 3:3; 6:17), an adjective (see Gen 5:5; 25:7), or an adverb (see Gen 19:11; 35:5; 41:48). Thus, with regard to resumption in the nominative (subject) position, it is clear that we must find a different explanation from the one that Parunak offers. Concerning resumption in the genitive slot within the relative, Parunak asserts that resumption is the norm. (It is important to note that, though Hebrew does not morphologically mark cases, Parunak considers all prepositional complements in Hebrew to be in the genitive case, since the complements of prepositions are marked as genitive in Semitic languages that have morphological case-marking such as Akkadian and Arabic.) First, he observes that possessive pronouns and pronominal complements of prepositions are obligatorily resumed when possession is required or when the preposition is present in the relative clause. Thus, Hebrew does not permit relative clauses such as the English example I saw the man whose hat I took, where the relative pronoun also indicates possession within the relative (compare the nonrelativized version, I saw the man. I took his hat.). Nor does Hebrew permit dangling prepositions within relative clauses, such as the English The girl who I talked with (see above, §5.2.2 on preposition stranding). Parunak does, however, suggest that the presence or absence of the entire PP within Hebrew relatives is “optional” and may have a discourse-based explanation, a proposal that I develop below in §5.2.3.3. 45.  Parunak’s assessment regarding resumption by the ancient Hebrew deictic ‫שׁם‬ ‘there’ is intriguing. While I do not agree with his assessment of the syntactic role fulfilled by a resumptive ‫שׁם‬, that it is a nominative predicate in a null-copula clause, I do believe that Parunak’s analysis is a move in the right direction. The presence of ‫ שׁם‬in many null-copula clauses within relatives is forced by the grammar so that the other constituent in the relative is not interpreted as an equative predication. In other words, taking Gen 2:11 as an example (‫ר־שׁם ַהזָּ ָהב‬ ָ ‫ל־א ֶרץ ַה ֲחוִ ָילה ֲא ֶשׁ‬ ֶ ‫‘ ָכּ‬the whole land of Havilai, whi that the gold is therei’), the resumptive ‫ שׁם‬allows a semantically acceptable interpretation. Without ‫שׁם‬, the equative predication within the relative clause, ‘the whole land of Havilai, whi that ___i is gold’, while grammatical, is nonsensical.

168

Chapter 5

Finally, we reach Parunak’s proposal for resumption in the accusative position—the most difficult type of resumption to explain. His solution is twofold. First, he suggests that some cases of resumption in the accusative depend on the semantics of the verb within the relative clause: The various accusative slots in the corpus represent a number of different (deep structure) cases. In some examples, including Patient [10×, 3 resumed], Place [16×, 9 resumed], Discourse [31×, 2 resumed], and Object of Transfer [34×, 1 resumed], both resumed and nonresumed examples can be found. However 79 slots that represent deep structure cases of Percept [9×, 0 resumed], Time [12×, 0 resumed], Action [24×, 0 resumed] or Product [34×, 0 resumed] never resume their pivots. (Parunak 1996: 111)

Second, Parunak suggests that the remaining cases of resumption (i.e., those that are not required by the semantics of the governing verb) are connected to the restrictiveness of the relative clause. He illustrates this connection by comparing the minimal pair given in (368) and (369). (368) ‫סוּרים‬ ִ ‫ל־בּית ַהסּ ַֹהר ְמקֹום ֲא ֶשׁר־אסורי ַה ֶמּ ֶלְך ֲא‬ ֵ ‫וַ יִּ ְתּנֵ הוּ ֶא‬ ‘and he put him in the round house, the placei whi that the prisoners of the king were confined ___i’ (Gen 39:20, Kethiv) (369) ‫יֹוסף אָסוּר‬ ֵ ‫ל־בּית ַהסּ ַֹהר ְמקֹום ֲא ֶשׁר‬ ֵ ‫וַ יִּ ֵתּן א ָֹתם ְבּ ִמ ְשׁ ַמר ֵבּית ַשׂר ַה ַט ָבּ ִחים ֶא‬ ‫ָשׁם׃‬ ‘and he put them [the chief cupbearer and the chief baker] in prison (at) the house of the captain of the guards, the round house, the placei whi that Joseph was confined therei’ (Gen 40:3) According to Parunak, the absence of resumption marks the relative clause in (368) as nonrestrictive, whereas presence of the resumptive adverb ‫‘ שׁם‬there’ in (369) marks that ‫ אשׁר‬clause as a restrictive relative (Parunak 1996: 112–13). Based on the context, he argues that, concerning example (368), “the reader needs to understand simply that Joseph is being placed in a royal prison. Any royal prison will do. But in 40:3 [example (369)], the stories about the baker and the butler depend on the fact that they are placed in the very same prison where Joseph is” (Parunak 1996: 112). While Parunak’s proposal is intriguing, since few have addressed the issue of restrictiveness in Hebrew relative clauses, the connection between resumption and restrictiveness in Hebrew is difficult to maintain. First, with regard to example (368), it is not clear how an indefinite noun (albeit cliticized to the following relative) can serve as the head of a nonrestrictive relative clause. Nonrestrictive clauses only modify heads that are referential by themselves— that is, their referent does not depend on the information provided by the relative clause (see below, §6.3). Yet, it is clear that ‘place’ (or even ‘the place’, if the noun in this instance is definite by virtue of being cliticized to the relative

Relative Syntax

169

clause) cannot be referential within the context of Gen 39:20. The fact that the head in (368) stands in apposition to ‘the round house’ suggests that the entire phrase ‘the place where the royal prisoners were confined’ serves to specify further ‘the round house’. Certainly, a bare NP such as ‘place’ or even ‘the place’ cannot serve this purpose apart from a heavily deictic or implicatureladen context. As a final note, relative clauses such as those given in (370) and (371) further suggest that there is no direct connection between the presence of resumption and restrictive relatives in Hebrew. (370) ‫ֹלהים ֲא ֵח ִרים‬ ִ ‫אַח ֵרי ֱא‬ ֲ ‫ן־ה ֶדּ ֶרְך ֲא ֶשׁר אָנ ִֹכי ְמ ַצוֶּ ה ֶא ְת ֶכם ַהיֹּום ָל ֶל ֶכת‬ ַ ‫וְ ַס ְר ֶתּם ִמ‬ ‫ֲא ֶשׁר לֹא־יְ ַד ְע ֶתּם׃‬ ‘and you shall depart from the way that I am commanding you today by walking after other godsi whi that you do not know ___i’   (Deut 11:28) (371) ‫ל־פּנִ ים׃‬ ָ ‫א־קם נָ ִביא עֹוד ְבּיִ ְשׂ ָר ֵאל ְכּמ ֶֹשׁה ֲא ֶשׁר יְ ָדעֹו יְ הוָ ה ָפּנִ ים ֶא‬ ָ ֹ ‫וְ ל‬ ‘and there has not risen another prophet in Israel like Mosesi, whi that Yhwh knew himi face to face’ (Deut 34:10) In both (370) and (371), the governing verb within the relative clause is ‫ידע‬ ‘to know’, presumably what Parunak would categorize as a verb of “Percept.” It is clear from the relative clause in (370) that the verb ‫ ידע‬does not require resumption in the accusative. However, we have resumption in the accusative position inside the relative clause in (371), a relative that is unquestionably nonrestrictive since the head is the personal name ‫משׁה‬. These two examples show a pattern exactly the opposite of what one would expect from Parunak’s study of the book of Genesis. In summary, while some of Parunak’s solutions may be accurate for the Genesis data, they do not explain the rest of the data in the Hebrew Bible; thus, even though his proposals are intriguing and have moved the discussion of relative clause resumption in a valuable direction, they cannot be taken as accurately descriptive of the grammar of Hebrew. 5.2.3.2.  Resumption and the Noun Phrase Accessibility Hierarchy Given that Hebrew reference works understandably abbreviate discussion of phenomena such as relative clause resumption and that the focused studies treated in the previous section inadequately describe resumption, it is important to consider the full range of resumptive strategies exhibited within the biblical corpus before investigating the finer syntactic and semantic issues. Let us begin with a simple set of minimal pairs using ‫ אשׁר‬clauses: (372) and (374) do not contain resumption (or, more accurately, they contain null resumption), while (373) and (375) do contain (overt) resumption.

170

Chapter 5

(372) ‫ר־ּד ֶּבר ִאּתֹו‬ ִ ‫וַ ּיַ ֵּצב יַ ֲעקֹב ַמ ֵּצ ָבה ַּב ָּמקֹום ֲא ֶׁש‬ ‘and Jacob set up a pillar in the placei whi that he [God] had spoken with him ___ i’ (Gen 35:14) (373) ‫ית־אל‬ ֵ ‫ֹלהים ֵּב‬ ִ ‫ת־ׁשם ַה ָּמקֹום ֲא ֶׁשר ִּד ֶּבר ִאּתֹו ָׁשם ֱא‬ ֵ ‫וַ ּיִ ְק ָרא יַ ֲעקֹב ֶא‬ ‘and Jacob named the placei whi that God had spoken with him therei “Bethel”’ (Gen 35:15) In (372), the head of the relative clause, ‫מקום‬, is not resumed within the relative clause, leaving a gap or null resumption in the verbal adjunct position. In contrast, the head NP ‫ מקום‬in (373) is resumed by the deictic adverb ‫ׁשם‬. Similar to the minimal pair in (372) and (373), which illustrates promotion from an adjunct position within the relative, the pair in (374) and (375) illustrates promotion (and thus overt and null resumption) from the verbal complement position. (374) ‫ר־ּת ְר ִּתי ָל ֶהם‬ ַ ‫ל־א ֶרץ ֲא ֶׁש‬ ֶ ‫ֶא‬ ‘to bring them from the land of Egypt to a landi whi that I explored ___i for them’ (Ezek 20:6) (375) ‫ל־ּבנֵ י יִ ְׂש ָר ֵאל‬ ְ ‫ּיֹוציאּו ִּד ַּבת ָה ָא ֶרץ ֲא ֶׁשר ָּתרּו א ָֹתּה ֶא‬ ִ ַ‫ו‬ ‘and they brought to the Israelites an evil report of the landi whi that they explored iti’ (Num 13:32) The relative head, ‫ארץ‬, in (374) is not resumed within the relative clause, leaving a gap (or null resumptive) as the verbal complement. In (375), however, the head ‫ הארץ‬is resumed at the verbal complement position within the relative clause, manifested as the 3fs clitic pronoun attached to the object marker ‫את‬. Although Hebrew disallows overt resumption in -‫ ה‬relatives, it allows it with all other relative elements. The remaining examples in this section illustrate the attested strategies with the three remaining relative elements: ‫אשׁר‬‎, ‫שׁ‬, and the ‫ז‬-series. There are three basic resumptive strategies in Hebrew: resumption by a (clitic or free) pronoun (376)–(378); resumption by a semantically related adverb (379)–(380) or noun (381); and, very rarely, resumption by a full copy of the head NP (382). (376) Resumption by a Pronoun in an ‫אשׁר‬-Relative Clause ‫וְ ִכ ֶּפר ַהּכ ֵֹהן ֲא ֶׁשר־יִ ְמ ַׁשח אֹתֹו‬ ‘and the priesti whi that one anoints himi . . . shall make atonement’ (Lev 16:32) (377) Resumption by a Pronoun in a ‫שׁ‬-Relative Clause ‫ֹלהיו‬ ָ ‫ַא ְׁש ֵרי ָה ָעם ֶׁשיֲ הוָ ה ֱא‬ ‘happy is the peoplei whi that Yhwh is itsi god’ (Ps 144:15)

Relative Syntax

171

(378) Resumption by a Pronoun in a ‫ז‬-Series Relative Clause ‫ר־צּיֹון זֶ ה ָׁש ַכנְ ָּת ּבֹו‬ ִ ‫ ַה‬. . .‫זְ כֹר‬ ‘remember . . . Mount Zioni, whi that you dwelt in iti’ (Ps 74:2) (379) Resumption by a Semantically Related Adverb in an ‫אשׁר‬-Relative Clause ‫ר־היָ ה ָׁשם ָא ֳהֹלה ַּב ְּת ִח ָּלה‬ ָ ‫ד־ה ָּמקֹום ֲא ֶׁש‬ ַ ‫ַע‬ ‘up to the placei whi that his tent was therei at the beginning’   (Gen 13:3 Kethiv) (380) Resumption by a Semantically Related Adverb in a ‫שׁ‬-Relative Clause ‫ ֶׁש ָּׁשם ָעלּו ְׁש ָב ִטים ִׁש ְב ֵטי־יָ ּה‬. . . ִ‫רּוׁש ַלם‬ ָ ְ‫י‬ ‘Jerusalemi, whi that the tribes, tribes of Yah would ascend therei’   (Ps 122:3–4) (381) Resumption by a Semantically Related Noun in an ‫אשׁר‬-Relative Clause ‫ל־ה ָא ָדם ְלכֹל ַע ְּמָך יִ ְׂש ָר ֵאל ֲא ֶׁשר יֵ ְדעּון‬ ָ ‫ל־ּת ִחּנָ ה ֲא ֶׁשר ִת ְהיֶ ה ְל ָכ‬ ְ ‫ל־ּת ִפ ָּלה ָכ‬ ְ ‫ָּכ‬ ‫ִאיׁש נֶ גַ ע ְל ָבבֹו‬ ‘every prayer, every supplication that comes from any man, any of your peoplei, Israel, whi that eachi knows the affliction of his heart’   (1 Kgs 8:38) (382) Resumption by a Full Copy of the Head NP in an ‫אשׁר‬-Relative Clause 46 ‫אָרץ‬ ֶ ‫ת־ע ַפר ָה‬ ֲ ‫ם־יוּכל ִאישׁ ִל ְמנֹות ֶא‬ ַ ‫אָרץ ֲא ֶשׁר ׀ ִא‬ ֶ ‫וְ ַשׂ ְמ ִתּי ֶאת־זַ ְר ֲעָך ַכּ ֲע ַפר ָה‬ ‫יִמּנֶ ה׃‬ ָ ‫גַּ ם־זַ ְר ֲעָך‬ 46. De Vries notes that, while there are rare cases in which postnominal relatives allow an appositive epithet within the relative (presumably appositive to a null resumptive), there exists an implicational universal such that postnominal relatives do not allow a full copy of the head noun to be repeated as the resumptive within the relative (2002: 36; 2005: 138–39). Hebrew confirms this universal, since resumption by a full copy of the relative head is extremely rare. For instance, out of the 5,500 ‫ אשׁר‬relatives, I have identified only three good candidates: the example in (382), Josh 22:31 (which some treat as an example of a causal ‫ ;אשׁר‬cf. Soggin 1972: 211; njps; but see below, §7.4) 1 Sam 2:23 (in which the apparent full copy resumption may reflect a textual conflation; see McCarter 1980: 81), and Isa 54:9. Besides the complexity of the conditional clause within the ‫ אשׁר‬relative in (382), the textual issues in 1 Sam 2:23, and the case for emendation in Isa 54:9 (see Blenkinsopp 2002: 357–59), it is clear that full copy resumption is highly dispreferred in Hebrew. In Holmstedt 2002, I also listed Gen 50:13, Jer 32:32, and Eccl 9:9 as possible examples, but I no longer consider these to be likely cases. Gen 50:13, ‫ת־ה ָשּׂ ֶדה ַל ֲא ֻחזַּ ת־‬ ַ ‫אַב ָר ָהם ֶא‬ ְ ‫ִבּ ְמ ָע ַרת ְשׂ ֵדה ַה ַמּ ְכ ֵפּ ָלה ֲא ֶשׁר ָקנָ ה‬ ‫ל־פּנֵ י ַמ ְמ ֵרא׃‬ ְ ‫ק ֶבר ֵמ ֵאת ֶע ְפר ֹן ַה ִח ִתּי ַע‬,ֶ is best understood as modifying the PN ‫המכפלה‬, which is not resumed overtly inside the relative, leaving the ‫ את השׂדה‬in the relative unrelated to the relative head. This results in better symmetry with the obviously similar example in Gen 49:30, ‫ת־ה ָשּׂ ֶדה‬ ַ ‫אַב ָר ָהם ֶא‬ ְ ‫י־מ ְמ ֵרא ְבּ ֶא ֶרץ ְכּנָ ַען ֲא ֶשׁר ָקנָ ה‬ ַ ֵ‫ל־פּנ‬ ְ ‫ַבּ ְמּ ָע ָרה ֲא ֶשׁר ִבּ ְשׂ ֵדה ַה ַמּ ְכ ֵפּ ָלה ֲא ֶשׁר ַע‬ ‫ת־ק ֶבר׃‬ ָ ַ‫מ ֵאת ֶע ְפר ֹן ַה ִח ִתּי ַל ֲא ֻחזּ‬. ֵ Kautzsch (GKC 445 n. 1) suggested that 49:30 was an example

172

Chapter 5 ‘and I will make your seed like the dirti of the land whi that if a man is able to count the dirti of the land, so your seed will be counted’   Gen 13:16

The resumptive constituent may serve in the subject 47 position within the relative, (383)–(384), in the complement position, (385)–(386), or as (or within) an adjunct phrase (387)–(390). (383) ‫ר־הוא ע ָֹׂשה ִע ְּמָך ִמ ְל ָח ָמה‬ ִ ‫ל־ה ִעיר ֲא ֶׁש‬ ָ ‫ית ָמצֹור ַע‬ ָ ִ‫ּובנ‬ ָ ‘and you shall build a siege-work against the cityi whi that iti is making war with you’ (Deut 20:20) (384) ‫ל־ּת ְראּונִ י ֶׁש ֲאנִ י ְׁש ַח ְרח ֶֹרת‬ ִ ‫ַא‬ ‘do not look at mei whi that Ii am dark’ (Song 1:6) (385) ‫ר־ה ַס ָּתה אֹתֹו ִאיזֶ ֶבל ִא ְׁשּתֹו‬ ֵ ‫ ֲא ֶׁש‬. . . ‫א־היָ ה ְכ ַא ְח ָאב‬ ָ ֹ ‫ַרק ל‬ ‘certainly there was no one like Ahabi . . . whi that Jezebel, his wife, incited himi’ (1 Kgs 21:25) of resumption by means of a fully copy of the head NP ‫שׂדה‬, but as with 50:13, the relative clause in 49:30, ‫אׁשר קנה אברהם את־הׂשדה‬, should be taken as modification of the nearest antecedent, ‫( ארץ כנען‬or perhaps ‫ )ממרא בארץ כנען‬rather than as modification of the distant NP ‫שׂדה המכפלה‬. Thus, the sequence of relatives is better analyzed and translated as follows (note the different subscripts to indicate the different heads and relatives): ‘In the cavei whi that ___i is in the field of Machpelahj, whj that ___j is near Mamre, in the land of Canaank, whk that Abraham bought the field ___k from Ephron the Hittite as a burial site’. Finally, I indicated in Holmstedt 2002 that the second relative in Jer 31:32 should be understood as a second, stacked relative modifying the head ‫ ברית‬and resumed by a full copy, ‫בריתי‬‎: ‫לֹא‬ ‫יתי‬ ִ ‫ת־בּ ִר‬ ְ ‫ר־ה ָמּה ֵה ֵפרוּ ֶא‬ ֵ ‫הֹוציאָם ֵמ ֶא ֶרץ ִמ ְצ ָריִם ֲא ֶשׁ‬ ִ ‫יקי ְביָ ָדם ְל‬ ִ ִ‫בֹותם ְבּיֹום ֶה ֱחז‬ ָ ‫ת־א‬ ֲ ‫ַכ ְבּ ִרית ֲא ֶשׁר ָכּ ַר ִתּי ֶא‬ ‫וְ אָנ ִֹכי ָבּ ַע ְל ִתּי ָבם נְ ֻאם־יְ הוָ ה׃‬. However, I have since decided that it is less likely that the head of the second relative is ‫ ;ברית‬rather, it is more likely an extraposed relative modifying the NP ‫אבותם‬, which is resumed by the pronoun ‫המה‬. 47.  It is important to distinguish between resumptive elements that are the subject constituents within the relative, as in (383) and (384), versus resumption by clitic pronouns attached to the subjects within the relative, as in (357). In these latter examples, the resumptive is in NP-internal position and is typically a possessive modifier; this kind of subject resumption is not uncommon. However, full resumption by means of a free subject pronoun is extremely rare in Hebrew (there are only 35 examples in the Hebrew Bible), and most of the examples exist within null-copula or participial relatives. The 40 examples break down in the following way. There are 30 cases of resumption in null-copula clauses within a relative (Gen 7:2, 8; 9:3; 17:12; 30:33; Lev 11:29, 39; Num 9:13; 17:5; 35:31; Deut 17:15; 20:15; 29:14; 1 Kgs 8:41; 9:20; 2 Kgs 25:19; Jer 40:7; Ezek 12:10; 20:9; 43:19; Hag 1:9; Ps 16:3; Song 1:6; Ruth 4:15; Eccl 4:2; 7:26; Neh 2:13, 18; 2 Chr 6:32; 8:7). There are 4 apparent examples of subject resumption in finite verbal clauses (1 Kgs 8:38; 2 Kgs 22:13; Ezek 14:5; Ruth 4:11). However, only 1 of these is resumption by subject pronoun (2 Kgs 22:13), whereas the other 3 exhibit resumption by a quantifier phrase (‫ אישׁ‬as ‘each’, ‫ כלם‬as ‘all of them’, and ‫ שׁתיהם‬as ‘both of them’, respectively) in apposition to a null resumptive. Finally, there are only 4 examples of subject resumption in participial relatives: Num 14:8; Deut 20:20; 1 Sam 10:19; Neh 2:13.

Relative Syntax

173

(386) ‫יחּנּו ָל ָא ָדם ֶׁשּיִ ְהיֶ ה ַא ֲח ָרי‬ ֶ ִ‫ ֶׁש ַאּנ‬. . . ‫ל־ע ָמ ִלי‬ ֲ ‫ת־ּכ‬ ָ ‫אתי ֲאנִ י ֶא‬ ִ ֵ‫וְ ָׂשנ‬ ‘and I hate all my acquisitionsi, . . . whi that I must leave iti to the man who comes after me’ (Eccl 2:18) (387) ‫ֹלהים עֹׁשֶר ּונְ ָכ ִסים וְ ָכבֹוד‬ ִ ‫ִאיׁש ֲא ֶׁשר יִ ֶּתן־לֹו ָה ֱא‬ ‘a mani whi that God gives to himi riches and wealth and honor’   (Eccl 6:2) (388) ‫ׁשּועה‬ ָ ‫ן־א ָדם ֶׁש ֵאין לֹו ְת‬ ָ ‫ל־ּת ְב ְטחּו ִבנְ ִד ִיבים ְּב ֶב‬ ִ ‫ַא‬ ‘do not trust in nobles, in a mani whi that there does not belong to himi (the ability to accomplish) deliverance’ (Ps 146:3) (389) ‫כּו־בּה‬ ָ ‫ּטֹובה ֲא ֶׁשר יֵ ְל‬ ָ ‫ת־ה ֶּד ֶרְך ַה‬ ַ ‫תֹורם ֶא‬ ֵ ‫ִּכי‬ ‘when you teach them the good wayi whi that they shall walk in iti’   (1 Kgs 8:36) (390) ‫ר־צּיֹון זֶ ה ָׁש ַכנְ ָּת ּבֹו‬ ִ ‫ ַה‬. . .‫זְ כֹר‬ ‘remember . . . Mount Zioni, whi that you dwelt in iti’ (Ps 74:2) While it may seem that resumption within the Hebrew relative clause may occur in any position, the appearance of most resumptive pronouns corresponds to the “Noun Phrase Accessibility Hierarchy” (NPAH) (391). (391) The Noun Phrase Accessibility Hierarchy (Keenan and Comrie   1977: 66) Subject > Direct Object > Indirect Object > Oblique > Genitive >    Object of Comparison The NPAH primarily addresses the positions from which languages can access a noun for relativization; Keenan and Comrie summarize the hierarchy as follows: On the basis of data from about fifty languages, we argue that languages vary with respect to which NP positions can be relativized, and that the variation is not random. Rather, the relativizability of certain positions is dependent on that of others, and these dependencies are, we claim, universal. The Accessibility Hierarchy (AH) . . . expresses the relative accessibility to relativization of NP positions in simplex main clauses. Here, “>” means ‘is more accessible than’. . . . The positions on the AH are to be understood as specifying a set of possible grammatical distinctions that a language may make. We are not claiming that any given language necessarily distinguishes all these categories, either in terms of RC formation or in terms of other syntactic processes. For example, some languages (e.g., Hindi) treat objects of comparison like ordinary objects of prepositions or postpositions. In such cases we treat these NPs as ordinary Obl[iques], and the O[bject of ] Comp[arison] position on the AH is unrealized (Keenan and Comrie 1977: 66).

174

Chapter 5

The hierarchy also strongly predicts the positions in which a language may use the resumptive pronouns. That is, Keenan and Comrie note that, if a language employs overt resumptive pronouns, it will use them first and more often in the less accessible positions “down” the NPAH, i.e., rightwards in (391) (Keenan and Comrie 1977: 92). Furthermore, however high on the NPAH a language uses resumption, such as in the indirect object position, it will also use overt resumption at each of the less accessible positions below the indirect object. The Hebrew patterns for overt resumption in relatives closely match those predicted by the NPAH. Beginning toward the right on the NPAH, overt resumption appears obligatory in Hebrew relatives when the head has been promoted from possessive (an NP-internal) position within the relative, as in (392). (392) ‫ל־א ִחינּו ֲא ֶׁשר ָר ִאינּו ָצ ַרת נַ ְפׁשֹו ְּב ִה ְת ַחנְ נֹו ֵא ֵלינּו וְ לֹא‬ ָ ‫ֲא ָבל ֲא ֵׁש ִמים ֲאנַ ְחנּו ַע‬ ‫ָׁש ָמ ְענּו‬ ‘surely we are guilty concerning our brotheri whi that we saw the distress of hisi life when he implored us but we did not listen’   (Gen 42:21) Similarly, when the verb within the relative selects a PP as its complement (Keenan and Comrie’s “oblique” position), and the relative head corresponds to the complement of the preposition, overt resumption is strongly preferred. This requirement also overlaps with the constraint on preposition stranding (see §5.2.2), since null resumption would leave the preposition “stranded” within the relative. Thus, in Hebrew either the whole PP is overt (393) or, much less commonly, the whole PP is null (394). Bare prepositions (i.e., prepositions without a pronominal or full NP object) are disallowed in Hebrew. (393) ‫ֹלהיָך ּבֹו‬ ֶ ‫ַּב ָּמקֹום ֲא ֶׁשר ְיִב ַחר יְ הוָ ה ֱא‬ ‘in the placei whi that Yhwh your god chooses iti’ (Deut 12:18) 48 48.  Most bivalent or trivalent verbs that take a PP complement exhibit much greater regularity (and a higher frequency of overt resumption) than the verb ‫‘ בחת‬to choose’. This verb exhibits nearly equal occurrences of overt resumption when the head is promoted from the complement position—either with ‫את‬+clitic pronoun (1 Kgs 11:34) or with the preposition ‫ב‬+clitic pronoun (16×: Gen 26:29; Num 16:5; 17:20; Deut 12:11, 18; 14:25; 16:7; 17:8, 15; Judg 10:14; 1 Sam 10:24; 1 Kgs 8:44; 11:32, 34; Ps 105:26; 2 Chr 6:34)—and null resumption (37×: Gen 6:2; Num 16:7; Deut 12:5, 14, 21, 26; 14:23, 24; 15:20; 16:2, 6, 11, 15, 16; 17:10; 18:6; 23:17; 26:2; 31:11; Josh 9:27; 1 Sam 8:18; 12:13; 2 Sam 15:15; 16:18; 19:39; 1 Kgs 3:8; 8:48; 11:13, 36; 14:21; 2 Kgs 21:7; 23:27; Neh 1:9; 9:7; 2 Chr 6:38; 12:13; 33:7). Some of the cases of null resumption are complicated by the presence of a resumptive adverb corresponding to the relative head within a further embedded (infinitival) clause inside the relative (12×: Deut 12:5, 21; 14:23, 24; 16:2, 6, 11; 26:2; 1 Kgs 11:36; 14:21; Neh 1:9;

Relative Syntax

175

(394) ‫ר־יִב ַחר יְ הוָ ה‬ ְ ‫ל־ה ָּמקֹום ֲא ֶׁש‬ ַ ‫ֶא‬ ‘to the placei whi that Yhwh chooses ___ i’ (Deut 12:26) Moving up (leftward on) the NPAH, overt resumption at the indirect object position does not occur (at least, I have not identified any clear examples). Further left along the NPAH is the direct object position, for which the Hebrew resumption evidence is complicated. Descriptively, it is clear that, when the object marker ‫ את‬is present, a resumptive noun or clitic pronoun is obligatory, as in (395), and when overt resumption is lacking, so also must the object marker be absent, as in (396). (395) ‫ת־מ ְׁש ָּפ ַטי ֲא ֶׁשר יַ ֲע ֶׂשה א ָֹתם ָה ָא ָדם וָ ַחי ָּב ֶהם ֲאנִ י יְ הוָ ה‬ ִ ‫ת־חּק ַֹתי וְ ֶא‬ ֻ ‫ּוׁש ַמ ְר ֶּתם ֶא‬ ְ ‘and you shall keep my statutes and my ordinancesi, whi that (if) a man does themi, he shall live by them’ (Lev 18:5) (396) ‫אּו־בם ֲא ֶׁשר־יַ ֲע ֶׂשה ָא ָדם‬ ָ ‫ּוב ִמ ְׁש ָּפ ֶטיָך ָח ְט‬ ְ ‘and against your ordinances—they sinned against themi, whi that (if) a man does ___ i, he will live by them’ (Neh 9:29) From this short survey, we are left with the following facts about resumption in Hebrew relatives: (1) overt resumption is obligatory when the resumptive pronoun is possessive; and (2) overt resumption may be present or absent when the head has been promoted from within a PP or from the NP complement position. What I will investigate in the next section is why overt resumption in Hebrew is present sometimes and absent elsewhere, as we have seen in (393)–(396). 5.2.3.3. Function In order to address one or more likely motivations for overt resumption in Hebrew relative clauses, I will commence by examining relatives containing null-copula clauses. The examples in (397)–(401) establish that null-copula relative clauses may consist of both a null copula and a null subject, resulting in the presence of a single overt constituent, the copular complement (contra Pat-El and Treiger 2008: 276). Though the data for one-part relative clauses are relatively few, it is interesting to note that 2 Chr 12:13). It is possible in these that the lower overt resumption blocks or makes the use of overt resumption in the main relative clause less felicitous, leaving only 25 cases of null resumption for ‫ בחר‬relative clauses. Though overt resumption may be required to disambiguate the head in some examples (e.g., Num 16:5; 1 Sam 10:24; 1 Kgs 11:32 [cf. 1 Kgs 11:36]; 11:34) and in others the resumption may be pragmatically motivated (e.g., to mark the head for Focus within the relative, e.g., Judg 10:14; Isa 41:8 [cf. Isa 43:10]; Jer 33:24), this does not account for all the examples, leaving ‫ בחר‬as a rare verb for which no clear pattern emerges regarding resumption and verbal valency within the context of a relative clause. On ‫בחר‬, see also n. 7 above.

176

Chapter 5

they exhibit significant diversity, with PPs, adverbs, adjectives, and NPs as the single overt constituent. 49 (397) Relative Consisting Solely of PP ‫ּוב ַא ְר ַּבע ֶע ְׂש ֵרה ָׁשנָ ה ָּבא ְכ ָד ְר ָלע ֶֹמר וְ ַה ְּמ ָל ִכים ֲא ֶׁשר ִאּתֹו‬ ְ ‘and in the fourteenth year, Chedorlaomer and the kingsi whi that ___i were with him, came’ (Gen 14:5) (398) Relative Consisting Solely of Adverb ‫ר־ׁשם‬ ָ ‫ת־ה ֱאמ ִֹרי ֲא ֶׁש‬ ָ ‫ויורׁש ֶא‬ ‘and he dispossessed the Amoritesi whi that ___i were there’   (Num 21:32 Qere) (399) Relative Consisting Solely of Adjective ‫רֹובים ַליהוָ ה‬ ִ ‫ר־ק‬ ְ ‫ַהּכ ֲֹהנִ ים ֲא ֶׁש‬ ‘the priestsi whi that ___i are near to Yhwh’ (Ezek 42:13) (400) Relative Consisting Solely of Adjective (with negative) ‫א־טֹובים‬ ִ ֹ ‫יכם ֲא ֶׁשר ל‬ ֶ ‫ּומ ַע ְל ֵל‬ ַ ‫יכם ָה ָר ִעים‬ ֶ ‫ת־ּד ְר ֵכ‬ ַ ‫ּוזְ ַכ ְר ֶּתם ֶא‬ ‘and you shall remember your evil ways and your deedsi whi that ___i were not good’ (Ezek 36:31) (401) Relative Consisting Solely of NP ‫ר־ח ְטא‬ ֵ ‫אּו־לי ָעֹון ֲא ֶׁש‬ ִ ‫יִמ ְצ‬ ְ ‫יעי לֹא‬ ַ ִ‫ָּכל־יְ ג‬ ‘(in) all my labors, they do not find in me an iniquityi whi that ___i is sin’ (Hos 12:9) In each of the cases, the single constituent within the relative clause must be able to combine with the relative head to form a predicational (property) or identificational (identity) statement. Thus, reversing the relativization should produce a grammatical construction—e.g., the kings are with him (397), the Amorites are there (398), the priests are near Yhwh (399), your deeds are not good (400), an iniquity is sin (401). There are many cases of null-copula relative clauses, however, that contain a constituent that does not represent an appropriate property or identity for the relative head. In these cases, Hebrew uses a resumptive constituent (e.g., a subject pronoun, cliticized pronoun attached to a PP, or Adverb such as ‫‘ שׁם‬there’) in order to “save” the relative clause from semantic failure (cf. Peretz 1967: 90; Shlonsky 1992; Parunak 1996). Consider the examples in (402)–(404). 50 49.  For examples of one-part null-copula relative clauses, see Gen 3:3 (PP); 14:15 (PP); Num 21:32 (AdvP); 2  Sam 3:19 (AdjP); 2  Kgs 7:7 (Pron), 10 (Pron); 23:16 (AdvP), 20 (AdvP); Ezek 36:31 (AdjP); 42:13 (AdjP); Hos 12:9 (NP). 50.  See Gen 1:11, 12; 2:11; 13:14; 21:7; 30:35; 34:14; 38:25; Exod 9:26; 12:13, 30; 20:21; Deut 4:7, 8; 31:13; Judg 18:10; 1 Sam 3:3; 9:10; 10:5; 19:3; 1 Kgs 8:21; 2 Kgs 7:7, 10; Ezek 8:3; 2 Chr 6:11.

Relative Syntax

177

(402) ‫ר־ׁשם ַהּזָ ָהב‬ ָ ‫ל־א ֶרץ ַה ֲחוִ ָילה ֲא ֶׁש‬ ֶ ‫ָּכ‬ ‘all the land of Havilai, whi that gold is therei’ (Gen 2:11) (403) ‫ר־א ָּתה ָׁשם‬ ַ ‫ן־ה ָּמקֹום ֲא ֶׁש‬ ַ ‫ְּור ֵאה ִמ‬ ‘and look from the placei whi that you are therei’ (Gen 13:14) (404) ‫ת־אח ֵֹתנּו ְל ִאיׁש ֲא ֶׁשר־לֹו ָע ְר ָלה‬ ֲ ‫נּוכל ַל ֲעׂשֹות ַה ָּד ָבר ַהּזֶ ה ָל ֵתת ֶא‬ ַ ‫לֹא‬ ‘we are not able to do this thing, to give our sister to a mani whi that a foreskin belongs to himi’ (Gen 34:14) With this type of resumption, if the resumptive constituent were not present and we dropped the relative word, the relationship between the head of the relative and the remaining constituent could not possibly be that of subjectpredicate. For instance, if resumption were absent in (402), while the (un­ attested) phrase the land of Havila that is gold is interpretable as a proposition, it is certainly not the intended proposition; rather than describing Havila as a land in which gold is the basic compositional substance, the intent is clearly to indicate that the substance gold can be found in the location referred to as the land of Havila. For examples such as (403) and (404), the necessity of the resumptive constituent is even clearer. For (403) the (unattested) phrase *the place that you are would be contextually nonsensical; similarly, for (404), the (unattested) phrase *to give our sister to a man who is a foreskin would be equally bad. Thus, the resumptive constituent is for grammatical semantic acceptability in these cases—so that the head of the relative is not misconstrued as the subject within the relative. For the majority of resumptive constituents in null-copula relative clauses, as in the examples discussed in (402)–(404), resumption is a device to save the construction from being semantically uninterpretable. There are only 30 cases out of approximately 1,200 null-copula relatives for which resumption does not seem to be required. For example, in (405) the subject pronoun within the relative resumes the head, though it would have been grammatical to omit it. 51 (I will return to this example below.) (405) ‫ן־ה ְּב ֵה ָמה ֲא ֶׁשר‬ ַ ‫ּומ‬ ִ ‫ח־לָך ִׁש ְב ָעה ִׁש ְב ָעה ִאיׁש וְ ִא ְׁשּתֹו‬ ְ ‫הֹורה ִּת ַּק‬ ָ ‫ִמּכֹל ַה ְּב ֵה ָמה ַה ְּט‬ ‫לֹא ְטה ָֹרה ִהוא ְׁשנַ יִ ם ִאיׁש וְ ִא ְׁשּתֹו‬ ‘from every clean animal you shall take for yourself seven pairs, a male and its mate, and from the animali whi that iti is not clean, two, a male and its mate’ (Gen 7:2) In relative clauses that contain participles (as the complement of a null copula) or finite verbs, the distribution of overt resumption is more complex 51.  In all 30 examples that exhibit nonobligatory overt resumption, the resumptive is the subject pronoun: Gen 7:2, 8; 9:3; 17:12; 30:33; Lev 11:29, 39; Num 9:13; 17:5; 35:31; Deut 17:15; 20:15; 29:14; 1 Kgs 8:41; 9:20; 2 Kgs 25:19; Jer 40:7; Ezek 12:10; 20:9; 43:19; Hag 1:9; Ps 16:3; Song 1:6; Ruth 4:15; Eccl 4:2; 7:26; Neh 2:13, 18; 2 Chr 6:32; 8:7.

178

Chapter 5

simply because there often are more constituents in the clause, whether the complements of bivalent and trivalent verbs or even adjuncts with mono­valent verbs. The first and perhaps primary factor for resumption when a lexical verb is present within the relative is verbal valency (see above, §4.3, n. 15). The examples in (406)–(409) illustrate that many verbs within relatives are bivalent and thus require NP complements. (406) ‫יֹוׁשב ְּב ַׁש ַער ַה ֶּמ ֶלְך‬ ֵ ‫הּודי‬ ִ ְ‫ת־מ ְר ֳּד ַכי ַהּי‬ ָ ‫ל־עת ֲא ֶׁשר ֲאנִ י ר ֶֹאה ֶא‬ ֵ ‫ְּב ָכ‬ ‘in every timei whi that I see Mordecai, the Jew, ___i sitting in the gate of the king’ (Esth 5:13) (407) ‫יהם‬ ֶ ‫אֹותָך ֲא ֵל‬ ְ ‫ל־הּגֹויִ ם ֲא ֶׁשר ָאנ ִֹכי ׁש ֵֹל ַח‬ ַ ‫ָּכ‬ ‘all the nationsi whi that I send you to themi’ (Jer 25:15) (408) ‫ץ־ּכנַ ַען ֲא ֶׁשר ֲאנִ י ֵמ ִביא ֶא ְת ֶכם ָׁש ָּמה‬ ְ ‫ּוכ ַמ ֲע ֵׂשה ֶא ֶר‬ ְ ‘and like the deed of the land of Canaani whi that I am bringing you therei’ (Lev 18:3) (409) ‫ם־א ֶּתנְ ָך ְּביַ ד ָה ֲאנָ ִׁשים ָה ֵא ֶּלה ֲא ֶׁשר ְמ ַב ְק ִׁשים ֶאת־נַ ְפ ֶׁשָך‬ ֶ ‫יתָך וְ ִא‬ ֶ ‫ם־א ִמ‬ ֲ ‫ִא‬ ‘and I shall surely not give you into the hand of these meni whi that ___i seek your life’ (Jer 38:16) Many bivalent or trivalent Hebrew verbs require not only NP complements but also PP complements, since Hebrew does not allow preposition stranding (i.e., leaving a bare preposition behind within a relative) or pied-piping (i.e., the relativization of the entire PP), in order for the verbal valency to be fulfilled, a PP is required inside the relative clause (cf. Peretz 1967: 90–91; Parunak 1996). When the (pronominal) complement of the PP complement happens to be co-referential with the relative head, as in examples (410)–(413), the result is overt relative resumption. (410) ‫ם־ּבּה‬ ָ ‫ץ־מ ְצ ַריִ ם ֲא ֶׁשר יְ ַׁש ְב ֶּת‬ ִ ‫ְּכ ַמ ֲע ֵׂשה ֶא ֶר‬ ‘like the deed of the land of Egypti, whi that you dwelt in iti’   (Lev 18:3) 52 (411) ‫ל־ה ִּמ ְׁש ָּכב ֲא ֶׁשר יִ ְׁש ַּכב ָע ָליו ַהּזָ ב יִ ְט ָמא‬ ַ ‫ָּכ‬ ‘all of the bedi whi that the discharging one lays upon iti will be unclean’ (Lev 15:4) 53 52. For ‫ יׁשב‬with resumption within a ‫ ב‬or ‫ על‬PP or by the adverb ‫שׁם‬, see Gen 19:29; 24:3, 37; Lev 15:4, 6, 20, 22, 23, 26; 18:3; Num 13:19; 21:34; 33:55; 35:34; Deut 1:4; 3:2; 4:46; 29:15; Josh 24:15; Judg 6:10; 1 Sam 27:7, 11; 1 Kgs 7:8; 17:19; 21:11; 2 Kgs 6:1; 17:29; Ezek 3:15 (excluding the textually problematic first case indicated by the Qere), 37:25; Neh 8:14; 11:3; 1 Chr 17:1. 53.  See Gen 28:13; Lev 15:4, 18, 20, 24, 26; 1 Kgs 3:19. The verb ‫ ׁשכב‬also takes complements with the prepositions ‫‘ עם‬with’ and ‫‘ את‬with’ and the adverb ‫‘ שׁם‬there’.

Relative Syntax

179

(412) ‫עֹומד ָע ָליו ַא ְד ַמת־ק ֶֹדׁש הּוא‬ ֵ ‫ִּכי ַה ָּמקֹום ֲא ֶׁשר ַא ָּתה‬ ‘because the placei whi that you are standing upon iti—it is holy ground’ (Exod 3:5) 54 (413) ‫חֹמ ֶֹתיָך ַהּגְ בֹהֹות וְ ַה ְּב ֻצרֹות ֲא ֶׁשר ַא ָּתה ּב ֵֹט ַח ָּב ֵהן‬ ‘your high and fortified wallsi whi that you trust in themi’   (Deut 28:52) 55 In these relative clauses, it would be ungrammatical for the resumptive PP complement to be absent. If we compare Hebrew to English, we see a similar issue. Unlike Hebrew, English can leave a preposition stranded (the man who[m] I spoke to), and it allows pied-piping (the man to whom I spoke); however, the preposition, wherever it resides, is obligatory (compare *the man who[m] I spoke). In Hebrew, the entire PP, including of course the resumptive, must be present for a grammatically felicitous construction; the absence of the PP results either in a grammatical failure or a marginally interpretable result (similarly, Peretz 1967: 90, though without any biblical examples): (414) ‫וַ ֵתּ ְשׁבוּ ְב ָק ֵדשׁ יָ ִמים ַר ִבּים ַכּיָּ ִמים ֲא ֶשׁר יְ ַשׁ ְב ֶתּם׃‬ ‘and you dwelt at Kadesh many days, like the (number of) daysi whi that you dwelt (there) ___(then)i’ (Deut 1:46) The occurrence of ‫ ישׁב‬in (414) is the sole case (out of 36) within a relative clause that does not contain overt resumption of the head by a clitic pronoun attached to the preposition ‫ ב‬or ‫ על‬or by the locative adverb ‫( שׁם‬see above, n. 52). In Deut 1:46, the lack of resumption is explicable, even though the sense of the idiom used may not be altogether clear—within a relative clause, the verb ‫ ישׁב‬is used everywhere else to refer to something or someplace that can be resumed by a locative PP or adverb. In this singular case, the head of the relative is the temporal NP “days,” which cannot be resumed by a locative phrase. Thus Peretz is partially correct (“‫הכינוי המוסב נשמט תכופות כשהזוקק‬ ‫ ;”מציין מקום או זמן‬Peretz 1967: 92)—the example in (414) illustrates the omission of overt resumption with a temporal head (“‫ זמן‬. . . ‫)”כשׁהזוקק מציין‬, though the presence of overt resumption in every other ‫ ישׁב‬example suggests quite the opposite: resumption is not dropped when it refers back to a place. In addition to fulfilling the valency requirements of the verb within a relative, resumptives may also serve to clarify the nuance of the verb’s lexical meaning. It is a well-known feature of Hebrew grammar that a single verb may have multiple lexical entries, even in a single binyan, and that these distinctions 54.  See Gen 19:27 (with ‫‘ שׁם‬there’); Exod 3:5; 8:18; Josh 5:15. The verb ‫ עמד‬also takes the PP ‫‘ לפני‬before’. 55. For ‫בטח‬, see Deut 28:52; 2 Kgs 19:10; Isa 37:10; Jer 5:17; 7:14; 17:5, 7; Ps 41:10; 115:8; 135:18.

180

Chapter 5

are often connected to the type of complement used with the verb. In these cases, the resumptive constituent within the relative clause is obligatory not only to fulfill the basic verbal valency but also to specify the lexical semantics of the verb. The examples in (415)–(416) present minimal pairs concerning the verb+complement combinations (where possible, I have used a relative clause example). 56 (415) [‫‘ ]דרשׁ‬to seek; to inquire’ a. ‫ֹלהיָך ּד ֵֹרׁש א ָֹתּה‬ ֶ ‫ֶא ֶרץ ֲא ֶׁשר־יְ הוָ ה ֱא‬ ‘a landi whi that Yhwh your God is SEEKING (‫ )את‬iti’   (Deut 11:12) 57 b. ‫ה־ּבּה‬ ָ ‫יה וְ ֶא ְד ְר ָׁש‬ ָ ‫ׁשּו־לי ֵא ֶׁשת ַּב ֲע ַלת־אֹוב וְ ֵא ְל ָכה ֵא ֶל‬ ִ ‫אמר ָׁשאּול ַל ֲע ָב ָדיו ַּב ְּק‬ ֶ ֹ ‫וַ ּי‬ ‘and Saul said to his servants: seek for me a medium-mistress, and I shall go to her and INQUIRE of (‫ )ב‬her’ (1 Sam 28:7) 58 (416) [‫‘ ]שׁרת‬to officiate; to serve’ a. ‫ר־ה ָּמה ְמ ָׁש ְר ִתם ָּבם‬ ֵ ‫יהם ֲא ֶׁש‬ ֶ ‫ת־ּבגְ ֵד‬ ִ ‫ֶא‬ ‘their garmentsi whi that they OFFICIATE with (‫ )ב‬themi’   (Ezek 44:19) 59 b. ‫יֹוסף ֵחן ְּב ֵעינָ יו וַ יְ ָׁש ֶרת אֹתֹו‬ ֵ ‫וַ ּיִ ְמ ָצא‬ ‘and Joseph found favor in his eyes and he [i.e., Joseph] SERVED (‫ )את‬him’ (Gen 39:4) 60 Overt resumption is also often critical for clearly identifying the relative head when multiple possible antecedents exist in the context. For example, in (417)–(419), the resumptive element identifies precisely which constituent among available NPs is in fact the head of the relative. 61 56.  For both verbs, ‫ דרשׁ‬and ‫שׁרת‬, see the standard lexica (BDB, HALOT, DCH), all of which indicate, with greater or lesser clarity, the semantic distinctions produced by verb+complement combinations. 57.  See Deut 11:12; 18:11; 2 Sam 11:3; 2 Kgs 1:2, 3, 6, 16; Isa 8:19; 19:3; 65:1; 2 Chr 17:3. 58.  See 1 Sam 28:7; 2 Kgs 1:2, 3, 6, 16 (2×); 1 Chr 10:14; 2 Chr 34:26. 59.  See Num 3:3; 4:9, 12, 14; 2 Kgs 25:14; Jer 52:18; Ezek 42:14; 44:19. 60.  See Gen 39:4; 40:4. 61.  See Exod 6:5; Lev 18:5; 26:54; Num 26:59; 35:25; Josh 2:10; 1 Kgs 21:25; Isa 28:4; 29:11; 37:4; 41:8–9; Jer 13:25; 19:4; 27:19–20; 29:22; Ezek 4:10; 5:16; 20:11, 13, 21; 32:9; Ps 107:2; Ruth 4:15; Ezra 9:11. There are also a few examples of this “head-disambiguation” use of resumption with null-copula clauses: see Ruth 4:15, where the second relative clause contains a subject pronoun—this resumption is required in order to specify which noun the relative clause is modifying, the nearest antecedent (‘him’, referring to the boy that was born), or a more distant antecedent (Ruth, the daughter-in-law). The 3fs resumptive pronoun clearly specifies that the latter is the head.

Relative Syntax

181

(417) ‫ּולעֹוג ֲא ֶׁשר‬ ְ ‫יתם ִל ְׁשנֵ י ַמ ְל ֵכי ָה ֱאמ ִֹרי ֲא ֶׁשר ְּב ֵע ֶבר ַהּיַ ְר ֵּדן ְל ִסיחֹן‬ ֶ ‫וַ ֲא ֶׁשר ֲע ִׂש‬ ‫אֹותם‬ ָ ‫ֶה ֱח ַר ְמ ֶּתם‬ ‘and what you did to the two Amorite kings who are in the region across the Jordan, to Sihon and to Ogi, whi that you put themi to the ban’ (Josh 2:10) (418) ‫אּצר ֶמ ֶלְך ָּב ֶבל‬ ַ ֶ‫בּוכ ְדנ‬ ַ ְ‫א־ל ָק ָחם נ‬ ְ ֹ ‫ּנֹות ִרים ָּב ִעיר ַהּזֹאת ֲא ֶׁשר ל‬ ָ ‫וְ ַעל יֶ ֶתר ַה ֵּכ ִלים ַה‬ ‘and concerning the remainder of the vesselsi left in this city whi that Nebuchadnezzar, king of Babylon, had not taken themi’   (Jer 27:19–20) (419) ‫עֹודּה ְּב ַכּפֹו ְיִב ָל ֶענָּ ה‬ ָ ‫אֹותּה ְּב‬ ָ ‫ּכּורּה ְּב ֶט ֶרם ַקיִ ץ ֲא ֶׁשר יִ ְר ֶאה ָהר ֶֹאה‬ ָ ‫ְּכ ִב‬ ‘like a first ripe figi before summer, whi that the one who looks will see iti, (and) while it is still in his palm he will swallow it’ (Isa 28:4) In (417), if ‫‘ אותם‬them’, which is the 3ms plural clitic pronoun attached to the direct object marker, were not present, it would be most natural to understand the nearest antecedent, just king ‫עוג‬, as the head. The plural resumptive indicates that both kings, ‫ סיחן‬and ‫עוג‬, are the intended antecedent for the relative. In (418), the nearest antecedent for the relative is the femininesingular NP ‘this city’ in the PP ‫בעיר הזאת‬. However, the masculine-plural object pronoun attached to the verb indicates that the intended antecedent is the more distant masculine-plural NP ‫הכלים‬. Finally, note the difficult example in (419), in which the apparent redundancy of ‫ אשׁר יראה הראה אתו‬is typically ignored, re­arranged, or reduced to something simpler (e.g., Delitzsch 1889: 312; Watts 2004: 425; Blenkinsopp 2000: 385; Wildberger 2002: 2). 62 And yet, the sequence of words can be straightforwardly analyzed without any change of meaning vis-à-vis the sense assigned to the passage in the commentaries cited above. As my translation indicates, there is a semantic distinction between monovalent ‫‘ ראה‬to look’ and bivalent ‫‘ ראה‬to see something’; the subject of the relative clause, ‫( הראה‬itself a null-head ‫ה‬-relative) is an example of monovalent ‫ראה‬, and the finite verb ‫ יראה‬with its complement ‫ אתה‬is an example 62.  Delitzsch: “Und es ergeht der welkenden Blume seines prangenden Schmuckes, die auf dem Haupte des üppigen Thales, wie einer Frühfeige ehe Ernte ist, die ansieht wer sie sieht und kaum ist sie in seiner Hand, so verschlingt er sie.” Watts: “And she became a drooping flower, the beauty of its glory, which (was) at the head of a fertile valley. Like a firstborn (fig) before summer which, when one sees it, no sooner is it in his hand than he swallows it.” Blenkinsopp: “[A]nd the flower of its splendid beauty doomed to fade at the head of a fertile valley will be like figs that ripen before the summer; those who see them will swallow them as soon as they have them in their hand.” Wildberger: “[T]he shriveled flowers, its gorgeous ornamentation upon the head above the lush valley. It will happen for her as for the “early fig” before the harvest in the summertime: Someone will see one, and when he sees it—hardly is it in his hand, he has gulped it down right away.”

182

Chapter 5

of bivalent ‫ראה‬. Note that the ‫ אשׁר‬relative is also extraposed, i.e., at a distance from its head, ‫( בכורה‬see below, §5.2.4). In addition to fulfilling the requirements of verbal valency, Hebrew relative resumption may also serve to disambiguate the syntactic function of the head within the relative clause. In other words, without resumption in relatives such as those in (420)–(421), it would not be clear whether the head of the relative was serving as the subject of the clause within the relative or as the complement (cf. Tsujita 1991). 63 (420) ‫וְ ִכ ֶּפר ַהּכ ֵֹהן ֲא ֶׁשר־יִ ְמ ַׁשח אֹתֹו‬ ‘and the priesti whi that one anointed himi shall make atonement’ (Lev 16:32) (421) ‫ת־ּבית ַא ְח ָאב‬ ֵ ‫‏אל־יֵ הּוא ֶבן־נִ ְמ ִׁשי ֲא ֶׁשר ְמ ָׁשחֹו יְ הוָ ה ְל ַה ְכ ִרית ֶא‬ ֶ ‘to Jehu, son of Nimshii, whi that Yhwh had anointed himi to cut off the house of Ahab’ (2 Chr 22:7) In (420), if the resumptive 3ms object pronoun attached to the object marker ‫ את‬were not present, it would be possible, if not most natural, to interpret the head of the relative as the subject within the relative, producing ‘the priest who anoints’ rather than ‘the priest who [someone else] anoints him’. Likewise, in (421) if the 3ms clitic pronoun attached to the verb ‫ משׁח‬were not present within the relative, it would become ambiguous—is Jehu the subject of the verb or is Yhwh? That is, the reader might make the mistake of reading the relative as ‘Jehu who (he) anointed Yhwh to cut off the house of Ahab’— the exact opposite of what is specified by the text. To summarize my discussion of resumption to this point, the majority of relative clauses that contain resumption do so because the resumptive constituent is necessary in order for the clause to be either syntactically or semantically acceptable. Within a generative linguistic framework, Shlonsky’s analysis of resumption in Modern Hebrew as a “last resort” strategy is also compelling for Biblical Hebrew—that overt resumption represents the insertion of the phonologically overt resumptive constituent when without it the clause would be ungrammatical or at least highly infelicitous (and thus difficult to interpret). 64 For the very small minority (e.g., only 30 cases in null-copula relatives) 65 that contain resumption for which I can determine no obligatory motivation, I fol63.  See Exod 25:2; 28:3; 35:21, 26, 29; 36:2; Lev 16:32; Num 25:35; Deut 29:25; Isa 66:13; Jer 8:2 (3×); 28:9; 44:3; Ezek 36:32; Esth 5:11; 7:5; 10:2; 2 Chr 8:8; 22:7. 64.  For a fuller discussion of Shlonsky’s technical argument, see Holmstedt 2002: 273– 77 and above, §5.1.2.1. 65.  See Gen 7:2, 8; 9:3; 17:12; 30:33; Lev 11:29, 39; Num 9:13; 17:5; 35:31; Deut 17:15; 20:15; 29:14; 1 Kgs 8:41; 9:20; 2 Kgs 25:19; Jer 40:7; Ezek 12:10; 20:9; 43:19; Hag 1:9; Ps 16:3; Song 1:6; Ruth 4:15; Eccl 4:2; 7:26; Neh 2:13, 18; 2 Chr 6:32; 8:7

Relative Syntax

183

low the lead of Ellen Prince, who has argued that Yiddish and English allow the use of nonobligatory resumptive pronouns in order to influence the shape of the information structure of the discourse (Prince 1990; cf. Tsujita 1991; Parunak 1996). In the 30 cases of nonobligatory resumption in null-copula relative clauses, I propose that resumptive pronouns may fulfill a pragmatic role in one of two ways. Either the resumptive pronoun indicates that the head is focused, or the resumptive pronoun is present so that the word order (e.g., predicate-subject order in a null-copula clause) may mark the predicate for Focus. An example of the resumptive carrying Focus-marking is given in (422). (422) ‫ל־הּכ ֲֹהנִ ים ַה ְלוִ ּיִ ם ֲא ֶׁשר ֵהם ִמּזֶ ַרע ָצדֹוק ַה ְּקר ִֹבים ֵא ַלי נְ ֻאם ֲאד ֹנָ י יְ הוִ ה‬ ַ ‫וְ נָ ַת ָּתה ֶא‬ ‫ן־ּב ָקר ְל ַח ָּטאת‬ ָ ‫ְל ָׁש ְר ֵתנִ י ַּפר ֶּב‬ ‘and you shall give to the Levitical priestsi whi that theyi (are) from the seed of Zadok, those who draw near to me—an utterance of the Lord Yhwh—to serve me a bull, a son of cattle, for a sin-offering’   (Ezek 43:19) For (422), since priests could have multiple modifiers, the passage is specific: it is only to the precise Levitical Zadokite near-to-God priests that an offering should be given. In this case, the resumptive pronoun indicates how strong the instruction is—that the offering should be brought to none other than the specified priests; the nonobligatory pronoun helps in this by focusing the head within the relative. The nonobligatory resumption in the null-copula relative clauses in (423), repeated from (405), is a case in which the resumptive constituent is necessary in order for the word order to indicate that the predicate carries Focus-marking. (423) ‫ן־ה ְּב ֵה ָמה ֲא ֶׁשר‬ ַ ‫ּומ‬ ִ ‫ח־לָך ִׁש ְב ָעה ִׁש ְב ָעה ִאיׁש וְ ִא ְׁשּתֹו‬ ְ ‫הֹורה ִּת ַּק‬ ָ ‫ִמּכֹל ַה ְּב ֵה ָמה ַה ְּט‬ ‫לֹא ְטה ָֹרה ִהוא ְׁשנַ יִ ם ִאיׁש וְ ִא ְׁשּתֹו‬ ‘from every clean animal you shall take for yourself seven pairs, a male and its mate, and from the animali whi that iti (is) not clean, two, a male and its mate’ (Gen 7:2) Gen 7:2 is widely attributed to the P(riestly) layer of the flood narrative (see, e.g., Wenham 1987: 167–69). The P layer adds to the description of animals to be brought into the ark the further specification of clean (seven pairs) and unclean (one pair). Though the resumption of sacrifice after the flood (Gen 8:20) would appear to necessitate the additional clean animals, in fact the ritual notion of clean and unclean as it relates to sacrifice is anachronistic to the pre-Mosaic setting. Thus, as the sacerdotal concerns of the P tradents color their telling of the narratives of Genesis 1–11, so the distinction of clean and unclean animals inform the P layer of the flood story. The second half of the

184

Chapter 5

verse in (423) functions are a clarification, perhaps in anticipation of any audience questions or objections. Lest the audience think that only clean animals are brought into the ark, the Priestly storyteller marks the negated adjective phrase ‫ לא טהרה‬for Focus (by virtue of using overt resumption and then raising the negative and adjective to a Focus position). The Focus-marking serves to contrast the ‫ לא טהרה‬animals with ‫ טהרה‬animals and so clarifies that, though ritual concerns are very important, Yhwh is concerned with saving a representation of all of his creation, not just the ritually clean. In verbal clauses, the function of nonobligatory resumption serves simply to mark the resumptive constituent and thus also the head with Focus. The clauses provided in (424) and (425) illustrate how resumption of the head as the verbal complement signals that the head carries Focus (in the translations, I have added contextual information to aid in interpreting the Focus). (424) ‫יִּמ ְכרּו ִמ ְמ ֶּכ ֶרת ָע ֶבד‬ ָ ‫אתי א ָֹתם ֵמ ֶא ֶרץ ִמ ְצ ָריִ ם לֹא‬ ִ ‫ר־הֹוצ‬ ֵ ‫י־ע ָב ַדי ֵהם ֲא ֶׁש‬ ֲ ‫ִּכ‬ ‘because my servants are theyi whi that I brought themi [not just you slaveowners] out from the land of Egypt, they will not be sold (at) a slave sale’ (Lev 25:42) 66 In Lev 25:42 (424), the relative clause appears in the middle of a discourse about the social implications of the Jubilee year among Israelite society. Because God had delivered all of the Israelites from slavery in Egypt, so Israelite slaveowners are required to remember the slave history they share with their indentured servants and to be merciful to them in the Jubilee year. The resumptive pronoun in (425) is used similarly. (425) ‫גֹורל‬ ָ ‫זֹאת ָה ָא ֶרץ ֲא ֶׁשר ִּת ְתנַ ֲחלּו א ָֹתּה ְּב‬ ‘this is the landi whi that you shall inherit iti [no more, no less] by lot’ (Num 34:13) 67 The extent of Canaan within the theology of the Hebrew Bible is delineated in Num 34:3–12. However, from both biblical and extrabiblical references it is clear that the boundaries of the geographical area referred to by the name Canaan changed over time (see Schmitz 1992: 828–31). Thus, the focused resumptive pronoun in (425) serves to inform the reader/listener that the precise extent of the land described in the preceding geographical list, nothing more, nothing less constitutes their inheritance from God. We have arrived at the last type of pragmatically motivated resumption: resumption at the subject position within a relative clause containing a finite verb. Because Hebrew is a pro-drop language (see 2.2.3; Holmstedt 2013a)— 66.  See Lev 25:42, 55. 67.  See also Lev 24:55; Num 13:32 Deut 4:19; 18:14; 1 Kgs 11:34; Ezek 15:6; Pss 1:4; 94:12.

Relative Syntax

185

i.e., the verbs are fully inflected for agreement features—subject pronouns within a relative are not required. Accordingly, there are no examples of a relative clause including a finite verb in which the subject was promoted to the relative head and overtly resumed. 68 The only type of verb for which we find subject resumption inside relatives is the participle, which is typologically closer to adjectives (and thus null-copula clauses) than finite verbs (see Cook 2008; 2012: 223–33). There are four examples of subject resumption with a participial predicate, as in (426). (426) ‫מֹוׁש ַיע ָל ֶכם‬ ִ ‫יכם ֲא ֶׁשר־הּוא‬ ֶ ‫ֹלה‬ ֵ ‫ֱא‬ ‘your godi whi that hei saves you’ (1 Sam 10:19) 69 In example (426), the presence of the free subject pronoun serves to indicate that the referent of the pronoun ‫‘ הוא‬he’ is focused. Indeed, all of 1 Sam 10:19 sums up the contrast that underlies the larger discourse: But today you have rejected your God, who he saves you from all your calamities and your distresses; and you have said, “No! but set a king over us.” Now therefore present yourselves before Yhwh by your tribes and by your clans. In the context of the larger discourse of 1  Samuel 8–11, it is clear that the Israelites’ wish for a king is contrary to the wish of their god, Yhwh. The rare presence of the personal pronoun in the highest subject position within the relative clause, as in 1 Sam 10:19, instructs the listener/reader to establish a membership set based on the referent of the pronoun—e.g., for 1 Sam 10:19 the set might be composed of ‘he/God, me/Samuel, a (human) king, . . .’. The Focus that the pronoun in 1 Sam 10:19 establishes is meant to highlight that the people are rejecting God (the only character who has the ability to deliver them from their calamities) in favor of a king—a mere human and an unknown quantity. Compare (426) with the more common null subject resumption in participial relatives, illustrated in (427). (427) ‫יֹוׁשב ְּב ֶח ְׁשּבֹון‬ ֵ ‫ִסיחֹן ֶמ ֶלְך ָה ֱאמ ִֹרי ֲא ֶׁשר‬ ‘Sihon, king of the Amoritesi, whi that ___i is dwelling in Heshbon’ (Deut 1:4) 70 68.  In 2 Kgs 22:13 there is what appears to be the sole example of subject resumption with a finite verb: ‫ר־היא נִ ְצּ ָתה ָבנוּ‬ ִ ‫דֹולה ֲח ַמת יְ הוָ ה ֲא ֶשׁ‬ ָ ְ‫כּי־ג‬. ִ However, the 3fs perfect verb ‫נִ ְּצ ָתה‬ can easily be repointed as a fs participle, ‫נִ ָּצ ָתה‬. Given that there are no other examples of subject resumption in a finite verbal relative clause, it is likely that we should take 2 Kgs 22:13 as a participial relative clause. See above, n. 47. 69.  See Num 14:8; Deut 20:20; 1 Sam 10:19; Neh 2:13 (reading the Qere). 70.  See Gen 7:8; 39:22; 40:5; Num 21:34; Deut 1:4; 3:2; 4:46; 28:61; Judg 18:7; 1 Sam 22:2; 1 Kgs 5:13; Isa 11:10; 24:2; 49:7; Jer 38:16; Ezek 9:2; 13:3; 43:1; Zech 11:5; Pss 115:8;

186

Chapter 5

Based on the kind of contrast between (427) and (426) and similar examples, I suggest that phonologically null constituents (e.g., traces left by relativization, or null resumption) in Hebrew cannot carry Focus. This is consistent with the nature of overt resumption in left-dislocation structures (see Holmstedt 2014b). It appears that a constituent within a Hebrew relative that is marked for pragmatic Focus must be spelled out phonologically. In summary, I have proposed that overt resumption in Hebrew relatives has two functions. First, an overt resumptive is inserted in order to save the grammaticality of a given relative clause when, for example, the valency of the verb within the relative would otherwise be unfilled. Or, in the case of nullcopula relatives, a resumptive constituent may be inserted in order to avoid a semantically uninterpretable result. Similarly, the use of overt resumption to disambiguate the relative head is related to semantic interpretability and discourse felicitousness. Thus, overt resumption is a last-resort phenomenon (see above, §5.1.2.1). Second, overt resumption that is not obligatory for one of the last-resort reasons above signals a pragmatic function: the resumptive either marks the relative head for Focus or by its word-order position allows another constituent within the relative to be marked for Focus. 5.2.4. Extraposition In an insightful 1949 article, Moshe Goshen-Gottstein argued that the placement of some relative clauses at a distance from their antecedent is the result of “afterthought.” Compare the examples below: in (428), the relative clause follows immediately after its antecedent, but in (429) the relative clause is removed from its antecedent; the relative clause is on the other side of the null-copula predication. (428) ‫[אנַ ְחנוּ] שׁ ְֹל ִחים א ְֹתָך ֵא ָליו נִ ְשׁ ָמע‬ ֲ ‫ֹלהינוּ ֲא ֶשׁר אנו‬ ֵ ‫ְבּקֹול ׀ יְ הוָ ה ֱא‬ ‘to the voice of Yhwhi our God, whi that we are sending you to himi, we shall listen’ (Jer 42:6 [Kethiv /Qere]) (429) ‫יתם ְבּ ֵעינֵ י יְ הוָ ה‬ ֶ ‫י־ר ַע ְת ֶכם ַר ָבּה ֲא ֶשׁר ֲע ִשׂ‬ ָ ‫ְוּדעוּ ְוּראוּ ִכּ‬ ‘and know and see that your evili ___j (is) great [whi that you have done ___i in the eyes of Yhwh]j’ (1 Sam 12:17) In addition to the necessity of recognizing the existence of relative clauses that are removed from their antecedent, Goshen-Gottstein also hints that this type of relative has particular semantic status: he suggests that all

135:18; Eccl 4:1; 8:12, 14; Esth 8:8; Neh 5:2, 3, 4; 2 Chr 34:10. The paucity of examples highlights the preference for using ‫ה‬- or zero-relative strategies when the verb within the relative clause is a participle.

Relative Syntax

187

“afterthought” relative clauses are nonrestrictive in BH (Goshen-Gottstein 1949: 39). Setting aside the psychologically influenced term afterthought, 71 the linguistic phenomenon that Goshen-Gottstein so insightfully described is known as extraposition (see above, 5.1.2.2, and also n. 8 with regard to terminology in Hebrew studies). Extraposition in relativization describes the placement of the relative clause at a distance from the relative head, further down the clause. I have identified 133 clear examples of relative clause extraposition in the Hebrew Bible, illustrated again in (430). 72 (430)  ‫ן־ה ִא ָשּׁה ַהזֹּאת ָליְ ָלה ֲא ֶשׁר ָשׁ ְכ ָבה ָע ָליו׃‬ ָ ‫וַ יָּ ָמת ֶבּ‬ ‘and the soni of this woman ___j died (at) night, [whi that (she) lay upon himi]j’ (1 Kgs 3:19) 73 In (430), the head that the extraposed relative modifies is the NP ‫בן האשׁה‬ ‫הזאת‬. The head and its modifying relative clause are separated by the temporal adverb phrase, ‫(‘ לילה‬at) night’. 74 Notice that the Hebrew examples exhibit the 71.  The language and approach of Goshen-Gottstein’s analysis coincide with a shift taking place in the American Descriptivist tradition of linguistics during the 1930s–40s. The study of language began to borrow notions from the branch of psychology called behaviorism; these behavioristic ideas found their way into linguistic analysis by virtue of Leonard Bloomfield’s influence. Although Goshen-Gottstein’s use of “afterthought” to characterize a syntactic construction may in fact have been a coincidence, it is interesting that his choice of terminology mirrors these new impulses in language study. See Sampson 1980 for more information on the Descriptivists and the influence of behaviorism on Bloomfield’s work. 72.  See Gen 22:14; 24:15; 30:2; 41:50; 48:9, 22; Exod 1:8; 4:17; 11:6; 15:17; 20:2; 29:42; 32:4; Lev 1:5; 7:35–36; 13:20, 39; 17:11; 22:2; 25:42, 55; Deut 8:16; 11:10 (2×); 19:9; 23:16; 25:17–18; Josh 6:26; 18:7; Judg 2:10; 9:16–17; 18:16; 21:19; 1  Sam 3:11; 10:16; 12:17; 14:45; 15:2; 2 Sam 2:5, 6; 3:8; 7:12, 23; 1 Kgs 3:19; 5:21; 6:12; 8:33; 10:3, 9, 10; 12:31; 13:14; 15:13; 2 Kgs 9:36, 37; 10:10; 12:3; 17:4; 21:12; Isa 28:4; 29:22; 30:24; 31:4; 54:9; 56:5; 63:7; Jer 5:22; 19:3; 24:3; 37:1; 42:14; 43:1; 49:12; Ezek 5:16 (2×); 6:9; 11:12; 12:2 (2×); 14:5; 15:6; 16:45 (2×); 20:11, 13; 20:21, 32; 47:14; Amos 2:9; Jonah 1:9; Hab 1:5; 3:16; Mal 3:19; Ps 26:9–10 (2×); 78:5; 84:4; 132:1–2; 137:7; 139:19–20; 140:2–3, 5; 147:7–8 (3×); Job 6:4; 12:9–10; 22:15; 31:12 (2×); 34:17–19; 38:22–23; 39:6; Ruth 1:22; 2:6; 4:1, 15; Song 6:5; Eccl 3:14; 4:9; 7:20; 9:9; Lam 1:10; 4:20; Neh 6:11; 9:26; 1  Chr 21:17; 2 Chr 9:2, 8; 15:16; 22:9; 36:13. 73.  There are two possible heads for this extraposed relative: ‫ בן האשׁה הזאת‬or just ‫האשׁה‬ ‫הזאת‬. The NP ‫ בן‬is the subject of the clause and the initial logical choice for the head of the relative. If we analyze the relative clause as modifying ‫בן‬, the nonrestrictive (and extraposed) relative provides further information about the son that would appear to be necessary in order to place blame upon the boy’s mother for his death. Although it is possible to analyze the relative as modifying the second half of the larger bound construction, ‫האשׁה הזאת‬, this is rather awkward in that the relative clause would modify a DP-internal, nonargument constituent. 74.  The extraposed relative serves nonrestrictively to provide additional information unnecessary for identifying the referent of ‫( בן‬the head ‫ בן‬is already identified by the clitic host

188

Chapter 5

same type of movement as the English example in (431) (repeated from [292] with more complete co-indexation). (431) [Some friendsi] ___j visited [whomi we hadn’t seen ___i for years]j. How diverse are the contexts in which extraposition of the relative clause occurs in Hebrew? The five Hebrew examples presented (432)–(436) are representative and demonstrate that quite different types of constituents can intervene between a head and its relative clause. (432) ‫יתם ְבּ ֵעינֵ י יְ הוָ ה‬ ֶ ‫י־ר ַע ְת ֶכם ַר ָבּה ֲא ֶשׁר ֲע ִשׂ‬ ָ ‫ְוּראוּ ִכּ‬ ‘and see that your evili ___j (is) great [ ___i whi that you did ___i]j in the eyes of Yhwh’ (1 Sam 12:17) (433) ‫תוּאל‬ ֵ ‫ִהנֵּ ה ִר ְב ָקה י ֵֹצאת ֲא ֶשׁר יֻ ְלּ ָדה ִל ְב‬ ‘and behold, Rebekahi ___j was coming out [ ___i whi that ___i was born to Bethuel]j’ (Gen 24:15) (434) ‎‫יתם לֹו׃ ֲא ֶשׁר־‬ ֶ ‫ם־כּגְ מוּל יָ ָדיו ֲע ִשׂ‬ ִ ‫ם־בּיתֹו וְ ִא‬ ֵ ‫יתם ִעם־יְ ֻר ַבּ ַעל וְ ִע‬ ֶ ‫ם־טֹובה ֲע ִשׂ‬ ָ ‫וְ ִא‬ ‫יכם וַ יַּ ְשׁ ֵלְך ֶאת־נַ ְפשֹׁו ִמנֶּ גֶ ד וַ יַּ ֵצּל ֶא ְת ֶכם ִמיַּ ד ִמ ְדיָ ן׃‬ ֶ ‫אָבי ֲע ֵל‬ ִ ‫נִ ְל ַחם‬ ‘and if according to the deed of his hands ∅(you)i ___j do to him [whi that my father fought for youi, and risked his life, and rescued you from the hand of Midian]j (Judg 9:16–17) (435) ‫אָסנַ ת‬ ְ ‫יֹוסף יֻ ַלּד ְשׁנֵ י ָבנִ ים ְבּ ֶט ֶרם ָתּבֹוא ְשׁנַ ת ָה ָר ָעב ֲא ֶשׁר יָ ְל ָדה־לֹּו‬ ֵ ‫וּל‬ ְ ‘and to Joseph, two sonsi ___j were born before the two famine years came, [whi that Asenath bore ___i for him]j’ (Gen 41:50) (436) ‫ֹלהיָך יִ ְשׂ ָר ֵאל‬ ֶ ‫אמרוּ ֵא ֶלּה ֱא‬ ְ ֹ ‫וַ יִּ ַקּח ִמיָּ ָדם וַ יָּ ַצר אֹתֹו ַבּ ֶח ֶרט וַ יַּ ֲע ֵשׂהוּ ֵעגֶ ל ַמ ֵסּ ָכה וַ יּ‬ ‫ֲא ֶשׁר ֶה ֱעלוָּך ֵמ ֶא ֶרץ ִמ ְצ ָריִ ם׃‬ ‘and he took [the gold earings (v. 3)] from their hand and bound it in the bag and he made it (into) a calf of metal and he said, “These are your godsi ___j, O Israel, [whi that ___i brought you up from the land of Egypt!]j”’ (Exod 32:4) Example (432) represents a case of ‫אשׁר‬-relative clause extraposition with an intervening null copula and the adjectival copular complement; (433) is an example of ‫אשׁר‬-relative clause extraposition with an intervening null copula and its participial complement; whether one places the null 2mp subject pronoun before or after the finite verb ‫ עשׂיתם‬in (434), minimally NP ‫)האשׁה הזאת‬. While the relative clause is in fact providing the cause of death, the point is that syntactically it is more economical to analyze the ‫ אשׁר‬clause as a relative, even if the needs of the target language prompt us to render it a causal clause in translation.

Relative Syntax

189

the PP ‫ לו‬separates the null head from its ‫אשׁר‬-relative; the ‫אשׁר‬-relative clause extraposition (435) exhibits an intervening temporal clause containing a finite verb; and in (436) the vocative phrase ‫ ישׂראל‬splits the relative head from its clause. The identification of the example in (436) as extraposition highlights a structural gray area. Since vocatives can be inserted at phrasal edges, 75 does (436) really qualify as extraposition: has the relative clause been moved rightward, or has the vocative simply been inserted between the head and the relative without implying any movement of the relative? If one’s view of vocative syntax is that this “extrasentential” element may be inserted between almost any clausal constituent after the host clause has been fully developed, then there is no need to propose rightward movement of the relative clause in (436), and it is thus not an example of extraposition. A similar line of argumentation can be given for the example in (437), in which the quotative frame ‫אמר יהוה‬ ‫ צבאות‬is the intervening constituent: (437) ‎‫י־הנֵּ ה ַהיֹּום ָבּא בּ ֵֹער ַכּ ַתּנּוּר וְ ָהיוּ ָכל־זֵ ִדים וְ ָכל־ ע ֵֹשׂה ִר ְשׁ ָעה ַקשׁ וְ ִל ַהט א ָֹתם‬ ִ ‫ִכּ‬ ‫אָמר יְ הוָ ה ְצ ָבאֹות ֲא ֶשׁר לֹא־יַ ֲעזֹב ָל ֶהם שׁ ֶֹרשׁ וְ ָענָ ף׃‬ ַ ‫ַהיֹּום ַה ָבּא‬ ‘“because look—the day is coming, burning like the oven and all the presumptuous and every doer of wickedness shall become chaff; and burn them up shall the dayi that comes ___j,” said Yhwh of Hosts, “[whi that ___i will leave them neither root nor branch”]j’   (Mal 3:19) Given the complex syntactic and semantic relationship between quotative frames and quotations (see Miller 1996), it may be that we should not view the separation of the head from its relative clause in (437) as extraposition, per se, but as the insertion of a constituent into a position between the major constituents of a phrase at a later point in the derivation of the clause. 75. See Miller 2010 for a recent examination of vocative syntax in Hebrew. Although Miller does not list any examples in which a vocative is situated between a relative head and its relative clause, the data she adduces lead her to observe that “the vocative may find a niche between any of the clausal constituents” (Miller 2010: 359). However, she later adds the important qualification, “[T]he linguistic pieces left by the insertion of a vocative are never less than a phrase—the vocative never divides a construct phrase, a noun and its adjective, or a preposition and its complement” (2010: 364). The example in (436) suggests otherwise, or at least that Miller’s description requires further nuancing; for example, the cliticization of an NP to another NP (the so-called construct phrase) or the attachment of a proclitic preposition to its complement differ in phrase structure from that of a noun and a modifying adjective or, more to the point, than a noun and a modifying relative. I suggest that this difference in phrase structure allows the presence of a vocative in (436) between the relative head and its relative clause and that the absence of any noun-vocative-adjective sequences in Miller’s corpus (Genesis, Joshua, Judges, Samuel, Kings, and Psalms) must be coincidental (and, admittedly, examples of a vocative intervening are very rare). See also Exod 15:17 for an example with a zero-relative.

190

Chapter 5

Related to this is the need to distinguish extraposition as rightward movement from relative clause stranding, in which the movement concerns the raising of the relative head to a higher position in the clause (for, e.g., Focus), leaving the relative clause in its lower position. The example in (438) is arguably such an example—the relative head ‫( יהוה‬with the ‫ תחת‬PP in which it resides) may be compellingly read as Focus-fronted, stranding the ‫ אשׁר‬relative modifying ‫ יהוה‬in its original place and thus separated from its head. (438) ‫י־ב ֶטן׃‬ ָ ‫ר־מנַ ע ִמ ֵמְּך ְפּ ִר‬ ָ ‫ֹלהים אָנ ִֹכי ֲא ֶשׁ‬ ִ ‫אמר ֲה ַת ַחת ֱא‬ ֶ ֹ ‫וַ יִּ ַחר־אַף יַ ֲעקֹב ְבּ ָר ֵחל וַ יּ‬ ‘and Jacob was angry at Rachel and said: In the place of Godi ___j am I, [whi ___i has withheld from you the fruit of the womb?]j’   (Gen 30:2) A final structural observation comes in the form of a general constraint on extraposition. In Hebrew, extraposition is not simply rightward movement to any site further down in the clause. In her study of apposition, Thorion-Vardi (1987) observes that extraposed appositives must move to the rightmost position within its clause. That is, when a constituent is extraposed, it is not moved rightward to an arbitrary position in the clause’s linear structure but all the way to the end of the clause. I have found Thorion-Vardi’s observation to be mostly accurate, both for nonrelative extraposition (see Holmstedt 2014b: 134–35) and for relative clause extraposition. Throughout the list of relative extrapositions that I have collected, a more accurate description is that extraposed relatives move as far down the larger clause in which they reside as they can go until they reach another clausal edge. This clausal edge, whether of the main clause or another subordinate clause, is a boundary restricting further extraposition movement. 76 There is but one example that stretches this general constraint, given in (439). (439) ‫ת־שׁם ָק ְד ִשׁי‬ ֵ ‫ל־בּנָ יו וְ יִ נָּ זְ רוּ ִמ ָקּ ְד ֵשׁי ְבנֵ י־יִ ְשׂ ָר ֵאל וְ לֹא יְ ַח ְלּלוּ ֶא‬ ָ ‫ל־אַהר ֹן וְ ֶא‬ ֲ ‫ַדּ ֵבּר ֶא‬ ‫ֲא ֶשׁר ֵהם ַמ ְק ִדּ ִשׁים ִלי‬ ‘speak to Aaron and his sons and let them keep separate from the holy donationsi of the children of Israel ___j (so they do not profane my holy name) [whi that they are consecrating ___i to me]j’   (Lev 22:2) In (439) the relative head is separated from its relative clause by a parenthetical clause, ‫ולא יחללו את שׁם קדשׁי‬. Parentheses are typically identifiable 76. This constraint makes the proposal in Stackert 2010 difficult to accept. Stackert argues that the ‫ אשׁר‬clause in Deut 13:2 does not modify the NPs immediately preceding it (‫ )האות והמופת‬but, rather, the very distant ‫ נביא‬from v. 1. The presence of not one but two (!) full clauses between ‫ נביא‬and the relative ‫ אשׁר דבר אליך‬indicates that this proposal is too creative with the grammar of Hebrew, the literary merits of the resulting interpretation notwithstanding.

Relative Syntax

191

by their departure from the surrounding syntax, whether a change in person or verbal form or, as here, the interruption of a relative clause structure (see Holmstedt 2014b: 146–47). Since parentheses share many “extrasentential” features with other interruptive elements, such as vocatives and exclamatives (‫)הנה‬, the same questions regarding the identification of (436) as a case of extraposition, strictly speaking, also applies here to (439). That is, since the parenthetical element is probably inserted into its position after the derivation of the relative phrase structure, it is unlikely that this is a case where the relative clause moves rightward. 77 The point here, as it was above concerning (436) and (438), is that a relative clause may be at a distance from its relative head for multiple reasons, only one of which should be, properly speaking, labeled extraposition. 5.2.4.1. Function Identifying the presence of extraposition with some Hebrew relative clauses is only part of the task. The question remains: what is its function? In a recent study of clause edge phenomena, including extraposition, I argued that extraposition either reflects a right-branching Focus-marking on the moved constituent or a discourse-processing relinearization (Holmstedt 2014b: 138–40). 78 The most common use of extraposition relates to an automatic “processing” linear reordering, whereby complex entities, especially embedded constituents such as relative clauses, are moved rightward to allow the simpler syntactic constituents to be cognitively processed first and thus more easily. Consider, for example, the extraposed relative in (440). (440) ‫תּואל‬ ֵ ‫וְ ִהּנֵ ה ִר ְב ָקה י ֵֹצאת ֲא ֶׁשר יֻ ְּל ָדה ִל ְב‬ ‘and behold, Rebekahi ___j was coming out [whi that ___i was born to Bethuel]j’ (Gen 24:15) Since the basic word order in null-copula participial clauses is overwhelmingly subject-predicate (see §2.3.1), I am unable to find any reason for suggesting that ‫ רבקה‬was Topic- or Focus-fronted from a lower position, stranding the ‫ אשׁר‬relative. Similarly, associating ‫ אשׁר ילדה לבתואל‬with right-branching Focus makes no contextual sense. Thus, the likeliest explanation of the extraposition in (440) is the cognitive preference to process simpler constituents before more-complex constituents. Though this preference does not often dic77.  See also Job 34:17–19, for which I take v. 18 as parenthetical; see Appendix A.7 for the example with translation. 78.  My reference to “right-branching” Focus reflects my position that linear order further within the clause does not necessarily correspond to a lower hierarchical order within the clause’s phrase structure. Rather, rightward movement in extraposition, as well as resumption within right-dislocation, situates the constituent in question in a position much higher in the clause hierarchy, in a right-branching Focus Phrase within the CP. See Holmstedt 2014b for further discussion.

192

Chapter 5

tate constituent order within a clause, there are clearly environments in which extraposition (including HNPS) may operate, and (430), (432), (433), (435), and (440) are good examples of this in Hebrew. In some cases, relative extraposition and right dislocation are also associated with Topic and Focus. Consider the two cases of extraposition in (441) and (442): (441) ‫ֶת־יֹוסף׃‬ ֵ ‫ל־מ ְצ ָריִ ם ֲא ֶשׁר לֹא־יָ ַדע א‬ ִ ‫ְך־ח ָדשׁ ַע‬ ָ ‫‏ וַ יָּ ָקם ֶמ ֶל‬ ‘and a new kingi ___j arose over Egypt [whi that ___i did not know Joseph]j’ (Exod 1:8) (442) ‫ר־א ֵה ַב ֶתְך יְ ָל ַדתּוּ ֲא ֶשׁר־‬ ֲ ‫ת־שׂ ָיב ֵתְך ִכּי ַכ ָלּ ֵתְך ֲא ֶשׁ‬ ֵ ‫וּל ַכ ְל ֵכּל ֶא‬ ְ ‫‏וְ ָהיָ ה ָלְך ְל ֵמ ִשׁיב נֶ ֶפשׁ‬ ‫טֹובה ָלְך ִמ ִשּׁ ְב ָעה ָבּנִ ים׃‬ ָ ‫ִהיא‬ ‘because your daughter-in-lawi, that you love ___j, has borne him, [whi that ___i has been better for you than seven sons]j’ (Ruth 4:15) While both extraposed relatives in (441) and (442) could be cases of heavy relatives shifted rightward, in the discourse context a right-branching Focus reading is also felicitous for both. In each case, the delay that the right-branching Focus causes allows the constituent to be set over against the alternatives in the contextually based membership set initiated by the Focus with the addition of the cognitive tension produced by delayed articulation in the linguistic structure (see Francis 2010; also Guéron 1980; Huck and Na 1990). In (441), the fact that the new pharaoh arose is important, but even more important is that the new pharaoh ‫לא ידע את יוסף‬. Thus, the extraposed relative may carry Focus, and thus the contrast with the contextually established set of related variables (e.g., [the previous pharaohs] did know, [this current pharaoh] does not know); the Focus signals the ominous conditions that will result in a shift from favor to oppression for the Hebrew. Similarly, in (442), it cannot be that the presence of a relative clause itself motivates extraposition, since the first relative modifying ‫ כלתך‬resides in its expected postnominal position. The second relative, ‫לך משׁבעה בנים אשׁר היא טובה‬, is a more complex, “heavier” relative than the first one. But it is also fitting in the discourse context that the culturally unexpected assertion that a daughter-in-law has been more profitable to Naomi than seven sons is marked for Focus, implicitly contrasting this assertion with the expected assertion, that even one son is more beneficial than seven daughters (much less a daughter-in-law). Given that clear cases of Focus-associated extraposition involve simple, non-heavy constituents (e.g., single words) whereas relative clauses are invariably complex and thus heavy constituents, I offer the Focus-based explanations above as reasonable possibilities, not firm conclusions.

Chapter 6

Relative Semantics 6.1.  Relative Semantic Typology Among the typological universals and tendencies for relative clauses that I listed in §1.1, those pertaining to the semantic features or relatives are repeated below in (443) and (444): (443) Semantics-Related Typological Universals b. All languages have semantically headed RCs (AU). c. All languages have restrictive or maximalizing RCs (AU). i. A nonrestrictive RC must contain a relative element (IU). (444) Semantics-Related Typological Tendencies d. For most languages, there is an incomplete participial relativization strategy that is restricted to subjects (GT). In table 1.1, I reduced these typological observations to two categories: (1) inflectional completeness of the relative and (2) kind of modification or relation. In the remainder of this chapter, I discuss how the Hebrew relative operates with regard to these two issues. 6.1.1.  Verbal Inflection within the Relative In addition to relative clauses formed with finite verbs, many languages also use a non-finite verbal relativization strategy. Consider the English relatives in (445)–(448): (445) The dogi that ___i ran down the street was a stray. (446) The dogi thati ___i is running down the street is a stray. (447) The dogi thati ___i was injured by the car was recovering well. (448) The dogi thati ___i was black with dirt was washed with a hose. (449)  The dogi thati ___i was a golden retriever was beautiful. The relative in (445) illustrates the use of a finite verb, while the relatives in (446)–(448) show the use of a copular verb plus present participle, past participle, and adjective, respectively. While the configurations of the relatives in (445)–(448) appear similar, structurally they differ in the use of the finite 193

194

Chapter 6

lexical verb versus the copula. First, the use of a participle or adjective requires the insertion of an inflected lexical copula (is, was, etc.). Second, the examples employing a copula have a completely different zero-relative configuration, as in (450). Third, English disallows the NP example (450c) as a prenominal zero-relative. (450) English Non-Finite Zero-Relatives a. The running dog b. The injured dog c. The black dog d. *The golden retriever dog In the zero-relatives formed with the participles of adjectives, the relative may be placed before the head (prenominal) in English. However, this placement is typically much less felicitous, if not unacceptable, if the participle or adjective is accompanied by further modification, as in (451). (451) English Non-Finite Zero-Relatives, with Modifiers a. The running (??down the street) dog b. The injured (??by the car) dog c. The black (??with dirt) dog The addition of modifiers motivates the expanded structure of the marked relatives in (446)–(448). As I discussed above in §1.1.3, these examples illustrate the very close relationship between relative clauses and other forms of noun modification (see also de Vries 2002: 58–59). 1 Significantly, this kind of “incomplete” non-finite relativization (de Vries 2002: 59), which accords with the general tendency in (444d)—languages generally restrict non-finite relativization to subjects—differs from true participial relativization, which may be from any position within the relative clause, as in the Telugu example in (452). (452) True Participial Relativization: Telugu (Lehmann 1984: 50) [Mīru nāku ic-cin-a] pustukamu cirigipǒ-yin-adi you.pl me give-past-ptcp book.nom tear.up-past-3.sg ‘the booki you gave __i me has been torn up’ (lit., ‘the booki you giving __i me . . .’) 1.  Note that the non-finite relatives in (446)–(448), such as the man (who was) running down the street bumped into the mayor’s dog, are syntactically and semantically different from “reduced” or “small” clauses, such as I heard the man running down the street. In the latter, the small clause is not simply a noun-modification strategy; it is the “reduced” form (i.e., lacking certain tense-aspect-mood features) of a full clause, such as that the man was running down the street. The reduced clause as a whole is, in the example above, the complement of the verb heard and thus the entire event or activity of the man running is what is heard. A study of reduced, or “small,” clauses in Biblical Hebrew is a desideratum; for small clauses in Modern Hebrew, among others see Rothstein 1995.

Relative Semantics

195

As I will illustrate below in §6.2, Hebrew is similar to English in that participial relatives are restricted to subject relativization (that is, only the subject may be promoted as the relative head); all full relatives include either a finite verb or a null copula, which carries full tense-aspect-mood features and licenses a participial, adjectival, or NP complement. 6.1.2.  Restrictive, Nonrestrictive, and Maximalizing Relatives In addition to the finite versus non-finite semantic distinction, many languages exhibit more than one type of semantic relationship between a relative head and its modifying relative clause. These relationships are typically referred to as restrictive, nonrestrictive (or, appositive), and maximalizing (Grosu and Landman 1998; Bianchi 2001; de Vries 2002: chap. 6), illustrated in (453). (453) Three Semantic Kinds of Relatives a. Restrictive: I took the three books that were on the table. b. Nonrestrictive: I took the three books, which were on the table. c. Maximalizing: I took the three books that were there on the table. The relative clause in (453a) defines the head the three books—i.e., it is only the three books lying on the table that are of concern in this clause, regardless of how many books are in the room. A restrictive relative, therefore, provides information necessary for the head to be correctly identified among a set of possible discourse referents, which in (453a) would be from among all the contextually likely books (e.g., books on the shelf, books on the desk). The nonrestrictive relative clause in (453b) does not define or limit the DP head, the three books, but simply adds information that the speaker/writer has deemed salient to the narrative. A helpful addition to (453b), which would drive home its appositional nature, is to include the phrase by the way—i.e., I took the three book, which, by the way, were on the table. This phrase functions as an informal text to determine the restrictiveness of a relative clause: if by the way may be added without any change of meaning, the relative is nonrestrictive; if it changes the meaning, the relative is restrictive. The maximalizing relative clause in (453c) specifies that all three books on the table were taken and that there were no more than three books on the table. Moreover, a maximalizing relative does not define the head over against some other set relating to the head DP (e.g., the books on the shelf); such a contrast is the semantic domain of a restrictive relative clause. Rather, a maximalizing relative has only the amount or number of relative-clause head items in view. In the case of (453c), then, the statement is true if and only if the three books on the table both were taken and constituted all of the books on the table. I have not identified any maximalizing relatives in Biblical Hebrew, though both restrictive and nonrestrictive strategies are used.

196

Chapter 6

Some languages distinguish restrictive and nonrestrictive relatives by morphological, lexical, orthographic, or intonational means, though other languages that have the semantic distinction do not signal it by grammatical means (Alexiadou et al. 2000: 30–34; Nikolaeva 2006: 502). English is a language that offers disambiguation strategies, even if they are not used or required. For example, a comparison of (453a) and (453b) demonstrates that English nonrestrictive relative clauses may be distinguished intonationally by a pause and orthographically by commas. Additionally, the relative complementizer that cannot be used to introduce a nonrestrictive relative; however, the nonuse of that in favor of the appropriate wh-word results in potential ambiguity (which, hopefully, is offset by speaker/author’s use of the other differentiation strategies—i.e., pause or commas). 6.1.2.1.  Syntax of Nonrestrictive Relatives Besides the semantic difference connected to restrictiveness, the two types of relatives are syntactically different. In fact, though I will treat nonrestrictive relatives as true relative clauses, the semantic and syntactic differences often lead linguistics to set aside nonrestrictives as something not quite fitting the generally accepted definition of relative clauses, which includes delimitation of the head NP (Andrews 2007: 207–8), or the universal quoted at the outset of this book, that “[a]ll languages make use of restrictive relative clauses” (Downing 1978: 381, italics added). I addressed the essential semantic distinction in the previous section. In this section, I describe some of the typical syntactic features that distinguish a nonrestrictive relative from its restrictive counterpart, as well as provide a sketch of how the syntactic difference may be accounted for in phrase structure terms. Examples (454)–(459) below illustrate the various different uses of restrictive relatives and nonrestrictive relatives in English. First, nonrestrictive relatives are rarely separated from their heads—e.g., by extraposition, as in (454) (Emonds 1979: 234–35; Mallinson and Blake 1981; McCawley 1998: 447). 2 (454) *My father came in, who runs his own business. vs. My father, who runs his own business, came in. Additionally, proper names, which are tied to unique referents that need no further identification, cannot be modified by restrictive relatives but only by nonrestrictive relatives, as in (455) (Mallinson and Blake 1981: 359–66). (455) I saw Benjamin, who studies daily. vs. *I saw Benjamin who studies daily. 2.  There are occasionally nonrestrictive relatives that exhibit extraposition and appear to be grammatical; Fabb presents the following as such an example (1990: 59): I met John yesterday, who I like a lot.

Relative Semantics

197

Similarly, restrictive relatives can only modify NPs or DPs (456a), whereas non­restrictive relatives can modify any XP—e.g., NPs, DPs, VPs, PPs, and CPs—as in (456b), (457), (458), and (459), respectively (Fabb 1990: 60; Demirdache 1991: 108–9; McCawley 1998: 447). (456) Relative Clause Modifying an NP/DP a. I called a friend that was working. b. I called Rachel, who was working. (457) Relative Clause Modifying a VP a. Benjamin weight-lifted, which I was having a problem doing. b. *Benjamin weightlifted that I was having a problem doing. (458) Relative Clause Modifying a PP a. Rachel put the box in the car, where I had put the others. b. *Rachel put the box in the car that I had put the others. (459) Relative Clause Modifying a CP a. Benjamin angered his wife, which was not a good idea. b. *Benjamin angered his wife that was not a good idea. Finally, (460) demonstrates that null-head relatives are, by nature, restrictive since the semantic content of the null head cannot be recovered without, minimally, the information contained in the following relative clause (in many cases, the discourse context is also necessary in order to determine what the null head represents). (460) Null-Head Relative a. Rachel saw what I did to the car. b. *Rachel saw what, I did to the car. By these examples and brief descriptions, I have illustrated just some of the features that often lead linguists to study nonrestrictive relatives as a different construction from restrictive relatives. Indeed, while there remain some points of debate about the structure of restrictive relatives, particularly regarding where the relative clause proper is attached to the head constituent in the phrase structure, nonrestrictive relatives pose a different set of problems. The article “Appositive Relatives Have No Properties” is a seminal generative analysis of nonrestrictive relatives in which Emonds (1979) proposes that his Main Clause Hypothesis can account for the syntactic characteristics of these relatives. The essence of the Main Clause Hypothesis is that non­restrictive relatives “are derived from [main] clauses which are deep structure coordinate right sisters to the [main] clause containing the modified antecedent” (Emonds 1979: 212). It is through two transformations

198

Chapter 6

that the second main clause becomes a nonrestrictive relative, illustrated in (461) (Emonds 1979: 211–15). (The symbol E stands for Expression, which Emonds uses for the highest category in a sentence, that is, a clause that cannot be subordinated.) (461) Too much sun made these tomatoes rot on the vine, and we paid a lot for them. E

E E



̄ S



Too much sun . . . and we paid . . . The first transformation (“Parenthetical Formation”) serves to move any constituent (the VP in [462]) from the left main clause and attach it to the right of the right main clause. (462) Too much sun made these tomatoes, and we paid a lot for them, rot on the vine. E E E VP S S





rot on the vine

Too much sun made and we paid a lot these tomatoes __i for them The second transformation (“S̄-Attachment”) deletes the coordinating conjunction between the two clauses and attaches the right main clause to the left main clause.

199

Relative Semantics

(463) Too much sun made these tomatoes, which we paid a lot for, rot on the vine. E E VP S S



Too much sun made these tomatoes __i



rot on the vine

which we paid a lot for __i

Significantly, in this analysis nonrestrictive relatives and their antecedents do not form a single constituent at any stage in the derivation. Demirdache (1991) builds on Emonds’s earlier analysis, but contra Emonds she argues that syntactically nonrestrictive relatives are subordinate clauses (thus, they do not differ syntactically from restrictive relatives). In order to identify the structural differences between restrictive and nonrestrictive relatives (since both are syntactically subordinate), Demirdache returns to the distinguishing features I presented above: restrictive relatives can only modify NPs or DPs, while nonrestrictive relatives can modify NPs, DPs, APs, VPs, PPs, CPs, etc.; restrictive relatives modify nonreferential NPs, while nonrestrictive relatives can modify referential NPs (i.e., DPs). Therefore, Demirdache concludes that nonrestrictive relatives are inserted from the lexicon into a position that is adjoined to their antecedent, illustrated in (464)–(468). Example (464) provides Demirdache’s basic structure for nonrestrictives (where the XP stands for any type of phrasal constituent), while (465)–(468) illustrate how non­restrictives syntactically modify various types of heads. (464) Syntactic Structure on Nonrestrictive Relatives XP XP CP In (465), the antecedent is the DP Rachel. Thus, the nonrestrictive relative CP is adjoined to the DP, necessitating that a higher DP is projected. The entire structure remains a DP.

200

Chapter 6

(465) I called Rachel, who was working. DP DP CP Rachel who was working In (466), the nonrestrictive relative CP is adjoined to the VP weightlifted, creating the larger VP that includes both the lower VP and the relative CP. (466)  Don weight-lifted, which I was having a problem doing. VP VP CP Weight- which I was having lifted a problem doing In (467), the relative CP is adjoined to the PP in the car. The result is a larger PP, in the car, where I had put the others, that includes both the lower PP and the relative CP. (467)  Rachel put the box in the car, where I had put the others. PP PP CP in the car

where I had put the others

Finally, (468) shows a relative CP that modifies and is adjoined to another CP Don angered his wife. The adjunction of the relative CP causes the main CP to project, creating the higher CP that includes both the lower, main CP and the relative CP.

Relative Semantics

201

(468)  Don angered his wife, which was not a good idea. CP CP CP Don which was angered not a good his wife idea The fundamental difference between a nonrestrictive relative and a restrictive relative is where the attachment site is. Demirdache illustrates this with the two trees in (469), after which I provide her explanation. (469)  Nonrestrictive Relative Restrictive Relative DP DP CP D NP

DP D′ D NP NP CP

The primary difference between these two types of modification is that non­ restrictives can modify referential expressions like proper names, whereas restrictives cannot. . . . [U]nder the DP hypothesis (Abney 1987), DPs are referential categories, NPs are not (they are predicates). Since a nonrestrictive relative can be adjoined to any XP, its head can be a referential category (DP) or any predicate including NP (i.e., it can modify an indefinite). In contrast, a restrictive relative always attaches to NP. Thus, its head will never be referential. (Demirdache 1991: 111)

Demirdache also notes that the distinction she makes in (469) between restrictive and nonrestrictive relatives correctly predicts that nonrestrictive relatives follow the complements and modifiers of a noun, as in (470). Thus, the non­ restrictive cannot precede the restrictive relative in (471). (470)  The girl that I like, who(m) John dislikes (471)  *The girl, who(m) I saw, that John dislikes In (470), the restrictive relative, being a constituent within the NP, is positioned after the head and before the nonrestrictive relative, which is not a constituent

202

Chapter 6

within the NP (but a constituent within the DP). Example (471) illustrates that the nonrestrictive, since it is a constituent of the larger DP structure, cannot precede the restrictive relative, which is a constituent of the smaller NP structure. Demirdache does not follow Emonds’s syntactic analysis of nonrestrictive relatives by analyzing them and their antecedents as a single constituent. However, she does follow Emonds in proposing that nonrestrictive relatives be interpreted as independent clauses. Nonrestrictive relatives “are lifted out of the clause in which they were base-generated and adjoined to the root clause at L[ogical] F[orm]” (Demirdache 1991: 113). Thus, the nonrestrictive, which is syntactically adjoined to the DP in (472), is converted into the structure in (473) at Logical Form in order to receive an interpretation. (472) A Nonrestrictive Relative before Logical Form CP IP DP VP Lisa

CP

is late

whoi __i is my friend (473)  A Nonrestrictive Relative at Logical Form CP

CP CP



IP whoi __i is my friend

DP VP Lisa

__ i

is late

Relative Semantics

203

In summary, Demirdache combines Emonds’s basic proposal that nonrestrictive relatives receive an independent clause interpretation with the position that nonrestrictive relatives do not differ from restrictive relatives in their syntactic relationship to their antecedent: they are both subordinate modifiers that form a single constituent with their heads. 6.2.  Hebrew Relatives and Verbal Semantics Hebrew relatives exhibit the use of the two primary verb conjugations—perfective (qatal), and imperfective (yiqtol)—as well as the null copula with participial, adjectival, prepositional, and nominal complements (also see §5.2.3.3). (474) ‫ל־כּל־יִ ְשׂ ָר ֵאל ְבּ ֵע ֶבר ַהיַּ ְר ֵדּן‬ ָ ‫‏אלּה ַה ְדּ ָב ִרים ֲא ֶשׁר ִדּ ֶבּר מ ֶֹשׁה ֶא‬ ֵ ‘these are the wordsi whi that Moses spoke ___i to all Israel in the region beyond the Jordan’ (Deut 1:1) (475) ‫וּשׁ ַמ ְע ִתּיו‬ ְ ‫‏וְ ַה ָדּ ָבר ֲא ֶשׁר יִ ְק ֶשׁה ִמ ֶכּם ַתּ ְק ִרבוּן ֵא ַלי‬ ‘and the matteri whi that ___i is too difficult for you you shall bring to me and I shall hear it’ (Deut 1:17) (476) ‫יֹושׁב ְבּ ֶח ְשׁבֹּון‬ ֵ ‫אַח ֵרי ַהכֹּתֹו ֵאת ִסיחֹן ֶמ ֶלְך ָה ֱאמ ִֹרי ֲא ֶשׁר‬ ֲ ‫‏‬ ‘after he smote Sihoni, king of the Amorites, whi that ___i was dwelling in Heshbon’ (Deut 1:4) (477) ‫רֹובים ַליהוָ ה‬ ִ ‫ר־ק‬ ְ ‫ַהּכ ֲֹהנִ ים ֲא ֶׁש‬ ‘the priestsi whi that ___i are near to Yhwh’ (Ezek 42:13) (478) ‫ּוב ַא ְר ַּבע ֶע ְׂש ֵרה ָׁשנָ ה ָּבא ְכ ָד ְר ָלע ֶֹמר וְ ַה ְּמ ָל ִכים ֲא ֶׁשר ִאּתֹו‬ ְ ‘and in the fourteenth year, Chedorlaomer and the kingsi whi that ___i were with him, came’ (Gen 14:5) (479) ‫ר־ח ְטא‬ ֵ ‫אּו־לי ָעֹון ֲא ֶׁש‬ ִ ‫יִמ ְצ‬ ְ ‫יעי לֹא‬ ַ ִ‫ָּכל־יְ ג‬ ‘(in) all my labors, they do not find in me an iniquityi whi that ___i is sin’ (Hos 12:9) There are no examples of an ‫אשׁר‬-relative clause with its primary subordinate verb as a past-narrative wayyiqtol, an infinitive (excepting the Kethiv of Eccl 6:10), or an imperative (excepting the textually difficult Ps 8:2). This is also the case for zero-relatives, which may include any of the same predicate types as ‫אשׁר‬-relatives and relativize from any position—subject, complement, or adjunct (see above, §5.2). 3 3.  For zero-relatives with a finite verb, see Gen 1:1; 26:10; 29:25; 42:28; 49:27; Exod 4:13; 6:28; 14:11; 15:17 (2×); 18:20; Lev 7:35; 13:20, 39; 17:11; Deut 4:15; 32:11 (5×), 17 (3×), 37; Judg 8:1; 15:11; 16:15; 1 Sam 6:9; 10:11; 14:21; 26:14; 2 Sam 22:44; 23:1, 9; 1  Kgs 13:12; 17:14 (K); 22:24; 2  Kgs 3:8; Isa 6:6; 10:3; 14:16–17 (2×); 15:7; 23:13; 28:16; 29:1; 30:5, 9; 31:1; 32:13; 40:20 (2×); 41:24; 42:1 (2×), 11, 16 (2×); 44:1, 2; 48:17; 50:9; 51:1 (2×), 2, 12 (2×); 53:7 (2×); 54:1 (2×); 55:13; 56:2 (2×); 62:1; 63:19 (2×); 64:2;

204

Chapter 6

The null-copula examples in (476)–(479) establish that Hebrew does allow certain non-finite predicates with relatives, as with the English examples in (446)– (448). As I discussed for the English examples in (450)–(451), the acceptability of the ‫ אשׁר‬Hebrew examples in (476)–(478) or the zero relative in (480), (480) ‫יתים‬ ִ ‫‏ּגֹוי ח ֵֹטא ַעם ֶּכ ֶבד ָעֹון זֶ ַרע ְמ ֵר ִעים ָּבנִ ים ַמ ְׁש ִח‬ ‘a nationi whi (that) ___i (is) sinning, a peoplej whj (that) ___j (is) heavy (with) iniquity, a seedk whk (that) ___k (is) doing evil, sonsl whl (that) ___l (are) acting corruptly’ (Isa 1:4), illustrates a similar ambiguity on the cline of internal complexity (§1.1.3). Excepting example (479) with the NP complement of the null copula, the relatives in (476)–(478) would be acceptable even if the ‫ אשׁר‬were removed. This is also accurate for ‫ה‬-relatives, as I demonstrated in §3.2.3, illustrated by (481) and (482): (481) ‫ית־אל ִמזְ ְר ָחה ַה ֶּׁש ֶמׁש ִל ְמ ִס ָּלה‬ ֵ ‫ימה ֲא ֶׁשר ִמ ְּצפֹונָ ה ְל ֵב‬ ָ ‫ִהּנֵ ה ַחג־יְ הוָ ה ְּב ִׁשלֹו ִמּיָ ִמים יָ ִמ‬ ‫ית־אל ְׁש ֶכ ָמה‬ ֵ ‫ָהע ָֹלה ִמ ֵּב‬ ‘see—the festival of Yhwh (is) at Shiloh from year to year, which is north of Bethel, on the east of a highwayi whi that ___i goes up from Bethel to Shechem’ (Judg 21:19)

(482) ‫רּוח ָה ָר ָעה‬ ַ ‫וְ ָס ָרה ֵמ ָע ָליו‬ ‘and a spiriti whi that ___i (was) evil would depart from him’   (1 Sam 16:23)

65:1 (3×); Jer 2:6 (2×), 8, 11; 5:15; 13:20; 15:14; 23:9, 29; 30:21; 31:2; 36:2; 48:36; 49:12; 51:43 (2×); Ezek 20:25b; Hos 4:14; 6:3; Jonah 1:10; Mic 5:2; Hab 1:5; 3:16; Zeph 2:2; 3:3, 17; Mal 2:16; Pss 4:8; 7:16; 9:16; 12:6; 15:3–5 (8×); 16:4; 18:3, 44; 21:5; 25:12b; 32:2a; 33:12; 34:9; 35:8; 38:14; 44:2; 46:2; 51:10; 52:9; 56:4, 10; 58:5; 65:5 (3×); 66:9; 68:31; 71:18; 74:2 (2×); 78:6, 8 (2×); 78:39b; 81:6; 83:15; 90:5, 15 (2×); 103:5; 104:9, 32; 118:22; 141:9; Job 3:3 (2×), 16; 4:7 (2×); 6:10, 17; 7:2 (2×); 9:26; 11:16; 13:28; 17:3; 18:21; 19:3; 21:27; 28:1, 7 (2×); 29:16; 31:12 (2×); 32:19; 33:21; 34:7–8 (2×); 35:5; 36:27; 38:19, 24 (2×); Prov 8:32; 9:5; 23:32 (2×); 30:15 (2×), 17 (2×), 18 (2×); Eccl 1:13, 18; 5:12; 10:5; 12:11; Lam 1:14, 21; 3:1, 25, 57; 4:14, 17; Ezra 1:5; 1 Chr 12:24; 15:12; 16:15; 29:3 (2×); 2 Chr 1:4; 15:11; 18:23; 20:22; 24:11; 28:9; 29:21; 30:19; 31:19. For zero-relatives with a null complement and participle, see Isa 1:4a, c, d, 30; 31:8 (2×); Jer 2:2; Ezek 22:24a; 38:11a; Hos 6:3; Hab 1:8, 14; Zeph 2:1; Pss 5:5; 78:39; Job 29:12; Prov 6:16; 30:29 (2×); Neh 8:10. For zero-relatives with a null copula and prepositional, adjectival, or NP complement, see Gen 15:13; 24:22 (2×); Lev 22:4; Num 7:13; Deut 4:38; 7:1; 9:1, 14; 11:23; Josh 7:21; 2 Sam 20:21; Isa 1:4b; 51:7; 66:1; Jer 5:21 (2×); Ezek 20:25a; 22:24b; 38:11b; Amos 2:13; 3:9; Hab 1:6; Zech 6:12; Ps 7:9; 9:18; 17:1; 25:12a; 32:2b; 84:6 (2×); 107:40; 109:16; Job 3:15; 27:3 (2×); Prov 9:16; 22:29; 26:17; 30:24; Eccl 2:1, 26; Lam 1:1, 10; 3:41; 2 Chr 16:9.

Relative Semantics

205

Both zero- and ‫ה‬-relatives with participles and adjectives are often understood as nonrelative structures in Hebrew (see, for example, GBH 514, in which both types are described simply as “attributive” noun modification). However, as I argued above in §3.2.3, this is not the most insightful or accurate analysis of the Hebrew patterns. In terms of verbal semantics, what is clear from the Hebrew data is that Hebrew has only one general constraint: ‫ה‬-relatives strongly prefer a null copula with a participle or an adjective and strongly disprefer the lexical finite verbs (qatal and yiqtol). The Hebrew data (see also my discussion of zero-relatives in §5.2) also strongly accord with the general tendency in (444d): both zerorelatives and ‫ה‬-relatives in which the predicate consists of a null copula and participle or adjective complement overwhelmingly reflect the promotion of the subject as the relative head. 6.3.  Hebrew Relatives and Restrictiveness In §1.1.3, I indicated that Hebrew uses both restrictive and nonrestrictive relative clauses. I also asserted that the basic semantic difference between the two types of relative clauses has not been adequately discussed in Hebrew scholarship. For instance, Waltke and O’Connor (IBHS; see also BHRG) never mention the difference between restrictive and nonrestrictive relatives, and Joüon and Muraoka (GBH 592) direct attention to the issue in a single paragraph (using the terms “limiting” and “non-limiting”) but do not pursue how (or even if) the two types are manifested in Hebrew syntax (see also Arnold and Choi 2003: 184–85). This neglect of the restrictiveness issues in Hebrew relatives may be due to the fact that the majority of relative clauses in the Hebrew Bible are restrictive, as is the pair in (483). (483) ‫ּובין ַה ַּמיִ ם ֲא ֶׁשר‬ ֵ ‫ת־ה ָר ִק ַיע וַ ְּיַב ֵּדל ֵּבין ַה ַּמיִ ם ֲא ֶׁשר ִמ ַּת ַחת ָל ָר ִק ַיע‬ ָ ‫ֹלהים ֶא‬ ִ ‫וַ ּיַ ַעׂש ֱא‬ ‫ֵמ ַעל ָל ָר ִק ַיע‬ ‘and God made the firmament; and he divided between the waters that were below the firmament and the waters that were above the firmament’ (Gen 1:7) The two relative clauses in (483) delimit their referents, distinguishing the waters above the firmament from those below. There is no question that these are restrictive relatives because each one supplies crucial identifying information about its respective head that enables the reader/listener to distinguish between the two groups of waters being mentioned. Similarly, the relative clause in (484) is restrictive: (484) (= [48]) ‫ן־ה ָא ָדם ְל ִא ָּׁשה‬ ָ ‫ר־ל ַקח ִמ‬ ָ ‫ת־ה ֵּצ ָלע ֲא ֶׁש‬ ַ ‫ֹלהים ֶא‬ ִ ‫‏וַ ֶּיִבן יְ הוָ ה ֱא‬ ‘and Yhwh God built the ribi whi that he had taken ___i from the man into a woman’ (Gen 2:22)

206

Chapter 6

As I explained in §1.1.3, the relative clause in (484) unambiguously identifies the ‫ צלע‬that is referenced—that is, it “limits” this ‫ צלע‬among all possible ‫צלע‬s relevant to the discourse/author/audience to the specific ‫ צלע‬that God took from the man. In contrast, the relative head in (485) is the proper noun ‫יהוה‬, which typically cannot be further defined since proper nouns need no delimitation to identify the intended referent (indeed, that is the purpose of naming). 4 (485) (= [49]) ‫יאָך ֵמ ֶא ֶרץ ִמ ְצ ַריִ ם ִמ ֵּבית‬ ֲ ‫הֹוצ‬ ִ ‫ן־ּת ְׁש ַּכח ֶאת־יְ הוָ ה ֲא ֶׁשר‬ ִ ‫ִה ָּׁש ֶמר ְלָך ֶּפ‬ ‫ֲע ָב ִדים‬ ‘Take care, lest you forget Yhwhi, whi that ___i brought you out of the land of Egypt’ (Deut 6:12) The relative is clearly nonrestrictive in that it provides information about what ‫ יהוה‬has done for the Hebrews, but it does not aid in identifying the god ‫יהוה‬ from among a set of deities. The relatives in (483)–(485), the first two restrictive and the third non­ restrictive, suggest that Hebrew does not use any simple lexical or orthographic means to distinguish the two semantic relative-clause types. Unfortunately, due to the written nature of our ancient data, if ancient Hebrew once used intonational means to demarcate a nonrestrictive relative, we no longer have access to this feature. The remaining relative words, ‫שׁ‬, the ‫ז‬-series, and ‫ה‬, also appear in both types of relatives, as in (486)–(491). (486) ‫י־הכֹּל ֶה ֶבל‬ ַ ‫ת־ה ַחיִּ ים ִכּי ַרע ָע ַלי ַה ַמּ ֲע ֶשׂה ֶשׁנַּ ֲע ָשׂה ַתּ ַחת ַה ָשּׁ ֶמשׁ ִכּ‬ ַ ‫אתי ֶא‬ ִ ֵ‫וְ ָשׂנ‬ ‫רוּח׃‬ ַ ‫וּרעוּת‬‎ ְ ‘and I hated life because the worki whi that ___i was done under the sun was grievous to me’ (Eccl 2:17) (487) ‫‏עם־זוּ יָ ַצ ְר ִתּי ִלי ְתּ ִה ָלּ ִתי יְ ַס ֵפּרוּ׃‬ ַ‎ ‘a peoplei whi that I formed ___i for myself will recount my praise’   (Isa 43:21) 4.  Accordingly to McCawley (1998: 481 n. 12; also see below, table 6.1, feature a), the constraint on nonrestrictive modification of proper nouns “relates only to proper nouns that are used as proper nouns. A proper noun that is used as a common noun can host a restrictive relative as well as any inherently common noun can: The Harry Smith who took your phonetics course last year has transferred to Cornell.” For a rare restrictive relative with a proper name, see Gen 31:49. There were many places called ‫‘ המצפה‬the watchtower’ in the Hebrew Bible, and the common noun appears to have become a proper noun (IBHS 249, §13.6); thus, in order to identify precisely which ‫ המצפה‬was being mentioned, a restrictive relative was often needed. For a personal name used in this way, see also Gen 36:24, in which ‫ ענה‬is identified as “(the) ʿAnah who found the hot springs in the wilderness.” Similarly, the use of the ‫ אשׁר‬relative with the name in Esth 1:1 is likely restrictive: ‫ימי ֲא ַח ְׁשוֵ רֹוׁש הּוא‬ ֵ ‫וַ יְ ִהי ִּב‬ ‫ּומ ָאה ְמ ִדינָ ה‬ ֵ ‫‘ ֲא ַח ְׁשוֵ רֹוׁש ַהּמ ֵֹלְך ֵמהֹּדּו וְ ַעד־ּכּוׁש ֶׁש ַבע וְ ֶע ְׂש ִרים‬and it was in the days of Ahasuerus (he is [the] Ahasuerus who was ruling from India to Ethiopia . . .)’ (Esth 1:1).

Relative Semantics

207

(488) ‫וּמי ֶבן־יִ ָשׁי ַהיֹּום ַרבּוּ ֲע ָב ִדים ַה ִמּ ְת ָפּ ְר ִצים ִאישׁ ִמ ְפּנֵ י ֲאד ֹנָ יו׃‬ ִ ‫‏מי ָדוִ ד‬ ִ‎ ‘Who is David and who is the son of Jesse? Today there are many servantsi whi that ___i are breaking away, each from his master’   (1 Sam 25:10) In each of the examples (486)–(488), the relative clauses are clearly restrictive. The information provided in each of the relative clauses is necessary in order to identify the precise referents of the respective heads. For instance, in (488), the participle with the relative ‫ ה‬follows an indefinite noun; thus, the only way to narrow the rather broad referential possibilities (the essential function of restrictive relatives) for ‘servants’ is on the basis of the information given within the following relative. In contrast to (486)–(488), the three examples in (489)–(491) present relatives that are unnecessary for the identification of their antecedents. (489) ‫יהם׃‬ ֶ ֵ‫ ָ‏‏בּרוְּך יְ הוָ ה ֶשׁלֹּא נְ ָתנָ נוּ ֶט ֶרף ְל ִשׁנּ‬‎ ‘blessed is Yhwhi, whi that ___i has not given us (as) prey for their teeth’ (Ps 124:6) (490) ‫שֹׁוסה יַ ֲעקֹב וְ יִ ְשׂ ָר ֵאל ְלבֹזְ זִ ים ֲהלֹוא יְ הוָ ה זוּ ָח ָטאנוּ לֹו‬ ֶ ‫‏מי־נָ ַתן ִל ְמ‬ ִ‎ ‘Who gave Jacob for spoil and Israel to the plunderers? Was it not Yhwhi, whi we sinned against himi’ (Isa 42:24 Kethiv) (491) ‫יתי־לֹו׃‬ ִ ֵ‫יתי ַליהוָ ה ַה ַמּ ְס ִתּיר ָפּנָ יו ִמ ֵבּית יַ ֲעקֹב וְ ִקוּ‬ ִ ‫‏‏‏ וְ ִח ִכּ‬‎ ‘and I shall wait for Yhwhi, whi that ___i hides his face from the house of Jacob’ (Isa 8:17) In examples (489)–(491), the noun being modified in each case is a proper noun. As I indicated above (see n. 4), proper nouns in English (unless two referents use the same proper noun designation, e.g. John A. and John B.) cannot be modified by restrictive relatives. Hebrew relative clauses are similarly constrained. Thus, in (489)–(491), since the proper name ‫ יהוה‬is certainly not an ambiguous reference in the Hebrew Bible, the relative clauses serve to provide additional information about, e.g., his character, actions, or words. Note that the Hebrew relative clause data given in (483)–(491) illustrate that all the Hebrew relative words may be used in either restrictive or nonrestrictive relatives, thereby eliminating the choice of the relative word as a distinguishing feature vis-à-vis restrictiveness. Furthermore, we should note that neither the ‫אשׁר‬- nor the ‫שׁ‬-relatives given in (484)–(486) and (489) overtly resume their respective heads. 5 In addition, the absence of a pattern that connects 5.  There is a difference in resumption between the ‫זו‬-restrictive (487) and the ‫זו‬-non­ restrictive (490) examples, but due to the paucity of data for ‫ז‬-relatives in the biblical text it is difficult to explain the differences. Resumption does not occur in ‫ה‬-relatives; see above in §3.2.3.

208

Chapter 6

resumption to restrictiveness suggests that resumption alone is not a distinguishing criterion with respect to restrictiveness (see above, §5.2.3). Note that the restrictive versus nonrestrictive distinction applies to overt relative words—in §3.2.5, I suggested that zero-relative clauses are always restrictive. In the 298 zero-relative clauses that I have identified in the Hebrew Bible, the zero-relatives present information that is necessary in order to identify the referent of the head precisely, as in (492). (492) ‫ת־ה ֶּד ֶרְך יֵ ְלכּו ָבּה‬ ַ ‫הֹוד ְע ָּת ָל ֶהם ֶא‬ ַ ְ‫ו‬ ‘and you shall make known to them the wayi whi ∅(that) they will walk in iti’ (Exod 18:20) Out of the 298 zero-relatives, only 2 of the examples appear to be nonrestrictive. However, each case can be either given an alternative syntactic explanation or interpreted restrictively. The first example is given in (493). (493) ‫יכָך‬ ֲ ‫הֹועיל ַמ ְד ִר‬ ִ ‫ֹלהיָך ְמ ַל ֶּמ ְדָך ְל‬ ֶ ‫ֹה־א ַמר יְ הוָ ה ּג ַֹא ְלָך ְקדֹוׁש יִ ְׂש ָר ֵאל ֲאנִ י יְ הוָ ה ֱא‬ ָ ‫ּכ‬ ‫ְּב ֶד ֶרְך ֵּת ֵלְך‬ ‘thus has said Yhwh, your redeemer, the Holy One of Israel: I am Yhwh your god, ∅(the one)i whi ∅(that) ___i teaches you to profit, ∅(the one)i whi ∅(that) ___ i leads you on the way you should walk’ (Isa 48:17) The string of constituents ‫ מדריכך‬. . . ‫ מלמדך‬. . . ‫ אני יהוה אלהיך‬is often analyzed and thus translated as a verbless clause, ‘I am Yhwh your god’ followed by two zero-relatives that modify the proper noun Yhwh. Thus, the nrsv provides the following translation: “I am the Lord your God, who teaches you for your own good, who leads you in the way you should go” (emphases mine). However, it is possible to analyze the constituents ‫ מלמדך‬and ‫ מדריכך‬not only as the participial predicates within zero-relative clauses but also as the predicates within null-headed zero-relative clauses. Thus, the relatives no longer directly modify the proper noun ‫ יהוה‬but, rather, the null head of the relative (which in turn refers back to ‫)יהוה‬. A possible translation (which illustrates a null-head zero-relative analysis) of these two relative clauses is then ‘I am Yhwh your god; (the one) (who) teaches you . . . , (the one) (who) leads you . . .’. The second example is more difficult: it appears in Isa 51:2, provided in (494). (494) (= [316]) ‫חֹול ְל ֶכם‬ ֶ ‫ל־ׂש ָרה ְּת‬ ָ ‫יכם וְ ֶא‬ ֶ ‫ל־א ְב ָר ָהם ֲא ִב‬ ַ ‫ַה ִּביטּו ֶא‬ ‘look to Abraham your father and to (the) Sarahi whi (that) ___i bore you’ (Isa 51:2) Given that the zero-relative clause data overwhelmingly indicate that such relative clauses are only used for restrictive modification, it is possible that we

Relative Semantics

209

Table 6.1.  Criteria for Determining Restrictiveness Relative Head a. A head cannot be a unique referent to support a restrictive RC. b. Restrictive RCs only modify NPs, whereas nonrestrictive RCs may modify any type of head. c. An indefinite head must be specific to support a nonrestrictive RC. Relative Element d. Zero-relativization is only restrictive.

should interpret the clause in (494) as referring to that particular Sarah who engendered the Hebrew people (see above, n. 4). Finally, a zero-relative is often provided for the string of constituents ‫יהוה‬ ‫ קוינו לו‬in (495). (495) ‫יׁשּועתֹו‬ ָ ‫יענּו זֶ ה יְ הוָ ה ִקּוִ ינּו לֹו נָ גִ ָילה וְ נִ ְׂש ְמ ָחה ִּב‬ ֵ ‫יֹוׁש‬ ִ ְ‫ֹלהינּו זֶ ה ִקּוִ ינּו לֹו ו‬ ֵ ‫ִהּנֵ ה ֱא‬ ‘Look—our god who we have waited for him that he would save us, this (is) Yhwh; we have waited for him, let us rejoice and be joyful in his salvation’ (Isa 25:9) However, the relative in (495) clearly cannot be restrictive since the head is the unambiguously referential proper noun ‫יהוה‬. Thus, if ‫ יהוה קוינו לו‬does contain a relative, it must be restrictive. But since everywhere else zero-relatives are restrictive, whatever the syntax of ‫ יהוה קוינו לו‬in (495), it does not likely include a relative. 6 6.3.1.  How to Identify Relative Restrictiveness in Hebrew Although they are not as common as restrictive relatives, Hebrew non­ restrictive relative clauses, such as (485) and (489)–(491), can be detected by using the cross-linguistic tendencies listed in table 6.1 as identifying criteria (from de Vries 2002: 182–96). 7 The nonrestrictive relative in (485) and (489)– (491) fit criterion (a) in table 6.1 by modifying a unique referent, ‫הוהי‬. While 6.  Note that a “nonrelative” analysis of ‫ יְ הוָ ה ִקּוִ ינּו לֹו‬appears to accord with the prosodic contour of the text according to the Masoretic accents: ‫ֹלהינּו ֶז֛ה ִ ִקּו֥ ינּו ֖לֹו‬ ֥ ֵ ‫וְ ָא ַמ ֙ר ַבּי֣ ֹום ַה ֔הּוא ִה ֵ֨נה ֱא‬ ‫יׁשּוע ֽתֹו׃‬ ָ ‫יענּו ז֤ ה יְ הוָ ֙ה ִ ִקּו֣ ינּו ֔לֹו ִָנג֥ ָילה וְ נִ ְש ְמ ָ ֖חה ִב‬ ֑ ֵ ‫ֹוש‬ ִ ֽ‫וְ י‬. Notice the Ole-weyored accent, which is on ‫יְ הוָ ה‬. This indicates that the Masoretes’ reading tradition reflected a prosodic break between ‫ יְ הוָ ה‬and ‫קּוִ ינּו לֹו‬, ִ and this prosodic break probably mirrors a syntactic break, as I have interpreted the verse. 7.  The above list of distinguishing criteria has been culled from the much longer discussion in de Vries 2002: 182–96. The criteria that I have included are those that are relevant for Hebrew relative clauses. It is also worth noting that the features of Hebrew restrictive and nonrestrictive relative clauses are in accordance with the typological statements in (3c, d, e, and i).

210

Chapter 6

they also modify an NP (b), as restrictive Hebrew relative clauses do, there is also the rare Hebrew relative clause that does not modify an NP, as in (496). (496) (= [222]) ‫א־ראּו ֲאב ֶֹתיָך‬ ָ ֹ ‫ל־מ ְצ ַריִ ם ֲא ֶׁשר ל‬ ִ ‫ּוב ֵּתי ָכ‬ ָ ‫ל־ע ָב ֶדיָך‬ ֲ ‫ּוב ֵּתי ָכ‬ ָ ‫ּומ ְלאּו ָב ֶּתיָך‬ ָ ‫וַ ֲאבֹות ֲאב ֶֹתיָך‬ ‘and your houses and the houses of all of your servants and the houses of all of Egypt shall be filled [with locusts, v. 4]i—whi that your fathers and your ancestors have not seen ___i. . . .’   (Exod 10:6) There is no appropriate NP head for the relative clause in (496) to modify; instead, the relative clause modifies an entire clause (CP). Given criterion (b) in table  6.1 and the natural reading of the relative clause in (496), it must be nonrestrictive. Criterion (c) asserts that nonrestrictive relative clauses can modify an indefinite NP, but only if the NP has a specific reading. This, too, is accurate for Hebrew, as example (497) demonstrates. (497) ‫ֹלהיָך‬ ֶ ‫א־ת ִקים ְלָך ַמ ֵּצ ָבה ֲא ֶׁשר ָׂשנֵ א יְ הוָ ה ֱא‬ ָ ֹ ‫‏וְ ל‬ ‘and do not erect for yourself a memorial stonei, whi that Yhwh your god hates ___i’ (Deut 16:22) The head of the relative clause in (497), ‫מצבה‬, is an indefinite NP, but it also clearly has a contextually specific reading. If the relative clause were restrictive, which might be the more natural reading at first sight, the clause would prohibit the precise ‫ מצבה‬that ‫ יהוה‬hates (versus others that he does not hate). But within the context of Deuteronomy, ‫ מצבות‬are prohibited in general (elsewhere, some ‫ מצבות‬are condoned, but this is not the view of Deuteronomy). Thus, the best reading of the relative clause in (497) is nonrestrictive. Finally, as I have already asserted, zero-marked relative clauses, such as the example in (498), cannot have a nonrestrictive interpretation. This is also in accordance with criterion (d) in table 6.1. (498) ‫ן־ה ָּׁש ָלל ֵה ִביאּו‬ ַ ‫‏וַ ּיִ זְ ְּבחּו ַליהוָ ה ַּבּיֹום ַההּוא ִמ‬ ‘and they sacrificed to Yhwh on that day from the bootyi whi ∅(that) they had brought __i’ (2 Chr 15:11) In (498), the NP head ‫ השׁלל‬is only fully identified by the restrictive relative clause—i.e., it is the booty they brought that they sacrificed, not the booty that, for example, they left home. Relative-clause stacking is one feature relating to restrictiveness in which Hebrew relative clauses differ from English relative clauses. Above in §5.2.2, I indicated that multiple nonrestrictive relatives cannot stack on the same head, illustrated again in (499). The Hebrew data do not exhibit the same restriction for nonrestrictive relative clauses; example (500) presents a string of rela-

Relative Semantics

211

tive clauses modifying the proper name David, indicating that the successive, stacked relative clauses are nonrestrictive. (499) Nonrestrictive Relative Clauses and Stacking in English *I saw John, who was yelling, who you dislike. vs. I saw the boy that was yelling that you dislike. (500) Nonrestrictive Relative Clauses and Stacking in Hebrew ‫ֹותי וְ ֻחּק ָֹתי‬ ַ ‫ְל ַמ ַען ָּדוִ ד ַע ְב ִּדי ֲא ֶׁשר ָּב ַח ְר ִּתי אֹתֹו ֲא ֶׁשר ָׁש ַמר ִמ ְצ‬ ‘for the sake of Davidi, my servant, whi that I chose himi, whi that ___i has kept my commandments and my statutes’ (1 Kgs 11:34) In Hebrew, then, relative-clause stacking may occur with both restrictive and nonrestrictive semantics. There is only one clear feature in Hebrew that distinguishes restrictive relatives from nonrestrictive ones, and it is neither syntactic nor is it always employed: the presence of a head noun in the bound or cliticized form. There are 217 examples in the Hebrew Bible of nouns cliticized to a relative clause introduced by ‫אשׁר‬, and all of them have restrictive interpretations, as in (501) and (502). 8 (501) ‫י־ׁשם ְּב ֵבית ַהּס ַֹהר‬ ָ ‫סּורים וַ יְ ִה‬ ִ ‫סּורי ַה ֶּמ ֶלְך ֲא‬ ֵ ‫ר־א‬ ֲ ‫ל־ּבית ַהּס ַֹהר ְמקֹום ֲא ֶׁש‬ ֵ ‫וַ ּיִ ְּתנֵ הּו ֶא‬ ‘and he put him into the round house, (the) place that the prisoners of the king were confined’ (Gen 39:20 Kethiv) (502) ‫ד־עֹולם ַּב ֲעֹון ֲא ֶׁשר־יָ ַדע‬ ָ ‫ת־ּביתֹו ַע‬ ֵ ‫וְ ִהּגַ ְד ִּתי לֹו ִּכי־ׁש ֵֹפט ֲאנִ י ֶא‬ ‘and I told him that I am (about to) judge his house forever on account of a/the iniquity that he knew’ (1 Sam 3:13)

8.  For examples of heads cliticized to ‫אשׁר‬, see Gen 1:31; 12:20; 13:1; 14:23; 20:7; 24:2, 36; 25:5; 31:1, 12, 21; 34:29; 35:2; 39:5, 6, 20, 22; 40:3; 41:56; 45:10, 11, 13; 46:1, 32; 47:1; Exod 6:29; 7:2; 9:19, 25; 10:12; 18:1, 8, 14; 19:18; 20:11; 25:22; 31:6; 34:32; 35:10; 38:22; 40:9; Lev 4:24, 33; 6:18; 7:2; 8:10; 13:46; 14:13, 32, 36; 18:29; 27:8, 28; Num 1:50; 4:16, 26; 9:17, 18; 15:23; 16:26, 30, 33; 18:13; 19:14; 22:2; Deut 3:21; 5:27, 28; 10:14; 12:11; 13:16; 18:18; 20:20; 22:24; 23:5; 29:1, 8; Josh 1:16; 2:13; 6:17, 21, 22, 23, 24, 25; 7:15, 24; 9:9, 10; 22:2; 23:3; Judg 3:1; 7:18; 9:25, 44; 11:24; 1 Sam 2:22; 3:12, 13; 14:7; 15:3; 19:18; 25:21, 22; 30:18, 19; 2 Sam 3:19, 25; 6:12; 11:22; 13:22; 14:20; 15:21; 16:21; 1 Kgs 2:3; 10:2; 11:38, 41; 14:29; 15:7, 23, 31; 16:14; 19:1; 20:4; 21:19; 22:39; 2 Kgs 8:6; 8:23; 10:34; 12:20; 13:8, 12; 14:28; 15:6, 16, 21, 26, 31; 18:12; 20:13, 15, 17; 21:17; 23:28; 24:5; Isa 39:2, 4, 6; Jer 1:7, 17; 22:12, 25; 26:8; 31:37; 32:23; 38:9; 44:23; Ezek 6:13; 14:22, 23; 16:37, 63; 21:35; 23:28; 40:4; 44:5; 47:9; 48:22; Hos 2:1; Mal 2:9; Pss 96:12; 109:11; 119:63; 146:6; Job 1:10, 11, 12; 42:10; Prov 17:8; 21:1; Ruth 3:16; 4:9; Eccl 1:13, 16; 3:14; 8:3; Esth 2:13; 3:12; 4:1, 3, 7; 5:11; 6:13; 8:9, 17; Neh 4:14; 9:6; 1 Chr 10:11; 13:14; 16:32; 2 Chr 9:1; 33:8.

212

Chapter 6

There are also a number of these “bound-relatives” with the relative words -‫שׁ‬‎ (503) 9 and -‫ה‬‎ (504). 10 (503) ‫ל־מקֹום ֶׁש ַהּנְ ָח ִלים ה ְֹל ִכים ָׁשם ֵהם ָׁש ִבים ָל ָל ֶכת‬ ְ ‫ֶא‬ ‘to (the) place that the rivers go, there they continually return’   (Eccl 1:7) (504) ‫וְ ר ַֹחב ְמקֹום ַה ֻּמּנָ ח ָח ֵמׁש ַאּמֹות ָס ִביב ָס ִביב‬ ‘and the width of (the) place that was open was five cubits all around’ (Ezek 41:11) In (501)–(504), the phonological shapes of the heads ‫‘ ְמקֹום‬place of’ and ‫ֲעֹון‬ ‘iniquity of’ (cf. ‫ ָמקֹום‬and ‫עֹון‬, ָ the respective free forms) indicate that they are cliticized to the respective relative words. All four relatives are also restrictive in nature, specifying which place (in Gen 39:20; Eccl 1:7; Ezek 41:11) and which iniquity (in 1 Sam 3:13) is being discussed. Attaching the relative head as a cliticized phrase to the modifying relative clause to mark the construction as restrictive is quite logical: the bound relationship is often described as one in which the first noun phonologically “leans” on the second noun in order to illustrate the syntactic and/or semantic “closeness” of the two; thus, when it is used with relatives, it signifies the close (i.e., restrictive) nature of the relative clause to its head. Returning to the possible role of resumption in indicating restrictiveness, note that the following two examples, with head NPs cliticized to their relatives (indicating that they are both restrictive), provide further evidence that resumption alone does not distinguish restrictiveness, since one does not contain resumption (505), and one does (506). (505) ‫סּורים‬ ִ ‫סּורי ַה ֶּמ ֶלְך ֲא‬ ֵ ‫ר־א‬ ֲ ‫ל־ּבית ַהּס ַֹהר ְמקֹום ֲא ֶׁש‬ ֵ ‫יֹוסף אֹתֹו וַ ּיִ ְּתנֵ הּו ֶא‬ ֵ ‫וַ ּיִ ַּקח ֲאד ֹנֵ י‬ ‘and he put him into the round house, (the) placei whi that the prisoners of the king were confined ___i’ (Gen 39:20 Kethiv) (506) ‫יֹוסף ָאסּור‬ ֵ ‫ל־ּבית ַהּס ַֹהר ְמקֹום ֲא ֶׁשר‬ ֵ ‫וַ ּיִ ֵּתן א ָֹתם ְּב ִמ ְׁש ַמר ֵּבית ַׂשר ַה ַט ָּב ִחים ֶא‬ ‫ָׁשם‬ ‘and he put them under the guard of the house of the captain of the guards, the round house, (the) placei whi that Joseph was confined therei’ (Gen 40:3) 9.  See Eccl 1:7; 5:15; 7:14; 11:3, 8. 10.  See Gen 21:6; 32:20; 42:29; 50:14; Exod 16:23; 19:12; 29:37; 30:13, 29; 31:14, 15; 35:2; Lev 11:24, 25, 26, 27, 31; 15:10, 19, 21, 22, 27; 20:5; 24:14; Num 4:23, 30, 35, 37, 39, 41, 43, 47; 19:13, 14; 30:3; Deut 25:18; Josh 2:23; 8:25; Judg 3:19; 19:30; 20:46, 48; 1 Sam 2:36, 5:5; 2 Sam 2:23; 18:31; 20:12; 1 Kgs 10:15; 2 Kgs 10:11, 17; 18:21; 21:24; Isa 7:22; 13:15; 36:6; 66:10; Jer 5:6; Ezek 12:19; 16:44; 34:21; 41:11; Zeph 1:8, 9; Zech 5:3; 14:16, 21; Pss 31:25; 34:23; 63:12; 145:14; Prov 6:29; Esth 9:27; Dan 12:1; Ezra 1:4; 3:8; 9:13; Neh 9:33; 10:29; 1 Chr 25:7; 2 Chr 7:11; 13:9; 29:29; 31:16; 33:25; 34:32, 33; 35:7, 25.

Relative Semantics

213

Examples (505) and (506) are a minimal pair in the sense that they both use a cliticized head NP (‫)מקֹום‬ ְ with its relative, they are both part of the same discourse context, and the first (505) does not include overt resumption of the head, whereas the second (506) does resume its head overtly. Contrary to the argument of Parunak (1996), who proposes that the difference in resumption between these two examples is related to a distinction in restrictiveness (nonrestrictive in Gen 39:20 and restrictive in 40:3), the use of the bound construction to connect both heads with the relative words strongly suggests that both are restrictive (see also above, §5.2.3.1). As a final point in the present discussion of restrictiveness in Hebrew relatives, we must note that the use of the bound relationship (i.e., NP cliticization) to mark restrictive relatives is not limited to overtly marked relative clauses; it occurs with zero-relatives as well. Consider the following pair of relatives: (507) ‫ת־הע ָֹלה ִל ְפנֵ י יְ הוָ ה‬ ָ ‫וְ ָׁש ַחט אֹתֹו ִּב ְמקֹום ֲא ֶׁשר־יִ ְׁש ַחט ֶא‬ ‘and he shall slaughter it in (the) placei whi that he slaughters the burnt offering before Yhwh ___i’ (Lev 4:24) (508) ‫ע־אל‬ ֵ ‫ְך־א ֶּלה ִמ ְׁש ְּכנֹות ַעּוָ ל וְ זֶ ה ְמקֹום לֹא־יָ ַד‬ ֵ ‫ַא‬ ‘surely these are the dwellings of unjust (people) and this is a/the placei whi ∅(that) ___i does not know God’ (Job 18:21) Example (507) presents the type of restrictive ‫אׁשר‬-relative I discussed above, with the head, ‫מקֹום‬, ְ cliticized to the relative clause. The example in (508) exhibits a similar restrictive relative with the same cliticized noun, ‫מקֹום‬, ְ with the crucial difference that the clitic host is a zero-relative. 11 Although many reference grammars note that heads may be bound to the following relative, as in examples (507)–(508) and also (509), they have not identified this as a semantically restrictive relative strategy. (509) ‫ת־הע ָֹלה ִל ְפנֵ י יְ הוָ ה‬ ָ ‫ִּב ְמקֹום ֲא ֶׁשר־יִ ְׁש ַחט ֶא‬ ‘in (the) placei whi that he slaughters the burnt offering before Yhwh ___i’ (Lev 4:24) The clause in (510) adds a final interesting linguistic feature of Hebrew relativization: a bound head and a zero-marked relative. 12

11.  On the bound form (the “construct”) and zero-relatives, see GKC 488; IBHS 155–56; GBH 562; Gibson 1994: 12. 12.  See Gen 1:1; 39:4; Exod 4:14; Lev 7:35; Deut 4:15; 32:37; 1 Kgs 17:14; Isa 29:1; 48:17; 51:1; Jer 2:8; 36:2; 48:36; Amos 3:9; Hos 1:2; Mic 5:2; Hab 3:16; Pss 4:8; 7:16; 9:16; 18:44; 25:12b; 44:2; 56:2, 10; 71:18; 81:6; 90:15; 118:22; 141:9; Job 3:3a; 6:17; 18:21; Lam 1:14, 21; 3:25, 57; 1 Chr 29:3b; 2 Chr 20:22; 24:11; 29:27; 30:19; 31:16, 19.

214

Chapter 6

(510) ‫יכם ְּבח ֵֹרב ִמּתֹוְך ָה ֵאׁש‬ ֶ ‫‏ּביֹום ִּד ֶּבר יְ הוָ ה ֲא ֵל‬ ְ ‘on (the) dayi whi ∅(that) Yhwh spoke to you __i at Horeb from the midst of the fire’ (Deut 4:15) Such examples are, in fact, redundantly marked as restrictive relatives. Perhaps the most famous of these bound-form zero-relatives is provided in (511). (511) ‫ֹלהים ֵאת ַה ָּׁש ַמיִ ם וְ ֵאת ָה ָא ֶרץ‬ ִ ‫אׁשית ָּב ָרא ֱא‬ ִ ‫ְּב ֵר‬ ‘in a/the beginningi whi (that) God created the heavens and the earth ___i’ (Gen 1:1) Gen 1:1 is most often interpreted incorrectly either as something other than a relative (‘In the beginning God created . . .’) or as a nonrestrictive relative (e.g., ‘In the beginning, when God created . . . ,’). It is neither of these; rather, it is very clearly a restrictive relative clause, which means that the head ‘beginning’ cannot be identified apart from the information in the relative clause (see Holmstedt 2008; 2014b: 143–47).

Chapter 7

“Relative” Diachrony 7.1.  Nominalization and Relativization In §3.2, I noted the similarity between English that and Hebrew ‫ אשׁר‬and ‫—שׁ‬that is, the same word in both languages is used to introduce relative clauses, (512)–(514), and complement clauses, (515)–(517). (512) He saw the dogi whi that he wanted to buy ___i. (513) ‫ת־ה ֶּׁש ֶמׁש‬ ַ ‫ת־ה ַּמ ֲע ֶׂשה ֲא ֶׁשר נַ ֲע ָׂשה ַת ַח‬ ַ ‫‏ ֶא‬ ‘the deedi whi that ___i was done under the sun’ (Eccl 8:17) (514) ‫ל־ה ַּמ ֲע ִׂשים ֶׁשּנַ ֲעׂשּו ַּת ַחת ַה ָּׁש ֶמׁש‬ ַ ‫ת־ּכ‬ ָ ‫‏א‬ ֶ ‘all the deedsi whi that ___i were done under the sun’ (Eccl 1:14) (515) He saw that the dog had chewed up his slippers. (516) ‫ֹלהים‬ ִ ‫ם־יֹוד ַע אָנִ י ֲא ֶשׁר יִ ְהיֶ ה־טֹּוב ְליִ ְר ֵאי ָה ֱא‬ ֵ ַ‫ִכּי גּ‬ ‘because I know that it will go well for those who fear God’   (Eccl 8:12) (517) ‫ת־כּ ָלּם‬ ֻ ‫וְ יָ ַד ְע ִתּי גַ ם־אָנִ י ֶשׁ ִמּ ְק ֶרה ֶא ָחד יִ ְק ֶרה ֶא‬ ‘and I know also that one fate befalls all of them’ (Eccl 2:14) Based on these multiple functions, as well as the cases of ‫ אשׁר‬and ‫ שׁ‬used to fulfill prepositional valency (§4.3.2), I advocated classifying the Hebrew function words as “nominalizers,” used to recategorize a clause into, for syntactic purposes, an NP-substitute. The use of ‫ אשׁר‬and ‫ שׁ‬in multiple capacities is itself not an issue. The question it raises is a diachronic one: did both functions of ‫( אשׁר‬and also ‫ )שׁ‬exist side-by-side during the entirety of the Hebrew linguistic history covered by the Hebrew Bible, or was there a diachronic change? Indeed, the close connection between the function words used for relative clauses and those used for complement clauses is often noted in the typological literature (Noonan 2007; Hendery 2012: 108–15), and the grammaticalization of relative > complementizer is cited in Heine and Kuteva’s lexicon of grammaticalization (2002: 254).  1 1.  Heine and Kuteva include Hebrew ‫( שׁ‬from Song 1:6) as an example of the relative > complementizer path; they cite a 1998 article by Sonia Cristofaro, who in turn cites Givón

215

216

Chapter 7

Similarly, Hendery notes a number of sources as well as extensions of relative words, including pronouns, classifiers and generic nouns, discourse markers, possessives, comparatives, adverbial clauses, general subordination, adjective markers, and general “linkers” (Hendery 2012: 43–132). 2 More specifically, the typological linguist Talmy Givón has argued that ‫ אשׁר‬underwent grammaticalization from an initial relative word to a function word introducing verbal complement clauses at a later stage of the language (Givón 1974; 1991). This chapter is the syntactic companion to the etymological discussion of ‫ אשׁר‬and ‫ שׁ‬that I provided in §3.3. Whereas there I considered the arguments for relating or distinguishing the two relative words, especially within historical-comparative Semitic and general grammaticalization perspectives, here I investigate the diachrony of ‫ אשׁר‬and ‫ שׁ‬clauses in Hebrew. I argue that, though Givón is mostly wrong in his analysis, there is, indeed, slight evidence of some diachronic change. 7.2.  The Story of ‫ אשׁר‬Clauses Rather than linguistic change in grammar or the function of, say, ‫אשׁר‬ clauses, the abundant literature treating the history of ‫ אשׁר‬and ‫ שׁ‬has focused on the etymological question (see above, §3.3). Even those newer studies that invoke grammaticalization are primarily concerned with the history of the words before Hebrew; and within Hebrew, the focus is on whether ‫ אשׁר‬was abbreviated as ‫( שׁ‬which, as I have argued, should be understood as a folk etymology). I first took up the question of diachronic development in the function of ‫ אשׁר‬in Holmstedt 2006b and ‫ שׁ‬in Holmstedt 2013d and will summarize the relevant data below, before partially correcting my conclusions in 2006b and elaborating on my 2013d conclusions. (Though the focus of this chapter is diachronic, the argument requires covering some of the same examples and even repeating many of the same descriptive points from the preceding chapters; for this redundancy, I ask the reader’s patience.) 7.2.1.  Pre-Hellenistic Period Nonbiblical Data There are 32 interpretable ‫אשׁר‬-relatives from the published corpus of Epigraphic Hebrew, mostly from the Arad and Lachish ostraca. The examples in 1991. The common citation of Givón’s works as representative of Hebrew grammar has led to many inaccurate or questionable representations of Hebrew in the typological literature. To wit, I disagree with this analysis of Song 1:6, since it can be much more simply understood as a relative clause modifying the 1cs clitic pronoun attached to the preceding verb; see above, example (220b). 2. As with Heine and Kuteva (see previous note), Hendery often cites Givón 1974 and 1991 concerning Biblical Hebrew ‫ אשׁר‬and ‫ שׁ‬data. As with my criticism of Heine and Kuteva, I consider Hendery’s use to be inherently flawed by Givón’s problematic arguments. See below, §7.4 and nn. 22–23.

“Relative” Diachrony

217

(518)–(520) are representative of the variation in the corpus (see Gogel 1998: 168–72). 3 (518) ‫ולדבר אשר צותני‬ ‘and regarding the matteri whi that you commanded ___i me’   (Arad 18.6–8) (519) ‫ככל האתת אשר נתנ אדני‬ ‘according to all the signsi whi that my lord gave ___i’   (Lach 4.11–12) (520) ‫זאת [קברת שבנ]יהו אשר על הבית‬ ‘this is the [tomb of Sheban]iahi, whi that ___i (is) over the palace’   (Silw 2.1) (521) ‫ועת הטה [ע]בדכ [ל]בה אל אשר אמ[רת‬ ‘and now, your servant has inclined his heart to ∅(the thing)i whi that you said ___i’ (Arad 40.4–5) (522) ‫דבר אשר לא ידעתה‬ ‘a thingi whi that you do not already know iti’ (Lach 2.5–6; see also Lach 3.5–6, 4.3–4; and Mous 2.4–6) As the representative data show, Epigraphic Hebrew relatives show no differences from the biblical data adduced in previous chapters. The relatives are postnominal and externally-headed. The heads may be overt, (518)–(520) and (522), or null, (521). The promotion of the relative head from within the relative has left a gap, (518)–(521), though resumption might be possible (522) (see below on this example). The heads in (518) and (522) are indefinite, whereas the heads in (519)–(521) are definite. The relative word in all five clauses appears without discernible agreement features: the relative heads in (518) and (520)–(522) are masculine singular, and the relative head in (519) is feminine plural, but the relative element ‫ אשׁר‬in all three remains the same. Thus, the epigraphic ‫ אשׁר‬is like the biblical ‫ אשׁר‬in being a relative complementizer 3.  Excluding questionable readings, reconstructions, and unprovenanced texts, there are 22 epigraphic occurrences of ‫ אשׁר‬and none of ‫( שׁ‬see Dobbs-Allsopp et al. 2005 for the texts): Arad 8.9; 18.6–8; 29.7; 40.4–5; 71.2; Kuntillet Ajrud 16.1; Lachish 2.5–6; 3.4–6, 10–12; 4.2–3, 3–4, 11–12, 9.4–9; 18.1; Yavneh Yam/Mesad Hashavyahu 1.6–8, 8–9; Papyrus Murabbaʿat 17a 1.2; Nahal Yishai 1.1; Samaria Basalt 1.1; Silwan 2.1, 2–3; 3.2. Additionally, there are 8 examples of ‫ אשׁר‬in unprovenanced texts, mostly seals, in private collections (again, see Dobbs-Allsopp et al. 2005 for the full texts): Moussaieff Ostraca 1.1, 2.4–6; seals ##1, 20, 403, 404, 405, and 406 in Avigad and Sass 1997. Finally, to this latter group, we can add ostracon 2 (lines 6 and 9) in the unprovenanced texts published in Lemaire and Yardeni 2006.

218

Chapter 7

(not “pronoun”). 4 Finally, the relative in (520), with the unique referent “[Sheban]iah,” indicates that Hebrew also used nonrestrictive relatives in addition to restrictive relatives. 5 Relatives with relative heads in the bound form, as in (523), are also semantically restrictive. 6 (523) ‫ועת ככל אשר שלח אדני כנ עשה עבדכ‬ ‘and now, according to alli whi that my lord sent ___i, so your servant has done (Lach 4.2–3) The restrictive nature of cliticized relative heads is due to the semantic dependency created by the cliticization of the noun—viz., it receives its specificity from its clitic host (see above, §6.3). The possible case of relative resumption in (522) requires further comment. It is possible to take the relative in (522) as containing resumption by means of the clitic pronominal complement attached to the verb ‫ידעתה‬. Elsewhere in Epigraphic Hebrew the 2ms perfect verb is written without a final ‫ה‬, suggesting that the few cases with a ‫ ה‬represent the cliticized 3ms pronoun attached to the verb as its complement (see, among many others, Cross 1985: 43–46; Rollston 2006: 62 n. 42). 7 If the four examples with the final ‫ ה‬within a relative do represent resumption, they are then the only Iron Age cases of relative resumption in Hebrew outside the Bible. Another option is to take the forms with the final ‫ ה‬as alternate writings for the 2ms perfect verb without a clitic object pronoun (that is, the ‫ ה‬is a final mater lectionis for the final /a/ of the 2ms verb; see, among many others, Gogel 1998: 83–87; Schniedewind 2000: 160). If we invoke the use of resumption in Biblical Hebrew as circumstantial evidence (§5.2.3.3), a resumptive pronoun at the complement position in (522) is not likely. First, most resumption in 4. Even Gogel, in her otherwise excellent grammar of Epigraphic Hebrew, uses imprecise language to describe the Hebrew relative element: “The relative pronoun in epigraphic Hebrew is ʾšr. This particle of relation brings the clause introduced by it into relation with an antecedent phrase or clause” (Gogel 1998: 168; italics added). 5.  Gogel 1998 does not include any discussion of restrictive versus nonrestrictive relatives in Epigraphic Hebrew. For Biblical Hebrew, see above, §6.3. 6.  There are two possible interpretations of the quantifier ‫ כל‬in (523): either (1) it modifies a covert head, e.g., ‘every(thing)’ (see Gogel 1998: 168–69), or (2) it serves as a substantive and is itself the relative head. Given that the quantifier ‫ כל‬can serve as a substantive in the Hebrew Bible, as in ‫‘ יוצר הכל הוא‬he is the creator of everything’ (Jer 10:16), and also given that relative heads may be in the bound form before a relative clause, as in ‫כל ימי אשׁר ישׁכן‬ ‫‘ הענן על המשׁכן‬all the days that the cloud dwelled over the tabernacle’ (Num 9:18), it seems simpler to take ‫ כל‬as the bound-form relative head in (523). 7.  For the 2ms perfect verb without the final ‫ה‬, see Arad 2.5–6, 7–8; 3.5, 8; 17.3–4, 40.5; Yavneh Yam 1.14. For the four examples of 2ms perfect verbs with the final ‫ ה‬within a relative, see Lachish 2.5–6; 3.5–6, 4.3–4; and Moussaieff 2.4–6 (compare with examples not within a relative: Lach 3.8, 5.4; Arad 7.5–6, 40.9).

“Relative” Diachrony

219

the Hebrew Bible is in an NP-internal 8 or prepositional complement position within the relative. Second, resumption in the verbal complement position occurs less frequently, and its use is highly constrained: (1) to disambiguate verbal semantics, or (2) to signal that the complement carries Focus pragmatics within the relative (see above, §5.2.3). Concerning the relative in (522), it is not clear how either function explains the ostensible resumption. Furthermore, taking the object within the relative as pragmatically marked for Focus (i.e., ‘the thing that you didn’t already know it [versus something else]’) makes little contextual sense. Thus, the ‫ ה‬on the 2ms forms is better understood as a mater lectionis. 7.2.2.  Biblical Data Through chaps. 3–6, I have used or cited hundreds of examples of ‫ אשׁר‬used to introduce relative clauses in the Hebrew Bible, as in (524) and (525). (524) ‫אָבן‬ ֶ ‫ר־דּ ֶבּר ִאתֹּו ַמ ֶצּ ֶבת‬ ִ ‫וַ יַּ ֵצּב יַ ֲעקֹב ַמ ֵצּ ָבה ַבּ ָמּקֹום ֲא ֶשׁ‬ ‘and Jacob set up a pillar in the placei whi that he [God] spoke with him ___i, a pillar of stone’ (Gen 35:14) (525) ‫ית־אל׃‬ ֵ ‫ֹלהים ֵבּ‬ ִ ‫ת־שׁם ַה ָמּקֹום ֲא ֶשׁר ִדּ ֶבּר ִאתֹּו ָשׁם ֱא‬ ֵ ‫וַ יִּ ְק ָרא יַ ֲעקֹב ֶא‬ ‘and Jacob named the placei whi that God spoke with him therei Bethel’ (Gen 35:15) To summarize the analysis I provided in the preceding chapters, these two examples demonstrate that the biblical use of ‫ אשׁר‬is to form postnominal, externally-headed relative clauses, most of which are restrictive (as in both examples above). The majority also leave a gap (or null resumption) at the place from which the head was promoted, as in (524), though some use overt resumption, as in (525). Many ‫ אשׁר‬clauses have overt heads, as in both (524) and (525), though there are numerous examples of null-head ‫אשׁר‬-relatives. There is no doubt, and no disagreement, that this is the overwhelmingly dominant function within the almost 5,500 biblical occurrences of the function word ‫אשׁר‬. With that said, there is also a minority use of ‫ אשׁר‬that I first provided in example (12) in §1.1.1 (see also §3.2, n. 7, and appendix A): ‫ אשׁר‬may introduce a verbal complement clause, as in (526). 8.  For NWS languages, the genitive position is most often realized as an enclitic possessive pronoun attached to a host NP. This is also known as relativization from (or resumption in) the “NP-internal” position. This kind of genitive/NP-internal resumption is obligatory in most languages and is thus often excluded from general discussions of relative resumption; even English, a language that does not generally use resumptive pronouns, requires this type of resumption (e.g., I saw the mani whoi had run over hisi dog yesterday). For further discussion of NP-internal resumption, see above, §5.1.2.1.

220

Chapter 7

(526) ‫ֹלהיָך‬ ֶ ‫ית ֲא ֶשׁר נְ ָשׂ ֲאָך יְ הוָ ה ֱא‬ ָ ‫וּב ִמּ ְד ָבּר ֲא ֶשׁר ָר ִא‬ ַ ‘and in the wilderness where you saw that Yhwh, your god, carried you’ (Deut 1:31) Example (526) presents a complement clause introduced by the function word ‫אׁשר‬. The salient syntax of this construction is: a clause, headed by a function word, filling the syntactic role of the complement of a transitive verb—in this case, the verb ‫‘ ראה‬to see’. Thus, ‫אשׁר‬-relative clauses and ‫ אשׁר‬complement clauses differ in two significant ways. First, in the case of ‫ אשׁר‬complement clauses, if the ‫ אשׁר‬clause were not the verbal complement, the valency of the verb would go unfilled (resulting in an ungrammatical clause). Second, since there is no relative head that has been promoted from within the relative, in ‫ אשׁר‬complement clauses there is no discernible gap related to the relativization process. The complement use of ‫ אשׁר‬is perhaps set in greater relief by comparison with the similar syntax of ‫ כי‬complement clauses, as that in (527). (527) ‫י־היָ ה יְ הוָ ה ִע ָמְּך‬ ָ ‫אמרוּ ָראֹו ָר ִאינוּ ִכּ‬ ְ ֹ ‫וַ יּ‬ ‘and they said: We saw clearly that Yhwh was with you’   (Gen 26:28) In addition to the clear and uncontroversial relative and complement clause functions of ‫אשׁר‬, descriptions, interpretations, and translations of ‫ אשׁר‬in a few dozen verses in the Hebrew Bible reflect the position that ‫ אשׁר‬was at some point assigned a broader function as a general subordinator, introducing causal, result, purpose, and conditional clauses. In (528)–(531,) I provide typical examples used to support these functions. (528) Causal? ‫ן־ה ִא ָשּׁה ַהזֹּאת ָליְ ָלה ֲא ֶשׁר ָשׁ ְכ ָבה ָע ָליו׃‬ ָ ‫וַ יָּ ָמת ֶבּ‬ ‘then this woman’s son died in the night, because she lay on him’    (1 Kgs 3:19 nrsv) (529) Purpose? ‫ל־היָּ ִמים‬ ַ ‫ת־דּ ָב ָרי ֲא ֶשׁר יִ ְל ְמדוּן ְליִ ְראָה א ִֹתי ָכּ‬ ְ ‫אַשׁ ִמ ֵעם ֶא‬ ְ ְ‫ת־ה ָעם ו‬ ָ ‫ל־לי ֶא‬ ִ ‫ַה ְק ֶה‬ ‫ל־ה ֲא ָד ָמה‬ ָ ‫ֲא ֶשׁר ֵהם ַחיִּ ים ַע‬ ‘assemble the people for me, and I will let them hear my words, so that they may learn to fear me as long as they live on the earth’   (Deut 4:10 nrsv) (530) Result? ‫אָרץ גַּ ם־‬ ֶ ‫ת־ע ַפר ָה‬ ֲ ‫ם־יוּכל ִאישׁ ִל ְמנֹות ֶא‬ ַ ‫אָרץ ֲא ֶשׁר ִא‬ ֶ ‫וְ ַשׂ ְמ ִתּי ֶאת־זַ ְר ֲעָך ַכּ ֲע ַפר ָה‬ ‫יִמּנֶ ה׃‬ ָ ‫זַ ְר ֲעָך‬ ‘I shall make your offspring like the dust of the earth; so that if one can count the dust of the earth, your offspring also can be counted’   (Gen 13:16 nrsv)

“Relative” Diachrony

221

(531) Conditional? ‫א־ת ָע ֶשׂינָ ה‬ ֵ ֹ ‫ֹלהיו ֲא ֶשׁר ל‬ ָ ‫ל־מ ְצֹות יְ הוָ ה ֱא‬ ִ ‫אַחת ִמ ָכּ‬ ַ ‫ֲא ֶשׁר נָ ִשׂיא יֶ ֱח ָטא וְ ָע ָשׂה‬ ‫אָשׁם׃‬ ֵ ְ‫ִבּ ְשׁגָ גָ ה ו‬ ‘if a ruler sins and inadvertently does one of any of the commands of Yhwh, his god, which should not be done, and he incurs guilt . . .’ (Lev 4:22; see DCH 433; HALOT 99) I will return to the issue of ‫’אשׁר‬s manifold functions below, in §7.4. 7.2.3.  Hellenistic and Roman Period Nonbiblical Data The next major corpora of texts from which to draw ‫ אשׁר‬examples are the nonbiblical texts from Qumran and the Judean Desert (second century b .c.e. to first century c.e.), including the book of Ben Sira (second century b.c.e.), and the Mishnah (third century c.e.). In all the texts, the primary use of ‫ אשׁר‬is, unsurprisingly, as a relative word: (532) ‫‏כיא לוא בקשו ולוא דרשהו בחוקיהו לדעת הנסתרות אשר תעו‏ בם‬ ‫לאששמה‬ ‘because “they have not sought Him nor inquired of His statutes” [Zeph 1:6] in order to know the hidden thingsi whi that they err in themi to their shame’ (1QS 5:11–12) (533) ‫ואשר יקים בברית על נפשו להבדל מכול אנשי העול ההולכים‏ בדרך הרשעה‬ ‘and ∅(the man)i whi that ___i enters the covenant upon his life is to separate himself from all of the men of perversity who walk in the way of wickedness’ (1QS 5:10–11) (534) ‫‏ ומיא כעמכה ישראל אשר בחרתה לכה מכול עמי הארצות‬ ‘and who is like your people, Israeli, whi that you chose ___i for yourself from all the peoples of the lands?’ (1QM 10:9) (535) ‫אל תמאס בשמיעת שבים אשר שמעו מאבתם‬ ‘do not reject the report of the agedi whi that they heard ___i from their fathers’ (Sir 8:9A) (536) ‫‏והוא יבין לבך ואשר איותה יחכמך׃‬ ‘and he will make your heart understanding and he will make you wise ∅(the way)i whi that you desired ___i’ (Sir 6:37A) (537) ‫‏ייי הקשה את לב פרעה אשר לא לא ידעו׃ שמעשיו ~ ‏מגולין תחת השמים‬ ‘Yyy hardened the heart of Pharaohi, whi that ___i did know (the one) who his works that had been revealed under the heavens’   (Sir 6:18–19A) As with the epigraphic evidence, the ‫ אשׁר‬clauses from Judean Desert texts and Ben Sira manifest the same patterns as the Hebrew Bible: the ‫אשׁר‬-relatives are

222

Chapter 7

postnominal, externally-headed; show both overt (532) and null resumption (533)–(537) and appear with both overt ([532], [534], [535], and [537]) and null heads ([533] and [536]); and occur both restrictively ([532], [533], [535], and [536]) and nonrestrictively ([534] and [537]). Furthermore, both sets of texts exhibit cases of ‫ אשׁר‬that nominalize the following clause as the complement of prepositions, as in (538) and (539): 9 (538) ‫֯לעשות הטוב והישר לפניו כאשר צוה ביד מושה וביד כול עבדיו הנביאים‬ ‘to do the good thing and upright thing before him, like-that he commanded by the hand of Moses and by the hand of all his servants the prophets’ (1QS 1:2–3) (539) ‫‏כבד אל והדר כהן ות[ן ח]לקם כאשר צוותה׃‬ ‘glorify God and honor a priest and give their portion (to them) like-that you were commanded’ (Sir 7:31A) After the Judean Desert texts and Ben Sira, the next significant corpus of Hebrew texts is the Mishnah, the language of which ostensibly dates to the third century c.e. At this point, we find a diachronically significant change in the use of ‫אשׁר‬. According to Pérez Fernández (1999: 50), ‫ אשׁר‬is “reserved only for biblical quotations and liturgical texts.” Indeed, the majority of the 69 occurrences (in 54 verses 10) in the Mishnah appear in a biblical quotation, and the remainder are demonstrably liturgical, with “elevated and semi-Biblical” style (Segal 1927: 42). The examples in (540) and (541) illustrate each type. (540) Biblical Quotation ‫וכל ההרים אשׁר במעדר יעדרון‬ ‘and any hillsi whi that ___i are hoed with a hoe’ (m. Peʾah 2:2; quoting Isa 7:25) (541) Liturgical ‫ר׳ טרפון או׳ אשׁר גאלנו וגאל את אבותינו ממצ׳‬ ‘Rabbi Tarfon says: . . . whi that ___i redeemed us and redeemed our ancestors from Egypt’ (m. Pesaḥ. 10:6) 9.  Similarly, the 20 examples of ‫( פשׁרו אשׁר‬both certain and reconstructed), mostly from Pesher Habakkuk, represent the complementizer use of ‫אשׁר‬, though in these cases the ‫ אשׁר‬is nominalizing a clause to serve as the complement of a null copula—i.e., ‘its interpretation is that . . .’. See 1QpHab 1:4; 4:1, 14; 5:7; 6:3, 6; 7:7, 15; 4Q161 frgs. 8–10:22; 4Q163 frg. 28:2; 4Q166 2:12, 15; 4Q167 frgs.11–13:4; 16:1; 4Q171 frgs. 1+3–4 iii 3; 4Q173 frg. 1:3; frg. 3:2; 4Q177 frgs. 5–6:8; 4Q252 4:5. 10.  M. Peʾah 2:2; m. Maʿaś. Š. 5:11, 12, 13; m. Ḥal. 4:10; m. Bik. 1:3, 4, 5; m. Pesaḥ. 10:6; m. Yoma 5:5; m. Roš Haš. 2:9; 3:8; m. Taʿan. 2:3; 3:3; m. Ḥag. 1:5; m. Yebam. 12:6; m. Ned. 3:11; m. Soṭah 2:2; 4:1; 5:2; 7:4, 5, 8; 8:1, 2, 4; 9:5, 6; m. Qidd. 4:14; m. B. Qam. 9:12; m. Sanh. 10:1, 5; 11:2; m. Mak. 1:3, 6; 2:2; 3:15; m. ʾAbot 3:3, 6, 8; 5:18; m. Zebaḥ. 10:1; m. Menaḥ. 5:2, 6; m. Bek. 1:7; m. ʿArak. 7:5; 9:2, 3; m. Ker. 4:3; m. Neg. 12:5, 6; 14:10; m. Yad. 4:8.

“Relative” Diachrony

223

Given the constraints that Pérez Fernandez describes, it is not surprising, though diachronically fascinating, that the Mishnah evinces neither the complementizer nor any other nonrelative function of ‫אשׁר‬. As in the Hebrew Bible, the ‫ אשׁר‬examples in the Judean Desert texts and Ben Sira (though not the Mishnah) are predominantly clear examples of relative or complement clauses. That said, both scholarly commentaries and popular translations of these texts reflect nonrelative, noncomplement analyses of numerous other ‫ אשׁר‬clauses. For example, Fassberg 1997 and Kaddari 2005 both provide discussions of ‫ אשׁר‬clauses in the book of Ben Sira. 11 Both Fassberg and Kaddari assign to ‫ אשׁר‬relative, conditional, temporal, purpose, result, and causal subordinating functions, illustrated in (542)–(544). (542) Conditional ‫מה יתחבר פרור אל סיר אשר הוא נוקש בו הוא נשבר‬ ‘How can the pot go with the vessel? If they knock together, the pot will be smashed’ (Sir 13:2A; Fassberg 1997: 60) (543) Temporal a. ‫כאשר יבוא עמך לא יתגלה לך‬ ‘when your people comes, he will not reveal himself to you’   (Sir 12:15A; Kaddari 2005: 260) b. ‫בקנאו לאלוהי כל ויעמד בפרץ עמו אשר נדבו לבו ויכפר על בני ישראל‬ ‘in his zeal for the God of everything, and he stood when his people burst out, when his heart prompted him and he atoned for the children of Israel’ (Sir 45:23B; cf. Kaddari 2005: 258) (544) Purpose/Result/Causal a. ‫ועם כל אלה העתר אל אל אשר יכין באמת צעדך‬ ‘and with all of these, pray to God in order that (or: because) he shall direct your step in truth’   (Sir 37:15D; cf. Kaddari 2005: 258) 11.  Kaddari explicitly sets out to provide a “full and detailed” description of the relative clauses in Ben Sira, and he lists 64 occurrences of ‫אשׁר‬, although 2 are listed twice and 2 are listed by mistake (they include ‫ שׁ‬not ‫)אשׁר‬. This leaves 60 legitimate ‫ אשׁר‬clauses in Kaddari’s study, to which we must add 7 that he has overlooked: Sir 10:9A; 13:2A, 7A; 16:15A; 36:31D[2×]; 38:13margin. Taking all of the Ben Sira Hebrew manuscripts together, as in the edition prepared by the Academy of the Hebrew Language (1973), there are 67 occurrences of ‫ אשׁר‬in 48 verses: Sir 3:22C; 6:37A; 7:31A; 8:9A, 14A;10:9A; 12:15A; 13:2A, 7A; 15:11A[2×]B, 16AB, 17A; 16:7AB, 15A; 18:32C; 30:19B, 20B; 33:4B + margin, 5B; 34:15B[margin], 16B; 36:31B[2×]C[2×]D[2×]; 37:12B[2×]D, 15BD; 38:13B+ 1× margin, 14B, 15B, 27B; 40:11B; 44:9B[3×]M, 20B; 45:23B, 24B; 46:1B, 11B; 47:13B, 23B[2×]; 48:1B, 4B, 11B, 15B; 49:10B; 50:1B, 2B, 3B, 24B, 27B[2×]; 51:8B. This includes three conjectured reconstructions of the text: Sir 30:19B; 36:31C; 37:12D (see Academy of the Hebrew Language 1973: 99–100).

224

Chapter 7 b. ‫לכן גם לו הקים חק ברית שלום לכלכל מקדש אשר תהיה לו ולזרעו‬ ‫כהונה גדולה עד עולם‬ ‘therefore also for him he established a principle, a covenant of peace, in order to sustain the sanctuary, so that the great priesthood shall be for him and his descendants forever’   (Sir 45:24B; cf. Kaddari 2005: 258)

Similarly, taking established translations of the Qumran sectarian texts as representative of common grammatical analysis, we see that Qumran Hebrew also is thought to parallel the Hebrew Bible in the multivalent use of ‫אשׁר‬. Consider examples (545)–(549): (545) Causal ‫‏בעבור רחמיכה ולוא כמעשינו אשר הרעונו ועלילות פשעינו‬ ‘because of Your mercy; not according to our works, for we have acted wickedly, nor for the acts of our rebelliousness’    (1QM 11:4; WAC 157) (546) Conditional ‫לוא תואכל בשר שור ^ושה^ ועז בתוך עירי אשר אנוכי מקדש לשום שמי‬ ‫בתוכה אשר לוא יבוא לתוך מקדשי וזבחו שמה‬ ‘You must not eat the meat of an ox, sheep, or goat within My city (which I shall sanctify by establishing My name in its midst) if it has not been brought to My temple’ (11Q19 52:19–20; WAC 621) (547) Purpose/Result ‫מיא כמוכה אל ישראל בשמים ובארץ אשר יעשה כמעשיכה הגדולים‬ ‘who is like You, O God of Israel, in heaven and on earth, that he can do according to Your great works?’ (1QM 10:8; WAC 157) (548) Exceptive ‫וגם אל יביא איש על רעהו דבר לפני הרבים אשר לוא בתוכחת לפני עדים‬ ‘Also, no man is to bring a charge against his fellow before the general membership unless he has previously rebuked that man before witnesses’ (1QS 6:1; WAC 124) (549) Disjunctive ‫‏וכול איש אשר יש אתו דבר לדבר לרבים אשר לוא במעמד האיש השואל‬ ‫את עצת היחד ועמד האיש על רגליהו‬ ‘If any man has something to say to the general membership, yet is of a lower rank than whomever is guiding the deliberations of the party of the Yahad, let him stand up.’ (1QS 6:12–13; WAC 125) In each case above, the ‫ אשׁר‬has been interpreted and translated as something other than a relative or complement function word. The cases in (545)–(547) mirror the minority functions often assigned to biblical ‫אשׁר‬, whereas the two

“Relative” Diachrony

225

in (548)–(549) are additional, nonbiblical functions that at least Wise, Abegg, and Cook (2005) consider appropriate for ‫אשׁר‬. The apparent story that emerges from the data as I have presented them is a fairly straightforward case of grammaticalization. At some point before Hebrew is attested—that is, before the epigraphic evidence—the word ‫ אשׁר‬experienced reanalysis by which it moved from a nominal content lexical word to a function word introducing relative and complement clauses (on this first stage, see above, §3.3). 12 During the second half of the first millennium, as evidenced in the biblical and extrabiblical Hellenistic period evidence, ‫ אשׁר‬experienced further reanalysis such that it became a general subordinator, used for relative clauses as well as a variety of adverbial clauses. But this development was halted or interrupted by the time that Mishnaic Hebrew appears. In the Mishnah, all the examples of ‫ אשׁר‬that are not direct quotes from the Hebrew Bible reflect a straightforward relative function. The contributors to the Mishnah clearly saw ‫ אשׁר‬as a relative word and nothing more. This fact poses a problem for the grammaticalization approach to Hebrew ‫אשׁר‬, since one of the pillars of grammaticalization theory is the unidirectional nature of the process (see Hopper and Traugott 2003: 99–139; also Heine 2003; Haspelmath 2004.). But the story I sketched above means that ‫ אשׁר‬became more grammaticalized during the biblical period and then by the later Mishnaic period reversed the grammaticalization trend. This would be highly atypical. Before sorting this out, however, we must consider ‫ שׁ‬clauses. 7.3.  The Story of ‫ שׁ‬Clauses Whereas there are about 5,500 ‫ אשׁר‬clauses in the Hebrew Bible, only 139 occurrences of ‫ שׁ‬exist. Of these, 68 are in Ecclesiastes, and 32 are in Song of Songs. 13 Twenty-one are in various psalms from Psalm 122 onward, and the remaining 18 are scattered in the Hebrew Bible, literally from beginning to end. 14 The distribution of ‫ אשׁר‬and ‫ שׁ‬in nonbiblical texts is somewhat similar. In epigraphic texts from the first millennium, there are 32 interpretable occurrences of ‫ אשׁר‬and none of ‫שׁ‬. 15 The Hebrew text of Ben Sira contains 29 cases

12.  For a lone dissenting voice, see Schwarzschild 1990, in which the author argues that ‫“ אשׁר‬never ceased to be categorized as a noun in the syntactic component of the grammar of BH” (1990: 7–8). For my critique, see Holmstedt 2002: 8–17. 13.  Eccl 1:3, 7, 9 (4×), 10, 11 (2×), 14, 17; 2:7, 9, 11 (2×), 12, 13, 14, 15, 16, 17, 18 (3×), 19 (2×), 20, 21 (2×), 22, 24, 26; 3:13, 14, 15, 18, 22; 4:2, 10; 5:4, 14 (2×), 15 (2×), 17; 6:3, 10 (2×); 7:10, 14, 24; 8:7, 14 (2×), 17; 9:5, 12 (2×); 10:3, 5, 14, 16, 17; 11:3, 8; 12:3, 7, 9; Song 1:6 (3×), 7 (2×), 12; 2:7, 17; 3:1, 2, 3, 4 (4×), 5, 7, 11; 4:1, 2 (2×), 6; 5:2, 8, 9; 6:5 (2×), 6 (2×); 8:4, 8, 12. For the full context and translation of the biblical ‫ שׁ‬clauses, see appendix A.. 14.  Pss 122:3, 4; 123:2; 124:1, 2, 6; 129:6, 7; 133:2, 3; 135:2, 8, 10; 136:23; 137:8 (2×), 9; 144:15 (2×); 146:3, 5; Gen 6:3; Judg 5:7 (2×); 6:17; 7:12; 8:26; 2 Kgs 6:11; Jonah 1:7, 12; 4:10; Job 19:29; Lam 2:15, 16; 4:9; 5:18; Ezra 8:20; 1 Chr 5:20; 27:27. 15.  See above, n. 3.

226

Chapter 7

of ‫  שׁ‬16 (and also 67 of ‫)אשׁר‬. 17 In the Qumran nonbiblical texts, ‫( שׁ‬including ‫ )שׁל‬occurs 145 times, but 124 of these are in just two texts: 57 in the Copper Scroll (3Q15) 18 and 67 in 4QMMTB,C (4Q394–99); 19 the remaining 21 examples are so spread out that no one text uses ‫ שׁ‬more than twice. 20 The Bar Kokhba period texts from Nahal Hever, and Wadi Murabbaʿat contain 118 occurrences of ‫ שׁ‬and none of ‫אשׁר‬. 21 Finally, ‫ שׁ‬dominates in the Mishnah, where ‫ אשׁר‬is used only 69 times (see above, n. 10). As the ‫ שׁ‬data that I have cited throughout chaps. 3–6 indicate, grammatically ‫ שׁ‬functions like ‫—אשׁר‬the majority of uses are relative (550)–(551), with 16.  Sir 3:22A; 14:16A, 18A; 15:17B; 16:3A,B, 15A; 25:8C; 26:17C; 30:12B[2×], 19B, 34E, 36E; 34:10B, 15B, 16B[2×], 20B, 27B; 37:3B; 44:9M; 51:30B. This includes one conjectured reconstruction (30:19A) and two occurrences of ‫שׁל‬‎(13:5A, 30:28E). Also, the following cases exhibit alternation between the manuscripts: 3:22 (A = ‫ ;במה שׁהורשׁית‬B = ‫באשׁר שׁהורשׁיתה‬‎); 15:17 (A = ‫ ;אשׁר יהפץ‬B = ‫וכל שׁיחפץ‬‎); 34:15 (B = ‫ ;ובכל שׁשׂנאת‬margin = ‫וכל אשׁר שׂנאת‬‎); 44:9 (B = ‫ ;וישׁ מהם אשׁר אין לו זכר‬M = ‫)וישׁ מהם שׁאין לו זכר‬. And finally, the following is the one case of the two relative words used in the same verse: 16:15A: ‫ייי הקשׁה את לב פרעה אשׁר‬ ‫לא ידעו שׁמעשׂיו מגולין תחת השׁמים‬. 17.  Sir 3:22C; 6:37A; 7:31A; 8:9A, 14A;10:9A; 12:15A; 13:2A, 7A; 15:11A[2×]B, 16AB, 17A; 16:7AB, 15A; 18:32C; 30:19B, 20B; 33:4B + margin, 5B; 34:15B[margin], 16B; 36:31B[2×]C[2×]D[2×]; 37:12B[2×]D, 15BD; 38:13B + margin, 14B, 15B, 27B; 40:11B; 44:9B[3×]M, 20B; 45:23B, 24B; 46:1B, 11B; 47:13B, 23B[2×]; 48:1B, 4B, 11B, 15B; 49:10B; 50:1B, 2B, 3B, 24B, 27B[2×]; 51:8B. This includes four conjectured reconstructions of the text: 30:19B; 36:31C; 37:12D; see The Book of Ben Sira 1973: 99–100. See also Beentjes 2003. 18.  There are 32 occurrences of ‫שׁ‬‎: 3Q15 1:1, 6; 2:1, 5, 7, 9, 10, 13; 3:8, 11; 4:1, 6, 9, 11; 5:5, 12; 6:14; 7:8; 8:1, 2, 14; 9:1, 14, 16, 17; 10:3, 5, 9; 11:7, 8, 14; 12:10. There are also 25 cases of ‫שׁל‬‎: 3Q15 1:9, 10, 13; 2:11; 3:2; 4:13; 5:6, 8; 6:1, 7, 8; 7:3, 8, 10, 14; 8:8, 10, 14; 9:14; 10:8, 15; 12: 4, 6, 7, 8. 19.  In the B manuscript, there are 43 occurrences of ‫שׁ‬‎: 4Q394 frgs. 3–7 i 4, 5, 9, 12 (2×), 13, 14, 15, 19; frgs. 3–7 ii 14, 16; frg. 8 iv 2, 3, 5 (2×), 8, 11; 4Q395 frg. 1 6; 4Q396 frgs. 1–2 i 3, 5; frgs. 1–2 ii 1, 5, 7 (2×), 10; frgs. 1–2 iii 1, 6 (2×), 10; frgs. 1–2 iv 2 (2×), 5, 6, 7, 8, 9; 4Q397 frgs. 1–2 4; frgs. 6–13 11 (2×), 12, 13 (2×), 14. There are 3 cases of ‫שׁל‬‎: 4Q394 frgs. 3–7 i 15, 19; 4Q395 frg. 1 10. In the C manuscript, there are 19 occurrences of ‫שׁ‬: ‎4Q397 frgs. 14–21 2, 10, 12, 15, 16; frg. 22 2; frg. 23 2; 4Q398 frgs. 11–13 3, 4 (2×), 6, 7; frgs. 14–17 i 5; frgs. 14–17 ii 1, 3 (2×), 4, 6; 4Q399 frg. 1 i 11. There are 2 cases of ‫שׁל‬‎: 4Q397 frg. 23 2; 4Q398 frgs. 14–17 ii 6. 20.  There are 1 case of ‫שׁל‬‎(4Q385 frg. 6:9) and 20 occurrences of ‫( שׁ‬CD 15:11; 20:4; 4Q222 frg. 1 7; 4Q266 frg. 10 i 1; frg. 10 ii 2; 4Q302 frg. 8 3; 4Q322 frg. 1 3; 4Q322a frg. 1 9; 4Q324 frg. 1 6; 4Q332 frg. 2 3; 4Q333 frg. 1 3; 4Q448 3 5; 4Q468l frg. 1 2; 4Q521 frg. 2 ii + 4 11; 4Q522 frg. 9 i + 10 10; 11Q5 28:13; 11Q20 12:14; KhQ3 1:1 [2×], 4). 21.  From Naḥal Ḥever, there are 51 occurrences of ‫ שׁ‬in the texts collected in Yadin et al. 2002 (44.5, 6, 8, 9, 10, 11, 12 [2×], 13 [2×], 14, 15, 16, 18, 20, 23, 25, 26; 45.7, 8, 10, 15, 16, 18 [2×], 19; 46.3 [3×], 4, 5 [3×], 7 [3×], 9, 11; 49.5, 6, 7 [3×], 8 [3×], 11, 13; 51.5; 61.4) and 12 ‫שׁל‬‎(44.7, 10, 11; 45.7, 12, 20, 21; 46.7 [2×], 8; 51.6) and, from the Naḥal Ṣeʿelim (Cotton and Yardeni 1997), we may add 7 ‫שׁ‬‎(8.9 [2×], 8e–k.9, 30.7, 49.8 [2×], 12) and a single occurrence of ‫שׁל‬‎(8.9). From Wadi Murabbaʿat (Yardeni 2000), there are 39 occurrences of ‫שׁ‬‎ (22.5–6; 22h.1; 24b.4, 11, 12, 15; 24c.4, 7, 9; 24d.4; 24e.3, 6, 9; 30.3, 15, 19 [2×], 23 [3×], 36; 42.2, 3 [2×], 4 [2×], 5, 6; 43.5, 6; 44.2, 9; 45.3; 46.3, 4, 7, 8, 9; 47.5) and 8 of ‫שׁל‬‎(30.26; 22f.2, 5–6; 24.6; 42.1, 4; 46.7; 47.3).

“Relative” Diachrony

227

a number introducing verb (552) and prepositional (553) complements and then a handful that are typically assigned nonrelative, noncomplement uses, such as (554) and (555). (550) ‫ל־ה ַּמ ֲע ִׂשים ֶׁשּנַ ֲעׂשּו ַּת ַחת ַה ָּׁש ֶמׁש‬ ַ ‫ת־ּכ‬ ָ ‫יתי ֶא‬ ִ ‫ָר ִא‬ ‘I have seen all the deedsi whi that ___i have happened under the sun’ (Eccl 1:14) (551) ‫לֹא־יִ ְהיֶ ה ָל ֶהם זִ ָּכרֹון ִעם ֶׁשּיִ ְהיּו ָל ַא ֲחר ֹנָ ה‬ ‘(they) will not have a memory with ∅(those)i whi that ___i come later’ (Eccl 1:11) (552) ‫רּוח‬ ַ ‫יָ ַד ְע ִּתי ֶׁשּגַ ם־זֶ ה הּוא ַר ְעיֹון‬ ‘I knew that this also was chasing the wind’ (Eccl 1:17) (553) ‫א־תּד ֹר ִמ ֶּׁש ִּתּדֹור וְ לֹא ְת ַׁש ֵּלם‬ ִ ֹ ‫טֹוב ֲא ֶׁשר ל‬ ‘that you do not vow is better than-that you do vow and don’t fulfill (it)’ (Eccl 5:4) (554) ‫ַׁש ָּל ָמה ֶא ְהיֶ ה ְּכע ְֹטיָ ה ַעל ֶע ְד ֵרי ֲח ֵב ֶריָך‬ ‘because why should I be like (one who) wraps (herself) beside the flocks of your companions?’ (Song 1:7) (555) ‫ה־ּדֹודְך ִמּדֹוד ֶׁש ָּכ ָכה ִה ְׁש ַּב ְע ָּתנּו‬ ֵ ‫ַמ‬ ‘how is your beloved (better) than (another) beloved, so that you made us swear so?’ (Song 5:9) This same grammatical sketch applies to the postbiblical use of ‫ שׁ‬in the Qumran texts (3Q15 and 4Q394–99) and the Mishnah: (556) ‫‏בחרובא שבעמק עכור תחת‏ המעלות‬ ‘in the ruini whi that ___i is in the Valley of Achor, under the steps’   (3Q15 1:1–2) (557) ‫ושליׄ [ם‬ ׄ ‫וי]ר‬ ׄ . . . [‫שהמקדש‬ ׄ ‫חושבים‬ ׄ ‫‏ואנ֯ ֯חנו‬ ׄ ‘we determine that the sanctuary [. . . Je]rusale[m]’    (4Q394 frgs. 3–7 ii 16) (558) ‫‏כי שלוא ראה ולוא שמע לוא י[דע ֯]לעשות‬ ‘because ∅(the one)i whi that ___i has not seen or has not heard does not k[now what ]to do’ (4Q396 frgs. 1–2 ii 5) (559) ‫] דו֯ י֯ ׄד שהיא איש חסדים‬..[ ‫‏זכור‬ ‘remember [ ] Davidi, whi that hei was a man of loyalty’   (4Q398 frgs. 14–17ii.1)

228

Chapter 7

(560) ‫רוּמ ָתן‬ ָ ‫אכל ִבּ ְת‬ ַ ֹ ‫‏מ ָשּׁ ָעה ֶשׁ ַהכּ ֲֹהנִ ים נִ ְכנָ ִסים ל‬ ִ ‘from the houri whi that the priests enter ___i to eat of their heave offering’ (m. Ber. 1:1) (561) ‫‏עד ֶשׁיַּ ֲע ֶלה ַעמּוּד ַה ַשּׁ ַחר׃‬ ַ ‘until that the pillar of dawn rises’ (m. Ber. 1:1) (562) ‫ִמ ֶשּׁיַ ִכּירוּ ֵבין ְתּ ֵכ ֶלת ַל ָלּ ָבן‬ ‘from ∅(the hour)i whi that they can recognize between blue and white ___i’ (m. Ber. 1:2) (563) ‫תֹּורה‬ ָ ‫קֹורא ַבּ‬ ֵ ‫אָדם ֶשׁהוּא‬ ָ ‫‏כּ‬ ְ ‘like a mani whi that hei recites in the Torah’ (m. Ber. 1:2) (564) ‫אָמרוּ לֹו‬ ְ ‫יא׳ ֶשׁנָּ ָשׂא וְ ָק ָרא ְב ַליְ ָלה ֶשׁנָּ ָשׂא‬ ֵ ‫ַ‏ר ָבּן גַּ ְמ ִל‬ ‘Rabban Gamlieli, whi, that ___i married and recited (the Shemaʿ) on the night that he married, said to him . . .’ (m. Ber. 2:5) The few examples above illustrate that the postbiblical use of ‫ שׁ‬mirrors the biblical usage, with both overt relative heads ([556], [559], [560], [563], and [564]) and null heads ([558] and [562]) and occasional overt resumption ([559] and [563]). Additionally, the relative clauses modifying proper names in (559) and (564) indicate both a restrictive and nonrestrictive use of ‫שׁ‬. Finally, the word ‫ שׁ‬also introduces complement clauses for verbs (557) and prepositions (561). All of these features are found with both ‫ שׁ‬and ‫ אשׁר‬in the Hebrew Bible. There is, however, some indication of subtle differences in the use of ‫שׁ‬ versus ‫אשׁר‬. First, the examples I used to illustrate overt resumption in the postbiblical texts, (559) and (563), employ it at the subject position, which is extremely rare for ‫אשׁר‬. Second, the biblical examples (554) and (555), for which I cannot offer a relative explanation, suggest a possible further grammaticalization for ‫שׁ‬, to a more general subordinator. The biblical examples, though rare ( just the 2 I have cited), foreshadow the greater variety of usage for ‫ שׁ‬in the Mishnah (see Pérez Fernández 1999: 49–54). I will discuss how this may relate to the diachrony of ‫ אשׁר‬clauses below, in §7.5. With significant grammatical overlap with ‫אשׁר‬, the use and distribution of ‫ שׁ‬in the Hebrew Bible and nonbiblical sources raise a host of questions. For example, was ‫ שׁ‬an item native to the grammar of the authors of Ecclesiastes, Song of Songs, Ben Sira, and 4QMMT, as well as the Mishnah, but not for most biblical authors? If so, why did it not displace ‫ אשׁר‬entirely in Ecclesiastes and Ben Sira, as it did in Song of Songs and the Mishnah? Did the authors of Ecclesiastes and Ben Sira borrow ‫ שׁ‬from another dialect of Hebrew, and if so, why? Is it a case of archaizing in both books? If so, why? For Qoheleth, such an archaizing can perhaps be argued as a strategy for strengthening the

“Relative” Diachrony

229

feel of the Solomonic persona of the book’s primary voice. But this does not explain Ben Sira. 7.4.  Sorting Out the Many Proposed Functions of ‫אשׁר‬ The data I have presented in the last two sections on ‫ אשׁר‬and ‫ שׁ‬as well as the inadequate diachronic explanation I sketched at the end of 7.2.3 suggest that three interrelated diachronic lines of inquiry should be pursued. First, are the relative and complement uses of ‫ אשׁר‬related, and if so, did one proceed from the other? Second, did ‫ אשׁר‬develop other, more-general subordination functions? And third, how do ‫ אשׁר‬and ‫ שׁ‬interact diachronically? The first question concerns the relationship of the relative and complement functions of ‫אשׁר‬. In two separate studies within the framework of functionalist typology and grammaticalization, Talmy Givón argues that ‫ אשׁר‬developed from an initial relative marker to also introducing complement clauses at a later stage of the language (Givón 1974; 1991). Givón marshals data from biblical texts that he designates either “early” or “late” and asserts that the early biblical examples of ‫ אשׁר‬all represent relative clauses, while the later biblical examples exhibit first the additional function of introducing “subjunctive complements,” as in (565) below, and then finally the function of introducing verbal complements for verbs of cognition, as in (566). 22 (565) ‫אכלוּ‬ ְ ֹ ‫אמר ַה ִתּ ְר ָשׁ ָתא ָל ֶהם ֲא ֶשׁר לֹא־י‬ ֶ ֹ ‫וַ יּ‬ ‘The minister told them that they shouldn’t eat’   (Ezra 2:63; trans. Givón 1974: 15) (566) ‫הוּדי‬ ִ ְ‫י־הגִּ יד ָל ֶהם ֲא ֶשׁר־הוּא י‬ ִ ‫ִכּ‬ ‘because he had told them that he was a Jew’   (Esth 3:4; trans. Givón 1974: 16) The epigraphic data partially support Givón’s diachronic analysis since there are no examples of complement clauses among the occurrences of ‫ ;אשׁר‬however, the paucity of ‫ אשׁר‬examples in the epigraphic texts should discourage us from drawing any firm conclusions in this regard (i.e., the absence of an ‫ אשׁר‬complement clause could be coincidental, given the number and types of inscriptional texts). And if we set aside Givón’s problematic assumptions concerning the dating of the texts in his corpus and the size of his corpus, 23 and 22.  Note that Givón incorrectly categorizes constructions with the object marker ‫ את‬immediately preceding ‫ אשׁר‬as complement clauses (Givón 1974: 18). The fact that ‫ את‬never precedes the other complement clause subordinator, ‫כי‬, suggests that cases of ‫ את אשׁר‬are best understood as null-head relatives (on null-head ‫ אשׁר‬relatives, see §4.3). 23.  Givón’s limited corpus prohibits him from seeing precisely the type of examples he proceeds to argue are nonexistent in that material. For what he calls “early” Biblical Hebrew, Givón 1974 uses the first 20 chapters of Genesis and the book of Joshua, and Givón 1991 uses 35 chapters of Genesis and 20 chapters of 2  Kings. For what he identifies as “late”

230

Chapter 7

examine the distribution of complement ‫ אשׁר‬within the traditional three-stage framework that is held by the majority of scholars, a slight pattern does in fact emerge. First, the texts typically identified as archaic (i.e., Genesis 49, Exodus 15, Numbers 23–24, Deuteronomy 32–33, Judges 5, and Psalm 68) lack any examples of ‫ אשׁר‬used to introduce complement clauses (although, since even the few instances of relative-‫ אשׁר‬in these passages are actually in the prosaic frames that could be part of a later editorial layer, we should be careful not to make too much of this—that is, without good contrasts, we cannot draw any solid conclusions). Second, outside the so-called archaic Hebrew material, the ‫ אשׁר‬complement clauses are distributed throughout the remaining linguistic strata, illustrated in (567)–(569). (567) “Earlier Standard” Biblical Hebrew ‫ד־בּ ֵה ָמה ְל ַמ ַען ֵתּ ְדעוּן ֲא ֶשׁר‬ ְ ‫ץ־כּ ֶלב ְלשׁ ֹנֹו ְל ֵמ ִאישׁ וְ ַע‬ ֶ ‫וּלכֹל ְבּנֵ י יִ ְשׂ ָר ֵאל לֹא יֶ ֱח ַר‬ ְ ‫וּבין יִ ְשׂ ָר ֵאל׃‬ ֵ ‫יַ ְפ ֶלה יְ הוָ ה ֵבּין ִמ ְצ ַריִ ם‬ ‘and a dog shall not growl at any of the children of Israel, whether people or beasts, so that you may know that Yhwh makes a distinction between Egypt and Israel’ (Exod 11:7) (568) “Later Standard” Biblical Hebrew ‫ק־א ֶמת‬ ֱ ‫א־ת ַד ֵבּר ֵא ַלי ַר‬ ְ ֹ ‫ד־כּ ֶמּה ְפ ָע ִמים ֲאנִ י ַמ ְשׁ ִבּ ֶעָך ֲא ֶשׁר ל‬ ַ ‫אמר ֵא ָליו ַה ֶמּ ֶלְך ַע‬ ֶ ֹ ‫וַ יּ‬ ‫ְבּ ֵשׁם יְ הוָ ה׃‬ ‘and the king said to him: Up to how many times must I make you swear that you will say to me only the truth in the name of Yhwh?’   (1 Kgs 22:16) (569) “Late” Biblical Hebrew ‫ם־פּ ָע ִמים ַרבֹּות יָ ַדע ִל ֶבָּך ֲא ֶשׁר גַּ ם־אתה ִק ַלּ ְל ָתּ ֲא ֵח ִרים׃‬ ְ ַ‫ִכּי גּ‬ ‘because your heart also knows many times (over) that you yourself have cursed others’ (Eccl 7:22, Kethiv) At the same time, it has not escaped notice that the number of examples does increase in what are considered by many to be later biblical texts (see, Biblical Hebrew, both articles use Lamentations, Esther, Ecclesiastes, Song of Songs, Ezra and Nehemiah. For post-Biblical Hebrew, both the 1974 and 1991 study use the first 17 chapters of the Mishnah (the first two tractates, Berakot and Peʾah, of the first order, Zeraʿim). Besides the methodological shortcoming of such a limited corpus, Givón’s simplistic approach to dating the biblical texts should give biblical scholars pause. For example, the identification of Genesis and Joshua as “early” Biblical Hebrew represents an oversimplification of the complex dating issues for both of the texts, regardless of where one stands in the increasingly vigorous debate on the history of ancient Hebrew (on this issue, see above, §2.1.3, n. 15). Genesis 49 arguably derives from some early stage of Biblical Hebrew, but the remainder of the book is commonly considered “standard” or “classical” Hebrew dating to the late preexilic or early postexilic periods, and is even considered Hellenistic or later by some.

“Relative” Diachrony

231

e.g., Rooker 1990: 111–12, 123). 24 This increase is apparent in the list of examples I consider to be ‫ אשׁר‬verbal complement clauses in (570): (570)  Exod 11:7; Lev 5:5; Num 32:23; Deut 1:31; 3:24; Josh 4:7; 1 Sam 15:20; 18:15; 2 Sam 1:4; 1 Kgs 2:44; 22:16; Jer 28:9; Ezek 20:26; Zeph 2:8; Pss 10:6; 89:52 (2×); Eccl 5:4, 17; 6:10; 7:18, 22, 29; 8:12, 14; 9:1; Esth 1:19; 2:10; 3:4; 4:11; 6:2; Dan 1:8 (2×); Ezra 2:63; Neh 2:10; 7:65; 8:14–15 (2×); 10:31; 13:1, 19, 22; 2 Chr 2:7; 18:15 In light of the apparent data trend, Givón may be correct that Hebrew ‫אשׁר‬ followed the grammaticalization path relative > complementizer while also retaining the earlier relative function. 25 But we should set aside the absence of ‫ אשׁר‬complement clauses in the epigraphic corpus and “archaic” Biblical Hebrew on the grounds that the entire data context contains too few tokens for drawing statistically confident conclusions (on the one hand, there are too few archaic and epigraphic texts; on the other hand, there are only 46 examples of ‫ אשׁר‬verbal complement clauses in total). The increased use of the complement ‫ אשׁר‬in books typically classified as “later” or “late” Biblical Hebrew could simply be coincidental. It is noteworthy that Givón muses about whether the syntactic-semantic functions of ‫ אשׁר‬are “organically” related; I think that this is the kernel of a viable alternative model for understanding the function of ‫ אשׁר‬in Hebrew. Like English that and French que, the two functions of Hebrew ‫ אשׁר‬are reducible to a single semantic and syntactic classification: nominalizer. Whether used as a relative word or a verbal or nominal complementizer, ‫ אשׁר‬is a grammatical word that allows clauses to be categorized as nominal items. In other words, in relative clauses ‫ אשׁר‬nominalizes a clause so that it may function as an adjective-like modifier of a noun (e.g., the man that . . .), and in complement clauses ‫ אשׁר‬nominalizes a clause so that it may function as the complement of a noun (e.g., the fact that . . .), a preposition or conjunction (e.g., because that he sinned . . .’), or verb (e.g., he swore that . . .). If introducing both relative clauses and complement clauses was original to the grammaticalization of ‫אשׁר‬ before the written evidence for Hebrew, the diachronic question remains—was there a change in the frequency of usage that betrays an underlying historical process? To answer that question requires that we consider the second and 24.  Givón makes this point as well (1974: 15–17). Note that this tendency toward the increasing use of ‫ אשׁר‬to introduce complement clauses had been noticed by the beginning of the twentieth century (e.g., by Davidson 1901: 196). 25. Though relative > complementizer is listed as a known grammaticalization path in Heine and Kuteva (2002: 254), the fact that they build on Givón’s analysis and (inaccurately) cite biblical data (see above, n. 1) should give us serious pause regarding how strongly we may draw any grammaticalization conclusions on this point.

232

Chapter 7

third issues I am addressing in this section: the view that ‫ אשׁר‬developed into a general subordinator, and the diachronic relationship between ‫ אשׁר‬and ‫שׁ‬. While there is little disagreement that ‫ אשׁר‬functions as a relative word by the earliest stage of Hebrew for which we have data—of almost 5,500 occurrences of ‫ אשׁר‬in the Hebrew Bible, nearly 5,400 are unarguably relative in function—all analyses assume that further change has occurred within Hebrew. This approach is succinctly summarized in BDB 81: apparently ‫אשׁר‬ further changed so that it “weakened in Heb[rew] to a mere particle of relation.” Similarly, Gaenssle characterizes ‫ אשׁר‬as a “vague and indeterminate” particle of relation that becomes “conjunctional” (Gaenssle 1915: 71) and then proceeds to discuss examples where ‫ אשׁר‬ostensibly introduces a wide variety of subordinate clauses, including complement, causal, result, purpose, conditional, concessive, temporal, and modal clauses (1915: 71–104). It is not without significance, however, that the standard reference works (GKC, IBHS, JM, and BHRG) and lexica (BDB, DCH, and HALOT) together list only 97 examples of ‫ אשׁר‬used to introduce nonrelative and noncomplement clauses. In table 7.1, I provide a quick reference to the location of these discussions, with footnotes listing the combined examples they cite (including examples used in Gaenssle 1915: 71–104). Assuming that these 97 account for the most compelling examples, taken together with the 43 cases of ‫ אשׁר‬introducing a verbal complement clause, the ostensibly nonrelative uses of ‫ אשׁר‬account for 2.6% of the total number of ‫ אשׁר‬clauses. This basic numerical fact, by itself, highlights the statistically questionable nature of any strong claims about further grammaticalization in the use of ‫אשׁר‬. More to the point is that even the 97 examples are not without an alternative analysis as relative clauses. For instance, for each of the ‫ אשׁר‬clauses I provided as examples of nonrelative, noncomplement uses—in (528)–(531) for the Hebrew Bible, (542)–(544) for Ben Sira, and (545)–(549) for Qumran—I can provide a compelling relative analysis. Below, each of these examples is given again but reinterpreted as a relative. Consider first the biblical data. (571) (= [528]) Extraposed Relative ‫ן־ה ִא ָשּׁה ַהזֹּאת ָליְ ָלה ֲא ֶשׁר ָשׁ ְכ ָבה ָע ָליו׃‬ ָ ‫וַ יָּ ָמת ֶבּ‬ ‘then the soni ___j of this woman died (at) night, [whi that (she) laid upon himi]j’ (1 Kgs 3:19) 26 (572) (= [529]) Normal Relative ‫ל־היָּ ִמים‬ ַ ‫ת־דּ ָב ָרי ֲא ֶשׁר יִ ְל ְמדוּן ְליִ ְראָה א ִֹתי ָכּ‬ ְ ‫אַשׁ ִמ ֵעם ֶא‬ ְ ְ‫ת־ה ָעם ו‬ ָ ‫ל־לי ֶא‬ ִ ‫ַה ְק ֶה‬ ‫ל־ה ֲא ָד ָמה‬ ָ ‫ֲא ֶשׁר ֵהם ַחיִּ ים ַע‬ 26.  See above, example (430) and n. 73.

233

“Relative” Diachrony

Table 7.1.  Non-nominalizing Analyses of ‫ אשׁר‬in Representative Grammars Kautzsch 1910 (GKC)

Waltke and O’Connor 1990 (IBHS)

Van der Merwe, Naudé, and Kroeze 1999 (BHRG)

Joüon and Muraoka 2006 (JM)

Causala

§158b

§38.4

§40.6.5

§170e

Conditionalb

§159cc

§38.2

§40.6.7

§167j

Comparative

c

§161b

§174f

Temporald

§164d

§19.3b (esp. n. 11)

Purpose

§165b

§38.3

§40.6.4

§168f

§166b

§38.3

§40.6.3

§169f

Resultf

e

§158k

a.  For supposed causal ‫אׁשר‬, see (34×) Gen 30:18; 31:49; 34:13, 27; 42:21; Num 20:13; Deut 3:24; Josh 4:7, 23; 22:31; Judg 9:17; 1 Sam 2:23; 15:15; 20:42; 26:23; 2 Sam 2:5; 14:22; 1 Kgs 8:33; 15:5; 2 Kgs 12:3; 17:4; 23:26; Isa 19:24–25; Jer 16:13; Hos 14:4; Hab 3:16; Zech 1:15; Job 34:27; Eccl 4:9; 6:12; 8:11, 12; Dan 1:10; 2 Chr 6:24. b.  For supposed conditional ‫אׁשר‬, see (13×) Exod 21:13; Lev 4:22; 25:33; Num 5:29; 9:21, 21; Deut 11:26–28; 18:22; Josh 4:21; 1 Sam 16:7; 1 Kgs 8:31, 33; Isa 31:4. c.  For supposed comparative ‫אׁשר‬, see (4×) Exod 10:6, 14:13, 34:18; Isa 7:17. d.  For supposed temporal ‫אׁשר‬, see (15×) Gen 40:13; 45:6; Num 33:1; Deut 1:46; 9:7; 1 Sam 20:31; 2 Sam 19:25; 1 Kgs 8:9; 11:42; 22:25; Jer 29:19; Ps 8:4; 139:19; Job 1:5; 2 Chr 35:20. e.  For supposed purpose ‫אׁשר‬, see (17×) Gen 11:7; 24:3; Exod 20:26; Deut 4:10, 40; 6:3 (2×); 32:46; Josh 3:7; 1 Kgs 22:16; Jer 42:14; Ruth 3:11; Eccl 7:21; Neh 2:7, 8; 8:14; 2 Chr 1:11. f.  For supposed result ‫אׁשר‬, see (14×) Gen 13:16; 22:14; Deut 28:27, 35, 51; 1 Kgs 3:8, 12, 13; 2 Kgs 9:37; Isa 65:16; Jer 19:11; Mal 3:19; Ps 95:11; Esth 9:1.

‘assemble the people for me, and I will let them hear my wordsi whi that they must learn ___i in order to fear me as long as they live on the earth’ (Deut 4:10) (573) (= [530]) Normal Relative (resumption by full copy of the head) ‫אָרץ גַּ ם־‬ ֶ ‫ת־ע ַפר ָה‬ ֲ ‫ם־יוּכל ִאישׁ ִל ְמנֹות ֶא‬ ַ ‫אָרץ ֲא ֶשׁר ִא‬ ֶ ‫וְ ַשׂ ְמ ִתּי ֶאת־זַ ְר ֲעָך ַכּ ֲע ַפר ָה‬ ‫יִמּנֶ ה׃‬ ָ ‫זַ ְר ֲעָך‬ ‘I will make your offspring like the dust of the earthi whi that if one can count the dust of the earthi your offspring also can be counted’ (Gen 13:16) (574) (= [531]) Prepositional Complement with Assumed Preposition ‫א־ת ָע ֶשׂינָ ה‬ ֵ ֹ ‫ֹלהיו ֲא ֶשׁר ל‬ ָ ‫ל־מ ְצֹות יְ הוָ ה ֱא‬ ִ ‫אַחת ִמ ָכּ‬ ַ ‫ֲא ֶשׁר נָ ִשׂיא יֶ ֱח ָטא וְ ָע ָשׂה‬ ‫אָשׁם׃‬ ֵ ְ‫ִבּ ְשׁגָ גָ ה ו‬

234

Chapter 7 ‘(when)-that a ruler sins and inadvertently does one of any of the commands of Yhwh, his god, which should not be done, and he incurs guilt . . .’ (Lev 4:22)

The key is being able to identify the relative head accurately. In many cases, the head of a relative in Hebrew is null or extraposed. Does this alternative analysis work for the nonbiblical data such as those that I cited above in (542)–(547)? Yes, it does. All 67 of the ‫ אשׁר‬clauses in Ben Sira and 232 ‫ אשׁר‬clauses isolated in Qumran texts CD, 1QS, and 1QM can arguably be analyzed as relative clauses, as below. (575) (= [542]) Ben Sira Null-Headed Relative ‫מה יתחבר פרור אל סיר אשר הוא נוקש בו והוא נשבר‬ ‘how can the pot go with the vessel? ∅(The time)i whi that they knock together ___i, the pot will be smashed’ (Sir 13:2A) 27 (576) (= [543a]) Ben Sira Prepositional Complement ‫כאשר יבוא עמך לא יתגלה לך‬ ‘when-that your people comes, he will not reveal himself to you’   (Sir 12:15A) (577) (= [544b]) Ben Sira Normal Relative ‫לכן גם לו הקים חק ברית שלום לכלכל מקדש אשר תהיה לו ולזרעו כהונה‬ ‫גדולה עד עולם‬ ‘therefore also for him he established a principle, a covenant of peace, in order to sustain the sanctuaryi whi that the great priesthood shall be for him and his descendants forever ___i’ (Sir 45:24B) (578) (= [545]) Qumran Normal Relative (relative head is 1cp clitic pronoun attached to noun) ‫‏בעבור רחמיכה ולוא כמעשינו אשר הרעונו ועלילות פשעינו‬ ‘on account of your mercy and not according to ouri works, whi that ___i have acted wickedly and not for the acts of our rebelliousness’   (1QM 11:4) (579) (= [546]) Qumran Extraposed Relative (resumption of the head inside the relative) ‫לוא תואכל בשר שור ^ושה^ ועז בתוך עירי אשר אנוכי מקדש לשום שמי‬ ‫  לתוך מקדשי‬2 ‫בתוכה אשר לוא יביא‬ 27.  The Ben Sira example in (575) is admittedly complex: whereas I find a null-head relative analysis not just acceptable but also compelling, it would be a rare example in which the null head has a basic temporal value, in constrast to the nearly ubiquitous ‫ כאשׁר‬prepositional complement type, which I would expect here. 28.  I read a Hiphil ‫ יביא‬rather than a Qal ‫יבוא‬, contra Wise 1990: 40; Qimron 2010: 191. Not only are the ‫ ו‬and ‫ י‬in 11Q19 almost identical, many who read a Qal ‫ יבוא‬in 52:20 read a Hiphil ‫ יביא‬in the very similar context of 47:6, 8, and 9. Finally, the pattern in Leviticus is the

“Relative” Diachrony

235

‘you will not eat the meati ___j of an ox, sheep, or goat within my city, which I shall sanctify by establishing my name in its midst, [whi that one has not brought ___i to my temple]j’   (11Q19 52:19–20) (580) (= [547]) Qumran Extraposed Relative ‫מיא כמוכה אל ישראל בשמים ובארץ אשר יעשה כמעשיכה הדולים‬ ‘whoi ___j is like You, O God of Israel, in heaven and on earth, [whi that ___i can do according to Your great works]j?’ (1QM 10:8) (581) (= [548]) Qumran Extraposed Relative ‫וגם אל יביא איש על רעהו דבר לפני הרבים אשר לוא בתוכחת לפני עדים‬ ‘also, let a man not bring a word against his neighbori ___j before the many, [whi that ___i has not been in rebuke before witnesses]j’   (1QS 6:1) (582) (= [549]) Qumran Normal (Stacked) Relative ‫‏וכול איש אשר יש אתו דבר לדבר לרבים אשר לוא במעמד האיש השואל‬ ‫את עצת היחד ועמד האיש על רגליהו‬ ‘and any mani that has a word to say to the many, whi that ___i is not in the rank of the man who is asking the counsel of the Yahad— that man shall stand on his feet’ (1QS 6:12–13) As with the biblical examples, each of the ‫ אשׁר‬clauses that I have read within Ben Sira and the Judean Desert texts—which have been assigned a nonrelative or noncomplement interpretation by others—can more simply be interpreted as a relative clause. I say “more simply,” not because the relative analysis is perhaps the most natural reading for a modern interpreter, but because the relative analysis builds upon well-known, well-attested Hebrew grammatical patterns, whereas the causal, conditional, purpose, etc., values assigned to the ‫ אשׁר‬clauses are ad hoc and thus grammatically indefensible. To summarize, the data indicate that the word ‫ אשׁר‬encodes a single syntactic-semantic function, to nominalize clauses. This is manifested in two ways: as a relative clause strategy and as a verb and noun-complement clause strategy. For the assumed further grammaticalization of ‫ אשׁר‬into a general subordinator, the commonly cited cases constitute only 2.6% of the 5,500 ‫אשׁר‬ examples in the Hebrew Bible (and, though I have not calculated the percentage for the nonbiblical texts, based on my reading I would estimate the same range). Furthermore, most of the nonrelative, noncomplementizer examples can be transparently analyzed as relatives. Up to this point, I have discussed the first two questions of this subsection— the relationship of the relative and complementizer functions of ‫ אשׁר‬and the use of an impersonal subject with a bivalent verb, including Hiphil ‫בוא‬, and a sacrifice complement; also, the verb ‫ בוא‬rarely takes inanimate subjects and never a sacrifice (see DCH).

236

Chapter 7

potential for further diachronic change concerning the use of ‫אשׁר‬. I will now address the distribution of ‫ אשׁר‬and ‫שׁ‬. There are three reasonably plausible explanations, all of which have been advanced at some point in scholarship: dialectal difference, stylistic variation, and diachronic change. As I discussed in §3.3, since Bergsträsser’s 1909 argument, the dominant etymology of Hebrew ‫ שׁ‬is to derive it from the Akkadian ša. This etymological connection is taken as the starting point for a word’s entrance into Hebrew and history within the language. The entrance for ‫ שׁ‬in the dialectal explanation lies with contact with Phoenician, perhaps via the northern tribes, or more likely during the period of strong connection between the early Omrides and Phoenicia. This, then, accounts for the situation within Hebrew, since this northern entrance point explains the suggestion by some that a number of the cases of ‫ שׁ‬in the Hebrew Bible reflect northern, “Israelian” Hebrew. 29 This scenario is plausible, that ‫ שׁ‬became the relative word of choice, by change and diffusion, within some Hebrew grammar in the north, from which it influenced some southern Hebrew grammar, particularly after 722 b.c.e., so that eventually it replaced ‫אשׁר‬, the process and final result of which we see in Ecclesiastes and Song of Songs, respectively (see, among others, Gordon 1955; Kutscher 1982: 32, §45; Davila 1990; Rendsburg 2006). One problem with a simple northern-to-southern dialectal explanation is the complete absence of ‫ שׁ‬from any epigraphic text with a northern provenance. A second problem with this explanation, as Rendsburg notes, is that ‫ אשׁר‬is used in biblical texts often identified as northern in origin (e.g., Judges 5–8, Hosea; Rendsburg 1990: 116). These two reasons are weighty enough to compel us to look further than the simple northern-to-southern explanation. Another dialectal approach is social instead of geographic. As Waltke and O’Connor state concerning the Biblical Hebrew data, “[T]here are signs that the speech of men differs from that of women; speech addressed to young or old may vary from a standard. Speech itself often differs from narrative prose, and there are traces of dialect variation based on region in both” (IBHS §1.4.1a). In this vein, some have identified ‫ שׁ‬as the colloquial Hebrew relative word and ‫ אשׁר‬as the literary choice (Bendavid 1967: 77; see also Joüon 1923: 89; Segal 1927: 42–43). Taking the social variation a step further, Rendsburg argued that the biblical texts contain evidence of diglossia in the biblical period (1990). Diglossia is, following Ferguson’s classic definition, A relatively stable language situation in which, in addition to the primary dialects of the language (which may include a standard or regional standards), there is a very divergent, highly codified (often grammatically more complex) super29.  On the identification of Judges 5–8 as northern, or “Israelian,” Hebrew, see, among others, Driver 1913: 449; Rendsburg 1990: 114–15.

“Relative” Diachrony

237

posed variety, the vehicle of a large and respected body of written literature, either of an earlier period or in another speech community, which is learned largely by formal education and is used for most written or formal spoken purposes, but is not used by any sector of the community for ordinary conversation. (Ferguson 1959: 336)

In the diglossia explanation, ‫ שׁ‬represents the “low,” colloquial register and ‫ אשׁר‬the “high,” literary register. How the colloquial form made its way into literary products (i.e., Qoheleth, Song of Songs) is explained by Rendsburg as follows: During the period of the monarchy, 1000–586 b.c.e., a standard literary Hebrew was utilized in which ‫ אשׁר‬was the sole relative pronoun. The colloquial form, which existed side-by-side with the classical form, was ‫שׁ‬, which in a very few instances infiltrated literary composition. The upheaval of 586 b.c.e., with the resultant exile and restoration, effected changes in the Hebrew language, and one of these was the further penetration of ‫ שׁ‬into written records. (Rendsburg 1990: 116–17)

Clearly, a straightforward diglossic explanation is perceived to be inadequate, since Rendsburg, following Segal and Rabin before him, combines the diglossia analysis with a geographical and diachronic analysis. Thus, for him the full story is that ‫ שׁ‬was the relative in the north and ‫ אשׁר‬in the south, that throughout the monarchy and after, northern and southern Hebrew influenced each other and thus we see ‫ שׁ‬in both supposedly northern and southern texts, and that ‫ שׁ‬continued in colloquial Hebrew, but ‫ אשׁר‬maintained its preferred status in the literary register (Rendsburg 1990: 118). 30 It is worth pausing to note that the continued study of diglossic situations has highlighted a number of common features that are relevant for the ancient Hebrew discussion. First, the “high” variety is quite often associated with sacred writings and small priestly groups (Hudson 2002: 20, 26). Second, the decline of diglossia (Hudson 2002: 32–34) is often connected to social change, such as a move toward nationalism, the decline of existing class structures (which would include the literary elite), greater literacy, or catastrophic political and social events (such as the Babylonian Exile or Maccabean revolt). Third, if a change occurs regarding “high” and “low” varieties, it is always “high” that loses ground and “low” that becomes the new standard (Hudson 2002: 8, 30). This is essentially what is proposed for the shift from Biblical to Rabbinic Hebrew. Fourth, it is possible that some “high” varieties represent 30.  Rendsburg’s study is particularly relevant to an analysis of the language of Ecclesiastes: he asserts that from the exile onward “the loss of political independence brought an end to the heyday of classical Hebrew. It continued to be used during the Exilic and postExilic periods, but the decay is evident from the greater number of spoken forms which appear in literary compositions” (1990: 168). Of all the books that contain features that Rendsburg identifies as colloquial, Ecclesiastes has by far the greatest percentage (14%).

238

Chapter 7

early forms of their “low” counterparts, and thus “low” is a later stage on an expected path of grammatical development; the “high” is “frozen” by social convention whereas the “low” continues to change (Hudson 2002: 18–19, 22). Each one of these observations fits what we know of the biblical books and the history of Israel at various points and makes it likely that diglossia did exist in ancient Israel. 31 However, the fact remains that the evidence we have at our disposal is all literary, which for obvious methodological reasons makes it difficult to argue confidently for the diglossic proposal. One important principle underlying the diglossia analysis—as well as the geographical and diachronic views—concerns language usage: the use of ‫שׁ‬ versus ‫ אשׁר‬is dictated by location, time, or social context; it is not open to free manipulation by the language user. As Kaye notes, users in a diglossic situation do not mix the “high” and “low” varieties; rather, “native speakers possess an overall ‘communicative competence’ rather than a mere ‘grammatical competence’” (Kaye 2002: 120). 32 Thus, if the variation between ‫ שׁ‬and ‫אשׁר‬ reflects intentionality, as suggested in Young 1993 and Davila 1994, then the salient linguistic distinction is not dialect or diglossic variety but style. Whereas dialectal differences in the textual evidence represent a linguistic accident—the differences reflect the separate origins of the contrasting linguistic forms—stylistic differences are not accidental. That is, characters are often distinguished by their speech, in a range of genre, from plays to novels and other types of narrated literature (e.g., James Joyce’s works). Speech may color the characters as old or young, educated or not, wealthy or poor, respectful or rude, local or foreign (see Rendsburg 1996 for biblical examples). Coloring the characters’ speech achieves two related ends for the storyteller. First, it allows the storyteller to stay out of the story, because an intrusive narrator (e.g., “and he said, in a Cockney accent, ‘. . .’”) is cumbersome, disrupts the flow of the story itself, and is mostly reserved for subjective asides or background information (e.g., “he was a nasty sort of fellow” or “and this was the way it was done back then”). Second, it allows a character to be continually distinguished from others with different speech patterns, thus subtly keeping the differences in the audience’s mind without being explicit. This literary technique is not simply aesthetic, however: the differences are used to engage the reader and encourage the construction of a reader-identity vis-à-vis the characters. The specific type of language use I have been describing is called style-shifting. 33 31.  For criticism of Rendsburg’s hypothesis, see Ólafsson 1991 and Young and Rezetko 2008, 1:173–79; for a more positive (though still qualified) assessment, see Kaye 1993. 32.  Crystal makes a similar observation about the communicative competence of users (although he includes more distinctions than just diglossia): regional dialects, class dialects, and temporal dialects are “rarely consciously manipulated by the individual language-user” (Crystal 1970: 103). 33.  To my knowledge, Stephen Kaufman was the first to use the term style-switching in the context of ancient Hebrew studies (Kaufman 1988: 55), and it has since been adopted and developed by Gary Rendsburg in a number of articles (1992a; 1992b; 1996; among others);

“Relative” Diachrony

239

Style-shifting is similar to but distinct from the linguistic phenomenon of code-switching. The latter is typically associated with “the use of more than one language in the course a single communicative episode” (Gordon and Williams 1998: 75–76), whereas the former is used to refer to the presence of more than one register within the same dialect, such as formal versus informal (1998: 75, 92). Importantly, Gordon and Williams argue that “when we look at code-switching in literature we are looking at a very conscious language use, unlike some instances of code-switching in speech” (1998: 70). In their study, the authors identify three primary functions of literary code-switching: extrinsic, organic, and political (Gordon and Williams 1998: 80–81). The extrinsic type of code-switching simply supplies local color in the text, whereas the organic type intends to avoid alienating the reader by translations or explanatory glosses. The political usage has the opposite goal: it is intended to cause discomfort in the reader by creating a cultural boundary. The literary code-switching in the organic and political categories works to establish group membership between some characters and some readers; that is, “in-groups” and “out-groups” are created with parts of the audience. 34 I have argued elsewhere that linguistic forms are used to create an “in-group” between the characters of Jonah and Ruth and the audience (Holmstedt 2006; 2009; see above, §3.3.3). 35 But is this what the authors of Ecclesiastes and Ben Sira (not to mention the scattered examples elsewhere in the Hebrew Bible) were doing with ‫ שׁ‬and ‫ ?אשׁר‬That is precisely what Davila has suggested for Ecclesiastes: “[T]he impression we get of [the author of Ecclesiastes] is that he was a proud iconoclast, and it is not hard to imagine him as a sage who insisted on talking like real folks and not the highbrows in Jerusalem” (Davila 1994). In other words, there were social registers of Hebrew in ancient Israel (note that this is not the same phenomenon as diglossia), and the author of Ecclesiastes intentionally used a lower-class variety to identify with his intended audience. Young makes a similar suggestion: “Perhaps then, just as with Deir Alla, we should describe Qoheleth’s language as the local literary dialect, even if it be considered as the author’s own invention and filled with what were according to contemporary literary taste, non-literary forms” (Young 1993: 157; 2005: 347–48). 36 however, this general concept is not new and has long been noticed in Hebrew studies—for example, in discussions about intentional “archaizing” (albeit not from an explicitly linguistic framework). 34. Similarly, Herman argues that style-switching “functions to mark aspects of participants’ identity, thereby reinforcing patterns of co-operation and conflict encoded at other levels of narrative structure as well” (Herman 2001: 62). 35. Similarly, see Young 1995 on ‫ שׁ‬in Judges 6–8, 2 Kgs 6:11, and Lamentations. Rendsburg (1990: 123 n. 29) and Holmstedt (2006) also comment briefly on 2 Kgs 6:11. 36.  Young and Rezetko take the stylistic analysis a step further: the use of ‫ שׁ‬is identified as “sub-standard” Hebrew in the service of the “unconventional writing” of an

240

Chapter 7

The problem with the stylistic-variation argument, at least for ‫ שׁ‬and ‫אשׁר‬, is the lack of a clear pattern. If ‫ שׁ‬did represent the lower social register, then why did Ecclesiastes not use it exclusively (as in Song of Songs)? What type of group membership or reader identity can the ‫ שׁ‬and ‫ אשׁר‬variation help to establish? Additionally, the random variation between ‫ שׁ‬and ‫ אשׁר‬is not explained by the “lowbrow-highbrow” proposal. 37 Studies of style-switching and code-switching (whether literary or spoken) have observed distinct patterns of usage (see Gordon and Williams 1998), which do not match the case of ‫ שׁ‬and ‫ אשׁר‬in Qoheleth. As with the dialectal hypothesis, there is little compelling about the stylistic variation hypothesis when it is teased apart. Having found the dialectal and stylistic explanations wanting, we are left to investigate the diachronic explanation. Given that the great bulk of Hebrew texts in the Bible witness the use of ‫ אשׁר‬to the exclusion of ‫שׁ‬, and also given that the Mishnah exhibits nearly the opposite case (the use of ‫ שׁ‬and the relegation of ‫ אשׁר‬to biblical quotations and allusions), it seems on the surface a logical proposal that ‫ אשׁר‬and ‫ שׁ‬represent a case of diachronic change. It is likely that the diachronic story, if it is correct, unfolded like so: for some reason (language contact in Babylon is likely, as I have suggested), ‫שׁ‬ was borrowed into the Hebrew lexicon; as it became diffuse, it challenged ‫אשׁר‬ so that the latter was eventually marginalized, such that it could be completely ignored by the time of 3Q15 and 4Q394–399 and was used in very restricted, specialized contexts in the Mishnah. The nearly equal usage of both ‫ שׁ‬and ‫ אשׁר‬in Ecclesiastes and Ben Sira indicates that the grammars of these authors existed right in the middle of the linguistic process of change-and-diffusion. 38 To make sense of this sketch, we must recognize a few features of language change. First, it is only the formal grammar represented in the output of an individual—the I-language represented by an idiolect—that is a discrete object of scientific study (Hale 2007: chap. 1; see also Naudé 2003: 197). From an I-language perspective, “change results when transmission is flawed with respect to some features. When transmission is not flawed (with respect to some feature), there has been no change in the strict sense” (Hale 2007: 36). This feature, the product of imperfect transmission in the acquisition process, spreads or becomes diffuse when it is accurately acquired by another speaker. The change-and-diffusion approach has a number of implications for the way that we may even talk about the history of Hebrew. First, each I-language, “unconventional thinker” (Young and Rezetko 2008: 2.65) and denied any diachronic relevance (2008: 1.214, 227, 247). 37.  The variation occurs indiscriminately, sometimes in the same verse and in adjacent and parallel clauses: Eccl 1:10; 2:12; 3:14, 15, 22; 4:2; 5:4, 14, 17; 6:10; 8:7, 14; 10:14; 12:7. 38.  Fredericks adamantly opposes placing Ecclesiastes’ language midway between Late Biblical Hebrew and Mishnaic Hebrew due to its mixed use of ‫ שׁ‬and ‫אשׁר‬‎ (Fredericks 1988: 147). Although his study focuses on methodology, it does not reflect the statistically-based historical linguistic analysis I have presented above.

“Relative” Diachrony

241

represented by the idiolect that is itself represented in the language of the philologically reconstructed texts of the biblical books, is its own “stage,” as it were (see Naudé 2003). Another way to approach this same issue is to borrow and modify the dialectological dictum slightly: every change and the resulting diffusion (if it becomes diffuse) has its own history. It is unlikely that any two change-and-diffusion features will have the same origin. It is also unlikely that any two I-languages will reflect the same cluster of change-and-diffusion features, which implies that the relative order of texts may vary for each feature analyzed. But no single feature set can be determinative for a relative order, since the texts (or, the I-languages represented within the texts) do not stand in a two-dimensional line; rather, since each I-language is a unique constellation of features, some will stand to the “left” or “right” of any two-dimensional line of descent (see Andersen and Forbes 1986; Forbes 2012; Forbes and Andersen 2012 for discussions of statistical modeling on biblical texts and grammar, especially as it relates to diachrony). Second, it necessarily follows from the acquisition-related change-anddiffusion framework that a new form will coexist with the corresponding older form within the speech community, perhaps for many generations. 39 Indeed, the precise pattern that has been observed many times over—and this is the third relevant feature—often follows a Sigmoid, or “S”-shaped curve. CharlesJames Bailey describes this pattern succinctly: A given change begins quite gradually; after reaching a certain point (say, twenty per cent), it picks up momentum and proceeds at a much faster rate; and finally tails off slowly before reaching completion. The result is an S-curve: the statistical differences among isolects in the middle relative times of the change will be greater than the statistical differences among the early and late isolects. (Bailey 1973: 77; see also Kroch 1989; Pintzuk 2003)

Note that the change-and-diffusion framework and the S-curve recognize the coexistence of the new and old forms for what may be many generations of speakers. The process may unfold relatively quickly or quite slowly; however it unfolds, the statistical research has shown the S-curve to be the expected result of the change over time. 40 39.  A commonly cited syntactic example is the development of “do”-support in Middle English, in which “do” appears as an auxiliary verb (or better, as the finite verb carrying the bundle of inflectional features) in questions (“do you want?”), clauses with an initial adverb (“rarely did they want”), and other restricted environments. This development began in a restricted environment and then spread to other contexts. Moreover, non-“do”-support clauses coexisted with the newer construction for over 300 years, until finally being replaced entirely by the “do”-support construction. See Lightfoot 1979; Kroch 1989; 2001. 40.  Note the caveats provided from a statistician’s perspective in Forbes 2014. In private communcation, Forbes has concurred with me that, the potential weaknesses of a single Scurve nothwithstanding, numerous S-curves representing different data sets with sufficient tokens taken together make a statistically stronger model for determining the relative dating

242

Chapter 7

The question now becomes how we might analyze ‫ שׁ‬and ‫ אשׁר‬in light of the three features of language change I have presented. Table 7.2 is based on the raw data provided in the first section of this study and presents the occurrences of ‫ אשׁר‬and ‫ שׁ‬by increasing frequency of ‫שׁ‬. If the order by increasing frequency of ‫ שׁ‬is on the right diachronic track, the results can then be plotted tentatively on an idealized S-curve, as in fig. 7.1. 41 What do we do with the information in table 7.2 and fig. 7.1? For instance, must it reflect only diachronic change? Not necessarily. One or two, or even a few cases in which a borrowed word is used for style do not undermine its overall diachronic story. Consider the use of ‫ אשׁר‬in the majority of the Qumran texts but the use of ‫ שׁ‬in the Copper Scroll (3Q15) and the Halakic Letter (4QMMT). It is reasonable, given the stance towards “scripture,” that the use of ‫ אשׁר‬reflects a religiously oriented archaizing—a very specific stylistic choice. This parallels the Mishnah’s use of ‫ אשׁר‬only in biblical quotations or allusions. In contrast, the use of ‫שׁ‬, then, in 3Q15, 4QMMT, and the Mishnah, represents the “real” picture of diffusion: ‫ שׁ‬replaced ‫אשׁר‬. The same explanation can apply to books that other feature analyses suggest belong later on the scale, such as Daniel—the lack of ‫ שׁ‬in the Hebrew of Daniel does not necessarily mean that the book is early but simply that the feature had not yet become diffuse in the Hebrew author’s idiolect, or the author was intentionally mimicking the “scriptural” use of ‫אשׁר‬, just like the majority of the Qumran texts. In addition to stylistic archaizing, there also appear to be examples of ‫שׁ‬ used as a literary device to portray a character as “non-Hebrew.” This would include the isolated example in 2 Kgs 6:11, the 3 examples in Jonah, and possibly the 5 examples in Judges (Holmstedt 2006a: 16–17; see also Young 1993; 1995). 42 If this analysis is accurate, it implies first that ‫ שׁ‬was perceived by the author and intended audience as “foreign” (though intelligible) and second that, as a borrowing for a literary need, such examples may have contributed to the actuation of change and diffusion but do not represent the process of of biblical books. Forbes (forthcoming) has further qualified a statistically proper use of the S-curve. 41.  The books in parentheses are those for which the ‫ שׁ‬and ‫ אשׁר‬data more likely reflect the influence of intentional stylistics; see further the subsequent discussion in the main text. Note also that my use of the idealized S-curve does not represent a statistical analysis per se but an approximating diagnostic tool to check the plausibility of a diachronic analysis. 42.  In 2 Kings 6, ‫ שׁ‬is placed in the mouth of an Aramean king, even though ‫ שׁ‬is not used in Aramaic. In Jonah, ‫ שׁ‬is placed once in the mouth of the sailors, when they speak among themselves, once in Jonah’s mouth, when he addresses the sailors, and once in God’s mouth, along with the immediately preceding ‫—אשׁר‬suggesting that ‫ שׁ‬is used for rhetorical effect— to support one of the author’s theological points: that Yhwh is the God of non-Israelites as well as Israelites. For a more detailed discussion, see Holmstedt and Kirk forthcoming.

243

“Relative” Diachrony

Table 7.2.  ‫ אשׁר‬and ‫ שׁ‬in Ancient Hebrew By Frequency

‫אשׁר‬ (Old)

‫שׁ‬ (New) % New

Epigraphic texts

38

0

0

Qumran Majority Sectarian9

Exod

309

0

0

Lev

309

0

Num

295

0

Deut

584

Josh

By Frequency

‫אשׁר‬ (Old)

‫שׁ‬ (New) % New

1,864

0

0

King

696

1

0.14

0

Gen

411

1

0.24

0

Chr10

345

2

0.58

0

0

Ezra–Neh

120

1

0.83

265

0

0

CD

132

2

1.50

Sam

428

0

0

Job

40

1

2.40

Isa

171

0

0

Judg

177

5

2.70

Jer

460

0

0

4Q266

67

2

2.90

Ezek

342

0

0

11Q20

12

1

7.70

Hos

12

0

0

Qumran Set A (4Q385, 521, 522)

23

3

11.50

Joel

12

0

0

Jonah

12

3

20.00

Amos

18

0

0

Qumran Set B (4Q222, 11Q5)

6

2

25.00

Obad

4

0

0

Sir

71

26

26.80

Mic

16

0

0

Lam

9

4

30.80

Nah

2

0

0

Eccl

89

68

43.30

Hab

3

0

0

4Q448

1

1

50.00

Zeph

6

0

0

Ps B (135, 144, 146)

6

7

53.80

Hag

7

0

0

Song

1

32

97.00

Zech

44

0

0

4QMMT (394–99)

2

67

97.10

Mal

13

0

0

Mishnah

69

11,690

99.40

Ps A (1–121, 125–28, 130–32, 134, 138–43, 145, 147–50)

96

0

0

Ps C (122–24, 129, 133, 136–37)

0

14 100,00

Prov

12

0

0

3Q15

0

56 100,00

Ruth

42

0

0

Qumran Set C (4Q302, 322, 322a, 324, 332, 333, 468l)

0

7 100,00

Esth

99

0

0

Naḥal Ḥever (and N. Ṣeʿelim)

0

71 100.00

Dan

47

0

0

Wadi Murabbaʿat

0

47 100.00

244

Chapter 7

Figure 7.1.  The Diffusion of ‫ שׁ‬on an Idealized S-Curve. diffusion itself. Thus, they do not sit on the S-curve of change in the position that corresponds to the frequency of the new item. In fact, it is my view that ‫שׁ‬ more likely reflects the borrowing (followed by the change-and-diffusion) of ša from Late Babylonian by the exilic population rather than earlier Akkadian through northern Canaanite. This, of course, fits a primarily diachronic explanation for the ‫ שׁ‬and ‫ אשׁר‬variation in the ancient Hebrew corpora. 7.5. Summary After considering the various ways to analyze the ‫ שׁ‬and ‫ אשׁר‬data—dialectal, stylistic, grammatical, and diachronic—I have concluded that the likeliest interpretation of the relationship of ‫ אשׁר‬and ‫ שׁ‬clauses is primarily diachronic, while acknowledging the probable secondary influences of dialect and style (all of which reflect the complexity of real language). To summarize, the rarity of ‫ שׁ‬in early texts (e.g., Judg 5:7; 6:17; 7:12; 8:26; 2 Kgs 6:11; Jonah 1:7, 12; 4:10) reflects a literary strategy of borrowing an intelligible form from a nearby dialect or related language (e.g., northern Hebrew or Phoenician) in order to color characters’ speech, while later examples (i.e., Gen 6:3; Job 19:29; Lam 2:15, 16; 4:9; 5:18; Ezra 8:20; 1 Chr 5:20; 27:27, and the occurrences in Ecclesiastes and Song of Songs as well as in Ben Sira and 3Q15, 4QMMT, and the Mishnah) reflect a change-and-diffusion process in the exilic or postexilic period. If further change occurred in the function of ‫( שׁ‬perhaps represented in the earliest stage by the two outlying examples in Song of Songs, provided here again in [583] and [584]), it is critical to provide a typologically and grammatically plausible path.

“Relative” Diachrony

245

(583) (= [554]) ‫ַׁש ָּל ָמה ֶא ְהיֶ ה ְּכע ְֹטיָ ה ַעל ֶע ְד ֵרי ֲח ֵב ֶריָך‬ ‘because why should I be like (one who) wraps (herself) beside the flocks of your companions?’ (Song 1:7) (584) (= [555]) ‫ה־ּדֹודְך ִמּדֹוד ֶׁש ָּכ ָכה ִה ְׁש ַּב ְע ָּתנּו‬ ֵ ‫ַמ‬ ‘how is your beloved (better) than (another) beloved, so that you made us swear so?’ (Song 5:9) I suggest that, if ‫ שׁ‬did experience further grammaticalization, the mechanism for this change likely has its origin in the use of ‫ שׁ‬to nominalize clauses as the complement of a preposition and as a nominal clitic host for a conjunction. Consider the minimal pairs below: (585) a) ‫ם־ּב ֵה ָמה‬ ְ ‫ֹלהים וְ ִל ְראֹות ְׁש ֶה‬ ִ ‫ַל־ּד ְברַ ת ְּבנֵ י ָה ָא ָדם ְל ָב ָרם ָה ֱא‬ ִ ‫ָא ַמ ְר ִּתי ֲאנִ י ְּב ִל ִּבי ע‬ ‫ֵה ָּמה ָל ֶהם‬ ‘I said, I with my ‫לב‬,ֵַ  43 concerning humans, “God should test them and show that they are beasts, they themselves’ (Eccl 3:18) (586) ‫ל־ּד ְב ַרת ֶׁשּלֹא יִ ְמ ָצא ָה ָא ָדם ַא ֲח ָריו‬ ִ ‫ֹלהים ַע‬ ִ ‫ּגַ ם ֶאת־זֶ ה ְל ֻע ַּמת־זֶ ה ָע ָׂשה ָה ֱא‬ ‫אּומה‬ ָ ‫ְמ‬ ‘indeed–God has made this (day of prosperity) corresponding to that (day of adversity). (I have concluded this) because-that man does not find anything after him’ (Eccl 7:14) 44 (587) ‫וְ ַעד יִ ְת ַק ְדּשׁוּ ַהכּ ֲֹהנִ ים ִכּי ַה ְלוִ יִּ ם יִ ְשׁ ֵרי ֵל ָבב ְל ִה ְת ַק ֵדּשׁ ֵמ ַהכּ ֲֹהנִ ים׃‬ ‘or until (other) priests had consecrated themselves because the Levites were more undivided in consecrating themselves than the priests’ (2 Chr 29:34) (588) ‫אַה ָבה ַעד‬ ֲ ‫ת־ה‬ ָ ‫ה־תּע ְֹררוּ ֶא‬ ְ ‫וּמ‬ ַ ‫ה־תּ ִ ֧עירוּ ׀‬ ָ ‫רוּשׁ ָלםִ ַמ‬ ָ ְ‫ִה ְשׁ ַבּ ְע ִתּי ֶא ְת ֶכם ְבּנֹות י‬ ‫ֶשׁ ֶתּ ְח ָפּץ׃ ס‬ ‘I adjure you, O daughters of Jerusalem, do not stir up or awaken love until that it is ready!’ (Song 8:4) In (585), the complex preposition, which includes a cliticized NP, is bound to the NP ‫בני אדם‬, and in (587) the preposition ‫ עד‬introduces an unmarked 43.  On the syntax and rhetorical function of the ‘I-and-my-‫’לב‬ ֵ construction, see Holmstedt 2009b. 44.  Most commentators take ‫ על־דברת ׁש‬as a purpose or result construction, as do the standard lexica and grammars (e.g., Seow 1997: 230, 240; Longman 1998: 181; Fox 1999: 257; HALOT, s.v. ‫ ;דברה‬DCH, s.v. ‫ ;דברה‬GKC §165b). Although I suggest that the causal analysis reflected in my translation above makes better sense contextually (see also Schoors 1992: 147), even if the construction did signal purpose or result, the syntactic role of the ‫שׁ‬ would not change.

246

Chapter 7

nominalized temporal phrase. Those same prepositions are used with ‫ שׁ‬nominalized clauses in (586) and (588). Just as relative clauses may be unmarked (i.e., zero-relatives), so it seems that nominalized clauses need not always be introduced by an overt nominalizer. We see the same pattern with ‫אשׁר‬, as in (589) and (590): (589) ‫ת־ה ִעיר ְּב ָח ְכ ָמתֹו‬ ָ ‫ּומ ַּלט־הּוא ֶא‬ ִ ‫ּומ ָצא ָבּה ִאישׁ ִמ ְס ֵּכן ָח ָכם‬ ָ ‘but he finds a poor, wise man in it and he spares the city because of his wisdom’ (Eccl 9:15) 45 (590) ‫ל־ה ָא ָדם‬ ָ ‫ל־ּבית ִמ ְׁש ֶּתה ַּב ֲא ֶׁשר הּוא סֹוף ָּכ‬ ֵ ‫ית־א ֶבל ִמ ֶּל ֶכת ֶא‬ ֵ ‫ל־ּב‬ ֵ ‫טֹוב ָל ֶל ֶכת ֶא‬ ‘It is better to walk to the house of mourning than to walk to the house of feasting, because-that it is the end of every man’   (Eccl 7:2) In (589), the prepositional ‫ ב‬has a causal nuance, and the NP ‫ חכמתו‬fulfills the valency of the preposition. In (590), the valency of the same preposition is fulfilled by the ‫ אשׁר‬clause, which also serves as the clitic host for the proclitic preposition. What is clear from all the pairs in (585)–(590) is that, strictly speaking, it is the preposition that establishes the semantic nuance of the entire PP, not the ‫ שׁ‬or ‫אשׁר‬, since they may be absent. This is also the case for the less common but attested use of nominalized clauses with conjunctions, as in (591) and (592): (591) ‫יכם נ ֵֹתן ָל ֶכם‬ ֶ ‫ֹלהי ֲאב ֵֹת‬ ֵ ‫אָרץ ֲא ֶשׁר יְ הוָ ה ֱא‬ ֶ ‫ת־ה‬ ָ ‫אתם וִ ִיר ְשׁ ֶתּם ֶא‬ ֶ ‫וּב‬ ָ ‫ְל ַמ ַען ִתּ ְחיוּ‬ ‘in order (that) you will live and enter and possess the land that Yhwh, the God of your fathers, is giving to you’ (Deut 4:1) (592) ‫יהם‬ ֶ ‫אֹלה‬ ֵ ‫תֹּועב ָֹתם ֲא ֶשׁר ָעשׂוּ ֵל‬ ֲ ‫ְל ַמ ַען ֲא ֶשׁר לֹא־יְ ַל ְמּדוּ ֶא ְת ֶכם ַל ֲעשֹׂות ְכּכֹל‬ ‫יכם׃ ס‬ ֶ ‫ֹלה‬ ֵ ‫אתם ַליהוָ ה ֱא‬ ֶ ‫וַ ֲח ָט‬ ‘in order that they do not teach you to do (things) like all their abominations that they do for their gods and you sin against Yhwh, your God’ (Deut 20:18) Whereas the clause following the proclitic ‫ למען‬may be introduced by an overt nominalizer (592), the nominalizer may also be null (591). 46 It is with these preposition+complementizer-clause and conjunction+nominalizedclause uses of ‫ אשׁר‬and ‫ שׁ‬that most of the (very few) otherwise-difficult cases like them find plausible explanations. Consider the ‫ אשׁר‬clause in (593): 45.  On the causal use of ‫ב‬, see IBHS §11.2.5e. 46.  The null-complementizer pattern appears to become less frequent, since it appears very rarely in the Judean Desert texts, Ben Sira, and the Mishnah.

“Relative” Diachrony

247

(593) ‫יטב ְלָך‬ ַ ִ‫ֹותיו ֲא ֶשׁר אָנ ִֹכי ְמ ַצוְּ ָך ַהיֹּום ֲא ֶשׁר י‬ ָ ‫ת־מ ְצ‬ ִ ‫ת־ח ָקּיו וְ ֶא‬ ֻ ‫וְ ָשׁ ַמ ְר ָתּ ֶא‬ ‘and you shall keep my statutes and my commandments that I am commanding you today (in order) that it will go well for you’   (Deut 4:40) My translation indicates my view, that just as the ‫ אשׁר‬or ‫ שׁ‬may be null, so too the conjunction itself may be null, even if the nominalizer is overt. Such flexibility regarding the overtness of both the nominalizer and the conjunction suggests an attested pathway for the eventual reanalysis of ‫ שׁ‬and ‫ אשׁר‬from nominalizers to more general subordinators. As Hendery suggests, after briefly surveying a number of languages for which the relative and complement clauses overlap and in which the function words introducing these clauses are extended further, [The] relative chronology [of these changes] in each of these languages makes it clear that it was not a matter of what Givón (1974: 2) calls the ‘facile explanation’ of speakers using their ‘syntactic intuition’ to realise that these constructions have something in common and then making a blanket generalisation, but rather that the relative clause pattern crept into the subordinate clause paradigm through individual ambiguous constructions, possibly from several points at once, and then the whole set of clausal subordination constructions was eventually regularised. (Hendery 2012: 116; on syntactic reanalysis, also see Harris and Campbell 1995: 61–96)

It is plausible that both ‫ אשׁר‬and ‫ שׁ‬had been set on the path of further grammaticalization in the Hellenistic period (i.e., the period of the latest biblical books, the earlier Judean Desert texts, and Ben Sira). Unfortunately, for ‫אשׁר‬, the process was interrupted by its replacement by ‫שׁ‬. The examples of ‫ אשׁר‬in the Mishnah reflect its linguistic dormancy, since it appears in only liturgical passages and biblical quotations, and always as a relative word. If this had not occurred, perhaps ‫ אשׁר‬would have witnessed the fuller range of usage we find in the Mishnah with ‫שׁ‬. We will simply never know. Though the very few ‫אשׁר‬ data that cannot be given a compelling relative or complementizer analysis (see appendix B)—especially in light of the explanation I have given for the few cases like (593)—are often philologically and textually interesting, from a statistical point of view, they are also cases of extreme outliers and thus linguistically marginal.

Chapter 8

The Semitic Relative In this final chapter, I survey the relativization strategies of the primary ancient representatives of the Semitic language branches. These data provide a second context by which to filter the Hebrew data, beyond the typological contextualization that I have provided at the beginning of chaps. 3–6. I have assumed throughout this book that the primary reader has facility with Hebrew language, and so I have not provided interlinear glossing for the hundreds of examples I have cited. While I consider it likely that the same reader will be able to read the transliterated Canaanite (Moabite, Edomite, Phoenician), Aramaic, and Ugaritic examples with reasonable ease, I do not make the same assumptions for the other Semitic evidence I adduce in this chapter. Thus, the reader will note the addition of a line of interlinear glossing between the language data and the English translation for the Akkadian, Classical Arabic, Epigraphic South Arabian, and Geʿez examples. 8.1.  Brief Introduction to Semitic Languages The Hebrew language is a member of the Semitic language family, located on the Northwest Semitic branch of Central West Semitic (see Huehnergard 1995; 2005; 2013b). Within Northwest Semitic, Hebrew is part of the Canaanite dialect group, its closest linguistic relatives being Phoenician, Moabite, Ammonite, and Edomite (see Garr 1985). For an orientation to the Semitic languages, with an abundance of data, see Lipiński 2001. For recent discussions on the classification of Semitic languages, see Kouwenberg 2010: 584–98; Huehnergard and Rubin 2011. As I have done for the Hebrew data in chaps. 3–7, so in my very brief summaries of relativization in the other ancient Semitic languages I comment on the following features: the type of relative element, the position of the relative head vis-à-vis the relative, the linear order of the head and relative, resumption, and restrictive versus nonrestrictive semantics. 1 I begin with the oldest evidence, from Old Akkadian, simply because it provides some evidence about the origins of subordination and relativization in Semitic, and then move to 1.  Lipiński 2001 does not recognize any restrictiveness distinction in the Semitic relatives. This is consistent with the general lack of discussion on relative-clause semantics in Semitic grammars, particularly those of the ancient languages.

248

The Semitic Relative

249

Figure 8.1.  The Semitic Language Family Tree. the near genetic and geographical neighbors of ancient Hebrew—the other socalled Canaanite dialects—after which I work my way through the remaining Northwest Semitic languages before turning to brief overviews of Classical Arabic, Epigraphic South Arabian (ESA), and Classical Ethiopic (Geʿez). Whenever possible. I have chosen data from the earliest stages of the languages I cover. So, for instance, I discuss the features of the relative in Old Akkadian and Old Babylonian, not the later Assyrian or Babylonian dialects. Similarly, I cover Epigraphic Aramaic but not later Aramaic or Syriac texts, and I comment on ESA and Geʿez but not modern South Arabian and Ethiopic dialects. 8.2. Akkadian Akkadian 2 is the oldest attested Semitic language and the primary, if not sole, representative of East Semitic (whether Eblaite is a dialect of Akkadian, a distinct East Semitic language, or even a West Semitic language is yet unsettled). Akkadian personal names have been found in texts dating to ca. 2600 b.c.e., and full texts written in Akkadian appear by the middle of the twentyfourth century. These early dialects are referred to as Old Akkadian. From the twentieth century b.c.e., the dialects of Akkadian are divided into Assyrian (northern) and Babylonian (southern), reflecting discernible linguistic differences as well as the cultural and political history of Mesopotamia. The northern branch is further divided into Old, Middle, and Neo-Assyrian stages, ending with the fall of the Neo-Assyrian Empire in the late seventh century b.c.e. The southern branch is further divided into Old, Middle, Neo-, and Late 2. See Ravn 1941; von Soden 1995; Buccellati 1996; Deutscher 2001, 2002; Huehnergard 1995, 2005a, Johnson 2005.

250

Chapter 8

Table 8.1.  The Old Akkadian Relative Marker singular masc

nom gen acc

fem

nom gen acc

dual

šū šī šā

nom

šāt šāti šāt

nom

obl

obl

plural

šā —

nom

— —

nom

obl

obl

šūt šūti šât šât

Babylonian and is attested until the first century c.e. At some point in the late second millennium, both Assyrian and Babylonian scribes began using a specifically literary form of the language, mimicking features of Old Babylonian; this literary language is called Standard Babylonian, and it was used through the first millennium b.c.e. for literature. In keeping with the nature of this chapter, I will focus on Old Akkadian and Old Babylonian in this overview of the Akkadian relative. As the oldest examples of the Akkadian relative, the Old Akkadian and Old Babylonian relatives exhibit a feature that will also figure in the discussion of Ugaritic, Arabic, Epigraphic South Arabian, and Geʿez: the initial use of a relative marker based on š- (< *ṯ), which is cognate to ḏ- (> z-) of the West Semitic relative elements (see above, §3.3.2). This relative marker, which is summarized in table 8.1, is followed by the shift to the use of an uninflected relative complementizer, ša (< *ṯa), which appears to be the masculine singular accusative form (see Huehnergard 2006: 115–16). The syntax of the Old Akkadian relative appears to be consistent with Semitic relative syntax in general: externally-headed and postnominal, with promotion from any position within the relative and options for null heads and zero-marking. The use of resumption follows the Noun Phrase Accessibility Hierarchy just as we have seen with other Semitic languages: obligatory resumption in NP-internal and PP positions ([598]), optional complement resumption ([595]–[596] versus [597]), and no subject resumption ([594]). (594) xubull-um šu al X ibaššeʾ-u debt-nom rel.ms.nom on X exist.3ms-subj ‘the debti whichi ___i is upon X’ (HSS 10:109:22) (595) šeʾ-um šu ana še.ba a-šīt-u barley-nom rel.ms.nom to rations 1cs-leave.pret-subj ‘(concerning) the barleyi whichi I had left ___i for rations’   (Kienast/Volk SAB: Ga 3:4)

The Semitic Relative

251

(596) eql-am ša ká . . . n-ītiq-u li-šqiʾ-ū field-acc rel.ms.acc gate 1cp-pass.pret-subj prec-water-3mp ‘they should water the fieldi whichi we passed ___i (near the) gate’   (Kienast/Volk SAB: Pu 3:5) (597) šarru-kīn šar māt-im šu enlil māxir-a lā iddinu-šum Sargon king.of land-gen rel.ms.nom Enlil rival-acc neg give. pret.3ms-subj-him.dat ‘Sargoni, king of the land, whoi Enlil has not given himi a rival’    (RIME 2.1.1.6, caption 1:1–6) (598) šīb-ū šūt maxar-šunu enma X ana Y witness-mp.nom rel.mp.nom before-them quot X to Y ‘witnessesi whoi in front of themi “X said to Y”’   (Gelb OAIC 12:16) Each of the relatives in (594)–(598) is headed and introduced by the relative marker. The examples make the “marker” status of the relative element clear: it is inflected for gender, number, and case. Although I have come across no examples of NP-internal resumption yet, the resumption in (597) and (598) (in the indirect-object and locative-PP positions) suggests that Old Akkadian patterns as other Semitic languages do by requiring resumption up to (and perhaps including) the indirect-object position, allowing optional resumption at the direct-object position, and disallowing subject resumption. Akkadian employs both null-head relatives, illustrated in (599)–(600), and zero-marked relatives, as in (601)–(602). d (599) ša dub šu-a usasaku-ni utu rel.acc.ms tablet this-acc remove.dur.3cs-subj Shamash suxuš-šu l-issux u še.numun-šu l-ilqût foundation​-his prec-pull.up.3cs and progeny-his prec-destroy.3cs ‘∅(he)i whoi ___i removes this inscription—may Shamash pull up his foundations and destroy his progeny’ (RIME 2.1.1.1)

(600) u ša il-šu inaddin-u-šum and rel.acc.ms god-his give.dur.3cs-subj-him.dat ‘and ∅(the thing)i whichi his god will give ___i to him’   (RIME   2.1.4.26) (601) in rīmāt dinanna ta-rām-u-šu by love.fp.of Ishtar 3fs-love.pret-subj-him.acc ‘by the lovei whi (that) Ishtar showed ___i him’ (RIME 2.1.4.10)

252

Chapter 8

(602) tupp-i a-ddin-u-šum tablet-of 1cs-give.pret-subj-him.dat ‘the tableti whi (that) I gave ___i to him’   (Kienast/Volk SAB: Ki 2:13) Critically, both types of relatives are semantically restrictive, which is in contrast to the option for nonrestrictive modification with overtly-headed and marked relatives, as in (603). (603) (= [597]) Nonrestrictive Relative šarru-kīn šar māt-im šu enlil māxir-a lā iddin-u-šum Sargon king.of land-gen rel.ms.nom Enlil rival-acc neg give.pret.3ms-subj-him ‘Sargoni, king of the land, whoi Enlil has not given himi a rival’    (RIME 2.1.1.6, Caption 1:1–6) The basic syntax of relatives, including the use of resumption, does not change in Old Babylonian or, as far as I am aware, in any later stage of Akkadian. The most noticeable change is morphosyntactic: due to the redundancy of matrix clause agreement features on relative elements (in contrast to the syntactically useful information provided by relative agreement features on relative pronouns), relative elements seem to be inherently unstable, and systems that use them eventually replace them with a relative complementizer. This is precisely what we see between Old Akkadian and later stages of Akkadian: the inflected relative marker shifted to an uninflected relative complementizer, ša (Deutscher 2001; 2002). Consider the examples from the Code of Hammurapi in (604)–(607). (604) Old Babylonian awīl-um ša ana bull-îm illik-u man.ms-nom rel to extinguish.inf-gen go.dur.3ms-subj ‘the mani whi that ___i went to extinguish (the fire)’ (CH §25) (605) ina tupp-im ša išṭur-u-šim in tablet-gen rel write.pret.3ms-subj-her.dat ‘in the tableti whi that he wrote ___i to her’ (CH §179) (606) šeʾ-am ša imxur-u iriab barley.ms-acc rel receive.pret.3ms-subj replace.dur.3ms ‘he will replace the barleyi whi that he received ___i’ (CH §254) (607) almat-t-um ša mār-ū-ša ṣexxer-ū widow-fs-nom rel son-mp.nom-her young-mp.nom ‘a widowi whi that heri sons are very young’ (CH §177)

The Semitic Relative

253

In a pair of insightful studies, Guy Deutscher argues that zero relatives such as those in (601)–(602)—that is, the dīn idīnu type in which the relative head is bound to the zero-relative—represent the origin of relative modification in Akkadian (as well as other Semitic languages). From this zero-relative pattern, Deutsher argues, the type using the pronoun šu developed; this new construction began with the pronoun and was both in apposition to the relative head (and thus case-marked to match the head’s position in the matrix clause) and also bound to the following relative. At some stage, perhaps due to the perceived redundancy of the same case-marking on the relative head and the demonstrative šu as well as the perception that the šu introduced the relative, the construction experienced reanalysis by which the clause boundaries were changed and the appositional clause became an embedded (relative) clause (Deutscher 2002: 99). Moreover, “once the pronoun šu was reanalyzed as the relative particle and became part of the [relative clause], its case-marking became maladaptive, because it was assigned to the wrong clause” (2002: 100), and the case agreement between the relative particle and the head was dropped, leaving the indeclinable ša. In response to Deutscher’s analysis, J. Cale Johnson (2005) has recently claimed that the zero-marked relatives in Akkadian were originally internallyheaded relatives (see above, §§4.1.2–3). Johnson classifies Deutscher’s analysis as “a typological argument at its essence” (Johnson 2005: 88), whereas his own argument is his effort “to make sense of relativization in Akkadian through appeal to ongoing work in syntactic theory rather than typological abnormality” (Johnson 2005: 96). Besides reflecting an inaccurate characterization of Deutscher’s analysis, which is a combination of linguistic typology and generative syntactic analysis (much like my approach in this volume), Johnson not only builds on the type of subtle semantic judgments (e.g., definiteness effects, weak quantification, Focus-induced constituent fronting) that are highly difficult to make with confidence for an ancient, no-longer-spoken language (see above, §2.1.2), in his dismissal of typology he avoids addressing a fundamental question—why does a language that has a dominant head-initial syntax (e.g., preposition-object, noun-adjective) use a head-internal/head-final relativization strategy? 3 In many ways, Johnson’s study is a model application of emergent findings in general linguistics to the analysis of Semitic syntax, and for that it should be both applauded and emulated. And yet, Johnson’s antitypological argument illustrates the dangers of using cutting-edge syntactic analysis, focusing on a specific grammatical feature that is, in the end, disconnected from the other syntactic patterns in the language in question. And it is typology that often is 3.  Akkadian is strongly head-initial except for the well-known strict verb-final word order in main clauses, which is most often taken to be the result of contact with Sumerian (see, however, Michalowski 2000: 181–82, for alternative views).

254

Chapter 8

the best aid in maintaining this larger perspective even if (or perhaps especially when) the latest theoretical framework is employed. Perhaps more problematic regarding Johnson’s argument are the questionable (whether simply inaccurate or incomplete) analyses he makes in the process. For example, in relating the definiteness effect and internally-headed relatives for understanding Semitic syntax, he suggests that “the behavior of the definite and indefinite articles in IHRCs clearly resembles the behavior of the definite article in the construct state in that the definite article cannot appear on the head noun, while it may or may not appear on the modifier of the head noun” (Johnson 2005: 86–87). This is simply misleading, since the determiners in the internally-headed relatives clauses in non-Semitic languages are enclitic on the entire relative phrase, whereas the position of the determiner in the Semitic bound phrase (e.g., Hebrew) is as a prefix/proclitic on the modifier within the head NP (or in Aramaic, as an enclitic, but still on the modifier within the head NP); the determiner in Semitic never attaches to a phrase, unless it is the relative complementizer. This is a significantly different syntax for definiteness that may undermine his comparison. Similarly, Johnson’s discussion of Akkadian kalu and weak-versus-strong quantifications assumes that we can discern weak-versus-strong quantification in these data; this is an assumption I strongly question. In summary, lacking better evidence, Deutscher’s (2001; 2002) account of the origins of the relative clause is more compelling than Johnson’s 2005 argument. 8.3.  Edomite, Ammonite, Moabite, and Philistine The more poorly attested, and thus lesser known, languages among the Canaanite dialects provide us with very little relative clause data. 4 The Moabite royal inscription, the Mesha Stele, is the only example of a long narrative text; otherwise, our small data set comes from ostraca or inscribed objects (e.g., the incense altar from Khirbet al-Mudeyineh). The seven extant relatives, given in (608)–(614), come from texts dating to the first half of the first millennium b.c.e. (608) Moabite: Mesha (ca. 850 b .c.e.) wʾnk mlkt[y ʿl] mʾt bqrn ʾšr yspty ʿl hʾrṣ ‘and I ruled over hundreds in the townsi whi that I added ___i to the country’ (KAI 181.28–29) (609) Moabite: Kh. Mudeiyineh (seventh–sixth century b .c.e.) mqṭ[r] ʾš ʿš ʾlšmʿ lys[p] bt ʾwt ‘(an/the) incense altari whi that Elishama made ___i to add to the Oracle house’ (Dion and Daviau 2000; Rainey 2002) 4.  For more-extensive discussion of the data and analysis of relative clauses in these Canaanite languages, see Holmstedt 2008; see also Rosén 1959.

The Semitic Relative

255

(610) Edomite: Ḥorvat ʿUza (ca. 600 b.c.e.) wʾt tn ʾt hʾkl ʾšr ʿmd ʾḥʾmh ‘and now, give the foodi whi that ___i (is) with Ahiʾimmo’   (lines 3–4; Beit-Arieh and Cresson 1985) (611) Ammonite: Heshbon Ostracon I (ca. 600 b .c.e.) lbʿš[ʾ] ksp 20+20 ʾš ntn l[ . . . ‘to Baasha: 40 (pieces of) silveri whi that he gave ___i to . . .’   (Aufrecht 1989: #80.6; Jackson 1983: H1) (612) Ammonite: Seal of Abinadab (ca. 600 b .c.e.) ʾbndb šndr lʿš⟨tr⟩t bṣdn ‘Abinadabi, whi that ___i made a vow to Ashtarte in Sidon’   (Aufrecht 1989, #56, Jackson 1983, AS #49) (613) Philistine: Ashkelon (ca. 600 b .c.e.) —-]mʿbr.š.tš[—‘from the (cereal) cropi whi that you . . .’ (Cross 1996: 64) (614) Philistine: Tel Miqne–Ekron (seventh century b .c.e.) bt bn ʾkyš bn pdy bn ysd bn ʾdʾ bn yʿr šr ʿqrn lptgyh ʾdth ‘The templei whi (that) Akish, son of Padi, son of YSD, son of Ada, son of Yair, ruler of Ekron, built ___i for PTGYH, his mistress’    (KAI 286 1:1) Every relative in (608)–(614) is clearly externally-headed and postnominal. All but one are introduced by overt relative elements, ʾšr, ʾš, or š, with the Philistine relative clause in (614) being the only “zero-relative.” The examples in (608), (609), (611), (614), and presumably (613) show promotion of the relative head from the complement position, whereas the clauses in (610) and (612) show promotion of the subject as the relative head. None of the relatives in (608)–(614) exhibits overt resumption, nor do any of the relative elements show discernible agreement features (for instance, the relative ʾšr in (608) follows the masculine-plural head qrn ‘towns’ and the relative ʾš in (609) follows a masculine-singular noun). These two facts taken together suggest that first-millennium b.c.e. Moabite, Edomite, Ammonite, and Philistine did not use relative pronouns or markers but, rather, relative complementizers, as I have argued for Hebrew ‫( אשׁר‬see §3.2). Contextually, the relatives in (608)–(611) and (613)–(614) appear to be semantically restrictive, since the information within the relatives critically aids in the identification of each relative head. The contextually clear restrictive sense of the relative in (614) supports the typological universal that zerorelatives are semantically restrictive, not nonrestrictive (which also applies to

256

Chapter 8

Hebrew; see §6.3). Finally, example (612) confirms the use of nonrestrictive relative modification, since the proper noun Abinadab is a unique referent and needs no definition. 8.4. Phoenician Phoenician is the northernmost ancient Canaanite dialect. The term Phoenician 5 is used to cover the closely related dialects of ancient Byblos, Tyre, and Sidon as well as the texts related to their colonization and trade in the Mediterranean; the Phoenician dialect originating in Carthage, for which we have evidence from the fifth century b.c.e., is typically referred to as Punic. 6 In this study, I will survey and discuss the data about relatives in Byblian and Standard Phoenician, but I have omitted Punic due to space limitations (for some Punic analysis, see Schuster 1965; and Kerr 2010: 154–58; for Punic texts, see Jongeling 2008; Jongeling and Kerr 2005). The early Byblian Phoenician texts (see, e.g., KAI 1–12, 280), which provide us with our earliest datable texts and thus the earliest Canaanite relative examples, are distinct in attesting a relative z, which is almost certainly related to the ‫ז‬-series in Hebrew: 7 (615) Ahiram (ca. 1000 b.c.e.) ʾrn.z pʿl.[ʾ]tbʿl.bn ʾḥrm.mlk gbl.lʾḥrm.ʾbh ‘the coffini whi that Ittobaal, son of Ahiram, king of Gubl, made ___i for Ahiram his father’ (KAI 1.1) (616) Yahimilk (ca. 950 b.c.e.) bt.z bny.yḥmlk.mlk.gbl ‘the templei whi that Yahimilk, king of Gubl, built ___i’ (KAI 4.1) (617) Ilibaal (ca. 915 b.c.e.) mš.z pʿl.ʾlbʿl.mlk.gbl.byḥ[mlk mlk gbl lb]ʿlt.gbl.ʾdtw ‘the votive stelei whi that Elibaal, king of Gubl, son of Yehi[milk, king of Gubl,] made ___i for Baʿalat of Gubl, his mistress’   (KAI 6.1–2) (618) Shipitibaal (ca. 900 b.c.e.) qr.z bny.špṭbʿl.mlk gbl.bn ʾlbʿl mlk.gbl byḥmlk.mlk.gbl.lbʿlt gbl.ʾdtw 5.  For more-extensive discussion of the data and analysis of relative clauses in Phoenician, see Holmstedt 2008; see also Rosén 1959. 6.  Many discussions of Phoenician distinguish the dialects not only by geography but also along chronological lines—i.e., “early/old” and “late” stages (Segert 1976: 27–30; Friedrich and Röllig 1999: 4; cf. Krahmalkov 2001: 5–15). For discussion of the stages of Phoenician, see Röllig 1983; Amadasi Guzzo 2005; and Holmstedt forthcoming. For this discussion of the relative clause, there is no need to use chronological labels, since I indicate the generally accepted dates of the texts from which I take my examples. 7.  See KAI 1.1; 4.1; 6.1; 7.1; 9 A.3, B.3; 10.1–2, 3–4, 4–5, 5, 6, 7–8, 11–12; 11.

The Semitic Relative

257

‘the walli whi that Shiptibaal, king of Gubl, son of Elibaal, king of Gubl, son of Yehimilk, king of Gubl, built ___i for Baʿalat of Gubl, his Mistress’ (KAI 7.1–4 ) The examples in (615)–(618) demonstrate that early Byblian Phoenician relativization occurs precisely as in the other Canaanite languages we have covered: the relative is externally-headed and postnominal. The relative word is overt (z), and the there are no attestations of overt resumption. Since each head in these four examples is masculine singular, there is no information on whether the relative element in these early first-millennium texts carries agreement features matching that of the relative head. Thus, we may refer to the Byblian relative element as either a relative marker (if we assume it carried inflectional features) or a relative complementizer (if we assume it had no inflectional features), but not as a relative pronoun (contra Schuster 1965; Friedrich and Röllig 1999: 72–74, 208–201; Krahmalkov 2001: 93–103). The relative-clause data from later Byblian texts, sampled in (619)–(621), conform to the basic relative-clause pattern of the earlier texts and add our only Byblian examples of null heads and resumptive pronouns. (619) Son of Shipitibaal III (ca. 500 b.c.e.) a. bmškb zn ʾš ʾnk škb bn ‘in this resting-placei whi that I lie in iti’ (KAI 9 A.3) b. ]qr hmškb zn ʾš tp[tḥ ] ‘[ ] of this resting-placei whi that you op[en ___i ]’ (KAI 9 B.3) (620) Yahawmilk (ca. 450 b.c.e.) a. ʾnk yḥwmlk . . . ʾš pʿltn hrbt bʿlt gbl mmlkt ʿl gbl ‘I am Yehawmilki . . . whi that the Lady Baʿalat of Gubl made mei king over Gubl’ (KAI 10.1–2) b. whptḥ ḥrṣ zn ʾš ʿl pn ptḥy z ‘and this gold inscriptioni whi that ___i is opposite this inscription of mine’ (KAI 10.4–5) c. whʿpt ḥrṣ ʾš btkt ʾbn ʾš ʿl ptḥ ḥrṣ zn ‘and the gold winged-discsi whi that ___i are in the midst of the stonej whj that ___j is over this gold doorway’ (KAI 10.5) d. whʿrpt zʾ wʿmdh whktrm ʾš ʿlhm wmspnth ‘and this portico and its pillars and the capitalsi whi that ___i are upon them and its roof’ (KAI 10.6) e. km ʾš qrʾt ʾt rbty bʿlt gbl wšmʿ ql ‘when that I called my Lady, Baʿalat of Gubl, and she heard my voice’ (KAI 10.7–8)

258

Chapter 8 f. qnmy ʾt kl mmlkt wkl ʾdm ʾš ysp lpʿl mlʾkt ʿlt mzbḥ zn wʿlt ptḥ ḥrṣ zn wʿlt ʿrpt zʾ ‘Whoever you are, any king or any mani whi that ___i may again do work upon this altar and upon this gold doorway and upon this portico’ (KAI 10.11–12)

(621) Batnoʿam (early fourth century b.c.e.) kmʾš lmlkyt ʾš kn lpny ‘like ∅(those)i whi that ___i (belonged) to the queensj whj that ___j were before me’ (KAI 11) The majority of examples from these sixth–fourth-century b.c.e. texts lack overt resumption within the relative. More precisely, all subject-relatives lack resumption (620b, c, d, f), (621). We cannot say with any certainty whether the complement relative in (619b) included or lacked resumption, due to the textual break at the end of the line. However, (619a) and (620a) clearly show resumption with a PP and cliticized pronominal complement, respectively. (For further remarks on resumption in Phoenician, see below, §8.10.) The example in (621) is the only null-head relative extant in Byblian, and the relative is restrictive, which accords with both the typological and the Hebrew evidence (see §6.3). Finally, the example in (620e) shows that later Byblian used the relative complementizer also for prepositional complement clauses. The rubric of Standard (or Tyro-Sidonian) Phoenician (see, e.g., KAI 13– 61, 281–94) is often used to cover texts from Phoenicia proper, the larger Levant, and locations throughout the Mediterranean as far west as Spain. This larger corpus provides us with numerous relative clauses, all of which exhibit the relativization profile established so far: the relative head is generally overt but may be null, the head is always external to the relative clause, the relative is postnominal, and there is most often a gap in the relative (i.e., null resumption). The examples in (622)–(625) illustrate the variety of relative strategies in Standard Phoenician (see also Schuster 1965; Friedrich and Röllig 1999: 270–72; Krahmalkov 2001: 93–103). 8 (622) Zincirli (Kulamuwa; ca. 825 b.c.e.) a. wʾn[k] klmw br tm mʾš pʿlt bl pʿl hlpny{h}m 8.  See KAI 13.3; 14.3–4, 7 (2×), 7–8, 9, 10, 10–11, 16–17, 17–18, 19, 19–20; 15; 17.1– 2; 18.1–3; 19.1–4, 9–10; 24.4–5, 6–7, 13–14, 15, 16; 26 AI.2, 9, 14, 15, 18–19; AII.3–4, 4–5; AIII.7–8, 13, 15–16; AIV.1; BI.1, 5, 8, 9, 10–11, 12–13; BII.11–12; CI.2, 16; CII.2–3, 4, 9–10; CIII.1–3; CIV.7–8, 14, 19; 27; 30; 32; 33; 34; 37 A.4–5, 6, 7, 10, 13; B.5, 7, 8; 38.1– 2; 39.2–3; 40.2, 3 (2×); 43.1–3, 4–5, 6, 7–8, 9, 12–13; 45; 47.1–3; 50.3, 5, 6; 51.2; 54.1–2; 55; 57; 58; 59.1; 60.2, 3, 3–4, 7; 277.1–4; 281.1; 282; 287.2–3, 4–5, 6–7, 7; 288.1, 2–3; and 290.1–3; Cyprus-Idalion #12; Umm el-ʿAmed xiii 1–3; Magnanini #2.

The Semitic Relative

259

‘I am Kulamuwa, son of TM. Better than ∅(the things)i whi that I did ___i those [i.e., the rulers] before me did not do’   (KAI   24.4–5) b. wyšḥt rʾš bʿl hmn ʾš lbmh wrkbʾl bʿl bt ‘and may Baʿal Hamoni, whi that ___i belongs to BHM and RKBʿL, the masters of the house, destroy his head’ (KAI 24.16) The two relatives from the Kulamuwa text show gaps at the complement (622a) and subject (622b) positions. The second example (622b) illustrates the common use of an overt head, while the first example (622a) provides a rare case of a null head. In (622b), the null head is also preceded by the comparative preposition m-, which is prefixed directly onto the relative word ʾš. With a null head, the relative clause in (622a) is necessarily restrictive, but the relative in (622b) modifies a unique referent, Baʿal Hamon, and is thus nonrestrictive. Now consider a few examples from the slightly later Karatepe text. (623) Karatepe (Azatiwada; ca. 720 b.c.e.) a. wtrq ʾnk kl hrʿ ʾš kn bʾrṣ ‘I have removed all the evili whi that ___i was in the land.’    (KAI 26 AI.9) b. wysʿ hšʿr z ʾš pʿl ʾztwd ‘and [another king] pulls up this gatei whi that Azatiwada made ___i’ (KAI 26 AIII.15–16) c. wbmqmm ʾš kn lpnm nštʿm ʾš yštʿ ʾdm llkt drk ‘and in placesi whi that ___i had formerly been feared, whi that a man was afraid to walk on the road ___i’ (KAI 26 AII.3–5) These examples mirror the lack of resumption in the subject (623a, c #1) and complement (623b) positions that we saw in the Kulamuwa examples and add an example of a gap at a (locative) adjunct (623c #2) position. The Karatepe examples are all overtly-headed and (623c) illustrates that Phoenician relatives can be stacked—that is, sequential relatives can modify a single head. Late Phoenician texts provide few changes in the patterns of relative-clause formation. Consider examples (624) and (625), which date to the third–second centuries b.c.e. (624) Umm el-ʿAmed / Maʿṣub (222 b.c.e.) km ʾš bn ʾyt kl ʾḥry [hmqdš]m ‘like ∅(the way)i whi that they built all the others of the sanctuaries ___i’ (KAI 19.9–10)

260

Chapter 8

(625) Tyre (Throne of Ashtart / Abdi-ubasti; second century b.c.e.) ʿštrt ʾš bgw hqdš ‘Astartei, whi that ___i is in (the) holy community’ (KAI 17.1–2) The relative in (624) has a null head (and is thus restrictive) and a gap at a (manner) adjunct position. The relative in (625) has an overt head, the proper noun Astarte, which makes the relative nonrestrictive and shows a gap in the subject position. Note that with all of the overtly-headed examples in (622)– (625), the relative element ʾš does not show any agreement features, whether person, number, gender, or case. Thus, as with the Byblian relative words, the relative ʾš in Standard Phoenician must be considered a relative complementizer, not a pronoun. Finally, zero-relatives appear, albeit infrequently, throughout the history of Standard Phoenician, as the examples in (626) demonstrate (see also Schuster 1965: 443–44; Krahmalkov 2001: 100–101). (626) Zero-Relatives in Standard Phoenician a. Arslan Tash (seventh century b.c.e.) 9 bt ʾbʾ bl tbʾn wḥṣr ʾdrk bl tdrkn ‘any housei whi (that) I enter ___i, you should not enter, and any courtyardj whj (that) I tread ___j, you should not tread!’   (KAI   27.5–8) b. Umm el-ʿAmed (third–second century b.c.e.) ʿbdʾlm bn mtn bn ʿbdʾlm bn bʿlšmr bplg lʾdk ‘Abdilimi, son of Mittun, son of Abdilim, son of Baalsamor, whi (that) ___i is from the district of Laodicaea’ (KAI 18.1–3) c. Greece-Piraeus (96 b.c.e.) šmʿbʿl bn mgn ʾš nšʾ hgw ʿl bt ʾlm wʿl ḥṣr bt ʾlm ‘Shamaʿ-Baʿal, son of Magon, who (is) the chief of the communityi whi (that) ___i (is) over the temple of the gods and over the courtyard of the temple of the gods’ (KAI 60.2) 10 While the first relative clause (626a) establishes that the gap can occur in the object position of a zero-relative, much more common is the type of zero-relative we see in (626b–c): a subject relative with a non-finite predication. From the contexts, we can see that all three relatives in (626) are 9.  I find arguments against the authenticity of the Arslan Tash inscription to be compelling but list the text here simply for completeness. 10.  In this example, it is impossible to distinguish between a zero-relative analysis and an NP-internal PP analysis (see §1.1.3).

The Semitic Relative

261

clearly restrictive, which confirms the implicational universal concerning zerorelatives and restrictiveness. Cross-linguistically, many languages use a non-finite verbal relativization strategy in addition to a finite one. All of the examples we have examined up to this point have contained either finite verbs or null copulas. The relative clause in (627) is the lone Phoenician example of a participial relative. (627) Kulamuwa (ca. 825 b .c.e.) wkt byd mlkm kmʾš ʾklt zqn w[km]ʾš ʾklt yd ‘and I was in the hand of the kings like a firei whi (that) ___i consumes the beard or like a firej whj (that) ___j consumes the hand’ (KAI 24.6–7) 11 Note that the relative head was promoted from the subject position within the relative. There are often constraints on both zero-relativization and relatives with non-finite predicates. If Phoenician is like Hebrew, then it does not allow participial zero-relativization from anything other than the subject position within the relative. Among the epigraphic Canaanite texts of the first millennium, only Phoenician has left us with unambiguous cases of relative-clause resumption, but even in the Phoenician texts the examples are limited, though they appear to increase over time. The two Byblian examples, presented above in (619a) and (620a), are both from the second half of the first millennium. Standard Phoenician texts from later in the first millennium exhibit resumption at positions higher in the NPAH: (628) PP Resumption: Cyprus-Lapethos ii (274 b.c.e.) hdlt hnḥšt . . . ʾš bn mnḥt ḥny ‘the bronze plaquei whi that in iti (are) the details of my beneficence’ (KAI 43.12–13) (629) Subject Resumption a. Cyprus-Idalion (254 b.c.e.) bšnt 31 lʾdn mlkm ptlmys . . . ʾš hʾ št 57 lʾš kty ‘in yeari 31 of the Lord of Kings, Ptolemy, . . . whi that iti is year 57 of the Kitionite’ (KAI 40.2) b. Cyprus-Lapethos ii (274 b.c.e.) bšnt 11 lʾdn mlkm ptlmyš . . . ʾš hmt lʿm lpš šnt 33 11.  Alternatively: “and I was in the hand of the kings like a man who consumed his beard, and how one which consumed his hand (literally: I consumed my hand)” (Schuster 1965: 443).

262

Chapter 8 ‘in yeari 11 of the Lord of Kings Ptolemy, . . . whi that theyi (are) year 33 of the people of Lapethos’ (KAI 43.4–5)

The relative clause in (628) exhibits overt resumption as the complement of a preposition inside the relative clause. The two examples in (629) are the only two cases of subject resumption that I have found in any of the Phoenician texts. The dates of these last two are worth noting: they are both late first millennium b.c.e. 12 Assuming that the examples in (619a), (620a), and (628)–(629) are representative of Phoenician grammar, it is possible to draw some tentative diachronic conclusions. The chronological distribution of the extant Phoenician relatives suggests that resumption in the first half of the first millennium was only employed in the genitive/NP-internal and oblique positions. It was only later, toward the second half of the first millennium (and mostly in the late second half), that Byblian and Standard began to allow resumption higher up the NPAH, in the object and subject positions. Aside from influence via non-Semitic language contact (for example, the possible Cushitic influence in Geʿez; see §8.9), the same constraints in relativization I discussed for Hebrew apply to all the Semitic languages—they do not allow internally-headed relatives (§4.4) or pied-piping and prepositionstranding (§5.2.2.). The constraint on preposition-stranding is demonstrated in the Byblian (619) and Standard (628) Phoenician relatives: if a PP is required within the relative clause and its complement is co-referential with the relative head, a resumptive pronoun is required. Similarly, none of the Canaanite examples presented so far or listed in the footnotes exhibits pied-piping—that is, the raising of a preposition from within the relative clause along with the relative head. What, then, do we do with the rare Canaanite example (in this case, the only example in Phoenician) of pied-piping given in (630)? (630) Karatepe (Azatiwada; eighth–seventh century b .c.e.) bmqmm bʾš kn ʾšm rʿm bʿl ʾgddm ‘in placesi whi in which evil men were ___i gang leaders’    (KAI 26 AI.14–15; // BI.8 // CII.2–3)

12.  Segert (1976: 259) suggests that the Nora Stele (late ninth century b.c.e., Sardinian; KAI 46) contains an instance of subject resumption in a relative clause. His reading of lines 2–3, which follows KAI, is at odds with the readings of Peckham 1972 and Cross 1972. Based on my own reading of Peckham’s 1972 photograph as well as the high resolution West Semitic Research Project images, I concur with the readings and interpretation of Peckham 1972 and Cross 1972 against that of Segert 1976 and KAI: there is no relative clause and therefore no subject resumption in lines 2–3 of the text. (I am grateful to Brian Peckham for the use of his photograph of the Nora Stele.)

The Semitic Relative

263

Although my English translation is acceptable, it reflects a phenomenon that is used nowhere else in Canaanite epigraphic texts. To repeat the context for this constraint: even in those Semitic languages that exhibit pied-piping, it is extremely rare (e.g., four possible occurrences in Biblical Hebrew; see above, §5.2.2), a notably unexpected diachronic or dialectal development (e.g., Maghrebine and Judeo-Arabic; see Blau 1999: 62), or likely the result of language contact (e.g., pied-piping, preposition-stranding, and prenominal relatives in Geʿez may reflect the Cushitic [SOV] substratum). So also with the Karatepe example: it seems likely that the bilingual nature of the inscription (Phoenician and Luwian) may have been the catalyst for pied-piping in the Phoenician version of the text. Consider the Luwian version (631) of the Phoenician clause above in (630). (631) Luwian Version of Karatepe (Azatiwada; eighth–seventh cent. b.c.e.) (“FINES”) i+ra/i-há-za (MALUS)á-tu-wa/i-ri+i-zi-wa/i-ta | frontier.dat.pl evil.nom.pl-connective-local CAPUT-ti-zi | REL-ta-na | a-ta á-sa-ta u-sa-li-[zi] head.nom.pl that in be.3pl.past robbers.nom.pl ‘frontiersi that evil men were ini’ (Hawkins 2000: 51, §§19–20) Notice that the Luwian relative clause allows preposition-stranding, unlike Phoenician. The different syntax of relatives, including the syntax of adpositions, raises the possibility that Luwian relativization strategies influenced the syntax of the Phoenician text, particularly concerning the relative-clause piedpiping in (630) (on other features suggesting Luwian influence, if not a Luwian base-text, see Younger 1998). 13 8.5. Aramaic Aramaic 14 is the best attested of the Northwest Semitic languages, with linguistic evidence from the early first millennium b.c.e. to still-spoken modern dialects (see Kaufmann 1997; Creason 2004 for overviews of Aramaic). As with my discussion of Akkadian and Phoenician, for this survey I will consider data from the earlier stages of Aramaic: “Old Aramaic” (ninth century through 612 b.c.e.) and “Imperial Aramaic” (612–ca. 200 b.c.e.). Early 13.  Although our understanding of the grammatical system of Luwian remains rudimentary, from the available evidence it appears to be an SOV language, with an internallyheaded primary relativization strategy, although an externally-headed option exists. Moreover, Luwian adpositions, which are normally postpositions but may also be prepositions, are not strongly bound to their host NP. Thus, if the relative element is fronted, the adposition may move with it, resulting in pied-piping. On Luwian grammar, see Marangozis 2003; and Payne 2004; 2010. I thank Petra Goedegebuure for discussing the Luwian relative clause and the Luwian version of the Karatepe text with me. 14. See Degen 1969; Segert 1975; Kaufman 1997; Hug 1993; Folmer 1995; Creason 2004; Muraoka and Porten 2003 (Egyptian Aramaic); Nöldeke 1904 (Syriac)

264

Chapter 8

Aramaic relatives demonstrate no salient differences from what we have seen so far in the contemporaneous Northwest Semitic languages, whether in terms of the types of relative heads and relative elements or in terms of the syntax and semantics of the relative clause itself. The examples in (632)–(636) are thus representative of the majority of extant Aramaic relatives from the Old and Imperial stages. (632) Bar-Hadad (ninth century b.c.e.) nṣbʾ zy šm brhdd br[—] mlk ʾrm lmrʾh lmlqrt ‘The stelei whi that Bar-Hadad, son of [—] king of Aram, raised ___i for his lord, Melqart’ (KAI 201.1–3) (633) Zakkur (eighth century b.c.e.) [n]ṣbʾ zy šm zkr mlk [ḥ]mt wlʿš lʾlwr [mrʾh] ‘The stelei whi that Zakkur, king of Hamath and Luʾath, set up ___i for Ilwer, [his lord]’ (KAI 202 A.1) (634) Sefire (eighth century b.c.e.) wʿm ʾrm klh wʿm kl mṣr wʿm bnwh zy ysqn bʾšr[h] ‘with all Aram and with Musr and with his sonsi whi that ___i will come after [him]’ (KAI 222 A1.5) (635) Panamuwa I (eighth century b.c.e.) ʾnk pnmw br qrl mlk yʾdy zy hqmt nṣb zn lhdd ‘I am Panamuwai, son of Qarli, king of Yaudi, whi that ___i have raised this statue for Hadad’ (KAI 214.1) (636) Sefire (eighth century b.c.e.) bywm zy yʿb dkn ‘on the dayi whi that he does so ___i’ (KAI 222 A3.20) All four Old Aramaic examples, (632)–(636), exhibit external heads, postnominal relatives, an indeclinable relative complementizer, and syntactic gaps within the relatives (representing the promotion of the relative head). In (632) and (633) the gap is in the complement position, in (634) and (635) it is in the nominative subject position, and in (636) the gap is in a (temporal) adjunct position. Additionally, the relatives in (632)–(634) and (636) are restrictive in nature, while the relative clause in (635), since it modifies a proper noun (a unique referent), 15 is nonrestrictive. There are only a few zero-relatives in the Old Aramaic texts, such as the two in (637). 15.  This assumes, of course, that the ostensible author, Panamuwa, was not distinguishing himself from another Panamuwa.

The Semitic Relative

265

(637) Nerab 2 (ninth–eighth century b.c.e.) a. bywm mtt pmy lʾ tʾḥz mn mln ‘on the dayi whi (that) I died ___i, my mouth was not closed to words’ (KAI 226 1.4) b. mn ʾt tʿšq wthnsny ‘whoever are youi whi (that) ___i do wrong and drag me away’    (KAI 226 1.8–9) The content of the relatives in (637), both of which are restrictive, corroborates the link between zero-relativization and restrictive semantics of the relative clause. As for overt resumption, it is used very rarely in Old Aramaic texts, with one subject example in Sefire (638), one NP-internal possessive example in the Tell Fekhariyeh text (639), and a PP example in the Bar-Hadad text (640). (638) Sefire (eighth century b.c.e.) kl zy rḥm hʾ ly ‘anyonei whi that a friend is hei to me’ (KAI 224.8) (639) Tell Fekhariyeh (ninth century b.c.e.) ʾlh rḥmn zy tṣlwth ṭbh ysb ‘merciful godi, whi that hisi prayers are sweet’ (KAI 309.5;   Kaufman 1982) (640) Bar-Hadad (ninth century b.c.e.) mlqrt zy nzr lh ‘Melqarti, whi that he had made a vow to himi’ (KAI 201.4–5) With the Imperial Aramaic corpus, particularly the texts from Egypt, we admit a wealth of data to our knowledge of first-millennium b.c.e. Aramaic grammar (see Folmer 1995; and Muraoka and Porten 2003); however, few features of relativization change. The examples in (641) illustrate the range of overtly-headed relatives. (641) Overtly-Headed Relatives a. ʾgwrʾ zy yhw ‘the templei whi that ___i (is of/belongs to) Yhw’   (TAD II.3.12:18–19) b. yhh ʾlhʾ zy byb byrtʾ ‘Yhh, the Godi whi that ___i (is) in YB’ (TAD II.3.3:2) c. kspʾ zy hwh bydy ‘the silveri whi that ___i was in my hand’ (TAD I.2.2:4–5)

266

Chapter 8 d. ktnh zy ʾwšrty ly ‘the tunici whi that you sent ___i to me’ (TAD I.2.1:4) e. ʿd ywm zy ʾšlmnhy lk ‘on (the) dayi whi that I pay it to you ___i’ (TAD II.4.2:10) f. zk bytʾ zy tḥwmwhy ktybn mnʿl ‘that housei whi that itsi boundaries are recorded above’   (TAD II.2.10:8) g. gbr zy tzbnwn lh bytʾ zk ‘a mani whi that you will sell that house to himi’   (TAD II.2.10:11) h. yrḥʾ zy lʾ ʾntn lk bh mrbyt ‘the monthi whi that I will not give you interesting in iti’   (TAD II.4.2:4)

The first five relatives, in (641a–e), illustrate relativization with a gap at the subject (641a–c), complement (641c), and adjunct (temporal) (641e) positions. In contrast, the relatives in (641f–h) exhibit overt resumption but only at the NP-internal (641f), prepositional complement (indirect object PP in [641g]) and adjunct (temporal PP in [641h]) positions. The relative clause in (642) is the only case of resumption in the verbal complement position that I have found (and the only one that is cited in Muraoka and Porten [2003: 168]). (642) Complement Resumption ʾgrʾ zy hnpʾ zy bnhw mṣryʾ ‘the walli whi that protects, whi that the Egyptians built iti’   (TAD II.3.10:8–9) This example even has a textual difficulty (the form of the verb with suffix, which Muraoka and Porten suggest emending to bnwh [2003: 169 n. 768; also see p. 148 n. 694]). The sole, questionable use of resumption in the verbal complement position in (642) (I have not found one example in the subject position) confirms that even within Imperial Aramaic (at least within its Egyptian form) the syntactic change whereby resumption at the complement (and then subject) positions within the relative clause had only begun to occur and had not yet become widely diffuse, as we see in later stages, such as in Classical Syriac (643) (see Nöldeke 1904: 278–90; on resumption in Biblical Aramaic, see Naudé 1996). (643) Subject Resumption in Classical Syriac məlɛk mɔdɔy wəpɔrɛs dəhuw daryɔwɔš ‘the kingi of Media and Persia, whi that hei (is) Darius’   (Aphraates 83.5)

The Semitic Relative

267

Finally, Imperial Aramaic provides us with adequate data on null-head relatives (644) and zero-marked relatives (645) to allow us to make at least one significant observation. (644) Null-Head Relatives a. zy tʿbdwn lh lʾ ytkswn mn ʿnny ‘∅(the thing)i whi that [= what] you do ___i will not be hidden from Anani’ (TAD IV.4.3:10–11) b. mn hw zy yqwm qdmwhy lhn zy ʾl ʿmh ‘who is he that can stand before him but ∅(the one)i whi that El is with himi’ (TAD III.1.1:91) (645) Zero-Marked Relatives a. yhw ʾlhʾ byb byrtʾ ‘Yhw, the Godi whi (that) ___i (is) in YB’ (TAD II.3.11:2) b. ʾyš špyr mddh wlbbh ṭb ‘A mani whi (that) hisi stature is beautiful and ∅(who) his heart is good’ (TAD III.1.1:95) Notably, each one of the relatives in (644) and (645) is semantically restrictive. Thus, contrary to the assertion of Muraoka and Porten (2003: 169), we may say that early Aramaic, as with each of the other languages in this survey, does in fact have a formal strategy by which restrictive relatives are marked: if a relative clause has a null head or if it is zero-marked, it is restrictive. 8.6. Ugaritic Ugaritic texts 16 date to the late thirteenth and early twelfth centuries b.c.e. and thus predate any other NWS languages covered in this survey. At the same time, the corpus of Ugaritic texts is considerable, numbering approximately 2,000 (Pardee 2011). Thus, as we will see below, most of the features of Ugaritic relatives mirror what we have already witnessed. However, there are two features that deserve special attention: the form of the relative elements and the use of resumption. First, consider the relative elements in Ugaritic. It appears that Ugaritic had two relative elements: a relative marker d/dt, which carried agreement features matching the gender, number, and case of the relative head in the matrix clause, and a relative complementizer d, which carried no agreement features.

16.  See Gordon 1965; Segert 1984; Sivan 1997; Tropper 2000; Bordreuil and Pardee 2004, 2009; Pardee 2011.

268

Chapter 8

Table 8.2.  The Ugaritic Relative Marker (First View) Singular Nom Gen Acc

Plural Nom Obl

Masc







Fem

dātu

dāti

dāta

dūtu

dūti

Syllabic evidence suggests the reconstruction in table 8.2 for the declinable relative (Tropper 2000: 235): 17 In contrast to two apparent “systems” coexisting (or, as Pardee puts is, “the coexistence of a system and an absence of system”; 2003–4: 137), Pardee suggests a single paradigm: A simple explanation, but one that results in a rather high incidence of identical forms, would say that the original forms /dāti/ (feminine singular genitive) and /dūti/ (plural oblique) were confused with /dā/ and /dū/ + the enclitic particle /-ti/; the other forms in the paradigm would as a result of this re-analysis have been dropped while perception of /-ti/ as the particle would have permitted that element to be dropped or retained as a matter of style. (Pardee 200–4: 138)

Pardee suggests the paradigm in table 8.3 as a way to harmonize what only appears to be two systems for the relative marker in Ugaritic (Pardee 2003–4: 138; 2011: 464). There is good evidence for accepting Tropper’s two-system description over Pardee’s single-system, however. Deutscher (2001; 2002) makes a very strong case for the inherent instability of the relative-marker system in Old Akkadian—i.e., the system in which the relative marker carries agreement features that match those of the head noun in the matrix clause rather than the head’s role in the relative clause. As I have indicated, such relative markers must be distinguished from the type of relative pronouns used in many IndoEuropean languages. Between Old Akkadian and later stages of Akkadian, the inflected relative marker shifted to an uninflected relative complementizer. I suggest that what we see in Ugaritic is a language in the process of the same shift. The snapshot of the Ugaritic language that we have preserved from the late thirteenth century just happens to exhibit the relative element as it shifted systems. Interestingly, we see the same shift in process in the Geʿez corpus (see Dillmann 1907; Lambdin 1978) and, perhaps with more research on Epigraphic South Arabian (particularly with the minuscule texts), we will find the same shift to have occurred in that group of languages. 17.  It is impossible to determine the vocalization of the indeclinable relative word. Tropper suggests /dā̆/ (Tropper 2000: 236), although it is unclear if there is any the evidence for this.

269

The Semitic Relative

Table 8.3.  The Ugaritic Relative Marker (Second View) Masc Sg /dū/, /dā/, /dī/

Fem Sg /dā(ti)/

Common Pl /dū(ti)/

Syntactically, Ugaritic relatives are postnominal and modify externallyplaced heads, as all the examples below will show. They can also have either overt (646) or null (647) heads, can be zero-marked (648), and employ overt resumption, but only in NP-internal (649a) and PP complement (649b) positions. (646) Ugaritic Overtly-Headed Relatives a. ʾarbʿ ḥršm dt tbʿln b pḫn ‘four craftsmeni whoi ___i are working in PḪN’   (KTU 4.141.3.5–6; Sivan 1997: 218) b. ʾabn brq d l tdʿ šmm ‘hail stonesi whichi the heavens have not known ___i’   (KTU 1.3.3.26) c. nʿl ʾil d qblbl ‘a divine palanquini whichi carrying-poles (belong to) ___i’   (KTU 1.4.1.36) (647) Ugaritic Null-Headed Relatives a. p d ʾin b bty ‘and ∅(the thing)i whichi ___i is not in my house’   (KTU 1.14.3.38) b. d b ḥlmy ʾil ytn ‘∅(the woman)i whichi in my dream El gave ___i’   (KTU 1.14.3.46) (648) Ugaritic Zero-Relative a. k ʾirby tškn šd ‘like the locustsi whi ___i dwell on the steppe’   (KTU 1.14.2.50–51) b. y bn ʾašld ‘O sonsi whi I have begotten ___i’ (KTU 1.23.65)

270

Chapter 8

(649) Ugaritic Relative-Clause Resumption a. tn ly mṯt ḥry nʿmt špḥ bkrk d k nʿm ʿnt nʿmh ‘Give me lady HRYi, the fair, your first begotten, whoi like the fairness of ʿNT is heri fairness’ (KTU 1.14.3.39–41) b. [d]nʾil mt rpʾi ʾanḫ ǵzr mt hrnmy d ʾin bn lh km ʾaḫh w šrš km ʾaryh ʿDNʾLi, man of RPʾ, the sigh of the hero, man of HRNMY, whoi there is no son to himi like his brothers nor a scion like his kinsmen’ (KTU 1.17.1.17–19) The three relatives in (646) demonstrate the plural agreement of the relative marker (646a) as well as gaps at the subject (646a), complement (646b), and adjunct (646c) positions within the relative clause. The example in (646c) is particularly interesting in that the lack of resumption results in a one-part nullcopula clause in which the relationship of the noun qblbl ‘carrying-poles’ to the relative head is left unexpressed; in the other NWS languages surveyed above—and also example (649b) for Ugaritic—the normal strategy would demand the additional presence of a PP, such as lh ‘for, (belonging) to him/it’. This PP would not only provide a possible position for overt resumption of the head; more importantly, it would provide a second constituent by which the predication within the NP would be clarified. The two null-head relatives in (647) demonstrate gaps at both the subject (647a) and complement (647b) positions, as do the two zero-marked relatives in (648a) and (648b), respectively. Note, too, that the zero-relatives are semantically restrictive. Finally, the two relatives in (649) illustrate the only types of overt resumption that Ugaritic apparently employed, in the NP-internal (649a) and prepositional complement (649b) positions. 8.7.  Classical Arabic 18 The Semitic languages that emerged from the northern Arabian Peninsula can be divided into two groups: Arabic and Ancient North Arabian. The latter consists of epigraphic evidence dating from the sixth century b.c.e. to the fourth century c.e. and is further divided into a number of dialects corresponding to regions (see MacDonald 2004). Although these dialects are often identified as the predecessor to Old Arabic and its classical and modern forms, it is also possible that Ancient North Arabian was a sister language to Old Arabic rather than its mother language (MacDonald 2004: 488). Old Arabic is the term of the pre-Islamic poetic texts, out of which the language of the Koran developed in the seventh century c.e.: 18. See Wright 1898; Fischer 1997; 2002.

271

The Semitic Relative

Table 8.4.  The Classical Arabic “Determinative-Relative” Marker singular masc

nom gen acc

fem

nom gen acc

dual

ḏū ḏī ḏā

nom

ḏātu ḏāti ḏāta

nom

obl

obl

plural

ḏāwā ḏāway

nom

ḏ(aw)ātā ḏ(aw)ātay

nom

obl

obl

ḏawū ḏawī ḏawātu ḏawāti

Classical Arabic is the literary language of Islam. Inscriptions from the fourth century ce reveal a dialect or dialects similar to the later classical language, but the latter, reflecting the spoken language of the Hejaz region of central western Arabia, truly began with the Koran in the seventh century. The language was standardized by grammarians during the eighth and ninth centuries, with features of the early poetic koine, and has remained essentially unchanged since. (Huehnergard 1995: 2121)

The writing system of Arabic is a borrowing and modification (by the addition of six additional consonantal graphemes) of the Aramaic script of the Nabateans. In the eighth century c.e., a system of diacritic notation was adopted from Syriac for indicating vowels and other phonological information (e.g., gemination). Turning to the relative-clause data—pre-Classical Arabic manifests a relative complementizer ḏū, which carries no agreement features, as in (650). (650) Pre-Classical Arabic Relative Complementizer a. faḥasbiya min ḏūʿindahum mā kafāniyā then-sufficiency.ms-me from rel with-them what compensateme.ms ‘then enough for me of that which is with them is what suffices me’ (Wright 1898: 1.272) b. wabiʾrī ḏū ḥafartu waḏū ṭawaytu and-well-my rel dug.pfv.1cs and-rel line.pfs.1cs ‘and my well which I dug and which I lined (or cased)’   (Wright 1898: 1.272) However, Wright indicates that there is evidence suggesting that a fully inflected version existed in the pre-Classical form. This seems to be supported by the fully inflected “determinative pronoun” (table 8.4) used in Classical Arabic to express ‘one of, he of, she of’ (Wright 1898: 1.272–73; also Pennacchietti 1968; Lipiński 2001: 333; Huehnergard 2006: 112; Goldenberg 2014: 105).

272

Chapter 8

Table 8.5.  The Classical Arabic Relative Marker singular masc

ʾallaḏī

dual nom obl

fem

ʾallātī

nom obl

plural

ʾallaḏāni ʾallaḏayni

ʾallaḏīna

ʾallatāni ʾallatayni

ʾall(aw)āti

As with the similar construction in Epigraphic South Arabian and Geʿez (see below), I suggest that this is best understood as a null-head relative clause, ‘the one who [is or has the following attribute]’, as in (651). (651) falā waḏū baytuhu fissamaaʾi then-no and-rel house-his in-heaven ‘No! by Him whose residence is in heaven’ (Wright 1898: 1.272) The ḏ-series relative element was replaced by a compound form (table 8.5) that functions as what de Vries (2002: 171–75) calls a relative marker, in that it exhibits some agreement with the relative head but does not consistently correspond to the syntactic role/case of the position from which the head was promoted within the relative. 19 The syntax of Classical Arabic relative clause mirrors the patterns exhibited by all the languages surveyed so far. Arabic relatives are post-nominal and modify externally-placed heads; they can also have either overt or null heads, be zero-marked, and employ resumption, obligatory at the “lower” NPAH positions and optional at the accusative position (see Wright 1898: 2.317–24; Fischer 2002: 216–20). 8.8.  Epigraphic South Arabian (ESA) The ancient languages of southwest Arabia, particularly the eastern edges of the central highlands of Yemen, are known together as old or Epigraphic South Arabian. 20 Four dialects have been identified and named after the people groups for whom they were the primary languages: Sabaic, Minaic, Qatabanic, and Hadramitic. The best attested dialect of ESA is Sabaic, in which inscriptions date from the eighth century b.c.e. to the sixth century c.e. Epigraphic texts are also available in Minaic, Qatabanic, and Hadramitic, although there 19. De Vries notes that the origin of some relative markers may be pronouns that have undergone “attraction” of (in the case of Arabic) case features to the matrix case of the relative head rather than the subordinate case of the relative gap (de Vries 2002: 171–72). This is essentially the same explanation that Wright offers when he points out the similarity of some of the Classical Arabic examples to Greek and older forms of German (Wright 1898: 175) 20. See Beeston 1962; 1984; Kogan and Korotayev 1996; Nebes and Stein 2004.

273

The Semitic Relative

Table 8.6.  The Epigraphic South Arabian Relative Marker Singular

Dual

Plural

Masculine

ḏ-

ḏy-

ʾl (Early), ʾlw (Middle Nom.), ʾly (Middle Acc.), ʾlht (Late)

Feminine

ḏt, t- (Late)

ḏty

ʾlt

are fewer than in Sabaic. As is typical within ESA studies, I will concentrate on Sabaic in my overview of the relative clause in ESA, noting data from the other dialects only when linguistically significant. As with many other ancient Semitic languages, the “abjad” writing system used for ESA primarily represents consonants, with only rare hints of vowels by the use of the signs for the glides w and y. Thus, it is unclear whether Sabaic exhibited morphological case-marking on nouns and nominal modifiers, including the elements introducing relatives in ESA. However, what is clear is that, like Ugaritic and Arabic, ESA employed both a relative marker ḏ/ḏt, which carries agreement features matching the gender, number (and perhaps case) of the relative head in the matrix clause, and a relative complementizer ḏ, which carries no agreement features. The attested forms of the relative marker are provided in table 8.6 (Kogan and Korotayev 1996). Examples of both are given in (652) and (653). Note that ESA relatives follow the relative head (i.e., externally-headed and postnominal). (652) ESA Relative Marker a. ṣlm-n ḏ-s2ft-hw statuette.ms-det rel.ms-promise.pfv.3ms-him.ms ‘the statuei whichi he promised ___i him’ (J 664.8) b. nṭʿ-t-hmw tlfm ḏ-t b-fn-w hwr bastion-fs-their.mp TLFM rel-fs (is) in-front-mp.of cistern.ms. mḥfd-n of tower.ms-det ‘their bastioni TLFM whichi ___i is in front of the cistern of the tower’ (C 407.26) c. kl ʾs1d-n w-ʾnṯ-n ʾlw ys1tmy-nn all man.mp-det and-woman.fp-det rel.mp 3-call.ipfv.pass-p  ʾs1lm  ʾs1lm ‘all the men and the womeni whoi ___i are called ʾs1lm’   (F  76.2–3)

274

Chapter 8

(653) ESA Relative Complementizer ʾbdt-m ḏ-kwn-w byn ḥms1nhn irregular-fp-indet rel-be.pfv-3p between army-3du.det ‘irregularsi whi that ___i were between the two armies’ (J 633.7–8) In (652b–c), the relative marker is clearly inflected for gender and number and, although it is formally ambiguous, the relative ḏ in (652a) could be taken as inflected to match the 3ms features of the relative head ṣlmn ‘the statuette’. In contrast, the relative complementizer in (653) clearly lacks agreement features matching those of the 3fp relative head, ʾbdtm ‘irregulars’. Both relative elements were also used widely in relatives that consist of a one-part null-copula clause in which the single overt constituent is typically a noun used to qualify the relative clause head, as in (654). (654) ESA Relative Clause/Periphrastic Genitive Construction a. bkrt-n ḏt ḏhb-n young.camel-fs-det rel-fs bronze-det ‘the young cameli whichi ___i (was) bronze’ (C 579.4–5) b. ṣlm-n ḏḏhb-n statuette.ms-det rel-bronze-det ‘the statuettei whi that ___i (is) bronze’ (C 555.4) c. ʾmṯln ʾly ḏhb-n image-mp-the rel.mp bronze.ms-det ‘the imagesi whichi ___i (are) bronze’ (J 558.2) Additionally, the uninflected complementizer is used to nominalize clauses to be the complements of a preposition or to follow subordinating conjunctions, illustrated in (655). (655) ESA Prepositional Complementizer Use of ḏ a. b-ḏt hws2ʿ ʾlmqh mrʾy-hmw b-s2kr in-rel-fs grant.pfv.3ms ʾLMQH lord-du-their.mp in-defeat.inf w-nqm w-qtl w-hṯlʿn w-hs1ḥtn and-take.revenge.inf and-kill.inf and-subjugate.inf and-rout.inf ḏ-rydn w-mṣr-hw rel-Raydan and-t roop.mp-his.ms ‘in-that ʾLMQH granted to their two lords to defeat, take revenge on, kill, subjugate, and rout the one that is of Raydan and his troops’ (J 2107.8–9 = NAM 429.8–9) b. bʿd ḏ-s1tṣr after rel-summon.aid.pfv.3ms ‘after-that he summoned aid’ (C 314.17)

The Semitic Relative

275

c. ʿdy ḏt ḥmlhmw hgrn until rel-fs drive.pfv.3ms-them.mp city-det ‘until-that he drove them (to) the town’ (Ry 535.9) d. ln ḏ-ʾtw bn mqmn ḏ-lḥgm since rel-return.pfv.3ms from observation.post-det rel  LḤGM ‘since-that he returned from the observation post of LḤGM’ (J 633.4–6) e. l-qbl ḏt s1ʾl-hw ʾlmqh b-ms1ʾl-hw for-because rel-fs ask.pfv.3ms-him.ms ʾLMQH in-oracle.ms  his.ms ‘because-that ʾLMQH asked him in his oracle’ (C 79.1–4) f. ʿln ḏ-ʾl tqrʿ s1lṭ-m on-account.of rel-neg draw.pfv.3ms lots.mp-indet ‘on account-that he had not drawn [the appropriate] lots’   (NNAG 12.7–8) Such constructions are not relatives but are related since the complementizer ḏ nominalizes clauses both for relative and non-relative clauses. The syntax of ḏ in the latter clauses is poorly understood in grammatical sketches of ESA and, as with Hebrew ‫ אשׁר‬and ‫שׁ‬, the ESA relative element ḏ is often, and erroneously, categorized as a subordinating conjunction with many functions. However, all of the ḏ clauses in the ESA texts are easily described as expected uses of the nominalizing function of the complementizer ḏ. Since the relative elements in ESA are not of the pronoun variety, we do occasionally find a resumptive element within the relative clause. Consider the illustrative cases in (656). (656) Resumption in ESA Relatives a. ṣlm-n ḏ-ṣrf-n ḏ-mdlt-hw ʾrbʿ statuette.ms-det rel-silver.ms-det rel-value-fs-its.ms four.   ms mʾn-m w-ʾḥd ʾlf-m rḍy-m hundred-mp-indet and-one.ms thousand.ms-indet coins.   mp-indet ‘the silver statuettei whi that itsi value (is) 1400 coins’ (J 609.4–6) b. ṣlm-n ḏ-ḏ[h]b-n ḏ-b-hw ḥmd statuette.ms-det rel-bronze.ms-det rel-in-it.ms thank. lyḫ w-mqm ʾlmqh pfv.3ms power .ms.of and-might .ms.of ʾLMQH

276

Chapter 8 ‘the bronze statuettei whi that with iti he thanked the power and might of ʾLMQH’ (J 739.4–5) c. w-ʿtb bn s2n ʾl wḍʾt and-destine.pfv.3ms from NS2N rel came.forth.pfv.3ms s2ft-hmw ns1rn ʾlʾlt-n saying-fs-their.mp toward god.mp-det ‘and he [KRBʾL] destined from NS2Ni whi that theiri promise had come forth from the gods’ (R 3945.16) d. mrḍ b-hw ʾtw malady.ms with-it.ms come.pfv.3ms ‘(the) maladyi whi with iti he came back’ (J 642.5)

In the overtly-headed relatives in (656a–b), the resumptive constituent is in the NP-internal and PP positions, respectively. The null-head relative clause in (656c) demonstrates that NP-internal resumption is necessary even when the head is not overt, and the relative clause in (656d) indicates that overt resumption is used in zero-relatives as well as relatives with an overt relative element. Significantly, none of the relative-clause data adduced in Beeston 1962; 1984; Kogan and Korotayev 1996; or Nebes and Stein 2004 exhibit resumption in any other position but NP-internal and PP; that is, I have found no examples of NP complement or subject resumption. Finally, example (657) below shows that ESA relatives may be used for nonrestrictive modification. (657) ESA Nonrestrictive Relative Clause w-ḫmr-hw mrʾ-hw ʾlmqh ḥyw l-hw and-grant.pfv.3ms-him.ms lord.ms-his.ms ʾLMQH live.inf tohim.ms ɣlm-m d-ys1tmy-n mrs1ʿm boy.ms-indet rel-3-be.named-ipfv.ms MRSʿM ‘and his lord ʾLMQH granted him a boyi, whi that ___i is named MRSʿM, to live for him’ (J 655.7–10) A few more examples round out the full variety of ESA relativization: (658) ESA Zero-Marked Relative Clause w-hṯb ʾbḍʿ whbhw mlk and-return.pfv.3ms district.mp.of give.pfv.3ms-him.ms king.ms.of s1bʾ l-ʾlmqh w-l-s1bʾ S1Bʾ to-ʾLMQH and-to-S1Bʾ ‘and he [KRBʾL] returned to ʾLMQH and S1Bʾ the districtsi whi (that) the king of S1Bʾ had given ___i him’ (R 3945.14–15)

The Semitic Relative

277

(659) ESA Null-Headed Relative Clause a. ṯw y-qh-n mlk-n ḏ-y-rḍyn until 3-command-ipfv.ms king-det rel-3-please-ipfv.ms ‘until the king commands ∅(the thing)i whi that ___i would please [him]’ (Ry 507.9) b. ḏ-rydn w-mṣr-hw rel-Raydan and-t roop.mp-his.ms ‘∅(the one)i whi that (is of) Raydan and his troops’ (J 2107.8–9 =   NAM 429.8–9) It is noteworthy that the relative head in the zero-relative in (658) is in the bound state and also that the relative clause is restrictive. This suggests that ESA exhibits a pattern similar to Hebrew: when the head is in the bound state or when the relative clause is zero-marked (or both), the modification is restrictive. The two null-head relatives in (659) show that the relative clause itself may be verbal or verbless, and in the case of the verbless relative clause in (659b) the single constituent within the relative clause may even be a genitive noun (though this does not mean that the complementizer ḏ is itself related to “genitive” modification; see below, §8.10). 8.9.  Classical Ethiopic (Geʿez) Classical Ethiopic, or Geʿez, 21 was the language of the Kingdom of Aksum, which arose toward the end of the second century c.e. on the Ethiopian plateau. Before the earliest attestations of Geʿez, which date to the mid-fourth–late fifth centuries c.e., lie Epigraphic South Arabian texts from the sixth century b.c.e. These earlier Semitic texts presumably represent the linguistic ancestor of Geʿez. Although it has not been possible to derive Ethiopian Semitic directly from the language of the older ESA texts, the typical reconstruction supposes a genetic relationship, albeit complicated by non-Semitic influence: In all likelihood, Ethiopic Semitic evolved out of a South-Arabian-based trade lingua franca, perhaps passing through stages of pidginization and creolization familiar from differentiation and development of language families elsewhere in the world (e.g., Romance). The substratum languages in this development presumably belonged to the Cushitic language family, and a number of important early loanwords from Cushitic are evident in Geʿez—but at present it is not possible to reconstruct the mechanisms of this development. (Gragg 2004: 427–28)

After the Axumite monumental inscriptions of the fourth and fifth centuries c.e., the next primary corpus of texts consists of translations of the Bible and related Christian texts, reflecting the arrival and spread of Christianity within Ethiopia. It is from the Axumite epigraphic texts (taken from Bernand, Drewes, 21. See Dillmann 1907; Lambdin 1978; Gragg 1997; 2004; Tropper 2002.

278

Chapter 8

and Schneider 1991 [RIE]), the biblical texts, and the examples in Lambdin 1978 that the examples for the Geʿez relative clause in this survey are taken. As with Ugaritic, Arabic, and Epigraphic South Arabian, Geʿez witnesses the use of a relative markers inflected for number and gender: zæ (ms), ʾəntæ (fs), and ʾəllæ (cp). (660) Relative Markers a. bəʾsi zæ-tæšayæṭæ bet-əyæ man.ms rel.ms-buy.pfv-3ms house-my ‘the mani whoi ___i bought my house’ (Lambdin 1978: 106) b. bəʾsit ʾəntæ wælædæt wældæ woman-fs rel.fs birth.pfv-3fs child.ms-acc ‘the womani whoi ___i bore the child’ (Lambdin 1978: 106) c. næbiyʾat ʾəllæ tænæbbæyu həyyæ prophet-mp rel.cp prophesy.pfv-3cp there ‘the prophetsi whoi ___i prophesied there’ (Lambdin 1978: 106) Also like Ugaritic and ESA, there is a tendency in Geʿez for the ms zæ to replace the others as a relative complementizer, signaled by the lack of agreement between the relative head and the relative word, as in (661). (661) Relative Complementizer zæ ʾænəstəya zæ-ḫæræyu woman.fp rel.ms-choose.pfv-3mp ‘the womeni whi that they chose ___i’ (Gen 6:2) The examples in (662)–(665) illustrate basic Geʿez relatives, divided into categories based on where the relativized element has been promoted from (subject, NP complement, PP, or NP-internal positions) within the relative clause. (662) Subject Relativization a. bəʾsi zæ-tæšayæṭæ bet-əyæ man.ms rel.ms-buy.pfv-3ms house-my ‘the mani whoi ___i bought my house’ (Lambdin 1978: 106) b. bəʾsit ʾəntæ wælædæt wældæ woman-fs rel.fs birth.pfv-3fs child.ms-acc ‘the womani whoi ___i bore the child’ (Lambdin 1978: 106) c. næbiyʾat ʾəllæ tænæbbæyu həyyæ prophet-mp rel.cp prophesy.pfv-3cp there ‘the prophetsi whoi ___i prophesied there’ (Lambdin 1978: 106)

The Semitic Relative

279

d. ʾəgziʾæ sæmay zæ-wæhæbæ-ni ʾəgziʾæ kwəlu lord.ms-of heaven rel.ms-give.pfv-3ms-me lord-of all ‘the Lord of Heaveni whoi ___i set me (as) lord of all’   (RIE 189.5–6) (663) Complement Relativization a. mədr ʾəntæ wæhæbæ-kæ land.fs rel.fs give.pfv-3ms-you.ms ‘the landi whichi he gave ___i you’ (Josh 1:14) b. ʾæḥzab zæ-ʾæntəmu tətwærræs-əwwomu nation.mp rel.ms-you.mp 2-inherit.ipfv-mp-them.mp ‘the nationsi whichi you will inherit themi’ (Deut 12:2) c. ḥæwarya-næ zæ-fænæw-ku lot-u messenger.ms-our rel.ms-send.pfv-1cs to-him ‘our messengeri whoi I sent ___i to him’ (RIE 189.11–12) (664) PP Relativization a. mədr ʾəntæ bæ-wəstet-a tæwælda land.fs rel.fs in-to-it.fs be.born.pfv-3ms ‘the landi whichi he was born into iti’ (Gen 11:28) b. ʾəgziʾæ sæmay . . . zæ-bot-u ʾæmænku lord-of heaven rel-in-him believe.pfv-1cs ‘the Lord of Heaveni . . ., whoi I believed in himi’ (RIE 189.5–6) (665) NP-internal Relativization a. hægær ʾəntæ səm-a sikar city.fs rel.fs name-her Sychar ‘a cityi whoi itsi name is Sychar’ (John 4:5) b. bəʾsi zæ-ʾi-ḫwællæqwæ lotu ʾəgziʾæbḥer man.ms rel.ms-neg-reckon.pfv-3ms to-him.ms God ḫæṭiʾæt-o sin.ms-his.ms ‘the mani whoi God did not reckon hisi sin’ (Ps 31:2) Geʿez, like the other Semitic languages surveyed so far, allows the use of resumptive elements within the relative clause proper. As with the other languages surveyed in this chapter, the majority of resumptive elements in Geʿez are within a PP, as in (664), or an NP-internal position, as in (665). Geʿez also allows overt resumption of the verbal complement (663b), although less

280

Chapter 8

commonly than the other two resumptive strategies. I have noticed no cases of subject resumption in the Aksumite epigraphic texts (and no grammar that I have consulted mentions this as an option; even so, it is an issue worthy of further investigation). In addition to the typical selection of overtly-headed relatives in (662)–(665), Geʿez also allows null-headed relatives (666) and zero-marked relatives (667). (666) Null-Head Relatives a. məslæ ʾəllæ motu with rel.cp die.pfv-3cp ‘with ∅(those)i whoi ___i died’ (Ruth 1:8) b. zæ-ʾæmnæ yədḫən rel.ms-believe.pfv-3ms 3-be.saved.subj.ms ‘∅(the man)i whoi ___i believes shall be saved’ (Mark 16:16) (667) Zero-Marked Relatives bæ-ʿəlæt ʾi-tæḥæzzæbæ in-day.fs neg-expect.pfv.3ms ‘on a dayi whi (that) he did not expect (him) ___i’ (Matt 24:50) On the use of zero-marked relatives, Dillmann (1907: 528) notes that the relative head may be in the bound form (marked by the affix -æ), as in (668). (668) bæ-mæwāʿəlæ yəkwēnnu mæsāfənt in-day.pl-of ipfv-rule-3mp judge.pl ‘in the daysi whi (that) the judges ruled ___i’ (Ruth 1:1) This use of the bound form of the relative head corresponds to Hebrew, Epigraphic South Arabian, and Akkadian (below); moreover, as in the other languages that allow relative head cliticization, it marks the relative clause unambiguously as semantically restrictive. Contrary to constraints in all other ancient Semitic languages, Geʿez allows both pied-piping and preposition-stranding, as in (669) and (670), respectively (Dillmann 1907: 532–33; Lambdin 1978: 106). (669) Pied-Piping in Geʿez a. kwəllu wəstæ zæ-wædqæ bædn-omu all-it.ms in rel.ms-fall.pfv.3ms corpse.ms-their.mp ‘everythingi upon whichi their corpse had fallen ___i’   (Lev 11:32) b. mæwaʿel-ihu bæ-zæ ʾæstærʾæy-omu kokæb time-its.ms in-rel appear.pfv.3ms-them.mp star ‘its timei in thati the star appeared to them ___i’ (Matt 2:7)

The Semitic Relative

281

c. wæ-b-o læ-zæ-wæhæb-o ḫæməst-æ and-in-him.ms to-rel.ms-give.pfv.3ms-him.ms five.ms-acc mæklit-æ talent.ms-acc ‘and onei to whomi he gave himi five talents’ (Matt 25:15) (670) Preposition-Stranding in Geʿez mədr ʾəntæ ḫaba mæṣaʾnæ land.fs rel.fs to come.pfv-1cp ‘the landi whichi we came to ___i’ (Gen 47:4) Given the presumed Cushitic substratum of Geʿez (see above), it is possible that these very “un-Semitic” relative clause features reflect non-Semitic influence ( just as I suggested above, in §8.4, for the singular example of piedpiping in the Phoenician text from Karatepe). Similarly, the option to place relatives in Geʿez before the relative head, i.e., prenominal instead of the typical Semitic pronominal position, as in (671), must also be investigated as a result of a Cushitic substratum influence or perhaps language contact (see Dillmann 1907: 532–36; Gragg 2004: 427). (671) Prenominal Relative Clauses in Geʿez a. wæʾælbo həyyæ zæ-yəsætti māyæ and-neg.exst there rel.ms-drink.ipfv-3ms water.ms-acc ‘and there was there no wateri whichi one could drink ___i’   (Exod 17:1) b. ʾi-təʾrəyu zæ-wædqæ ʾəkəl neg-2-gather.subj-mp rel.ms-fall.pfv-3ms corn.ms ‘do not gather up the corni whichi ___i has fallen’ (Lev 19:9) c. wəstæ ʾəntæ boʾ-kəmu hægær in rel.fs enter.pfv-2mp city.fs ‘in the cityi whichi you entered ___i’ (Matt 10:11) 8.10.  Summary and Final Matters This survey has, above all, illustrated that the features of Hebrew relativization that I have identified in this work are shared throughout Semitic. And vice versa, my investigation of the Hebrew patterns suggests that there are three features of Semitic relativization that must be presented with greater clarity and linguistic nuance in the appropriate language-specific grammars or comparative works. First, the early Semitic languages used both restrictive and nonrestrictive relatives and, where there are data of zero-relatives, they corroborate the

282

Chapter 8

typological universal that this relativization strategy is semantically restrictive. Second, in the earliest material for each of the languages covered (for which chronological stages can be determined), overt resumption occurs only in the NP-internal and PP positions and appears to be grammatically obligatory. Resumption at the NP complement and subject positions within the relative clause begins to appear only in the later stages of the languages. It therefore appears that resumption for positions farther left on the NPAH was a development for Semitic languages only in the second half of the first millennium b.c.e. As such, there is potential in using this feature to refine diachronic analyses of the individual languages, including the complex ancient Hebrew corpus. For example, the possible diachronic change in the use of resumption I suggested for Canaanite—that is, the move up on the NPAH over time—is circumstantially corroborated by what we find in Ugaritic, Aramaic, and Ethiopic. In the late second millennium, Ugaritic allows resumptive pronouns but only in genitive/NP-internal (KTU 1.14.3.39–41) and oblique (KTU 1.17.1.17–19) positions. Given the number of relatives available in published Ugaritic texts ( just under 700), the lack of resumption higher than the oblique position on the NPAH supports the sketch I have offered: for the NWS languages of the Levant, this expansion of resumption did not occur until the second half of the first millennium b.c.e. In the Old Aramaic texts of the first millennium b.c.e., resumption is used very rarely, with one genitive/NP-internal example in the Tell Fekhariyeh text (KAI 309.5) and one oblique PP example in the Bar-Hadad text (KAI 201.4–5). Similarly, the wealth of Imperial Aramaic data lacks even one example of subject resumption and provides us with only one questionable example of object resumption (B3.10.8–9). The use of resumption higher on the NPAH is, however, clear in forms of Aramaic from later in the first millennium b.c.e. and onward. It appears in the fourth-century b.c.e. Imperial Aramaic of the book of Ezra (e.g., subject resumption in Ezra 4:9; 6:15), the second-century b.c.e. Middle Aramaic of the book of Daniel (e.g., object resumption in Dan 4:27; see Naudé 1996), and Classical Syriac (Noldeke 1904: 278–90). Finally, Ethiopic exhibits a change in the use of resumption between the classical form of the language, Geʿez, and modern Ethiopic languages, such as Tigrinya and Amharic. Specifically, whereas Geʿez prohibits the use of resumption in the object position within the relative clause, Tigrinya and Amharic allow it (Hailu Fulass 1983). While this diachronic development occurs much later than in the NWS languages, it provides further support that such a development appears to be a natural one in Semitic languages in general. The third and final relative-clause feature that deserves comment is the shift in Semitic from using an inflected relative marker, which provides redundant information (in contrast to actual relative pronouns). Thus, against Rosén 1959

The Semitic Relative

283

and almost every other description of the relative clause in NWS languages, we should immediately dispense with the label relative pronouns. The use of pronoun for any of the Semitic relative elements at any stage of development reflects the inaccurate importation of Indo-European grammatical terminology into Semitic studies. We have relative markers, relative complementizers, and resumptive pronouns but not relative pronouns. 22 The correct classification of the respective Semitic relative elements has long been recognized in, for instance, Akkadian studies (see Ravn 1941) and Ethiopic studies (see Palmer 1962; Hailu 1983): The term “relative particle” and not “relative pronoun” has been used to refer to the element zə-. The choice of terminology results from the fact that zə- indicates that a clause is a relative clause, but it does not in any way mark the kind of distinctions that are shown by relative pronouns, such as English “who”, “whom”, “whose”. Distinctions of this kind, which are here called “referential relations” are marked in Tigrinya by the concord of certain elements within the relative clause with the noun it modifies. (Palmer 1962: 38–39)

Additionally, we see the shift from inflected relative marker to uninflected relative complementizer occurring in the late third millennium/early second millennium b.c.e. in Old Akkadian (East Semitic), late second/early first millennium b.c.e. in the NWS languages and Epigraphic South Arabian (Central Semitic), and late first millennium c.e. in Arabic (Central Semitic) and Geʿez 22.  Lipiński’s (2001) description of the relative in Semitic starts well, with the statement that “Semitic languages do not have any real relative pronoun” (2001: 533), but he then later comments that ‫ אשׁר‬in Hebrew became “a generalized ‘relative pronoun’” (2001: 535), thereby obscuring the typological distinction between the different types of relative elements. Similarly, Goldenberg (2014) persists in calling the Semitic relative elements “pronouns (2014: 104–6). The use of pronoun for Semitic relative elements was excusable before linguistic typology provided an adequate set of terms, such as I provide above in fig. 3.1 and table 3.1. There has been little excuse for at least the last 20 years, and yet pronoun continues to be used in, for example, Biblical Hebrew studies: There is a fundamental difference between the Hebrew relative pronoun and its counterpart in the major Indo-European languages. The former is unchangeable and its main function is to signal that what follows it is an attribute, mostly in clause form, qualifying the preceding antecedent. . . . By contrast, in Indo-European relative clauses the relative pronoun is an integral and essential part of the relative clause, and it often displays a variety of morphologically inflected forms, the choice of which depends on the grammatical category or categories of the antecedent on the one hand, and on the syntactic status of the relative pronoun within the relative clause on the other. (GBH 558)

Much of Muraoka’s description is reasonably accurate (the degree of accuracy depends on one’s linguistic framework), but it also illustrates the confusion that a good typology of relative elements can clear up. The Hebrew relative element is not a pronoun in any formal sense of the term, and the continued use of this term in recent grammatical works perpetuates inaccurate linguistic description.

284

Chapter 8

(Southwest Semitic). 23 We can conclude from this temporal disparity that each language branch or subbranch experienced this process independently; in other words, it is unlikely that the shift in Akkadian was the catalyst for the shift in Ugaritic nearly a millennium later, and it is similarly unlikely that the shift in Ugaritic was the catalyst for the shift in Geʿez nearly two millennia later. Two more conclusions proceed from this conclusion: (1) these data confirm that comparative Semitists are correct in reconstructing the inflected relative marker system for proto-Semitic, and (2) the break between East and West Semitic was before the late third millennium. Neither reconstruction has recently been in doubt, to be sure, but further support is always valuable. Related to the issue of the linguistically accurate classification of the relative word in the early Semitic languages is the common association of the relative word with a “genitive” function. Thus, a wide geographic and chronological range of divine epithets are cited to support this view: Canaanite ḏt bṯn ‘the Serpent Lady’, as an epithet of Asherah, El’s consort; the phrase ḏ ṯb ‘the Merciful One’; Ugaritic dū paʾidi ‘the Compassionate One’; Assyro-Babylonian ši atānim ‘of the [man] of the she-ass’; Nabatean ḏū-Šara [Gk. Dusares] ‘the [god] of [mount] Šara’; Liḥyānite ḏū-Ġābat ‘the [god] of the Thicket; Classical Arabic ḏū l-qarnayn ‘the [man] of two horns’; Colloquial Arabic ḏū ʿilm ‘the [man] of learning’; Sabaic ḏāt-ḥamīm ‘the [goddess] of the heat’; and Hebrew ‫( זֶ ה ִסינַ י‬e.g., ‘the one of Sinai’; Cross 1962: 238–40; Lipiński 2001: 334; on the Hebrew ‫זֶ ה ִסינַ י‬, see Holmstedt 2014a). In a compelling argument concerning the origin of subordination, Deutscher (2009) points to this very structure in Old Akkadian, such as, šūt in TU.RA uḫirū-n līḫuz ‘he should take those.of/šūt (who) were delayed in illness’ (Deutscher 2009: 209, with slight modification), in which the šūt demonstrative is the East Semitic reflex of the West Semitic ḏ- demonstrative-cumrelative. In this example, the demonstrative šūt is the head of a zero-relative; moreover, it is bound to the zero-relative that modifies it (which, as I have shown, is a common Semitic structure). Deutscher then argues that relative clauses formed using the š-demonstratives, such as dīn-um šu idīn-u ‘the judgment that he rendered’, reflect the integration of an appositional structure, in which a noun (e.g., dīn-um) is modified by demonstratively-headed relative (e.g., šu idīn-u)—that is, ‘the judgment, that one.of (that) he rendered’. From this apposition structure, the “demonstrative pronoun was degraded from an 23.  A much later case of the shift from relative marker to relative complementizer occurs in Arabic—and not once but twice. The first case concerns the shift of the relative ḏ-, which is inflected in the pre-Classical poetic texts of one tribe but not others (see Wright 1898: 1.272–73; Huehnergard 2006: 112–13). The second case concerns the Classical relative marker ʾallaḏī, which often lacks agreement features in Middle Arabic (Blau 1961: 235–37; 1999: 87–88; 2002: 55).

The Semitic Relative

285

independent head of a relative to a mere marker of the onset of a relative” (Deutscher 2009: 209). Deutscher’s cogent analysis can be directly applied to most of the Semitic parallels adduced as examples of the “genitive” use of the relative word. In each of these demonstrative-noun constructions, the analytical options are either that the demonstrative retains its demonstrative force (and is bound to and so modified by a zero-relative, e.g., ḏū-Šara ‘that one of Šara’ or ‘that one [who is] of Šara’) or has become a relative word modifying a null head (e.g., ‘[the one] who is of Šara’). In either case, the demonstrative/relative element itself is not a genitive marker (contra, e.g., Allegro 1955; Pat-El 2010); rather, the “genitive” nature of the relationship is a product of the bound/cliticization construction in Semitic. In fact, calling this a “genitive” at all, especially for the Semitic languages that lost case-marking, makes the notion of “genitive” vacuous, at least with regard to its typical definitions. 24 24.  The mistaken (in my opinion) use of genitive for Semitic languages is one of the reasons that I have interacted very little with Goldenberg 2014, which I finally acquired at the very end of this project. The data are the same, but the analytical framework and descriptive language differ too greatly from my own—indeed, from those of most current linguistic theories and especially linguistic typology. The cost of translating Goldenberg’s analyses into more-common language simply to interact with it far outweighs any benefit.

Appendix A

Relative Clause Data This appendix contains the exhaustive lists of ‫שׁ‬‎-relatives, ‫שׁ‬-complement clauses, ‫ז‬‎-relatives, ‫מ‬‎-relatives, ‫ה‬-relatives, and zero-relative clauses that I identified in the Hebrew Bible and used for my analysis. For ‫ה‬-relatives, the list includes only nonparticipial and nonadjectival examples, since ‫ה‬ + participle and ‫ה‬ + adjective examples are simply too numerous to include. Additionally, though I obviously cannot list all ‫אשׁר‬-relatives, I do include here a list of all those ‫ אשׁר‬clauses I take to be complement clauses as well as those ‫אשׁר‬-relatives that are extraposed. For each data set, I include the Hebrew (the shaded-gray font indicates the beginning of the relative clause, for easier identification) and an explicit translation, in which I include co-indexation (subscript i, j, k, etc.) to mark the head and resumption (overt or covert/gap). Note that the resulting English translation is sometimes awkward; I ask the reader’s patience on this matter, since this work is about Hebrew syntax and not translation theory or practice. Finally, when it does not affect the identification or interpretation of the relative example, I pare the surrounding verse to the most relevant information. The data sets are presented in the following order: 1. 2. 3. 4. 5. 6. 7. 8.

‫שׁ‬-Relative Data ‫שׁ‬-Complementizer Data ‫ז‬-Relative Data ‫מ‬-Relative Data ‫ה‬-Relative Data Zero-Relative Data Extraposed Relative Data ‫אשׁר‬-Nonrelative, Nominalizer Data

1.  ‫שׁ‬-Relative Data (99×) ‫ל־שׂ ַפת ַהיָּ ם ָלר ֹב׃‬ ְ ‫יהם ֵאין ִמ ְס ָפּר ַכּחֹול ֶשׁ ַע‬ ֶ ‫ וְ ִלגְ ַמ ֵלּ‬Judg 7:12 ‘and concerning their camels, no number existed and (they) were like the sandi whi that ___i is upon the shore of the sea with regard to abundance’ ‫אַרגָּ ָמן ֶשׁ ַעל ַמ ְל ֵכי ִמ ְדיָ ן‬ ְ ‫וּבגְ ֵדי ָה‬ ִ ‫ן־ה ַשּׂ ֲהר ֹנִ ים וְ ַהנְּ ִטפֹות‬ ַ ‫ ְל ַבד ִמ‬Judg 8:26 ‘apart from the crescents and the pendants and the purple garmentsi whi that ___i are on the kings of Midian’

286

Relative Clause Data

287

‫ל־מ ֶלְך יִ ְשׂ ָר ֵאל׃‬ ֶ ‫ ֲהלֹוא ַתּגִּ ידוּ ִלי ִמי ִמ ֶשּׁ ָלּנוּ ֶא‬2 Kgs 6:11 ‘will you not tell me who from (those)i whi that ___i are among us is with the king of Israel?’ ‫ן־ליְ ָלה אָ‬ ַ ‫וּב‬ ִ ‫ן־ליְ ָלה ָהיָ ה‬ ַ ‫א־ע ַמ ְל ָתּ בֹּו וְ לֹא גִ ַדּ ְלתֹּו ֶשׁ ִבּ‬ ָ ֹ ‫ל־ה ִקּ ָיקיֹון ֲא ֶשׁר ל‬ ַ ‫אַתּה ַח ְס ָתּ ַע‬ ָ Jonah 4:10 ‫ָבד׃‬ ‘you are concerned about the qiqayon-planti that you did not labor and that you did not grow, whi that ___i was the product of a night and perished in a night?’ ‫ה־לּהּ יַ ְח ָדּו׃‬ ָ ‫רוּשׁ ַלםִ ַה ְבּנוּיָ ה ְכּ ִעיר ֶשׁ ֻח ְבּ ָר‬ ָ ְ‫ י‬Ps 122:3 ‘Jerusalem, who is built as a cityi whi that ___i is bound firmly together’ ‫ ֶשׁ ָשּׁם ָעלוּ ְשׁ ָב ִטים ִשׁ ְב ֵטי־יָ הּ‬Ps 122:4 ‘[Jerusalemi], whi that therei the tribes go up, the tribes of Yhwh’ ‘if it had not been Yhwhi, whi that ___i was for us’

‫לוּלי יְ הוָ ה ֶשׁ ָהיָ ה ָלנוּ‬ ֵ Ps 124:1

‫אָדם׃‬ ָ ‫לוּלי יְ הוָ ה ֶשׁ ָהיָ ה ָלנוּ ְבּקוּם ָע ֵלינוּ‬ ֵ Ps 124:2 ‘if it had not been Yhwhi, whi that ___i was for us, when people rose against us’ ‫יהם׃‬ ֶ ֵ‫ ָבּרוְּך יְ הוָ ה ֶשׁלֹּא נְ ָתנָ נוּ ֶט ֶרף ְל ִשׁנּ‬Ps 124:6 ‘blessed be Yhwhi, whi that ___i has not given us as prey to their teeth’ ‫ יִ ְהיוּ ַכּ ֲח ִציר גַּ גֹּות ֶשׁ ַקּ ְד ַמת ָשׁ ַלף ֵיָבשׁ׃‬Ps 129:6 ‘let them be like the grassi on the housetops whi that, before one pulls it up, ___i withers’ ‫קֹוצר וְ ִח ְצנֹו ְמ ַע ֵמּר׃‬ ֵ ‫ ֶשׁלֹּא ִמ ֵלּא ַכפֹּו‬Ps 129:7 ‘[grassi] whi that a reaper do not fill his hand (with) ___i and a gatherer (fill) his cloak (with) ___i’ ‫דֹּותיו׃‬ ָ ‫ל־פּי ִמ‬ ִ ‫ן־אַהר ֹן ֶשׁיּ ֵֹרד ַע‬ ֲ ‫ל־הזָּ ָקן זְ ַק‬ ַ ‫ל־הרֹאשׁ י ֵֹרד ַע‬ ָ ‫ ַכּ ֶשּׁ ֶמן ַהטֹּוב ׀ ַע‬Ps 133:2 ‘it is like the precious oili (that) is on the head, (that) runs down upon the beard, on the beard of Aaron, whi that ___i runs down over the collar of his robes’ ‫ל־ה ְר ֵרי ִציֹּון‬ ַ ‫ל־ח ְרמֹון ֶשׁיּ ֵֹרד ַע‬ ֶ ‫ ְכּ ַט‬Ps 133:3 ‘it is like the dewi of Hermon, whi that ___i falls on the mountains of Zion’ ‫ֹלהינוּ׃‬ ֵ ‫ ֶשׁע ְֹמ ִדים ְבּ ֵבית יְ הוָ ה ְבּ ַח ְצרֹות ֵבּית ֱא‬Ps 135:2 ‘[O servants of Yhwhi], whi that ___i that stand in the house of Yhwh, in the courts of the house of our God’ ‫ד־בּ ֵה ָמה׃‬ ְ ‫אָדם ַע‬ ָ ‫כֹורי ִמ ְצ ָריִ ם ֵמ‬ ֵ ‫ ֶשׁ ִה ָכּה ְבּ‬Ps 135:8 ‘[hei] whi that ___i struck down the firstborn of Egypt, both human beings and animals’ ‫צוּמים׃‬ ִ ‫ ֶשׁ ִה ָכּה גֹּויִ ם ַר ִבּים וְ ָה ַרג ְמ ָל ִכים ֲע‬Ps 135:10 ‘[hei] whi that ___i struck down many nations and killed mighty kings’ ‫ ֶשׁ ְבּ ִשׁ ְפ ֵלנוּ זָ ַכר ָלנוּ‬Ps 136:23 ‘[give thanks to hei] whi that ___i remembered us in our low estate’

288

Appendix A

‫מוּלְך ֶשׁגָּ ַמ ְל ְתּ ָלנוּ׃‬ ֵ ְ‫ם־לְך ֶאת־גּ‬ ָ ‫אַשׁ ֵרי ֶשׁיְ ַשׁ ֶלּ‬ ְ Ps 137:8a ‘blessed is (hei) whi that ___i repays deedj whj that you did ___j to us’ ‫מוּלְך ֶשׁגָּ ַמ ְל ְתּ ָלנוּ׃‬ ֵ ְ‫ם־לְך ֶאת־גּ‬ ָ ‫אַשׁ ֵרי ֶשׁיְ ַשׁ ֶלּ‬ ְ Ps 137:8b ‘blessed is (hei) whi that ___i repays deedj whj that you did ___j to us’ ‫ל־ה ָסּ ַלע׃‬ ַ ‫אחז וְ נִ ֵפּץ ֶאת־ ע ָֹל ַליִ ְך ֶא‬ ֵ ֹ ‫אַשׁ ֵרי ׀ ֶשׁיּ‬ ְ Ps 137:9 ‘blessed is (hei) whi that ___i takes your little ones and dashes them against the rock’ ‫ֹלהיו׃‬ ָ ‫אַשׁ ֵרי ָה ָעם ֶשׁיֲ הוָ ה ֱא‬ ְ ‫אַשׁ ֵרי ָה ָעם ֶשׁ ָכּ ָכה לֹּו‬ ְ Ps 144:15a ‘blessed is the peoplei whi such belongs to iti; blessed is the peoplej whj that Yhwh is itsj God’ ‫ֹלהיו׃‬ ָ ‫אַשׁ ֵרי ָה ָעם ֶשׁיֲ הוָ ה ֱא‬ ְ ‫אַשׁ ֵרי ָ ֭ה ָעם ֶשׁ ָכּ ָכה לֹּו‬ ְ Ps 144:15b ‘blessed is the peoplei whi such belongs to iti; blessed is the peoplej whj that Yhwh is itsj God’ ‫שׁוּעה׃‬ ָ ‫ן־אָדם ׀ ֶשׁ ֵאין לֹו ְת‬ ָ ‫אַל־תּ ְב ְטחוּ ִבנְ ִד ִיבים ְבּ ֶב‬ ִ Ps 146:3 ‘do not put your trust in princes, in a humani whi that deliverance is not hisi’ ‫ֹלהיו׃‬ ָ ‫אַשׁ ֵרי ֶשׁ ֵאל יַ ֲעקֹב ְבּ ֶעזְ רֹו ִשׂ ְברֹו ַעל־יְ הוָ ה ֱא‬ ְ Ps 146:5 ‘blessed is (hei) whi that the God of Jacob is hisi help, his hope is upon Yhwh his God’ Song 1:6a ‫רוּ־בי ָשׂ ֻמנִ י נ ֵֹט ָרה ֶאת־‬ ִ ‫אַל־תּ ְראוּנִ י ֶשׁ ֲאנִ י ְשׁ ַח ְרח ֶֹרת ֶשׁ ֱשּׁזָ ַפ ְתנִ י ַה ָשּׁ ֶמשׁ ְבּ ֵנ֧י ִא ִמּי נִ ֲח‬ ִ ‫ַה ְכּ ָר ִמים ַכּ ְר ִמי ֶשׁ ִלּי לֹא נָ ָט ְר ִתּי׃‬ ‘do not gaze at mei whi that Ii am dark, whi that the sun has browned mei; the sons of my mother were angry with me; they made me keeper of the vineyards—the vineyardj whj that ___j belonged to me I have not kept!’ Song 1:6b ‫רוּ־בי ָשׂ ֻמנִ י נ ֵֹט ָרה ֶאת־‬ ִ ‫אַל־תּ ְראוּנִ י ֶשׁ ֲאנִ י ְשׁ ַח ְרח ֶֹרת ֶשׁ ֱשּׁזָ ַפ ְתנִ י ַה ָשּׁ ֶמשׁ ְבּ ֵנ֧י ִא ִמּי נִ ֲח‬ ִ ‫ַה ְכּ ָר ִמים ַכּ ְר ִמי ֶשׁ ִלּי לֹא נָ ָט ְר ִתּי׃‬ ‘do not gaze at mei whi that Ii am dark, whi that the sun has browned mei; the sons of my mother were angry with me; they made me keeper of the vineyards—the vineyardj whj that ___j belonged to me I have not kept!’ Song 1:6c ‫רוּ־בי ָשׂ ֻמנִ י נ ֵֹט ָרה ֶאת־‬ ִ ‫אַל־תּ ְראוּנִ י ֶשׁ ֲאנִ י ְשׁ ַח ְרח ֶֹרת ֶשׁ ֱשּׁזָ ַפ ְתנִ י ַה ָשּׁ ֶמשׁ ְבּ ֵנ֧י ִא ִמּי נִ ֲח‬ ִ ‫ַה ְכּ ָר ִמים ַכּ ְר ִמי ֶשׁ ִלּי לֹא נָ ָט ְר ִתּי׃‬ ‘do not gaze at mei whi that Ii am dark, whi that the sun has browned mei; the sons of my mother were angry with me; they made me keeper of the vineyards—the vineyardj whj that ___j belonged to me I have not kept!’ ‫יכה ַתּ ְר ִבּיץ ַבּ ָצּ ֳה ָריִ ם ַשׁ ָלּ ָמה ֶא ְהיֶ ה ְכּע ְֹטיָ ה‬ ָ ‫יכה ִת ְר ֶעה ֵא‬ ָ ‫אָה ָבה נַ ְפ ִשׁי ֵא‬ ֲ ‫ ַהגִּ ָידה ִלּי ֶשׁ‬Song 1:7a ‫ַעל ֶע ְד ֵרי ֲח ֵב ֶריָך׃‬ ‘tell me, ∅(you)i whi that my soul loves ___i, where you pasture, where you have (your flock) lie down at noon; ∅(you)i whi that should I be like one who picks fleas beside the flocks of youri companions?’

Relative Clause Data

289

‫יכה ַתּ ְר ִבּיץ ַבּ ָצּ ֳה ָריִ ם ַשׁ ָלּ ָמה ֶא ְהיֶ ה ְכּע ְֹטיָ ה‬ ָ ‫יכה ִת ְר ֶעה ֵא‬ ָ ‫אָה ָבה נַ ְפ ִשׁי ֵא‬ ֲ ‫ ַהגִּ ָידה ִלּי ֶשׁ‬Song 1:7b ‫ַעל ֶע ְד ֵרי ֲח ֵב ֶריָך׃‬ ‘tell me, ∅(you)i whi that my soul loves ___i, where you pasture, where you have (your flock) lie down at noon; ∅(you)i whi that should I be like one who picks fleas beside the flocks of youri companions?’ ‫אתיו׃‬ ִ ‫אָה ָבה נַ ְפ ִשׁי ִבּ ַקּ ְשׁ ִתּיו וְ לֹא ְמ ָצ‬ ֲ ‫ל־מ ְשׁ ָכּ ִבי ַבּ ֵלּילֹות ִבּ ַקּ ְשׁ ִתּי ֵאת ֶשׁ‬ ִ ‫ ַע‬Song 3:1 ‘upon my bed at night I sought (the onei) whi that my soul loves ___i; I sought him, but found him not’ ‫אָה ָבה נַ ְפ ִשׁי ִבּ ַקּ ְשׁ ִתּיו‬ ֲ ‫וּב ְרחֹבֹות ֲא ַב ְק ָשׁה ֵאת ֶשׁ‬ ָ ‫סֹוב ָבה ָב ִעיר ַבּ ְשּׁוָ ִקים‬ ְ ‫אָקוּמה נָּ א וַ ֲא‬ ָ Song 3:2 ‫אתיו׃‬ ִ ‫וְ לֹא ְמ ָצ‬ ‘I will rise now and go around through the city, through the streets and through the squares; I will seek (the onei) whi that my soul loves ___i. I sought him, but found him not’ ‫יתם׃‬ ֶ ‫אָה ָבה נַ ְפ ִשׁי ְר ִא‬ ֲ ‫ ְמ ָצאוּנִ י ַהשּׁ ְֹמ ִרים ַהסּ ְֹב ִבים ָבּ ִעיר ֵאת ֶשׁ‬Song 3:3 ‘the watchers found me, those who go around through the city. Have you seen (the onei) whi that my soul loves ___i?’ ‫אַר ֶפּנּוּ ַעד־‬ ְ ‫אָה ָבה נַ ְפ ִשׁי ֲא ַחזְ ִתּיו וְ לֹא‬ ֲ ‫אתי ֵאת ֶשׁ‬ ִ ‫ ִכּ ְמ ַעט ֶשׁ ָע ַב ְר ִתּי ֵמ ֶהם ַעד ֶשׁ ָמּ ָצ‬Song 3:4c ‫הֹור ִתי׃‬ ָ ‫ל־ח ֶדר‬ ֶ ‫ל־בּית ִא ִמּי וְ ֶא‬ ֵ ‫יאתיו ֶא‬ ִ ‫ֶשׁ ֲה ֵב‬ ‘about a little that I had passed from them until that I found (the onei) whi that my soul loves ___i. I held him, and would not let him go until that I brought him into my mother’s house, and into the chamber of her that conceived me.’ ‫ ִהנֵּ ה ִמ ָטּתֹו ֶשׁ ִלּ ְשֹׁלמֹה ִשׁ ִשּׁים גִּ בּ ִֹרים ָס ִביב ָלהּ ִמגִּ בּ ֵֹרי יִ ְשׂ ָר ֵאל׃‬Song 3:7 ‘look—the litteri whi that ___i belongs to Solomon! Sixty warriors are around it, from the warriors of Israel’ ‫ ְצ ֶ ֧אינָ ה ׀ ְוּר ֶאינָ ה ְבּנֹות ִציֹּון ַבּ ֶמּ ֶלְך ְשֹׁלמֹה ָבּ ֲע ָט ָרה ֶשׁ ִע ְטּ ָרה־לֹּו ִאמֹּו ְבּיֹום ֲח ֻתנָּ תֹו וּ‬Song 3:11 ‫ְביֹום ִשׂ ְמ ַחת ִלבֹּו׃ ס‬ ‘come out and look, O daughters of Zion, at King Solomon, at the crowni whi that his mother crowned him (with) ___i on the day of his wedding, on the day of the gladness of his heart’ ‫ ִהנָּ ְך יָ ָפה ַר ְעיָ ִתי ִהנָּ ְך יָ ָפה ֵעינַ יִ ְך יֹונִ ים ִמ ַבּ ַעד ְל ַצ ָמּ ֵתְך ַשׂ ְע ֵרְך ְכּ ֵע ֶדר ָה ִעזִּ ים ֶשׁגָּ ְלשׁוּ‬Song 4:1 ‫ֵמ ַהר גִּ ְל ָעד׃‬ ‘look—you are beautiful, my companion; look—you are beautiful! Your eyes are doves behind your veil; your hair is like a flock of goatsi whi that ___i skip from the hill country of Gilead’ ‫ן־ה ַר ְח ָצה ֶשׁ ֻכּ ָלּם ַמ ְת ִאימֹות וְ ַשׁ ֻכּ ָלה ֵאין ָבּ ֶהם׃‬ ָ ‫ ִשׁנַּ יִ ְך ְכּ ֵע ֶדר ַה ְקּצוּבֹות ֶשׁ ָעלוּ ִמ‬Song 4:2a ‘your teeth are like a flock of sheared (ewes)i whi that ___i have come up from the washing, whi that all of themi bear twins and not one among them is barren’ ‫ן־ה ַר ְח ָצה ֶשׁ ֻכּ ָלּם ַמ ְת ִאימֹות וְ ַשׁ ֻכּ ָלה ֵאין ָבּ ֶהם׃‬ ָ ‫ ִשׁנַּ יִ ְך ְכּ ֵע ֶדר ַה ְקּצוּבֹות ֶשׁ ָעלוּ ִמ‬Song 4:2b ‘your teeth are like a flock of sheared (ewes)i whi that ___i have come up from the washing, whi that all of themi bear twins and not one among them is barren’

290

Appendix A

‫אשׁי‬ ִ ֹ ‫י־לי ֲאח ִֹתי ַר ְעיָ ִתי יֹונָ ִתי ַת ָמּ ִתי ֶשׁרּ‬ ִ ‫דֹופק ִפּ ְת ִח‬ ֵ ‫דֹּודי‬ ִ ‫ ֲאנִ י יְ ֵשׁנָ ה וְ ִל ִבּי ֵער קֹול ׀‬Song 5:2 ‫יסי ָליְ ָלה׃‬ ֵ ‫צֹּותי ְר ִס‬ ַ ֻ‫א־טל ְקוּ‬ ָ ‫נִ ְמ ָל‬ ‘I slept, but my heart was awake. The sound of my beloved knocking! “Open to mei, my sister, my companion, my dove, my perfect one, whi that myi head is filled with dew, my locks with the drops of the night’ ‫ה־דֹּודְך ִמדֹּוד ֶשׁ ָכּ ָכה ִה ְשׁ ַבּ ְע ָתּנוּ׃‬ ֵ ‫ה־דֹּודְך ִמדֹּוד ַהיָּ ָפה ַבּנָּ ִשׁים ַמ‬ ֵ ‫ ַמ‬Song 5:9 ‘what is your beloved more than another beloved, O fairest among women? What is your beloved more than another beloved, ∅(you)i whi that ___(you)i made us swear so?’ ‫ן־הגִּ ְל ָעד׃‬ ַ ‫ ָה ֵס ִבּי ֵעינַ יִ ְך ִמנֶּ גְ ִדּי ֶשׁ ֵהם ִה ְר ִה ֻיבנִ י ַשׂ ְע ֵרְך ְכּ ֵע ֶדר ָה ִעזִּ ים ֶשׁגָּ ְלשׁוּ ִמ‬Song 6:5a ‘turn away your eyesi from me whi that theyi overwhelm me! Your hair is like a flock of goatsj whj that ___j skips from Gilead’ ‫ן־הגִּ ְל ָעד׃‬ ַ ‫ ָה ֵס ִבּי ֵעינַ יִ ְך ִמנֶּ גְ ִדּי ֶשׁ ֵהם ִה ְר ִה ֻיבנִ י ַשׂ ְע ֵרְך ְכּ ֵע ֶדר ָה ִעזִּ ים ֶשׁגָּ ְלשׁוּ ִמ‬Song 6:5b ‘turn away your eyesi from me whi that theyi overwhelm me! Your hair is like a flock of goatsj whj that ___j skips from Gilead’ ‫ן־ה ַר ְח ָצה ֶשׁ ֻכּ ָלּם ַמ ְת ִאימֹות וְ ַשׁ ֻכּ ָלה ֵאין ָבּ ֶהם׃‬ ָ ‫ ִשׁנַּ יִ ְך ְכּ ֵע ֶדר ָה ְר ֵח ִלים ֶשׁ ָעלוּ ִמ‬Song 6:6a ‘your teeth are like a flock of ewesi whi that ___i have come up from the washing, whi that all of themi bear twins and not one among them is barren’ ‫ן־ה ַר ְח ָצה ֶשׁ ֻכּ ָלּם ַמ ְת ִאימֹות וְ ַשׁ ֻכּ ָלה ֵאין ָבּ ֶהם׃‬ ָ ‫ ִשׁנַּ יִ ְך ְכּ ֵע ֶדר ָה ְר ֵח ִלים ֶשׁ ָעלוּ ִמ‬Song 6:6b ‘your teeth are like a flock of ewesi whi that ___i have come up from the washing, whi that all of themi bear twins and not one among them is barren’ ‫ר־בּהּ׃‬ ָ ‫ אָחֹות ָלנוּ ְק ַטנָּ ה וְ ָשׁ ַדיִ ם ֵאין ָלהּ ַמה־נַּ ֲע ֶשׂה ַל ֲאח ֵֹתנוּ ַבּיֹּום ֶשׁיְּ ֻד ַבּ‬Song 8:8 ‘we have a little sister, and she has no breasts. What shall we do for our sister, on the dayi whi that one is spoken for her ___i?’ ‫ת־פּ ְריֹו׃‬ ִ ‫אתיִ ם ְלנ ְֹט ִרים ֶא‬ ַ ‫וּמ‬ ָ ‫ ָכּ ְר ִמי ֶשׁ ִלּי ְל ָפנָ י ָה ֶא ֶלף ְלָך ְשֹׁלמֹה‬Song 8:12 ‘my vineyardi whi that ___i belongs to me is before me; you, O Solomon, may have the thousand, and the keepers of the fruit two hundred!’ ‫ל־ע ָמלֹו ֶשׁיַּ ֲעמֹל ַתּ ַחת ַה ָשּׁ ֶמשׁ׃‬ ֲ ‫אָדם ְבּ ָכ‬ ָ ‫ ַמה־יִּ ְתרֹון ָל‬Eccl 1:3 ‘what profit belongs to a man in exchange for all his accomplishmenti whi that he accomplishes ___i under the sun?’ ‫ל־מקֹום ֶשׁ ַהנְּ ָח ִלים ה ְֹל ִכים ָשׁם ֵהם ָשׁ ִבים ָל ָל ֶכת׃‬ ְ ‫ ֶא‬Eccl 1:7 ‘to the placei whi that the rivers go ___i—there they go continually’ ‫ה־שׁנַּ ֲע ָשׂה הוּא ֶשׁיֵּ ָע ֶשׂה‬ ֶ ‫וּמ‬ ַ ‫ה־שּׁ ָהיָ ה הוּא ֶשׁיִּ ְהיֶ ה‬ ֶ ‫ ַמ‬Eccl 1:9a ‘whateveri whi ___i has been—it is (the thing)j whj that ___j will be, and whateverk whk ___k has happened—it is (the thing)l whl that ___l will happen’ ‫ה־שׁנַּ ֲע ָשׂה הוּא ֶשׁיֵּ ָע ֶשׂה‬ ֶ ‫וּמ‬ ַ ‫ה־שּׁ ָהיָ ה הוּא ֶשׁיִּ ְהיֶ ה‬ ֶ ‫ ַמ‬Eccl 1:9b ‘whateveri whi ___i has been—it is (the thing)j whj that ___j will be, and whateverk whk ___k has happened—it is (the thing)l whl that ___l will happen’

Relative Clause Data

291

‫ה־שׁנַּ ֲע ָשׂה הוּא ֶשׁיֵּ ָע ֶשׂה‬ ֶ ‫וּמ‬ ַ ‫ה־שּׁ ָהיָ ה הוּא ֶשׁיִּ ְהיֶ ה‬ ֶ ‫ ַמ‬Eccl 1:9c ‘whateveri whi ___i has been—it is (the thing)j whj that ___j will be, and whateverk whk ___k has happened—it is (the thing)l whl that ___l will happen’ ‫ה־שׁנַּ ֲע ָשׂה הוּא ֶשׁיֵּ ָע ֶשׂה‬ ֶ ‫וּמ‬ ַ ‫ה־שּׁ ָהיָ ה הוּא ֶשׁיִּ ְהיֶ ה‬ ֶ ‫ ַמ‬Eccl 1:9d ‘whateveri whi ___i has been—it is (the thing)j whj that ___j will be, and whateverk whk ___k has happened—it is (the thing)l whl that ___l will happen’ ‫אמר ְר ֵאה־זֶ ה ָח ָדשׁ הוּא‬ ַ ֹ ‫ יֵ שׁ ָדּ ָבר ֶשׁיּ‬Eccl 1:10 ‘there is a thingi whi that one might say (about) ___i, “See this—it is new!”’ ‫אַחר ֹנָ ה׃ פ‬ ֲ ‫אַחר ֹנִ ים ֶשׁיִּ ְהיוּ לֹא־יִ ְהיֶ ה ָל ֶהם זִ ָכּרֹון ִעם ֶשׁיִּ ְהיוּ ָל‬ ֲ ‫ וְ גַ ם ָל‬Eccl 1:11a ‘also, as for the latter (generations)i whi that ___i will be, there will be no remembrance of them among (those)j whj that ___j will be at the latter (time)’ ‫אַחר ֹנָ ה׃ פ‬ ֲ ‫אַחר ֹנִ ים ֶשׁיִּ ְהיוּ לֹא־יִ ְהיֶ ה ָל ֶהם זִ ָכּרֹון ִעם ֶשׁיִּ ְהיוּ ָל‬ ֲ ‫ וְ גַ ם ָל‬Eccl 1:11b ‘also, as for the latter (generations)i whi that ___i will be, there will be no remembrance of them among (those)j whj that ___j will be at the latter (time)’ ‫ל־ה ַמּ ֲע ִשׂים ֶשׁנַּ ֲעשׂוּ ַתּ ַחת ַה ָשּׁ ֶמשׁ‬ ַ ‫ת־כּ‬ ָ ‫יתי ֶא‬ ִ ‫ ָר ִא‬Eccl 1:14 ‘I have seen all the eventsi whi that ___i have happened under the sun’ ‫ירוּשׁ ָלםִ׃‬ ָ ‫ גַּ ם ִמ ְקנֶ ה ָב ָקר וָ צֹאן ַה ְר ֵבּה ָהיָ ה ִלי ִמכֹּל ֶשׁ ָהיוּ ְל ָפנַ י ִבּ‬Eccl 2:7 ‘also livestock—cattle and sheep—(belonged to me), more than alli whi that ___i were before me in Jerusalem’ ִ‫ירוּשׁ ָלם‬ ָ ‫הֹוס ְפ ִתּי ִמכֹּל ֶשׁ ָהיָ ה ְל ָפנַ י ִבּ‬ ַ ְ‫ וְ גָ ַד ְל ִתּי ו‬Eccl 2:9 ‘and I became greater and increased myself more than anyonei whi that ___i was before me in Jerusalem’ ‫וּב ָע ָמל ֶשׁ ָע ַמ ְל ִתּי ַל ֲעשֹׂות‬ ֶ ‫ל־מ ֲע ַשׂי ֶשׁ ָעשׂוּ יָ ַדי‬ ַ ‫יתי ֲאנִ י ְבּ ָכ‬ ִ ִ‫וּפנ‬ ָ Eccl 2:11a ‘and I faced, I, all my worksi whi that my hands had done ___i and the toilj whj that I had accomplished ___j by doing’ ‫וּב ָע ָמל ֶשׁ ָע ַמ ְל ִתּי ַל ֲעשֹׂות‬ ֶ ‫ל־מ ֲע ַשׂי ֶשׁ ָעשׂוּ יָ ַדי‬ ַ ‫יתי ֲאנִ י ְבּ ָכ‬ ִ ִ‫וּפנ‬ ָ Eccl 2:11b ‘and I faced, I, all my worksi whi that my hands had done ___i and the toilj whj that I had accomplished ___j by doing’ ‫אַח ֵרי ַה ֶמּ ֶלְך‬ ֲ ‫אָדם ֶשׁיָּבֹוא‬ ָ ‫הֹוללֹות וְ ִס ְכלוּת ִכּי ׀ ֶמה ָה‬ ֵ ְ‫יתי ֲאנִ י ִל ְראֹות ָח ְכ ָמה ו‬ ִ ִ‫וּפנ‬ ָ Eccl 2:12 ‘and I faced, I, to see wisdom and inanity and foolishness, that what the mani whi that ___i comes after the king (does)’ ‫ת־ה ַחיִּ ים ִכּי ַרע ָע ַלי ַה ַמּ ֲע ֶשׂה ֶשׁנַּ ֲע ָשׂה ַתּ ַחת ַה ָשּׁ ֶמשׁ‬ ַ ‫אתי ֶא‬ ִ ֵ‫ וְ ָשׂנ‬Eccl 2:17 ‘and I came to hate life, because the eventi whi that ___i has happened under the sun was terrible to me’ ‫אַח ָרי׃‬ ֲ ‫אָדם ֶשׁיִּ ְהיֶ ה‬ ָ ‫יחנּוּ ָל‬ ֶ ִ‫ל־ע ָמ ִלי ֶשׁ ֲאנִ י ָע ֵמל ַתּ ַחת ַה ָשּׁ ֶמשׁ ֶשׁאַנּ‬ ֲ ‫ת־כּ‬ ָ ‫אתי ֲאנִ י ֶא‬ ִ ֵ‫ וְ ָשׂנ‬Eccl 2:18a ‘and I came to hate, I, all my accomplishment(s)i whi that I accomplished ___i under the sun, whi that I must leave iti to the manj whj that ___j comes after me’

292

Appendix A

‫אַח ָרי׃‬ ֲ ‫אָדם ֶשׁיִּ ְהיֶ ה‬ ָ ‫יחנּוּ ָל‬ ֶ ִ‫ל־ע ָמ ִלי ֶשׁ ֲאנִ י ָע ֵמל ַתּ ַחת ַה ָשּׁ ֶמשׁ ֶשׁאַנּ‬ ֲ ‫ת־כּ‬ ָ ‫אתי ֲאנִ י ֶא‬ ִ ֵ‫ וְ ָשׂנ‬Eccl 2:18b ‘and I came to hate, I, all my accomplishment(s)i whi that I accomplished ___i under the sun, whi that I must leave iti to the manj whj that ___j comes after me’ ‫אַח ָרי׃‬ ֲ ‫אָדם ֶשׁיִּ ְהיֶ ה‬ ָ ‫יחנּוּ ָל‬ ֶ ִ‫ל־ע ָמ ִלי ֶשׁ ֲאנִ י ָע ֵמל ַתּ ַחת ַה ָשּׁ ֶמשׁ ֶשׁאַנּ‬ ֲ ‫ת־כּ‬ ָ ‫אתי ֲאנִ י ֶא‬ ִ ֵ‫ וְ ָשׂנ‬Eccl 2:18c ‘and I came to hate, I, all my accomplishment(s)i whi that I accomplished ___i under the sun, whi that I must leave iti to the manj whj that ___j comes after me’ ‫ל־ע ָמ ִלי ֶשׁ ָע ַמ ְל ִתּי וְ ֶשׁ ָח ַכ ְמ ִתּי ַתּ ַחת ַה ָשּׁ ֶמשׁ‬ ֲ ‫ וְ יִ ְשׁ ַלט ְבּ ָכ‬Eccl 2:19a ‘and he will rule over my accomplishment(s)i whi that I accomplished ___i and whi that I became wise under the sun (by) ___i.’ ‫ל־ע ָמ ִלי ֶשׁ ָע ַמ ְל ִתּי וְ ֶשׁ ָח ַכ ְמ ִתּי ַתּ ַחת ַה ָשּׁ ֶמשׁ‬ ֲ ‫ וְ יִ ְשׁ ַלט ְבּ ָכ‬Eccl 2:19b ‘and he will rule over my accomplishment(s)i whi that I accomplished ___i and whi that I became wise under the sun (by) ___i.’ ‫ל־ה ָע ָמל ֶשׁ ָע ַמ ְל ִתּי ַתּ ַחת ַה ָשּׁ ֶמשׁ׃‬ ֶ ‫ת־ל ִבּי ַ ֚על ָכּ‬ ִ ‫בֹּותי ֲאנִ י ְליַ ֵאשׁ ֶא‬ ִ ‫ וְ ַס‬Eccl 2:20 ‘and I turned around, I, to put my ‫ לב‬in a state of despair about all the accomplishment(s)i whi that I accomplished ___i under the sun’ ‫אָדם ֶשׁלֹּא ָע ַמל־בֹּו יִ ְתּנֶ נּוּ ֶח ְלקֹו‬ ָ ‫וּל‬ ְ ‫וּב ִכ ְשׁרֹון‬ ְ ‫וּב ַד ַעת‬ ְ ‫אָדם ֶשׁ ֲע ָמלֹו ְבּ ָח ְכ ָמה‬ ָ ‫ ִכּי־יֵ שׁ‬Eccl 2:21a ‘because there is a mani whi that hisi gain is by wisdom and by knowledge and by skill, but to a manj whj that ___j has not exerted himself for it—he must give him his portion’ ‫אָדם ֶשׁלֹּא ָע ַמל־בֹּו יִ ְתּנֶ נּוּ ֶח ְלקֹו‬ ָ ‫וּל‬ ְ ‫וּב ִכ ְשׁרֹון‬ ְ ‫וּב ַד ַעת‬ ְ ‫אָדם ֶשׁ ֲע ָמלֹו ְבּ ָח ְכ ָמה‬ ָ ‫ ִכּי־יֵ שׁ‬Eccl 2:21b ‘because there is a mani whi that hisi gain is by wisdom and by knowledge and by skill, but to a manj whj that ___j has not exerted himself for it—he must give him his portion’ ‫וּב ַר ְעיֹון ִלבֹּו ֶשׁהוּא ָע ֵמל ַתּ ַחת ַה ָשּׁ ֶמשׁ׃‬ ְ ‫ל־ע ָמלֹו‬ ֲ ‫אָדם ְבּ ָכ‬ ָ ‫ ִכּי ֶמה־הֹוֶ ה ָל‬Eccl 2:22 ‘because what endures for man in exchange for all his accomplish­ ment(s) and striving of his hearti whi that he accomplished ___i under the sun?’ ‫אָדם ֶשׁטֹּוב ְל ָפנָ יו נָ ַתן ָח ְכ ָמה וְ ַד ַעת וְ ִשׂ ְמ ָחה‬ ָ ‫ ִכּי ְל‬Eccl 2:26 ‘because to a mani whi that ___i is good before him he has given wisdom, knowledge, and joy’ ‘whateveri whi that ___i has been—it is already’

‫ה־שּׁ ָהיָ ה ְכּ ָבר הוּא‬ ֶ ‫ ַמ‬Eccl 3:15

‫אַח ָריו׃‬ ֲ ‫יאנּוּ ִל ְראֹות ְבּ ֶמה ֶשׁיִּ ְהיֶ ה‬ ֶ ‫ ִכּי ִמי ִיְב‬Eccl 3:22 ‘because who will bring him to see whati whi ___i will be after him’ ‫ת־ה ֵמּ ִתים ֶשׁ ְכּ ָבר ֵמתוּ‬ ַ ‫ וְ ַשׁ ֵ ֧בּ ַח ֲאנִ י ֶא‬Eccl 4:2 ‘and I praised the deadi whi that ___i had already died’ ‫ וְ ִאילֹו ָה ֶא ָחד ֶשׁיִּ פֹּול וְ ֵאין ֵשׁנִ י ַל ֲה ִקימֹו׃‬Eccl 4:10 ‘and woe to him, the onei whi that ___i falls and there is no second (person) to raise him up’

293

Relative Clause Data

‫אוּמה לֹא־יִ ָשּׂא ַב ֲע ָמלֹו ֶשׁיּ ֵֹלְך‬ ָ ‫וּמ‬ ְ ‫ ַכּ ֲא ֶשׁר יָ ָצא ִמ ֶבּ ֶטן ִאמֹּו ָערֹום יָ שׁוּב ָל ֶל ֶכת ְכּ ֶשׁ ָבּא‬Eccl 5:14b ‫ְבּיָ דֹו׃‬ ‘like he exited from the womb of his mother, naked he will again go, like he entered; nothing will he be able to take in exchange for his ac­ complishment(s)i whi that he may bring ___i into his [son’s] hand’ ‫רוּח׃‬ ַ ‫וּמה־יִּ ְתרֹון לֹו ֶשׁיַּ ֲעמֹל ָל‬ ַ ‫ל־ע ַמּת ֶשׁ ָבּא ֵכּן יֵ ֵלְך‬ ֻ ‫חֹולה ָכּ‬ ָ ‫ וְ גַ ם־זֹה ָר ָעה‬Eccl 5:15b ‘and this also is a sickly misfortune: like he came, so he will go; what is profit for himi whi that ___i works for wind?’ ‫ת־ה ֶשּׁ ֶמשׁ ִמ ְס ַפּר יְ ֵמי־חיו ֲא ֶשׁר־נָ ַתן־לֹו‬ ַ ‫ל־ע ָמלֹו ׀ ֶשׁיַּ ֲעמֹל ַתּ ַח‬ ֲ ‫טֹובה ְבּ ָכ‬ ָ ‫ וְ ִל ְראֹות‬Eccl 5:17 ‫ֹלהים‬ ִ ‫ָה ֱא‬ ‘and to see goodness in all his acquisition(s)i whi that he accomplished ___i under the sun, (for) the number of the days of his life that God has given him’ ‫א־יוּכל ָל ִדין ִעם שׁהתקיף ִמ ֶמּנּוּ׃‬ ַ ֹ ‫אָדם וְ ל‬ ָ ‫נֹודע ֲא ֶשׁר־הוּא‬ ָ ְ‫ה־שּׁ ָהיָ ה ְכּ ָבר נִ ְק ָרא ְשׁמֹו ו‬ ֶ ‫ ַמ‬Eccl 6:10a ‘whateveri whi that ___i has been—its name has already been called and that he is man is known, but he is not able to contend with (the one)j whj that ___j is mightier than he’ ‫א־יוּכל ָל ִדין ִעם שׁהתקיף‬ ַ ֹ ‫אָדם וְ ל‬ ָ ‫נֹודע ֲא ֶשׁר־הוּא‬ ָ ְ‫ה־שּׁ ָהיָ ה ְכּ ָבר נִ ְק ָרא ְשׁמֹו ו‬ ֶ ‫ ַמ‬Eccl 6:10b ‫[שׁ ַתּ ִקּיף] ִמ ֶמּנּוּ׃‬ ֶ ‘whateveri whi that ___i has been—its name has already been called and that he is man is known, but he is not able to contend with ∅(the one)j whj that ___j is mightier [Qere] than he’ ‘distant is whateveri whi that ___i has been’

‫ה־שּׁ ָהיָ ה‬ ֶ ‫ ָרחֹוק ַמ‬Eccl 7:24

‫ה־שּׁיִּ ְהיֶ ה‬ ֶ ‫י־אינֶ נּוּ י ֵֹד ַע ַמ‬ ֵ ‫ ִכּ‬Eccl 8:7 ‘that he does not know whateveri whi that ___i will be’ ‫יקים ֲא ֶשׁר ַמגִּ ַיע ֲא ֵל ֶהם ְכּ ַמ ֲע ֵשׂה‬ ִ ‫אָרץ ֲא ֶשׁר ׀ יֵ שׁ ַצ ִדּ‬ ֶ ‫ל־ה‬ ָ ‫שׁ־ה ֶבל ֲא ֶשׁר נַ ֲע ָשׂה ַע‬ ֶ ֶ‫ י‬Eccl 8:14a ‫אָמ ְר ִתּי ֶשׁגַּ ם־זֶ ה ָה ֶבל׃‬ ַ ‫יקים‬ ִ ‫ָה ְר ָשׁ ִעים וְ יֵ שׁ ְר ָשׁ ִעים ֶשׁ ַמּגִּ ַיע ֲא ֵל ֶהם ְכּ ַמ ֲע ֵשׂה ַה ַצּ ִדּ‬ ‘there is an absurdity that has happened on the earth, that there are righteous people who (a thing) befalls them just like the deed of the wicked, and there are wickedi whi that (a thing) befalls themi like the deed of the righteous. I say that this too is absurd!’ ‫אָדם‬ ָ ‫יוּק ִשׁים ְבּנֵ י ָה‬ ָ ‫צֹודה ָר ָעה וְ ַכ ִצּ ֳפּ ִרים ָה ֲא ֻחזֹות ַבּ ָפּח ָכּ ֵהם‬ ָ ‫ ַכּ ָדּגִ ים ֶשׁנֶּ ֱא ָחזִ ים ִבּ ְמ‬Eccl 9:12a ‫יהם ִפּ ְתאֹם׃‬ ֶ ‫ְל ֵעת ָר ָעה ְכּ ֶשׁ ִתּפֹּול ֲע ֵל‬ ‘like fishi whi that ___i are caught in a bad net and like birds that are snagged in a trap—like them humans are ensnared at a time of misfortune, when that it falls upon them suddenly’ ‫יתי ַתּ ַחת ַה ָשּׁ ֶמשׁ ִכּ ְשׁגָ גָ ה ֶשׁיּ ָֹצא ִמ ִלּ ְפנֵ י ַה ַשּׁ ִלּיט׃‬ ִ ‫ יֵ שׁ ָר ָעה ָר ִא‬Eccl 10:5 ‘there is a tragedy that I saw under the sun, like a mistakei whi that ___i proceeds from the ruler’ ‫ה־שׁיִּ ְהיֶ ה‬ ֶ ‫אָדם ַמ‬ ָ ‫ לֹא־יֵ ַדע ָה‬Eccl 10:14 ‘the man does not know whateveri whi that ___i will be’

294

Appendix A

‫אכלוּ׃‬ ֵ ֹ ‫י־לְך ֶא ֶרץ ֶשׁ ַמּ ְל ֵכְּך נָ ַער וְ ָשׂ ַריִ ְך ַבּבּ ֶֹקר י‬ ָ ‫ ִא‬Eccl 10:16 ‘woe belongs to youi, O Land, whi that youri king is a servant and your princes feast in the morning’ Eccl 10:17 ‫בוּרה וְ לֹא ַב ְשּׁ ִתי׃‬ ָ ְ‫אכלוּ ִבּג‬ ֵ ֹ ‫ן־חֹורים וְ ָשׂ ַריִ ְך ָבּ ֵעת י‬ ִ ‫אַשׁ ֵריְך ֶא ֶרץ ֶשׁ ַמּ ְל ֵכְּך ֶבּ‬ ְ ‘fortunate are youi, O Land, whi that youri king is a noble and whose princes eat at the (proper) time—with manliness and not with drinking!’ ‫ וְ ִאם־יִ פֹּול ֵעץ ַבּ ָדּרֹום וְ ִאם ַבּ ָצּפֹון ְמקֹום ֶשׁיִּ פֹּול ָה ֵעץ ָשׁם יְ הוּא׃‬Eccl 11:3 ‘and if a tree falls in the south or in the north, the placei whi that the tree falls ___i—there it remains’ ‫י־ה ְר ֵבּה יִ ְהי‬ ַ ‫אָדם ְבּ ֻכ ָלּם יִ ְשׂ ָמח וְ יִ זְ כֹּר ֶאת־יְ ֵמי ַהח ֶֹשְׁך ִכּ‬ ָ ‫ם־שׁנִ ים ַה ְר ֵבּה יִ ְחיֶ ה ָה‬ ָ ‫ ִכּי ִא‬Eccl 11:8 ‫ל־שׁ ָבּא ָה ֶבל׃‬ ֶ ‫וּ ָכּ‬ ‘indeed—if a man lives many years, in all of them he should be happy, and he should remember the days of darkness, that they will be many. Alli whi that ___i is coming is ephemeral’ ‫וּב ְטלוּ ַהטּ ֲֹחנֹות ִכּי ִמ ֵעטוּ וְ ָח ְשׁכוּ‬ ָ ‫ ַבּיֹּום ֶשׁיָּ זֻ עוּ שׁ ְֹמ ֵרי ַה ַבּיִ ת וְ ִה ְת ַעוְּ תוּ אַנְ ֵשׁי ֶה ָחיִ ל‬Eccl 12:3 ‫ָהר ֹאֹות ָבּ ֲא ֻרבֹּות׃‬ ‘on the dayi whi that the keepers of the house tremble ___i and the men of character are stooped over and those who grind cease because they have become few and those looking through the windows have become dark’ ‫רוּשׁ ָלםִ ֲהזֹאת ָה ִעיר‬ ָ ְ‫ל־בּת י‬ ַ ‫אשׁם ַע‬ ָ ֹ ‫ ָס ְפקוּ ָע ַליִ ְך ַכּ ַפּיִ ם ָכּל־ ע ְֹב ֵרי ֶד ֶרְך ָשׁ ְרקוּ וַ יָּ נִ עוּ ר‬Lam 2:15 ‫אָרץ׃ ס‬ ֶ ‫ל־ה‬ ָ ‫אמרוּ ְכּ ִל ַילת י ִֹפי ָמשֹׂושׂ ְל ָכ‬ ְ ֹ ‫ֶשׁיּ‬ ‘all who pass along the way clap their hands at you; they hiss and wag their heads at daughter Jerusalem; “Is this the cityi whi that they would say (about) ___i ‘perfection of beauty, the joy of all the earth’?”’ ‫אָמרוּ ִבּ ָלּ ְענוּ אְַך זֶ ה ַהיֹּום ֶשׁ ִקּוִּ ינֻ הוּ‬ ְ ‫קוּ־שׁן‬ ֵ ‫ל־אֹויְביִ ְך ָשׁ ְרקוּ וַ יַּ ַח ְר‬ ַ ‫יהם ָכּ‬ ֶ ‫ ָפּצוּ ָע ַליִ ְך ִפּ‬Lam 2:16 ‫ָמ ָצאנוּ ָר ִאינוּ׃ ס‬ ‘all your enemies open their mouths against you; they hiss, they gnash their teeth, they cry: “We have devoured her! Ah, this is the dayi whi that we waited for iti; we have found (it), we have seen it!”” ‫י־ח ֶרב ֵמ ַח ְל ֵלי ָר ָעב ֶשׁ ֵהם יָ זוּבוּ ְמ ֻד ָקּ ִרים ִמ ְתּנוּבֹת ָשׂ ָדי׃ ס‬ ֶ ‫טֹובים ָהיוּ ַח ְל ֵל‬ ִ Lam 4:9 ‘better were the slain by the sword than the slaini by hunger, whi that theyi flow out, being pierced without the produce of the field’ ‫שׁוּע ִלים ִה ְלּכוּ־בֹו׃ פ‬ ָ ‫ר־ציֹּון ֶשׁ ָשּׁ ֵמם‬ ִ ‫ ַעל ַה‬Lam 5:18 ‘because of Mount Zioni, whi, that ___i is desolate; jackals prowl over it’ ‫אתיִ ם וְ ֶע ְשׂ ִרים‬ ַ ‫ן־הנְּ ִתינִ ים ֶשׁנָּ ַתן ָדּוִ יד וְ ַה ָשּׂ ִרים ַל ֲעב ַֹדת ַה ְלוִ יִּ ם נְ ִתינִ ים ָמ‬ ַ ‫וּמ‬ ִ Ezra 8:20 ‘and from the temple servantsi whi that David and his officials had set apart ___i to attend the Levites were two hundred and twenty temple servants’ ‫יאים וְ כֹל ֶשׁ ִע ָמּ ֶהם‬ ִ ‫ וַ יִּ נָּ ְתנוּ ְביָ ָדם ַה ַהגְ ִר‬1 Chr 5:20 ‘and the Hagrites and alli whi that ___i was with them were given into their hand’

Relative Clause Data

295

‫ל־ה ְכּ ָר ִמים ִשׁ ְמ ִעי ָה ָר ָמ ִתי וְ ַעל ֶשׁ ַבּ ְכּ ָר ִמים ְלא ְֹצרֹות ַהיַּ יִ ן זַ ְב ִדּי ַה ִשּׁ ְפ ִמי׃ ס‬ ַ ‫ וְ ַע‬1 Chr 27:27 ‘and over the vineyards was Shimei the Ramathite and over (the pro­ ducei) whi that ___i was the vineyards for the wine cellars was Zabdi the Shiphmite’

2.  ‫שׁ‬-Complementizer Data (40×) ‫אָדם ְלע ָֹלם ְבּ ַשׁגַּ ם הוּא ָב ָשׂר‬ ָ ‫רוּחי ָב‬ ִ ‫אמר יְ הוָ ה לֹא־יָ דֹון‬ ֶ ֹ ‫ וַ יּ‬Gen 6:3 ‘and Yhwh said, “My spirit shall not abide in man forever, because that, indeed, he is flesh’ ‫בֹורה ַשׁ ַקּ ְמ ִתּי ֵאם ְבּיִ ְשׂ ָר ֵאל׃‬ ָ ‫ ָח ְד ֧לוּ ְפ ָרזֹון ְבּיִ ְשׂ ָר ֵאל ָח ֵדלּוּ ַעד ַשׁ ַקּ ְמ ִתּי ְדּ‬Judg 5:7a ‘the peasantry ceased in Israel, they ceased, until that I, Deborah, arose, (until) that I arose as a mother in Israel’ ‫בֹורה ַשׁ ַקּ ְמ ִתּי ֵאם ְבּיִ ְשׂ ָר ֵאל׃‬ ָ ‫ ָח ְדלוּ ְפ ָרזֹון ְבּיִ ְשׂ ָר ֵאל ָח ֵדלּוּ ַעד ַשׁ ַקּ ְמ ִתּי ְדּ‬Judg 5:7b ‘the peasantry ceased in Israel, they ceased, until that I, Deborah, arose, (until) that I arose as a mother in Israel’ ‫אַתּה ְמ ַד ֵבּר ִע ִמּי׃‬ ָ ‫ית ִלּי אֹות ָשׁ‬ ָ ‫אתי ֵחן ְבּ ֵעינֶ יָך וְ ָע ִשׂ‬ ִ ‫אמר ֵא ָליו ִאם־נָ א ָמ ָצ‬ ֶ ֹ ‫ וַ יּ‬Judg 6:17 ‘and he said to him: if I have found favor in your eyes, you should do for me a sign that you are speaking with me’ (cf. ‘you should signal that you . . .’) [see below, §8a, on Isa 38:7] ‫גֹורלֹות וְ נֵ ְד ָעה ְבּ ֶשׁ ְלּ ִמי ָה ָר ָעה ַהזֹּאת ָלנוּ‬ ָ ‫ ְלכוּ וְ נַ ִפּ ָילה‬Jonah 1:7 ‘come and let us cast lots and we shall know because that for whom this calamity has come to us’ ‫יכם׃‬ ֶ ‫יֹוד ַע אָנִ י ִכּי ְב ֶשׁ ִלּי ַה ַ ֧סּ ַער ַהגָּ דֹול ַהזֶּ ה ֲע ֵל‬ ֵ ‫ ִ ֚כּי‬Jonah 1:12 ‘because I know that because that for whom this great storm is upon us’ ‫יהם ְכּ ֵעינֵ י ִשׁ ְפ ָחה ֶאל־יַ ד גְּ ִב ְר ָתּהּ ֵכּן ֵעינֵ ינוּ ֶאל־יְ הוָ ה‬ ֶ ֵ‫ ִהנֵּ ה ְכ ֵעינֵ י ֲע ָב ִדים ֶאל־יַ ד ֲאדֹונ‬Ps 123:2 ‫ֹלהינוּ ַעד ֶשׁיְּ ָחנֵּ נוּ׃‬ ֵ ‫ֱא‬ ‘look—as the eyes of servants are to the hand of their master, as the eyes of a maid is to the hand of her mistress, so our eyes look to Yhwh our God, until that he has mercy upon us’ ‫[שׁדּוּן]׃ ס‬ ַ ‫י־ח ָמה ֲעֹונֹות ָח ֶרב ְל ַמ ַען ֵתּ ְדעוּן שׁדין‬ ֵ ‫י־ח ֶרב ִכּ‬ ֶ ֵ‫ גּוּרוּ ָל ֶכם ׀ ִמ ְפּנ‬Job 19:29 ‘be afraid for yourselves from the sword, for wrath is the punishment of the sword, in order that you may know that there is a judgment [Kethiv]’ ‫ד־שׁ ַה ֶמּ ֶלְך ִבּ ְמ ִסבֹּו נִ ְר ִדּי נָ ַתן ֵריחֹו׃‬ ֶ ‫ ַע‬Song 1:12 ‘while that the king was on his couch, my nard gave forth its fragrance’ ‫ם־תּ ִ ֧עירוּ ׀ וְ ִאם־‬ ָ ‫רוּשׁ ַלםִ ִבּ ְצ ָבאֹות אֹו ְבּאַיְ לֹות ַה ָשּׂ ֶדה ִא‬ ָ ְ‫ ִה ְשׁ ַבּ ְע ִתּי ֶא ְת ֶכם ְבּנֹות י‬Song 2:7 ‫אַה ָבה ַעד ֶשׁ ֶתּ ְח ָפּץ׃ ס‬ ֲ ‫ת־ה‬ ָ ‫עֹוררוּ ֶא‬ ְ ‫ְתּ‬ ‘I adjure you, O daughters of Jerusalem, by the gazelles or by the does of the field: do not stir up or awaken love until that it is ready!’

296

Appendix A

‫ל־ה ֵרי‬ ָ ‫דֹודי ִל ְצ ִבי אֹו ְלע ֶֹפר ָהאַיָּ ִלים ַע‬ ִ ‫ה־לָך‬ ְ ‫פוּח ַהיֹּום וְ נָ סוּ ַה ְצּ ָל ִלים סֹב ְדּ ֵמ‬ ַ ָ‫ ַעד ֶשׁיּ‬Song 2:17 ‫ָב ֶתר׃ ס‬ ‘until that the day blows and the shadows flee, turn, my beloved, be like a gazelle or a young stag on the cleft mountains’ ‫אַר ֶפּנּוּ ַעד־‬ ְ ‫אָה ָבה נַ ְפ ִשׁי ֲא ַחזְ ִתּיו וְ לֹא‬ ֲ ‫אתי ֵאת ֶשׁ‬ ִ ‫ ִכּ ְמ ַעט ֶשׁ ָע ַב ְר ִתּי ֵמ ֶהם ַעד ֶשׁ ָמּ ָצ‬Song 3:4a ‫הֹור ִתי׃‬ ָ ‫ל־ח ֶדר‬ ֶ ‫ל־בּית ִא ִמּי וְ ֶא‬ ֵ ‫יאתיו ֶא‬ ִ ‫ֶשׁ ֲה ֵב‬ ‘about a little that I had passed from them until that I found (the onei) whi that my soul loves ___i. I held him, and would not let him go until that I brought him into my mother’s house, and into the chamber of her that conceived me.’ ‫אַר ֶפּנּוּ ַעד־‬ ְ ‫אָה ָבה נַ ְפ ִשׁי ֲא ַחזְ ִתּיו וְ לֹא‬ ֲ ‫אתי ֵאת ֶשׁ‬ ִ ‫ ִכּ ְמ ַעט ֶשׁ ָע ַב ְר ִתּי ֵמ ֶהם ַעד ֶשׁ ָמּ ָצ‬Song 3:4b ‫הֹור ִתי׃‬ ָ ‫ל־ח ֶדר‬ ֶ ‫ל־בּית ִא ִמּי וְ ֶא‬ ֵ ‫יאתיו ֶא‬ ִ ‫ֶשׁ ֲה ֵב‬ ‘about a little that I had passed from them until that I found (the onei) whi that my soul loves ___i. I held him, and would not let him go until that I brought him into my mother’s house, and into the chamber of her that conceived me.’ ‫אַר ֶפּנּוּ ַעד־‬ ְ ‫אָה ָבה נַ ְפ ִשׁי ֲא ַחזְ ִתּיו וְ לֹא‬ ֲ ‫אתי ֵאת ֶשׁ‬ ִ ‫ ִכּ ְמ ַעט ֶשׁ ָע ַב ְר ִתּי ֵמ ֶהם ַעד ֶשׁ ָמּ ָצ‬Song 3:4d ‫הֹור ִתי׃‬ ָ ‫ל־ח ֶדר‬ ֶ ‫ל־בּית ִא ִמּי וְ ֶא‬ ֵ ‫יאתיו ֶא‬ ִ ‫ֶשׁ ֲה ֵב‬ ‘about a little that I had passed from them until that I found (the onei) whi that my soul loves ___i. I held him, and would not let him go until that I brought him into my mother’s house, and into the chamber of her that conceived me.’ ‫עֹורר‬ ְ ‫ם־תּ‬ ְ ‫ם־תּ ִ ֧עירוּ ׀ וְ ִא‬ ָ ‫רוּשׁ ַלםִ ִבּ ְצ ָבאֹות אֹו ְבּאַיְ לֹות ַה ָשּׂ ֶדה ִא‬ ָ ְ‫ ִה ְשׁ ַבּ ְע ִתּי ֶא ְת ֶכם ְבּנֹות י‬Song 3:5 ‫אַה ָבה ַעד ֶשׁ ֶתּ ְח ָפּץ׃ ס‬ ֲ ‫ת־ה‬ ָ ‫וּ ֶא‬ ‘I adjure you, O daughters of Jerusalem, by the gazelles or by the does of the field: do not stir up or awaken love until that it is ready!’ ‫ל־הר ַהמֹּור וְ ֶאל־גִּ ְב ַעת ַה ְלּבֹונָ ה׃‬ ַ ‫פוּח ַהיֹּום וְ נָ סוּ ַה ְצּ ָל ִלים ֵא ֶלְך ִלי ֶא‬ ַ ָ‫ ַעד ֶשׁיּ‬Song 4:6 ‘until that the day breathes and the shadows flee, I will hasten to the mountain of myrrh and the hill of frankincense’ ‫אַה ָבה‬ ֲ ‫חֹולת‬ ַ ‫ה־תּגִּ ידוּ לֹו ֶשׁ‬ ַ ‫ת־דֹּודי ַמ‬ ִ ‫ם־תּ ְמ ְצאוּ ֶא‬ ִ ‫רוּשׁ ָלםִ ִא‬ ָ ְ‫ ִה ְשׁ ַבּ ְע ִתּי ֶא ְת ֶכם ְבּנֹות י‬Song 5:8 ‫אָנִ י׃‬ ‘I adjure you, O daughters of Jerusalem, if you find my beloved, what should you tell him? (Tell him) that I am faint with love’ ‫אַה ָבה ַעד ֶשׁ ֶתּ ְח ָפּץ׃ ס‬ ֲ ‫ת־ה‬ ָ ‫ה־תּע ְֹררוּ ֶא‬ ְ ‫וּמ‬ ַ ‫ה־תּ ִ ֧עירוּ ׀‬ ָ ‫רוּשׁ ָלםִ ַמ‬ ָ ְ‫ ִה ְשׁ ַבּ ְע ִתּי ֶא ְת ֶכם ְבּנֹות י‬Song 8:4 ‘I adjure you, O daughters of Jerusalem, do not stir up or awaken love until that it is ready!’ ‫רוּח׃‬ ַ ‫הֹוללֹות וְ ִשׂ ְכלוּת יָ ַד ְע ִתּי ֶשׁגַּ ם־זֶ ה הוּא ַר ְעיֹון‬ ֵ ‫ וָ ֶא ְתּנָ ה ִל ִבּי ָל ַד ַעת ָח ְכ ָמה וְ ַד ַעת‬Eccl 1:17 ‘I set my ‫ לב‬to know wisdom; and knowing blindness and folly—I came to know that even this was wind-chasing’ ‫ן־הח ֶֹשְׁך׃‬ ַ ‫יתרֹון ָהאֹור ִמ‬ ְ ‫ן־ה ִסּ ְכלוּת ִכּ‬ ַ ‫יתי אָנִ י ֶשׁיֵּ שׁ יִ ְתרֹון ַל ָח ְכ ָמה ִמ‬ ִ ‫ וְ ָר ִא‬Eccl 2:13 ‘I saw, I, that the profit of wisdom is more (than the profit of) foolishness, like the profit of light is more than (the profit of) darkness’

Relative Clause Data

297

‫הֹולְך וְ יָ ַד ְע ִתּי גַ ם־אָנִ י ֶשׁ ִמּ ְק ֶרה ֶא ָחד יִ ְק ֶרה‬ ֵ ‫ ֶה ָח ָכם ֵעינָ יו ְבּרֹאשֹׁו וְ ַה ְכּ ִסיל ַבּח ֶֹשְׁך‬Eccl 2:14 ‫ת־כּ ָלּם׃‬ ֻ ‫ֶא‬ ‘as for the wise man—his eyes are in his head. But the fool in darkness is walking; yet I have come to know, even I, that one fate befalls them all’ ‫יֹותר וְ ִד ַבּ ְר ִתּי‬ ֵ ‫ם־אנִ י יִ ְק ֵרנִ י וְ ָל ָ֧מּה ָח ַכ ְמ ִתּי ֲאנִ י אָז‬ ֲ ַ‫אָמ ְר ִתּי אנִ י ְבּ ִל ִבּי ְכּ ִמ ְק ֵרה ַה ְכּ ִסיל גּ‬ ַ ְ‫ ו‬Eccl 2:15 ‫ְב ִל ִבּי ֶשׁגַּ ם־זֶ ה ָה ֶבל׃‬ ‘and I said, I with my ‫לב‬, “Just like the fool’s fate, even I—it will befall me! Why have I become wise, I, then, so much?” I spoke with my heart that this too was absurd’ ‫עֹולם ְבּ ֶשׁ ְכּ ָבר ַהיָּ ִמים ַה ָבּ ִאים ַהכֹּל נִ ְשׁ ָכּח וְ ֵאיְך יָ מוּת‬ ָ ‫ם־ה ְכּ ִסיל ְל‬ ַ ‫ ִכּי ֵאין זִ ְכ ֧רֹון ֶל ָח ָכם ִע‬Eccl 2:16 ‫ם־ה ְכּ ִסיל׃‬ ַ ‫ֶה ָח ָכם ִע‬ ‘because remembrance is never for the wise alongside the fool; because that already (in) the coming days, all is forgotten. How the wise dies with the fool!’ ‫אכל וְ ָשׁ ָתה וְ ֶה ְראָ֧ה ֶאת־נַ ְפשֹׁו טֹוב ַבּ ֲע ָמלֹו‬ ַ ֹ ‫אָדם ֶשׁיּ‬ ָ ‫ ֵאין־טֹוב ָבּ‬Eccl 2:24 ‘a better thing does not exist for man (than) that he eat and drink and show himself good in his toil’ ‫ֹלהים ִהיא׃‬ ִ ‫ל־ע ָמלֹו ַמ ַתּת ֱא‬ ֲ ‫אכל וְ ָשׁ ָתה וְ ָראָה טֹוב ְבּ ָכ‬ ַ ֹ ‫אָדם ֶשׁיּ‬ ָ ‫ל־ה‬ ָ ‫ וְ גַ ם ָכּ‬Eccl 3:13 ‘and also every man—that he eats and drinks and experiences goodness in all his acquisition(s)—it is the gift of God’ ‫ֹלהים ָע ָשׂה ֶשׁיִּ ְראוּ ִמ ְלּ ָפנָ יו׃‬ ִ ‫ וְ ָה ֱא‬Eccl 3:14 ‘God has done (it), (in order) that they should be afraid of him’ ‫ם־בּ ֵה ָמה ֵה ָמּה‬ ְ ‫ֹלהים וְ ִל ְראֹות ְשׁ ֶה‬ ִ ‫אָדם ְל ָב ָרם ָה ֱא‬ ָ ‫ל־דּ ְב ַרת ְבּנֵ י ָה‬ ִ ‫אָמ ְר ִתּי ֲאנִ י ְבּ ִל ִבּי ַע‬ ַ Eccl 3:18 ‫ָל ֶהם׃‬ ‘I said, I with my ‫לב‬, concerning humans, “God should test them and see that they are cattle, they to themselves”’ ‫א־תדּ ֹר ִמ ֶשׁ ִתּדֹּור וְ לֹא ְת ַשׁ ֵלּם׃‬ ִ ֹ ‫ טֹוב ֲא ֶשׁר ל‬Eccl 5:4 ‘that you do not vow is better than that you do and do not fulfill (it)’ ‫אוּמה לֹא־יִ ָשּׂא ַב ֲע ָמלֹו ֶשׁיּ ֵֹלְך‬ ָ ‫וּמ‬ ְ ‫ ַכּ ֲא ֶשׁר יָ ָצא ִמ ֶבּ ֶטן ִאמֹּו ָערֹום יָ שׁוּב ָל ֶל ֶכת ְכּ ֶשׁ ָבּא‬Eccl 5:14a ‫ְבּיָ דֹו׃‬ ‘like he exited from the womb of his mother, naked he will again go, like that he entered; nothing will he be able to take in exchange for his accomplishment(s)i whi that he may bring ___i into his [son’s] hand’ ‫רוּח׃‬ ַ ‫וּמה־יִּ ְתרֹון לֹו ֶשׁיַּ ֲעמֹל ָל‬ ַ ‫ל־ע ַמּת ֶשׁ ָבּא ֵכּן יֵ ֵלְך‬ ֻ ‫חֹולה ָכּ‬ ָ ‫ וְ גַ ם־זֹה ָר ָעה‬Eccl 5:15a ‘and this also is a sickly misfortune: like that he came, so he will go; what is profit for himi whi that ___i works for wind?’ ‫י־שׁנָ יו‬ ָ ‫ם־יֹוליד ִאישׁ ֵמאָה וְ ָשׁנִ ים ַרבֹּות יִ ְחיֶ ה וְ ַרב ׀ ֶשׁיִּ ְהיוּ יְ ֵמ‬ ִ ‫ ִא‬Eccl 6:3 ‘if a man begets a hundred and lives many years and complains* that the days of his years would occur . . .’ [emend: ‫ ַרב‬to ‫ָ]רב‬ ‫אַל ָתּ‬ ְ ‫טֹובים ֵמ ֵא ֶלּה ִכּי לֹא ֵמ ָח ְכ ָמה ָשׁ‬ ִ ‫אמר ֶמה ָהיָ ה ֶשׁ ַהיָּ ִמים ָה ִראשׁ ֹנִ ים ָהיוּ‬ ַ ֹ ‫ אַל־תּ‬Eccl 7:10 ‫ַעל־זֶ ה׃‬ ‘don’t say, “What has happened?”—that is, that the former days have been better than these’

298

Appendix A

‫אוּמה׃‬ ָ ‫אַח ָריו ְמ‬ ֲ ‫אָדם‬ ָ ‫ל־דּ ְב ַרת ֶשׁלֹּא יִ ְמ ָ ֧צא ָה‬ ִ ‫ֹלהים ַע‬ ִ ‫ גַּ ם ֶאת־זֶ ה ְל ֻע ַמּת־זֶ ה ָע ָשׂה ָה ֱא‬Eccl 7:14 ‘indeed, God made this (day of goodness) corresponding to that (day of adversity), for the purpose that man does not find anything after him’ ‫יקים ֲא ֶשׁר ַמגִּ ַיע ֲא ֵל ֶהם ְכּ ַמ ֲע ֵשׂה‬ ִ ‫אָרץ ֲא ֶשׁר ׀ יֵ שׁ ַצ ִדּ‬ ֶ ‫ל־ה‬ ָ ‫שׁ־ה ֶבל ֲא ֶשׁר נַ ֲע ָשׂה ַע‬ ֶ ֶ‫ י‬Eccl 8:14b ‫אָמ ְר ִתּי ֶשׁגַּ ם־זֶ ה ָה ֶבל׃‬ ַ ‫יקים‬ ִ ‫ָה ְר ָשׁ ִעים וְ יֵ שׁ ְר ָשׁ ִעים ֶשׁ ַמּגִּ ַיע ֲא ֵל ֶהם ְכּ ַמ ֲע ֵשׂה ַה ַצּ ִדּ‬ ‘there is an absurdity that has happened on the earth, that there are righteous people who (a thing) befalls them just like the deed of the wicked, and there are wickedi whi that (a thing) befalls themi like the deed of the righteous. I say that this too is absurd!’ ‫ת־ה ַמּ ֲע ֶשׂה ֲא ֶשׁר נַ ֲע ָשׂה‬ ַ ‫אָדם ִל ְמצֹוא ֶא‬ ָ ‫יוּכל ָה‬ ַ ‫ֹלהים ִכּי לֹא‬ ִ ‫ל־מ ֲע ֵשׂה ָה ֱא‬ ַ ‫ת־כּ‬ ָ ‫יתי ֶא‬ ִ ‫ וְ ָר ִא‬Eccl 8:17 ‫אמר ֶה ָח ָכם ָל ַד ַעת‬ ַ ֹ ‫אָדם ְל ַב ֵקּשׁ וְ לֹא יִ ְמ ָצא וְ גַ ם ִאם־י‬ ָ ‫מל ָה‬ ֹ ֧ ‫ת־ה ֶשּׁ ֶמשׁ ְבּ ֶשׁל ֲא ֶשׁר יַ ֲע‬ ַ ‫ַת ַח‬ ‫יוּכל ִל ְמצֹא׃‬ ַ ‫לֹא‬ ‘and I saw the whole work of God, that man is not able to “find out” the deed that has happened under the sun, so that man toils to seek but does not find. Even if the wise man intended to understand, he could not “find (it) out”’ ‫אוּמה וְ ֵאין־עֹוד ָל ֶהם ָשׂ ָכר ִכּי‬ ָ ‫יֹוד ִעים ְמ‬ ְ ‫יֹוד ִעים ֶשׁיָּ ֻמתוּ וְ ַה ֵמּ ִתים ֵא ָינ֧ם‬ ְ ‫ ִ ֧כּי ַה ַחיִּ ים‬Eccl 9:5 ‫נִ ְשׁ ַכּח זִ ְכ ָרם׃‬ ‘because the living know that they will die, but the dead—they do not know anything, and they have no more reward, because memory of them has been forgotten’ ‫אָדם‬ ָ ‫יוּק ִשׁים ְבּנֵ י ָה‬ ָ ‫צֹודה ָר ָעה וְ ַכ ִצּ ֳפּ ִרים ָה ֲא ֻחזֹות ַבּ ָפּח ָכּ ֵהם‬ ָ ‫ ַכּ ָדּגִ ים ֶשׁנֶּ ֱא ָחזִ ים ִבּ ְמ‬Eccl 9:12b ‫יהם ִפּ ְתאֹם׃‬ ֶ ‫ְל ֵעת ָר ָעה ְכּ ֶשׁ ִתּפֹּול ֲע ֵל‬ ‘because, indeed, man does not know his time. Like fish that are caught in a bad net and like birds that are snagged in a trap—like them humans are ensnared at a time of misfortune, when that it falls upon them suddenly’ ‫אָמר ַלכֹּל ָס ָכל הוּא׃‬ ַ ְ‫[כּ ֶשׁ ָסּ ָכל] ה ֵֹלְך ִלבֹּו ָח ֵסר ו‬ ְ ‫ם־ב ֶ ֛דרְך כשׁהסכל‬ ֶַ ַ‫ וְ ג‬Eccl 10:3 ‘and also, on the road, when that the fool [Kethiv] is walking, his ‫ לב‬is lacking, so that he says to everyone (that) he is a fool!’ ‫ֹלהים ֲא ֶשׁר נְ ָתנָ הּ׃‬ ִ ‫ל־ה ֱא‬ ָ ‫רוּח ָתּשׁוּב ֶא‬ ַ ‫אָרץ ְכּ ֶשׁ ָהיָ ה וְ ָה‬ ֶ ‫ל־ה‬ ָ ‫ וְ יָ שׁ ֹב ֶה ָע ָפר ַע‬Eccl 12:7 ‘and the dirt shall return to the land like that it was and the breath re­ turns to God who gave it’ ‫ת־ה ָעם וְ ִאזֵּ ן וְ ִח ֵקּר ִתּ ֵקּן ְמ ָשׁ ִלים ַה ְר ֵבּה׃‬ ָ ‫ד־דּ ַעת ֶא‬ ַ ‫ וְ י ֵֹתר ֶשׁ ָהיָ ה ק ֶֹה ֶלת ָח ָכם עֹוד ִל ַמּ‬Eccl 12:9 ‘and that Qoheleth was (only) wise (would be) excessive; he continually taught the people knowledge and would test and seek (and) arrange many proverbs’

3.  ‫ז‬-Relative Data (24×) ‫אתי ִמ ִמּ ְצ ָריִ ם׃‬ ִ ‫ ַבּ ֲעבוּר זֶ ה ָע ָשׂה יְ הוָ ה ִלי ְבּ ֵצ‬Exod 13:8 ‘on account of (the thing)i whi that Yhwh did ___i for me when I left Egypt’ (alt., ‘on account of thisi whi [that] Yhwh did ___i for me when I left Egypt’)

Relative Clause Data

299

‫אָל ָתּ‬ ְ ָ‫ית ְב ַח ְס ְדָּך ַעם־זוּ גּ‬ ָ ‫ נָ ִח‬Exod 15:13 ‘you lead in your faithfulness a peoplei whi that you redeemed ___i’ ‫ית׃‬ ָ ִ‫ ַעד־יַ ֲעבֹר ַעם־זוּ ָקנ‬Exod 15:16 ‘until a peoplei whi that you acquired ___i passes by’ ‫ל־מת׃‬ ֵ ‫אַבּ ֶלת ַע‬ ֶ ‫ וְ ָהיִ ית ְכּ ִא ָשּׁה ֶז֚ה יָ ִמים ַר ִבּים ִמ ְת‬2 Sam 14:2 ‘and you shall be like a womani whi that ___i has been mourning (for) many days about a dead person’ ‘Is it Yhwhi, whi that we sinned against himi?’

‫ ֲהלֹוא יְ הוָ ה זוּ ָח ָטאנוּ לֹו‬Isa 42:24

‫ ַעם־זוּ יָ ַצ ְר ִתּי ִלי ְתּ ִה ָלּ ִתי יְ ַס ֵפּרוּ׃ ס‬Isa 43:21 ‘let (the) peoplei whi that I formed ___i for myself declare my praise’ ‫אָשׁם זוּ כֹחֹו ֵלאֹלהֹו׃‬ ֵ ְ‫ ו‬Hab 1:11 ‘and (he)i whi that hisi might is his god became guilty’ ‫ ְבּ ֶר ֶשׁת־זוּ ָט ָמנוּ נִ ְל ְכּ ָדה ַרגְ ָלם׃‬Ps 9:16 ‘in the neti whi that they hid ___i their foot has been caught’ ‫ יִ ָתּ ְפשׂוּ ׀ ִבּ ְמזִ מֹּות זוּ ָח ָשׁבוּ׃‬Ps 10:2 ‘let them be caught in the schemesi whi that they have devised ___i’ ‫ ִמ ְפּנֵ י ְר ָשׁ ִעים זוּ ַשׁדּוּנִ י‬Ps 17:9 ‘because of wicked peoplei whi that ___ i destroy me’ ‫יאנִ י ֵמ ֶר ֶשׁת זוּ ָט ְמנוּ ִלי‬ ֵ ‫תֹּוצ‬ ִ Ps 31:5 ‘may you take me out of this neti whi that they have hidden ___i for me’ ‫אֹורָך ְבּ ֶד ֶרְך־זוּ ֵת ֵלְך‬ ְ ְ‫ ו‬Ps 32:8 ‘and I shall instruct you in the wayi whi that you should walk ___i’ ‫אֹלהים׃‬ ִ ‫ֹלהים ְשׁ ַתּיִ ם־זוּ ָשׁ ָמ ְע ִתּי ִכּי עֹז ֵל‬ ִ ‫אַחת ׀ ִדּ ֶבּר ֱא‬ ַ Ps 62:12 ‘one time (that) God spoke, two timesi whi that I heard ___i that power belongs to God’ ‫ֹלהים זוּ ָפּ ַע ְל ָתּ ָלּנוּ׃‬ ִ ‫ עוּזָּ ה ֱא‬Ps 68:29 ‘be strong, O God, (you)i whi that (you)i have worked for us’ ‘Mount Zioni, whi that you dwelt in iti’

‫ר־ציֹּון זֶ ה ׀ ָשׁ ַכנְ ָתּ בֹּו׃‬ ִ ‫ ַה‬Ps 74:2

‫יְמינֹו׃‬ ִ ‫ ַהר־זֶ ה ָקנְ ָתה‬Ps 78:54 ‘(the) mounti whi that his right hand had acquired ___i’ ‫ל־מקֹום זֶ ה ׀ יָ ַס ְד ָתּ ָל ֶהם׃‬ ְ ‫ ֶא‬Ps 104:8 ‘to a placei whi that you established ___i for them’ ‫ ָשׁם ֳאנִ יֹּות יְ ַה ֵלּכוּן ִלוְ יָ ָתן זֶ ה־יָ ַצ ְר ָתּ ְל ַשׂ ֶחק־בֹּו׃‬Ps 104:26 ‘there ships travel, (and) Leviathani, whi that you formed ___i to play in it’ ‫יתי וְ ֵעד ִֹתי זֹו ֲא ַל ְמּ ֵדם‬ ִ ‫ ְבּ ִר‬Ps 132:12 ‘my covenant and my lawsi whi that I will teach them ___i’ ‫ נְ ִת ָיב ִתי ְבּא ַֹרח־זוּ ֲא ַה ֵלְּך ָט ְמנוּ ַפח ִלי׃‬Ps 142:4 ‘in the pathi whi that I walk ___i they have hidden a trap for me’

300

Appendix A

‫יענִ י ֶדּ ֶרְך־זוּ ֵא ֵלְך‬ ֵ ‫הֹוד‬ ִ Ps 143:8 ‘make me know (the) wayi whi that I should walk ___i’ ‫יתי וַ ֲא ַס ֵפּ ָרה׃‬ ִ ִ‫ה־חז‬ ָ ֶ‫ וְ ז‬Job 15:17 ‘and (the thing)i whi that I have seen ___i I will declare’ (alt., ‘and this I have seen and I will declare [it]’) ‫כוּ־בי׃‬ ִ ‫ה־אָה ְב ִתּי נֶ ְה ְפּ‬ ַ ֶ‫ וְ ז‬Job 19:19 ‘and (those)i whi that I loved ___i have turned (their back) on me’ ‘listen to your fatheri, whi that ___i bore you’

‫אָביָך זֶ ה יְ ָל ֶדָך‬ ִ ‫ ְשׁ ַמע ְ ֭ל‬Prov 23:22

4.  ‫מ‬-Pronoun Relative Data (37×) For the ‫מ‬-relative data, I have included my judgment concerning whether the relative is a free choice (FC) relative or an identificational (IDN) relative. ‫י־ב ַעל ְדּ ָב ִרים יִ גַּ שׁ ֲא ֵל ֶהם׃‬ ַ ‫ ִמ‬Exod 24:14 FC ‘whoeveri whi ___i is an owner of issues shall draw near to them’ (≈ whoever has an issue . . .) ‫ל־בּנֵ י ֵלוִ י׃‬ ְ ‫אָספוּ ֵא ָליו ָכּ‬ ְ ֵ‫ ִמי ַליהוָ ה ֵא ָלי וַ יּ‬Exod 32:26 IDN ‘whoeveri whi ___i is for Yhwh, (come) to me’ (NR: ‘who is for Yhwh? [Come] to me!’) ‫א־לי ֶא ְמ ֶחנּוּ ִמ ִסּ ְפ ִרי׃‬ ִ ‫ ִמי ֲא ֶשׁר ָח ָט‬Exod 32:33 IDN ‘whoeveri whi that ___i has sinned against me—I will blot him from my book’ (NR: ‘who is the one who . . . ? I will blot him from my book’) ‫ ְוּד ַבר ַמה־יַּ ְר ֵאנִ י וְ ִהגַּ ְד ִתּי ָלְך‬Num 23:3 IDN ‘and the wordi-whati he shows ___i me I will tell you’ (≈ whatever he shows me . . .) ‫ת־ח ָדׁש וְ לֹא ֲחנָ כֹו יֵ ֵלְך וְ יָ ׁש ֹב ְל ֵביתֹו‬ ָ ִ‫י־ה ִאיׁש ֲא ֶׁשר ָּבנָ ה ַבי‬ ָ ‫ ִמ‬Deut 20:5 IDN ‘let whichever mani whi that ___i built a new house and has not dedicated it go and return to his house’ (NR: ‘who is the man . . . ? Let him go . . .’) ‫י־ה ִאיׁש ֲא ֶׁשר־נָ ַטע ֶּכ ֶרם וְ לֹא ִח ְּללֹו יֵ ֵלְך וְ יָ ׁש ֹב ְל ֵביתֹו‬ ָ ‫‏ּומ‬ ִ Deut 20:6 IDN ‘and let whichever mani whi that ___i has planted a vineyard but not used it go and return to his house’ (NR: ‘who is the man . . .? Let him go . . .’) ‎‫ר־א ַרׂש ִא ָּׁשה וְ לֹא ְל ָק ָחּה יֵ ֵלְך וְ יָ ׁש ֹב ְל ֵביתֹו‬ ֵ ‫י־ה ִאיׁש ֲא ֶׁש‬ ָ ‫ּומ‬ ִ Deut 20:7 IDN ‘let whichever mani whi that ___i has betrothed a woman and not married her go and return to his house’ (NR: ‘who is the man . . . ? Let him go . . .’) ‫ת־מי ַת ֲעבֹדוּן‬ ִ ‫ ַבּ ֲחרוּ ָל ֶכם ַהיֹּום ֶא‬Josh 24:15 IDN ‘choose for yourselves today whoeveri whi you will serve ___i’

Relative Clause Data

301

‫ ִמי־יָ ֵרא וְ ָח ֵרד יָ שׁ ֹב‬Judg 7:3 IDN ‘whoeveri whi ___i is afraid and trembling should return’ (NR: ‘who is the man . . . ? Let him return’) ‫יתי ַמ ֲהרוּ ֲעשׂוּ ָכמֹונִ י׃‬ ִ ‫יתם ָע ִשׂ‬ ֶ ‫ ָמה ְר ִא‬Judg 9:48 IDN ‘whati whi you saw ___i (that) I did—hurry, do like me’ (≈ whatever you saw me do . . .) ‫‏מי ָה ִאיׁש ֲא ֶׁשר יָ ֵחל ְל ִה ָּל ֵחם ִּב ְבנֵ י ַעּמֹון יִ ְהיֶ ה ְלרֹאׁש ְלכֹל י ְֹׁש ֵבי גִ ְל ָעד׃ פ‬‎ ִ Judg 10:18 IDN ‘whichever mani whi that ___i begins to fight with the Ammonites will be a leader for all the inhabitants of Gilead’ (NR: ‘who is the man . . . ? He shall be a leader’) ‫ה־לְּך׃ פ‬ ָ ‫אמר נַ ְפ ְשָׁך וְ ֶא ֱע ֶשׂ‬ ַ ֹ ‫ ַמה־תּ‬1 Sam 20:4 FC ‘whateveri whi your soul says ___i, I will do for you’ (NR: ‘what does your soul say? I will do [it] for you’) ‫אַח ֵרי יֹואָב׃‬ ֲ ‫ר־ל ָדוִ ד‬ ְ ‫וּמי ֲא ֶשׁ‬ ִ ‫ ִמי ֲא ֶשׁר ָח ֵ ֧פץ ְבּיֹואָב‬2 Sam 20:11a IDN ‘whoeveri whi that ___i delights in Joab . . . (follow) after Joab’ (NR: ‘who is the one who delights in Joab?’) ‫אַח ֵרי יֹואָב׃‬ ֲ ‫ר־ל ָדוִ ד‬ ְ ‫וּמי ֲא ֶשׁ‬ ִ ‫ ִמי ֲא ֶשׁר ָח ֵ ֧פץ ְבּיֹואָב‬2 Sam 20:11b IDN ‘whoeveri whi that ___i is for David, (follow) after Joab!’ (NR: ‘who is the one for David?) ‫יתן הוּא גֹּוי‬ ָ ‫יכם גּ֧ ֹוי ִמ ֶמּ ְר ָחק ֵבּית יִ ְשׂ ָר ֵאל נְ ֻאם־יְ הֹוָ ה גֹּוי ׀ ֵא‬ ֶ ‫ ִהנְ נִ י ֵמ ִביא ֲע ֵל‬Jer 5:15 FC ‫א־ת ַדע ְלשׁ ֹנֹו וְ לֹא ִת ְשׁ ַמע ַמה־יְ ַד ֵבּר׃‬ ֵ ֹ ‫עֹולם הוּא ֚גֹּוי ל‬ ָ ‫ֵמ‬ ‘It is an enduring nation, it is an ancient nation, a nation (that) you do not know its language and you cannot hear whateveri whi it says ___i’ ‫י־ה ִאישׁ ֶה ָח ָכם וְ ֵיָבן ֶאת־זֹאת‬ ָ ‫ ִמ‬Jer 9:11 FC ‘whoeveri whi ___i is the wise man—let him understand this’ (NR: ‘who is the wise man? Let him understand’) ‫ ִמי ָח ָכם וְ ֵיָבן ֵא ֶלּה‬Hos 14:10 FC ‘whoeveri whi ___i is wise—let him understand these things’ (NR: ‘who is wise? Let him understand’) ‫י־פ ִתי יָ ֻסר ֵהנָּ ה‬ ֶ ‫ ִמ‬Prov 9:4 FC ‘whoeveri whi ___i is simple should turn aside here’ (NR: ‘who is simple? He should turn aside’) ‫י־פ ִתי יָ ֻסר ֵהנָּ ה‬ ֶ ‫ ִמ‬Prov 9:16 FC ‘whoeveri whi ___i is simple should turn aside here’ (NR: ‘who is simple? He should turn aside’) ‫ה־שׁנַּ ֲע ָשׂה הוּא ֶשׁיֵּ ָע ֶשׂה‬ ֶ ‫וּמ‬ ַ ‫ה־שּׁ ָהיָ ה הוּא ֶשׁיִּ ְהיֶ ה‬ ֶ ‫ ַמ‬Eccl 1:9a FC ‘whateveri whi that ___i has been—it is (the thing)j whj that ___j will be, and whateverk whk that ___k has happened—it is (the thing)l whl that ___l will happen’

302

Appendix A

‫ה־שׁנַּ ֲע ָשׂה הוּא ֶשׁיֵּ ָע ֶשׂה‬ ֶ ‫וּמ‬ ַ ‫ה־שּׁ ָהיָ ה הוּא ֶשׁיִּ ְהיֶ ה‬ ֶ ‫ ַמ‬Eccl 1:9b FC ‘whateveri whi that ___i has been—it is (the thing)j whj that ___j will be, and whateverk whk that ___k has happened—it is (the thing)l whl that ___l will happen’ ‎‫ה־ּׁש ָהיָ ה ְּכ ָבר הּוא‬ ֶ ‫ ַמ‬Eccl 3:15 FC ’whateveri whi that ___i has been—it is already’ ‫אַח ָריו׃‬ ֲ ‫יאנּוּ ִל ְראֹות ְבּ ֶמה ֶשׁיִּ ְהיֶ ה‬ ֶ ‫ ִכּי ִמי ִיְב‬Eccl 3:22 FC ‘because who will bring him to see whati whi that ___i will be after him’ ‫וּמי־א ֵֹהב ֶבּ ָהמֹון לֹא ְתבוּאָה‬ ִ Eccl 5:9 FC ‘and whoeveri whi ___i loves abundance (will) not (be sated) by the product’ (NR: ‘who is he who loves abundance? He will not [be sated] by the product’) ‎‫ה־ּׁש ָהיָ ה ְּכ ָבר נִ ְק ָרא ְׁשמֹו‬ ֶ ‫‏מ‬ ַ Eccl 6:10 FC ‘whateveri whi that ___i has been—its name has already been called’ ‘distant is whateveri whi that ___i has been’

‎‫ה־ּׁש ָהיָ ה‬ ֶ ‫ ָרחֹוק ַמ‬Eccl 7:24 FC

‎‫ה־ּׁשּיִ ְהיֶ ה‬ ֶ ‫י־אינֶ ּנּו י ֵֹד ַע ַמ‬ ֵ ‫ ‏ ִּכ‬Eccl 8:7 FC ‘that he does not know whateveri whi that ___i will be’ ‎‫ל־ה ַחּיִ ים יֵ ׁש ִּב ָּטחֹון‬ ַ ‫י־מי ֲא ֶׁשר יבחר [יְ ֻח ַבּר] ֶאל ָּכ‬ ִ ‫‏ּכ‬ ִ Eccl 9:4 FC ‘because whoeveri whi that ___i is joined [Qere] to all the living has confidence’ ‎‫ה־ׁשּיִ ְהיֶ ה‬ ֶ ‫לֹא־יֵ ַדע ָה ָא ָדם ַמ‬ ‘the man does not know whateveri whi that ___i will be’ ְ ‫ֲצי ַהּמ ְַלכּות ְוִינָּתֵ ן ל‬ ‫ָך׃‬ ִ ‫ּשתֵ ְך עַד־ח‬ ָׁ ‫ּומה־ ַב ָּק‬ ַ ‘and whateveri whi ___i is your request, up to half of the kingdom, shall be given to you’ ְ ‫ה־ּׁש ֵאלָתֵ ְך ְוִיּנָתֵ ן ל‬ ‫ֲצי ַהּמ ְַלכּות ְותֵ עָׂש׃‬ ִ ‫ּשתֵ ְך עַד־ח‬ ָׁ ‫ּומה־ ַב ָּק‬ ַ ‫ָך‬ ְ ‫ַמ‬ ‘whateveri whi ___i is your petition shall be given to you; and whateveri whi ___i is your request, up to half of the kingdom, shall be done’ ְ ‫ֶסתֵ ּר ַהּמ ְַל ָכּה ְו ִתּנָתֵ ן ל‬ ‫ֲצי ַהּמ ְַלכּות‬ ִ ‫ּשתֵ ְך עַד־ח‬ ָׁ ‫ּומה־ ַב ָּק‬ ַ ‫ָך‬ ְ ‫ה־ּׁש ֵאלָתֵ ְך א‬ ְ ‫ַמ‬ ‫ְותֵ עָׂש׃‬ ‘whateveri whi ___i is your petition, O Esther, the queen, shall be given to you; and whateveri whi ___i is your request is, up to half of the kingdom, shall be done’ ְ ‫ה־ּׁש ֵאלָתֵ ְך ְוִיּנָתֵ ן ל‬ ‫שתֵ ְך עֹוד ְותֵ עָׂש׃‬ ָׁ ‫ּומה־ ַב ָּּק‬ ַ ‫ָך‬ ְ ‫ּומ‬ ַ ‘and whateveri whi ___i is your petition shall be given to you; and whateveri whi ___i is your additional request shall be done’

Eccl 10:14 FC Esth 5:3 FC

Esth 5:6 FC

Esth 7:2 FC

Esth 9:12 FC

Relative Clause Data

303

‫ֹלהיו ִעמֹּו‬ ָ ‫הו֧ה ֱא‬ ָ ְ‫ל־עמֹּו י‬ ַ ‫י־ב ֶכם ִמ ָכּ‬ ָ ‫ ִמ‬2 Chr 36:23 IDN ‘whoeveri whi ___i is among you from all his people—Yhwh his God be with him’ (NR: ‘who is among you from all his people? Yhwh be with him.’

5.  ‫ה‬-Relative Data (nonparticipial/nonadjectival) (23×) ‫ ֵא ֲר ָדה־נָּ א וְ ֶא ְר ֶאה ַה ְכּ ַצ ֲע ָק ָתהּ ַה ָבּאָה ֵא ַלי ָעשׂוּ ׀ ָכּ ָלה‬Gen 18:21 ‘I shall go down and see whether like her outcryi whi that ___i has come to me they have done completely’ ‫נֹּולד־לֹו ֲא ֶשׁר־יָ ְל ָדה־לֹּו ָשׂ ָרה יִ ְצ ָחק׃‬ ַ ‫ם־בּנ֧ ֹו ַה‬ ְ ‫ת־שׁ‬ ֶ ‫אַב ָר ָהם ֶא‬ ְ ‫ וַ יִּ ְק ָרא‬Gen 21:3 ‘and Abraham called the name of his soni whi that ___i was born to him that Sarah bore (him) for him “Isaac”’ ‫ל־ה ֶנּ ֶ֧פשׁ ְל ֵבית־יַ ֲעקֹב ַה ָבּאָה ִמ ְצ ַריְ ָמה ִשׁ ְב ִעים׃‬ ַ ‫ ָכּ‬Gen 46:27 ‘all the personsi belonging to the house of Jacob whi that ___i came into Egypt were seventy’ ‫ל־ק ִצ ֵ֞יני אַנְ ֵשׁי ַה ִמּ ְל ָח ָמה ֶה ָה ְלכוּא ִאתֹּו‬ ְ ‫אמר ֶא‬ ֶ ֹ ‫ וַ יּ‬Josh 10:24 ‘and he said to the chiefsi of the men of war whi that ___i had gone with him’ ‫יה‬ ָ ‫ת־השֹּׁוק וְ ֶה ָע ֶל‬ ַ ‫ וַ יָּ ֶרם ַה ַטּ ָבּח ֶא‬1 Sam 9:24 ‘and the cook took up the thigh and (the things)i whi that ___i were upon it ‫ֹלהי יִ ְשׂ ָר ֵאל ַהנִּ ְראָה ֵא ָליו ַפּ ֲע ָמיִ ם׃‬ ֵ ‫ וַ יִּ ְתאַנַּ ף יְ הוָ ה ִבּ ְשֹׁלמֹה ִכּי־נָ ָטה ְל ָבבֹו ֵמ ִעם יְ הוָ ה ֱא‬1 Kgs 11:9 ‘and Yhwh was angry with Solomon, because his heart had inclined away from Yhwhi, the God of Israel, whi that ___i had appeared to him twice’ ‫אוּלים׃‬ ִ ְ‫־היא ַה ַמּ ֲח ֶר ֶבת יָ ם ֵמי ְתּהֹום ַר ָבּה ַה ָשּׂ ָמה ַמ ֲע ַמ ֵקּי־יָ ם ֶדּ ֶרְך ַל ֲעבֹר גּ‬ ִ ‫אַתּ‬ ְ ‫ ֲהלֹוא‬Isa 51:10 ‘were you not the onei . . . whi that ___i made the depths of the sea a way for the redeemed to cross over?’ ‫ן־הנֵּ ָכר ַהנִּ ְלוָ ה ֶאל־יְ הוָ ה ֵלאמֹר ַה ְב ֵדּל ְיַב ִדּ ַילנִ י יְ הוָ ה ֵמ ַעל ַעמֹּו‬ ַ ‫אמר ֶבּ‬ ַ ֹ ‫ וְ אַל־י‬Isa 56:3 ‘and do not let the foreigneri whi that ___i was joined to Yhwh say, “Yhwh will surely separate me from his people”’ ‫ וְ ַה ִדּ ֵבּר ֵאין ָבּ ֶהם כֹּה יֵ ָע ֶשׂה ָל ֶהם׃ ס‬Jer 5:13 ‘and (the one)i whi that ___i spoke—nothing is in them’ ‫נֹושׁ ֶבת ִמיַּ ִמּים ָה ִעיר ַה ֻה ָלּ ָלה ֲא ֶשׁר ָהיְ ָתה ֲחזָ ָקה ַביָּ ם‬ ֶ ‫אָב ְד ְתּ‬ ַ ‫ ֵאיְך‬Ezek 26:17 ‘how you perished, One inhabited from the seas, the cityi whi that ___i was praised, who was strong on the sea’ ‫ל־ה ָר ָעה ַהזֹּאת ַה ָבּאָה ָע ָליו‬ ָ ‫ֹלשׁת ׀ ֵר ֵעי ִאיֹּוב ֵאת ָכּ‬ ֶ ‫ וַ יִּ ְשׁ ְמעוּ ְשׁ‬Job 2:11 ‘and Job’s three friends heard all this troublei whi that ___i had come upon him’

304

Appendix A

‫מֹּוא ִביָּ ה ַכ ָלּ ָתהּ ִע ָמּהּ ַה ָשּׁ ָבה ִמ ְשּׂ ֵדי מֹואָב וְ ֵה ָמּה ָ ֚בּאוּ ֵבּית ֶל ֶחם‬ ֲ ‫ וַ ָתּ ָשׁב נָ ֳע ִמי וְ רוּת ַה‬Ruth 1:22 ‫ִבּ ְת ִח ַלּת ְק ִציר ְשׂע ִֹרים׃‬ ‘and Naomi returned, and Ruth the Moabitei, her daughter-in-law with her, whi that ___i came back with her from the country of Moab’ ‫מֹוא ִביָּ ה ִהיא ַה ָשּׁ ָבה ִעם־נָ ֳע ִמי ִמ ְשּׂ ֵדה מֹואָב׃‬ ֲ ‫ נַ ֲע ָרה‬Ruth 2:6 ‘a Moabite maideni is she, whi that ___i came back with Naomi from the country of Moab’ ‫ימ ֶלְך ָמ ְכ ָרה נָ ֳע ִמי ַה ָשּׁ ָבה ִמ ְשּׂ ֵדה מֹואָב׃‬ ֶ ‫אָחינוּ ֶל ֱא ִל‬ ִ ‫ ֶח ְל ַקת ַה ָשּׂ ֶדה ֲא ֶשׁר ְל‬Ruth 4:3 ‘the parcel of land that belonged to our kinsman Elimelech Naomii, whi that ___i has come back from the country of Moab, is selling’ ‫אַח ֵרי ַהנִּ ְראָה‬ ֲ ‫אשׁ ַצּר ַה ֶמּ ֶלְך ָחזֹון נִ ְראָה ֵא ַלי ֲאנִ י ָדנִ יֵּ אל‬ ַ ‫ ִבּ ְשׁנַ ת ָשׁלֹושׁ ְל ַמ ְלכוּת ֵבּ ְל‬Dan 8:1 ‫ֵא ַלי ַבּ ְתּ ִח ָלּה׃‬ ‘in the third year of the reign of King Belshazzar a vision appeared to me, I, Daniel, after (the one)i whi that ___i had appeared to me at first’ ‫ֹלהינוּ ַה ֵה ִרימוּ ַה ֶמּ ֶלְך וְ י ֲֹע ָציו וְ ָשׂ ָריו וְ ָכל־יִ ְשׂ ָר ֵאל ַהנִּ ְמ ָצ ִאים׃‬ ֵ ‫ית־א‬ ֱ ‫רוּמת ֵבּ‬ ַ ‫ ְתּ‬Ezra 8:25 ‘the offeringi for the house of our God whi that the king, his counselors, his lords, and all Israel who were present had offered ___i’ ‫ וְ כֹל ׀ ֲא ֶשׁר ֶבּ ָע ֵרינוּ ַהה ִֹשׁיב נָ ִשׁים נָ ְכ ִריֹּות יָבֹא ְל ִע ִתּים ְמזֻ ָמּנִ ים‬Ezra 10:14 ‘and all thosei who are in our towns whi that ___i have taken foreign wives shall come at appointed times’ ‫ וַ יְ ַכלּוּ ַבכֹּל ֲאנָ ִשׁים ַהה ִֹשׁיבוּ נָ ִשׁים נָ ְכ ִריֹּות ַעד יֹום ֶא ָחד ַלח ֶֹדשׁ ָה ִראשֹׁון׃ פ‬Ezra 10:17 ‘and they finished with everything, (that is,) men who who had married foreign women, by the first day of the first month’ ‫ן־צרוּיָ ה‬ ְ ‫אַבנֵ ר ֶבּן־נֵ ר וְ יֹואָב ֶבּ‬ ְ ְ‫ן־קישׁ ו‬ ִ ‫מוּאל ָהר ֶֹאה וְ ָשׁאוּל ֶבּ‬ ֵ ‫ וְ כֹל ַה ִה ְק ִדּישׁ ְשׁ‬1 Chr 26:28 ‘also every (thing)i whi that Samuel the seer, and Saul son of Kish, and Abner son of Ner, and Joab son of Zeruiah had dedicated ___i’ ‫יאל ַהגֵּ ְר ֻשׁנִּ י׃‬ ֵ ‫אֹוצר ֵבּית־יְ הוָ ה ַעל יַ ד־יְ ִח‬ ַ ‫ וְ ַהנִּ ְמ ָצא ִאתֹּו ֲא ָבנִ ים נָ ְתנוּ ְל‬1 Chr 29:8 ‘and (the person)i whi that with himi were found precious stones gave them to the treasure of the house of Yhwh, by the hand of Jehiel the Gershonite’ ‫ב־לְך׃‬ ָ ‫יתי ְב ִשׂ ְמ ָחה ְל ִה ְתנַ ֶדּ‬ ִ ‫ וְ ַע ָתּה ַע ְמָּך ַהנִּ ְמ ְצאוּ־פֹה ָר ִא‬1 Chr 29:17 ‘and now your peoplei whi that ___i are present here I have seen with joy to offering freely to you’ ‫ֹלהים ֶה ֱע ָלה ָדוִ יד ִמ ִקּ ְריַ ת יְ ָע ִרים ַבּ ֵה ִכין לֹו ָדּוִ יד‬ ִ ‫ ֲא ָבל ֲארֹון ָה ֱא‬2 Chr 1:4 ‘but David had brought the ark of God up from Kiriath-jearim into (the place)i whi that David had prepared ___i for it’ ‫ֹלהים ָל ָעם‬ ִ ‫ל־ה ָעם ַעל ַה ֵה ִכין ָה ֱא‬ ָ ‫ וַ יִּ ְשׂ ַמח יְ ִחזְ ִקיָּ הוּ וְ ָכ‬2 Chr 29:36 ‘and Hezekiah and all the people rejoiced on account of (the thing)i whi that God had done ___i for the people’

Relative Clause Data

305

6.  Zero-Relative Data (299×) Since zero-relatives, by definition, have no overt relative word, I have highlighted the two words surrounding the beginning of the relative clause in a shaded-gray font. Note that the word preceding the relative may or may not be the relative head; moreover, for cases in which the head itself is null as well as cases of stacked relatives, I have highlighted only the first word of the relative clause. Finally, as with many other examples in this appendix, I sometimes pare the example to only the relevant data; however, since zero-relatives are often only identifiable within their larger context, in this section I include many whole verses. ‫אָרץ׃‬ ֶ ‫ֹלהים ֵאת ַה ָשּׁ ַמיִ ם וְ ֵאת ָה‬ ִ ‫אשׁית ָבּ ָרא ֱא‬ ִ ‫ ְבּ ֵר‬Gen 1:1 ‘in the beginningi whi (that) God created the heavens ___i’ ‫ יָ ד ַֹע ֵתּ ַדע ִכּי־גֵ ר ׀ יִ ְהיֶ ה זַ ְר ֲעָך ְבּ ֶא ֶרץ לֹא ָל ֶהם‬Gen 15:13 ’know that a stranger your offspring will be, in a landi whi (that) ___i is not theirs’ ‫יה ֲע ָשׂ ָרה זָ ָהב ִמ ְשׁ ָק ָלם׃‬ ָ ‫וּשׁנֵ י ְצ ִמ ִידים ַעל־יָ ֶד‬ ְ ‫ וַ יִּ ַקּח ָה ִאישׁ נֶ זֶ ם זָ ָהב ֶבּ ַקע ִמ ְשׁ ָקלֹו‬Gen 24:22 (2×) ‘the man took a gold nose-ringi whi (that) a half shekel was itsi weight, and two braceletsi for her arms whi (that) ten gold shekels was theiri weight’ ‘what is thisi whi (that) you have done ___i to us?’ ‘what is thisi whi (that) you have done ___i to me?’

‫ית ָלּנוּ‬ ָ ‫ ַמה־זֹּאת ָע ִשׂ‬Gen 26:10 ‫ית ִלּי‬ ָ ‫ ַמה־זֹּאת ָע ִשׂ‬Gen 29:25

‫ וְ ָכל־יֶ שׁ־לֹו נָ ַתן ְבּיָ דֹו׃‬Gen 39:4 ‘and every (thing)i whi (that) ___i belonged to him he put into his hand’ ‫ֹלהים ָלנוּ׃‬ ִ ‫ ַמה־זֹּאת ָע ָשׂה ֱא‬Gen 42:28 ‘what is thisi whi (that) God has done ___i to us?’ ‫אכל ַעד וְ ָל ֶע ֶרב יְ ַח ֵלּק ָשׁ ָלל׃‬ ַ ֹ ‫ ִבּנְ יָ ִמין זְ ֵאב יִ ְט ָרף ַבּבּ ֶֹקר י‬Gen 49:27 ‘Benjamin is a wolfi whi (that) ___i tears; in the morning he eats his prey and at the evening he divides spoil’ ‫ד־תּ ְשׁ ָלח׃‬ ִ ַ‫ ִבּי ֲאד ֹנָ י ְשׁ ַלח־נָ א ְבּי‬Exod 4:13 ‘by me, O Lord, send [the message] by the hand ∅(of the man)i whi (that) you shall send ___i’ ‫ וַ יְ ִהי ְבּיֹום ִדּ ֶ ֧בּר יְ הוָ ה ֶאל־מ ֶֹשׁה ְבּ ֶא ֶרץ ִמ ְצ ָריִ ם׃ פ‬Exod 6:28 ‘and it was, on the dayi whi (that) Yhwh spoke to Moses in the land of Egypt ___i’ ‫י־ש ַּל ְחנּו ֶאת־יִ ְש ָר ֵאל ֵמ ָע ְב ֵדנּו׃‬ ִ ‫ ַמה־ּזֹאת ָע ִשינּו ִכ‬Exod 14:5 ‘what is thisi whi (that) we have done ___i, that we sent Israel away from serving us’

306

Appendix A

‫הֹוציאָנוּ ִמ ִמּ ְצ ָריִ ם׃‬ ִ ‫ית ָלּנוּ ְל‬ ָ ‫ ַמה־זֹּאת ָע ִשׂ‬Exod 14:11 ‘what is thisi whi (that) you have done ___i to us by bringing us out of Egypt?’ ‫‏מכֹון ְל ִשׁ ְב ְתָּך ָפּ ַע ְל ָתּ יְ הוָ ה ִמ ְקּ ָדשׁ ֲאד ֹנָ י כֹּונְ נוּ יָ ֶדיָך׃‬ ָ Exod 15:17 ‘a placei whi (that) you made ___i for you to dwell, O Yhwh, a (2×) sanctuaryj, O Lord, whj (that) your hands established ___j’ ‫ת־ה ַמּ ֲע ֶשׂה ֲא ֶשׁר יַ ֲעשׂוּן׃‬ ַ ‫ת־ה ֶדּ ֶרְך יֵ ְלכוּ ָבהּ וְ ֶא‬ ַ ‫הֹוד ְע ָתּ ָל ֶהם ֶא‬ ַ ְ‫ ‏ו‬Exod 18:20 ‘and you shall make known to them the wayi whi (that) they are to go ___i and the things they are to do’ ‫ ְבּיֹום ִה ְק ִריב א ָֹתם ְל ַכ ֵהן ַליהוָ ה׃‬Lev 7:35 ‘on the dayi whi (that) one brought them near ___i to be priests for Yhwh’ ‫ע־צ ַר ַעת ִהוא ַבּ ְשּׁ ִחין ָפּ ָר ָחה׃‬ ָ ַ‫ נֶ ג‬Lev 13:20 ‘a leprous diseasei is it, whi (that) in boils ___i has broken out’ ‎‫עֹור־בּ ָשׂ ָרם ֶבּ ָהר ֹת ֵכּהֹות ְל ָבנֹת בּ ַֹהק הוּא ָפּ ַרח ָבּעֹור ָטהֹור‬ ְ ‫ וְ ָראָה ַהכּ ֵֹהן וְ ִהנֵּ ה ְב‬Lev 13:39 ‫הוּא׃ ס‬ ‘and the priest shall look, and—look, (if) on the skin of the body are dull white spots, a lesioni ___j is it [whi (that) ___i has broken out on the skin]j; he is clean’ ‫י־ה ָדּם הוּא ַבּנֶּ ֶפשׁ יְ ַכ ֵפּר׃‬ ַ ‫‏כּ‬ ִ Lev 17:11 ‘because the bloodi is it, whi (that) by means of life ___i atones’ ‫אכל‬ ַ ֹ ‫רוּע אֹו זָ ב ַבּ ֳקּ ָד ִשׁים לֹא י‬ ַ ‫אַהר ֹן וְ הוּא ָצ‬ ֲ ‫‏אישׁ ִאישׁ ִמזֶּ ַרע‬ ִ Lev 22:4 ‘any mani whi (that) ___i is from the seed of Aaron and he is leprous or has discharged may not eat of the sacred donations’ ‫וּמאָה ִמ ְשׁ ָק ָלהּ ִמזְ ָרק ֶא ָחד ֶכּ ֶסף ִשׁ ְב ִעים ֶשׁ ֶקל‬ ֵ ‫ֹלשׁים‬ ִ ‫אַחת ְשׁ‬ ַ ‫ת־כּ ֶסף‬ ֶ ‫ ‏וְ ָק ְר ָבּנֹו ַק ֲע ַר‬Num 7:13 ‫( ְבּ ֶשׁ ֶקל ַהקּ ֶֹדשׁ‬2×) ‘and his offering was one silver platei whi (that) itsi weight was one hundred thirty (shekels), one silver basinj whj (that) itsj weight was seventy shekels, according to the shekel of the sanctuary’ ‫יכם ְבּח ֵֹרב ִמתֹּוְך ָה ֵאשׁ׃‬ ֶ ‫הו֧ה ֲא ֵל‬ ָ ְ‫‏בּיֹום ִדּ ֶבּר י‬ ְ Deut 4:15 ‘on the dayi whi (that) Yhwh spoke to you ___i at Horeb out of the fire’ ‫הֹורישׁ גֹּויִ ם גְּ ד ִֹלים וַ ֲע ֻצ ִמים ִמ ְמָּך ִמ ָפּנֶ יָך‬ ִ ‫ ְל‬Deut 4:38 ‘by driving out nationsi whi (that) ___i are greater and mightier than you from before you’ ‫צוּמים ִמ ֶמּךָּ ׃‬ ִ ‫‏שׁ ְב ָעה גֹויִ ם ַר ִבּים וַ ֲע‬ ִ Deut 7:1 ‘seven nationsi whi (that) ___i are mightier and more numerous than you’ ‫וּב ֻצר ֹת ַבּ ָשּׁ ָמיִ ם׃‬ ְ ‫ ‏גֹּויִ ם גְּ ד ִֹלים וַ ֲע ֻצ ִמים ִמ ֶמּךָּ ָע ִרים גְּ ד ֹֹלת‬Deut 9:1 (2×) ‘nationsi whi (that) ___i are larger and mightier than you, citiesj whj (that) ___j are great and fortified to the heavens’

Relative Clause Data

307

‫גֹוי־עצוּם וָ ָרב ִמ ֶמּנּוּ׃‬ ָ ‫אֹותָך ְל‬ ְ ‫ ‏וְ ֶא ֱע ֶשׂה‬Deut 9:14 ‘and I will make you a nationi whi (that) ___i is mightier and more numerous than it’ ‫ ‏וִ ִיר ְשׁ ֶתּם גֹּויִ ם גְּ ד ִֹלים וַ ֲע ֻצ ִמים ִמ ֶכּם׃‬Deut 11:23 ‘and you will dispossess nationsi whi (that) ___i are larger and mightier than yourselves’ ‫ל־א ְב ָרתֹו׃‬ ֶ ‫ ‏ ְכּנֶ ֶשׁר יָ ִעיר ִקנֹּו ַעל־גֹּוזָ ָליו יְ ַר ֵחף יִ ְפר ֹשׂ ְכּנָ ָפיו יִ ָקּ ֵחהוּ יִ ָשּׂ ֵאהוּ ַע‬Deut 32:11 (5×, stacked) ‘like an eaglei whi (that) ___i arouses its nest, (that) ___i hovers over its young, (that) ___i spreads its wings, (that) ___i takes it up, (that) ___i lifts him upon his pinions’ ‫ֹלהים לֹא יְ ָדעוּם ֲח ָד ִשׁים ִמ ָקּר ֹב ָבּאוּ לֹא ְשׂ ָערוּם‬ ִ ‫ֹלה ֱא‬ ַ ‫ ‏יִ זְ ְבּחוּ ַל ֵשּׁ ִדים לֹא ֱא‬Deut 32:17 ‫יכם׃‬ ֶ ‫( ֲאב ֵֹת‬4×, 2 stacked) ‘they sacrificed to demonsi whi (that) ___i are not gods, deitiesj whj (that) they had never known themh, new onesk whk (that) ___k recently arrived, whk (that) your ancestors had not feared ___k’ ‘for the timei whi (that) their foot shall slip ___i’

‫ ְ‏ל ֵעת ָתּמוּט ַרגְ ָלם‬Deut 32:37

‫‏וּלשֹׁון זָ ָהב ֶא ָחד ֲח ִמ ִשּׁים ְשׁ ָק ִלים ִמ ְשׁ ָקלֹו‬ ְ Josh 7:21 ‘and a tongue(-shaped bar) of goldi whi (that) itsi weight was fifty shekels’ ‫ית ָלּנוּ ְל ִב ְל ִתּי ְקרֹאות ָלנוּ ִכּי ָה ַל ְכ ָתּ ְל ִה ָלּ ֵחם ְבּ ִמ ְדיָ ן‬ ָ ‫ה־ה ָדּ ָבר ַהזֶּ ה ָע ִשׂ‬ ַ ‫‏מ‬ ָ Judg 8:1 ‘what is this thingi whi (that) you have done ___i to us by not calling us when you went to fight against the Midianites?’ ‘what is thisi whi (that) you have done ___i to us?’

‫ית ָלּנוּ‬ ָ ‫‏וּמה־זֹּאת ָע ִשׂ‬ ַ Judg 15:11

‫א־הגַּ ְד ָתּ ִלּי ַבּ ֶמּה כּ ֲֹחָך גָ דֹול׃‬ ִ ֹ ‫ זֶ ה ָשֹׁלשׁ ְפּ ָע ִמים ֵה ַת ְל ָתּ ִבּי וְ ל‬Judg 16:15 ‘this is three timesi whi (that) you have mocked me ___i and you have not told me how your strength is great’ ‫ ‏וְ יָ ַד ְענוּ ִכּי לֹא יָ דֹו נָ גְ ָעה ָבּנוּ ִמ ְק ֶרה הוּא ָהיָ ה ָלנוּ׃‬1 Sam 6:9 ‘and we shall know that (it) was not his handi whi (that) ___i struck us’ ‫אמר ָה ָעם ִאישׁ‬ ֶ ֹ ‫ל־יֹודעֹו ֵמ ִא ְתּמֹול ִשׁ ְלשֹׁום וַ יִּ ְראוּ וְ ִהנֵּ ה ִעם־נְ ִב ִאים נִ ָבּא וַ יּ‬ ְ ‫ וַ יְ ִהי ָכּ‬1 Sam 10:11 ‫יאים׃‬ ִ ‫ן־קישׁ ֲהגַ ם ָשׁאוּל ַבּנְּ ִב‬ ִ ‫ל־ר ֵעהוּ ַמה־זֶּ ה ָהיָ ה ְל ֶב‬ ֵ ‫ֶא‬ ‘and it was, everyone who knew him from a previous time looked and, look—he was prophesying with prophets. And the people, each to the other, said, “What is thisi whi (that) ___i has happened to the son of Kish? Is also Saul among the prophets?”’ ‫ וְ ָה ִע ְב ִרים ָהיוּ ַל ְפּ ִל ְשׁ ִתּים ְכּ ֶא ְתמֹול ִשׁ ְלשֹׁום ֲא ֶשׁר ָעלוּ ִע ָמּם ַבּ ַמּ ֲחנֶ ה ָס ִביב וְ גַ ם־‬1 Sam 14:21 ‫ם־שׁאוּל וְ יֹונָ ָתן׃‬ ָ ‫ֵה ָמּה ִל ְהיֹות ִעם־יִ ְשׂ ָר ֵאל ֲא ֶשׁר ִע‬ ‘and the Hebrewsi whi (that) ___i previously had been with the Philistines and whi that ___i had gone up with them into the camp turned to be with the Israel who was with Saul and Jonathan’ [‫סביב‬ ‫ וגם המה‬is a textual difficulty; see McCarter 1980: 237]

308

Appendix A

‫ל־ה ֶמּ ֶלְך׃ פ‬ ַ ‫את ֶא‬ ָ ‫אַתּה ָק ָר‬ ָ ‫ ִמי‬1 Sam 26:14 ‘who are youi whi (that) ___(you)i call to the king?’ ‫ן־בּ ְכ ִרי ְשׁמֹו נָ ָשׂא יָ דֹו ַבּ ֶמּ ֶלְך ְבּ ָדוִ ד‬ ִ ‫‏כּי ִאישׁ ֵמ ַהר ֶא ְפ ַריִ ם ֶ ֧שׁ ַבע ֶבּ‬ ִ 2 Sam 20:21 ‘because a mani from the hill country of Ephraim whi (that) hisi name is Sheba son of Bichri, has lifted up his hand against King David’ ‫ה־אַתּם א ְֹמ ִרים ֶא ֱע ֶשׂה ָל ֶכם׃‬ ֶ ‫‏מ‬ ָ 2 Sam 21:4 ‘whati are you saying whi (that) I should do ___i for you?’ ‫‏עם לֹא־יָ ַד ְע ִתּי יַ ַע ְב ֻדנִ י׃‬ ַ 2 Sam 22:44 ‘peoplei whi (that) I did not know ___i serve me ’ ‫ ‏וּנְ ֻאם ַהגֶּ ֶבר ֻה ַקם ָעל‬2 Sam 23:1 ‘and the oracle of the mani whi (that) ___i was raised (to) a height’ ‫פוּ־שׁם ַל ִמּ ְל ָח ָמה‬ ָ ‫‏בּ ְפּ ִל ְשׁ ִתּים נֶ ֶא ְס‬ ַ 2 Sam 23:9 ‘against the Philistinesi whi (that) ___i were gathered there for battle’ ‫ת־ה ֶדּ ֶרְך ֲא ֶשׁר ָה ַלְך ִאישׁ‬ ַ ‫יהם ֵאי־זֶ ה ַה ֶדּ ֶרְך ָה ָלְך וַ יִּ ְראוּ ָבנָ יו ֶא‬ ֶ ‫ ‏וַ יְ ַד ֵבּר ֲא ֵל ֶהם ֲא ִב‬1 Kgs 13:12 ‫יהוּדה׃‬ ָ ‫ר־בּא ִמ‬ ָ ‫ֹלהים ֲא ֶשׁ‬ ִ ‫ָה ֱא‬ ‘and their father said to them, “Where is this, the roadi whi (that) he went ___i?” And his sons saw the road that the man of God who came from Judah went (on)’ ‫ֹלהי יִ ְשׂ ָר ֵאל ַכּד ַה ֶקּ ַמח לֹא ִת ְכ ָלה וְ ַצ ַפּ ַחת ַה ֶשּׁ ֶמן לֹא ֶת ְח ָסר‬ ֵ ‫אָמר יְ הוָ ה ֱא‬ ַ ‫‏כּי כֹה‬ ִ 1 Kgs 17:14 ‫ל־פּנֵ י ָה ֲא ָד ָמה׃‬ ְ ‫תתן־[תּת]־יְ הוָ ה גֶּ ֶשׁם ַע‬ ֵ ‫ַעד יֹום‬ ‘because thus said Yhwh the God of Israel, “The jar of meal will not end and the jug of oil will not lack until the dayi whi (that) you send [Kethiv], O Yhwh, rain on the earth ___i”’ [the Qere also presents a well-attested pattern, with the infinitive instead of the finite verb, ‘until the day of Yhwh’s sending rain . . .’] ‫רוּח־יְ הוָ ה‬ ַ ‫אמר ֵאי־זֶ ה ָע ַ ֧בר‬ ֶ ֹ ‫ל־ה ֶלּ ִחי וַ יּ‬ ַ ‫יכיְ הוּ ַע‬ ָ ‫ת־מ‬ ִ ‫ן־כּנַ ֲענָ ה וַ יַּ ֶכּה ֶא‬ ְ ‫ וַ יִּ גַּ שׁ ִצ ְד ִקיָּ הוּ ֶב‬1 Kgs 22:24 ‫אֹותְך׃‬ ָ ‫ֵמ ִא ִתּי ְל ַד ֵבּר‬ ‘and Zedekiah son of Chenaanah approached and struck Michaiah on the cheek, and said, “Where is this, ∅(wayi) whi (that) the spirit of Yhwh pass ___i from me to speak to you?”’ ‫אמר ֶדּ ֶרְך ִמ ְד ַבּר ֱאדֹום׃‬ ֶ ֹ ‫אמר ֵאי־זֶ ה ַה ֶדּ ֶרְך נַ ֲע ֶלה וַ יּ‬ ֶ ֹ ‫ וַ יּ‬2 Kgs 3:8 ‘and he asked, “Where is this, the roadi whi (that) we shall ascend ___i?” And Jehoram answered, “(It is) the road of the wilderness of Edom” ‫יתים ָעזְ בוּ ֶאת־יְ הוָ ה נִ ֲאצוּ‬ ִ ‫ ‏הֹוי ׀ גֹּוי ח ֵֹטא ַעם ֶכּ ֶבד ָעֹון זֶ ַרע ְמ ֵר ִעים ָבּנִ ים ַמ ְשׁ ִח‬Isa 1:4 ‫ת־קדֹושׁ יִ ְשׂ ָר ֵאל נָ זֹרוּ אָחֹור׃‬ ְ ‫ֶא‬ ‘alas, a nation (that) is sinning, a peoplei whi (that) ___i (is) heavy (with) iniquity, a seed (that) is doing evil, sons (that) are acting corruptly—they have abandoned Yhwh, spurned the Holy One of Israel, turned away’

Relative Clause Data

309

‫ר־מיִ ם ֵאין ָלהּ׃‬ ַ ‫וּכגַ נָּ ה ֲא ֶשׁ‬ ְ ‫‏כּי ִת ְהיוּ ְכּ ֵא ָלה נ ֶֹב ֶלת ָע ֶל ָה‬ ִ Isa 1:30 ‘because you shall be like an oaki whi (that) itsi leaf withers and like a garden that it has no water. ’ ‫וּביָ דֹו ִר ְצ ָפּה ְבּ ֶמ ְל ַק ַחיִ ם ָל ַקח ֵמ ַעל ַה ִמּזְ ֵבּ ַח׃‬ ְ ‫ן־ה ְשּׂ ָר ִפים‬ ַ ‫ ‏וַ יָּ ָעף ֵא ַלי ֶא ָחד ִמ‬Isa 6:6 ‘and one of the seraphs flew to me and in his hand was a live coali whi (that) with a pair of tongs he had taken ___i from upon the altar.’ ‫וּלשֹׁואָה ִמ ֶמּ ְר ָחק ָתּבֹוא‬ ְ ‫ה־תּ ֲעשׂוּ ְליֹום ְפּ ֻק ָדּה‬ ַ ‫‏וּמ‬ ַ Isa 10:3 ‘and what will you do at the day of punishment and at the calamityi whi (that) from afar ___i will come?’ ‫אָרץ ַמ ְר ִעישׁ ַמ ְמ ָלכֹות׃ ָשׂם‬ ֶ ‫ ר ֶֹאיָך ֵא ֶליָך יַ ְשׁגִּ יחוּ ֵא ֶליָך יִ ְתבֹּונָ נוּ ֲהזֶ ה ָה ִאישׁ ַמ ְרגִּ יז ָה‬Isa 14:16–17 ‫א־פ ַתח ָבּיְ ָתה׃‬ ָ ֹ ‫( ֵתּ ֵבל ַכּ ִמּ ְד ָבּר וְ ָע ָריו ָה ָרס ֲא ִס ָיריו ל‬2×, stacked) ‘those who see you will gaze at you, and they will stare at you, “Is this the mani (that) is making the earth tremble, (that) is shaking kingdoms, (that) is making the world like a desert and whi (that) ___i overthrew its cities, whi (that) ___i would not release his prisoners to home?”’ ‫וּפ ֻק ָדּ ָתם ַעל נַ ַחל ָה ֲע ָר ִבים יִ ָשּׂאוּם׃‬ ְ ‫ל־כּן יִ ְת ָרה ָע ָשׂה‬ ֵ ‫‏ע‬ ַ Isa 15:7 ‘therefore the abundancei whi (that) he made ___i and their goods— they carry them away over the Wadi Arabim’ ‫ ֵהן ׀ ֶא ֶרץ ַכּ ְשׂ ִדּים זֶ ה ָה ָעם לֹא ָהיָ ה‬Isa 23:13 ‘look—the land of the Chaldeans. This is the peoplei whi (that) ___i does not exist (any longer)’ [see Wildberger 1997: 405, 410–11; Blenkinsopp 2000: 341] ‫מוּסּד‬ ָ ‫מוּסד‬ ָ ‫אָבן ֶא ֶבן בּ ַֹחן ִפּנַּ ת יִ ְק ַרת‬ ֶ ‫אָמר ֲאד ֹנָ י יְ הוִ ה ִהנְ נִ י יִ ַסּד ְבּ ִציֹּון‬ ַ ‫ ָ‏ל ֵכן כֹּה‬Isa 28:16 ‘therefore thus says the Lord Yhwh, “Look—I am (hei) whi (that) ___i laid in Zion a foundation stone, a tested stone, a precious cornerstone, an established foundation’ ‫יאל ִק ְריַ ת ָחנָ ה ָדוִ ד‬ ֵ ‫יאל ֲא ִר‬ ֵ ‫ ‏הֹוי ֲא ִר‬Isa 29:1 ‘Alas, Ariel, Ariel, the cityi whi (that) David encamped ___i!’ ‫הֹועיל ִכּי ְלב ֶֹשׁת‬ ִ ‫א־יֹועילוּ ָלמֹו לֹא ְל ֵעזֶ ר וְ לֹא ְל‬ ִ ֹ ‫ל־עם ל‬ ַ ‫ כֹּל הבאישׁ [ה ִֹבישׁ] ַע‬Isa 30:5 ‫ם־ל ֶח ְר ָפּה׃ ס‬ ְ ַ‫וְ ג‬ ‘everyone comes to shame (Qere) because a peoplei whi (that) ___i does not profit them, neither for help nor for profit, but shame and disgrace’ ‫תֹּורת יְ הוָ ה׃‬ ַ ‫מֹוע‬ ַ ‫ ִכּי ַעם ְמ ִרי הוּא ָבּנִ ים ֶכּ ָח ִשׁים ָבּנִ ים לֹא־אָבוּ ְשׁ‬Isa 30:9 ‘For they are a rebellious people, faithless children, childreni whi (that) ___i are not willing to hear the instruction of Yhwh’ ‫ל־ר ֶכב ִכּי ָרב וְ ַעל ָפּ ָר ִשׁים‬ ֶ ‫ל־סוּסים יִ ָשּׁ ֵענוּ וַ ְיִּב ְטחוּ ַע‬ ִ ‫ הֹוי ַהיּ ְֹר ִדים ִמ ְצ ַריִ ם ְל ֶעזְ ָרה ַע‬Isa 31:1 ‫ל־קדֹושׁ יִ ְשׂ ָר ֵאל וְ ֶאת־יְ הוָ ה לֹא ָד ָרשׁוּ׃‬ ְ ‫י־ע ְצמוּ ְמאֹד וְ לֹא ָשׁעוּ ַע‬ ָ ‫ִכּ‬ ‘Alas for ∅(those)i who go down to Egypt for help (and) whi (that) ___i relied on horses and trusted in chariots because they are many and (trusted) in horsemen because they are very strong, but have not looked to the Holy One of Israel or consulted Yhwh!’

310

Appendix A

‫אכ ֶלנּוּ‬ ֲ ֹ ‫א־אָדם תּ‬ ָ ֹ ‫א־אישׁ וְ ֶח ֶרב ל‬ ִ ֹ ‫ וְ נָ ַפל אַשּׁוּר ְבּ ֶח ֶרב ל‬Isa 31:8 (2×) ‘Assyria shall fall by a swordi whi (that) ___i is not a man and a swordj whj (that) ___j is not human shall consume her’ ‫אַד ַמת ַע ִמּי קֹוץ ָשׁ ִמיר ַתּ ֲע ֶלה‬ ְ ‫‏על‬ ַ Isa 32:13 ‘because the soili of my people whi (that) ___i brings up in thorns and briers’ ‫רוּמה ֵעץ לֹא־יִ ְר ַקב ְיִב ָחר ָח ָרשׁ ָח ָכם ַיְב ֶקּשׁ־לֹו ְל ָה ִכין ֶפּ ֶסל לֹא יִ מֹּוט׃‬ ָ ‫‏ה ְמ ֻס ָכּן ְתּ‬ ַ Isa 40:20 (2×) ‘The one placed in charge of the offering one chooses woodi whi (that) ___i will not rot; he seeks out a skilled artisan to set up an imagej whj (that) ___j will not topple’ [on the difficult first phrase, see Blenkinsopp 2002: 188–90] ‫תֹּוע ָבה ְיִב ַחר ָבּ ֶכם׃‬ ֵ ‫אָפע‬ ַ ‫וּפ ָע ְל ֶכם ֵמ‬ ָ ‫ן־אַתּם ֵמאַיִ ן‬ ֶ ‫‏ה‬ ֵ Isa 41:24 ‘look—you are nothing and your work is nothing at all; an abomination is ∅(he)i whi (that) ___i chooses you’ ‫‏הן ַע ְב ִדּי ֶא ְת ָמְך־בֹּו ְבּ ִח ִירי ָר ְצ ָתה נַ ְפ ִשׁי‬ ֵ Isa 42:1 (2×) ‘look—my servanti whi (that) I uphold ___i, my choseni whi (that) my soul delights ___i’ ‫ יִ ְשׂאוּ ִמ ְד ָבּר וְ ָע ָריו ֲח ֵצ ִרים ֵתּ ֵשׁב ֵק ָדר‬Isa 42:11 ‘let the desert and its towns lift up (their voice), the villagesi whi (that) Kedar inhabits ___i’ ‫יכם‬ ֵ ‫אַד ִר‬ ְ ‫הֹול ְכ ִתּי ִעוְ ִרים ְבּ ֶד ֶרְך לֹא יָ ָדעוּ ִבּנְ ִתיבֹות לֹא־יָ ְדעוּ‬ ַ ְ‫ ‏ו‬Isa 42:16 (2×) ‘I will lead the blind by a roadi whi (that) they do not know ___i, by pathsj whj (that) they have not known ___j I will guide them’ ‫ ‏וְ ַע ָתּה ְשׁ ַמע יַ ֲעקֹב ַע ְב ִדּי וְ יִ ְשׂ ָר ֵאל ָבּ ַח ְר ִתּי בֹו׃‬Isa 44:1 ‘And now hear, O Jacob, my servant, O Israeli whi (that) I have chosen ___i’ ‫ישׁרוּן ָבּ ַח ְר ִתּי בֹו׃‬ ֻ ִ‫‏אַל־תּ ָירא ַע ְב ִדּי יַ ֲעקֹב ו‬ ִ Isa 44:2 ‘do not fear, O Jacob, my servant, O Jeshuruni whi (that) I have chosen ___i’ ‫יכָך ְבּ ֶד ֶרְך ֵתּ ֵלְך׃‬ ֲ ‫הֹועיל ַמ ְד ִר‬ ִ ‫ֹלהיָך ְמ ַל ֶמּ ְדָך ְל‬ ֶ ‫‏אנִ י יְ הוָ ה ֱא‬ ֲ Isa 48:17 ‘I am Yhwh, your God, who teaches you to profit, who leads you in the wayi whi (that) you should go ___i’ ‫יענִ י‬ ֵ ‫ר־לי ִמי־הוּא יַ ְר ִשׁ‬ ִ ָ‫‏הן ֲאד ֹנָ י יְ הוִ ה יַ ֲעז‬ ֵ Isa 50:9 ‘look—the Lord Yhwh will help me; who is hei whi (that) ___i will declare me guilty?’ ‫ל־מ ֶקּ ֶבת בֹּור נֻ ַקּ ְר ֶתּם׃‬ ַ ‫‏ה ִבּיטוּ ֶאל־צוּר ֻח ַצּ ְב ֶתּם וְ ֶא‬ ַ Isa 51:1 (2×) ‘look to the rocki whi (that) you were hewn ___i, and to the excavation pitj whj (that) you were dug ___j’ ‫חֹול ְל ֶכם‬ ֶ ‫ל־שׂ ָרה ְתּ‬ ָ ‫יכם וְ ֶא‬ ֶ ‫ל־אַב ָר ָהם ֲא ִב‬ ְ ‫‏ה ִבּיטוּ ֶא‬ ַ Isa 51:2 ‘look to Abraham your father and to Sarahi whi (that) ___i bore you’

Relative Clause Data

311

‫וּמגִּ ֻדּפ ָֹתם‬ ִ ‫אַל־תּ ְיראוּ ֶח ְר ַפּת ֱאנֹושׁ‬ ִ ‫תֹּור ִתי ְב ִל ָבּם‬ ָ ‫ ִשׁ ְמעוּ ֵא ַלי י ְֹד ֵעי ֶצ ֶדק ַעם‬Isa 51:7 ‫אַל־תּ ָחתּוּ׃‬ ֵ ‘listen to me, O those who know righteousness, O peoplei whi (that) my teaching is in theiri heart’ ‫ן־אָדם ָח ִציר יִ נָּ ֵתן׃‬ ָ ‫וּמ ֶבּ‬ ִ ‫ ‏וַ ִתּ ְיר ִאי ֵמ ֱאנֹושׁ יָ מוּת‬Isa 51:12 (2×) ‘you were afraid of a mani, whi (that) ___i will die, of a human beingj whj (that) grass will be given ___(from himj)?’ ‫יה נֶ ֱא ָל ָמה‬ ָ ֶ‫וּכ ָר ֵחל ִל ְפנֵ י גֹזְ ז‬ ְ ‫יוּבל‬ ָ ‫‏כּ ֶשּׂה ַל ֶטּ ַבח‬ ַ Isa 53:7 (2×) ‘like a lambi whi (that) ___i is led to the slaughter, and like a sheepj whj (that) before its shearers ___j is silent ’ ‫א־ח ָלה‬ ָ ֹ ‫ ָרנִּ י ֲע ָק ָרה לֹא יָ ָל ָדה ִפּ ְצ ִחי ִרנָּ ה וְ ַצ ֲה ִלי ל‬Isa 54:1 (2×) ‘ring out, O barren onei whi (that) ___i did not bear; burst into song and shout, ∅(you)j whj (that) ___j have not been in labor!’ ‫עֹולם לֹא יִ ָכּ ֵרת׃ ס‬ ָ ‫ וְ ָהיָ ה ַליהוָ ה ְל ֵשׁם ְלאֹות‬Isa 55:13 ‘and it shall become for Yhwh a memorial, a perpetual signi whi (that) ___i shall not be cut off’ ‫ן־אָדם יַ ֲחזִ יק ָבּהּ שׁ ֵֹמר ַשׁ ָבּת ֵמ ַח ְלּלֹו וְ שׁ ֵֹמר יָ דֹו‬ ָ ‫וּב‬ ֶ ‫אַשׁ ֵרי ֱאנֹושׁ יַ ֲע ֶשׂה־זֹּאת‬ ְ Isa 56:2 (2×) ‫ל־רע׃ ס‬ ָ ‫ֵמ ֲעשֹׂות ָכּ‬ ‘Happy is the mani whi (that) ___i does this, and (happy is) the humanj whj (that) ___j holds it fast, who keeps the sabbath from profaning it, and keeps his hand from doing any evil’ ‫ישׁוּע ָתהּ ְכּ ַל ִפּיד ְיִב ָער׃‬ ָ ִ‫ ַעד־יֵ ֵצא ַכנֹּגַ הּ ִצ ְד ָקהּ ו‬Isa 62:1 ‘until her vindication comes out like the bright light (of dawn), and her salvation (comes out) like a torchi whi (that) ___i burns’ ‫יהם‬ ֶ ‫א־מ ַשׁ ְל ָתּ ָבּם לֹא־נִ ְק ָרא ִשׁ ְמָך ֲע ֵל‬ ָ ֹ ‫עֹולם ל‬ ָ ‫ ָהיִ ינוּ ֵמ‬Isa 63:19 (2×) ‘we have been from old (like) ∅(those)i whi (that) you do not rule over themi, like ∅(those)j whj (that) your name is not called over themj’ ‫נֹוראֹות לֹא נְ ַקוֶּ ה‬ ָ ‫שֹׂותָך‬ ְ ‫ ַבּ ֲע‬Isa 64:2 ‘when you did awesome deedsi whi (that) we did not expect ___i’ ‫אָמ ְר ִתּי ִהנֵּ נִ י ִהנֵּ נִ י ֶאל־גֹּוי לֹא־ק ָֹרא‬ ַ ‫אתי ְללֹא ִב ְק ֻשׁנִ י‬ ִ ‫ נִ ְד ַר ְשׁ ִתּי ְללֹוא ָשׁאָלוּ נִ ְמ ֵצ‬Isa 65:1 (3×) ‫ִב ְשׁ ִמי׃‬ ‘I was sought out by ∅(those)i whi (that) ___i did not ask, I was found by ∅(those)j whj (that) ___j did not seek me. I said, “Here I am, here I am,” to a nationk whk (that) ___k was not called by my name’ ‫נוּח ִתי׃‬ ָ ‫נוּ־לי וְ ֵאי־זֶ ה ָמקֹום ְמ‬ ִ ‫ ֵאי־זֶ ה ַביִ ת ֲא ֶשׁר ִתּ ְב‬Isa 66:1 ‘where is this, a house that you would build for me, and where is this, a placei whi (that) ___i is my rest?’ ‫רוּעה׃‬ ָ ְ‫אַח ַרי ַבּ ִמּ ְד ָבּר ְבּ ֶא ֶרץ לֹא ז‬ ֲ ‫לוֹּלתיִ ְך ֶל ְכ ֵתְּך‬ ָ ‫אַה ַבת ְכּ‬ ֲ ‫עוּריִ ְך‬ ַ ְ‫ ‏זָ ַכ ְר ִתּי ָלְך ֶח ֶסד נ‬Jer 2:2 ‘I remember concerning you the devotion of your youth, your love as a bride, your walking after me in the wilderness, in a landi whi (that) ___i is not sown. ’

312

Appendix A

‫מֹּוליְך א ָֹתנוּ ַבּ ִמּ ְד ָבּר ְבּ ֶא ֶרץ ֲע ָר ָבה וְ שׁוּ‬ ִ ‫ אַיֵּ ה יְ הוָ ה ַה ַמּ ֲע ֶלה א ָֹתנוּ ֵמ ֶא ֶרץ ִמ ְצ ָריִ ם ַה‬Jer 2:6 (2×, ‫אָדם ָשׁם׃‬ ָ ‫א־ע ַבר ָבּהּ ִאישׁ וְ לֹא־יָ ַשׁב‬ ָ ֹ ‫ ָחה ְבּ ֶא ֶרץ ִציָּ ה וְ ַצ ְל ָמוֶ ת ְבּ ֶא ֶרץ ל‬stacked) ‘where is Yhwh who brought us up from the land of Egypt, who led us in the wilderness, in a land of deserts and pits, in a land of drought and deep darkness, in a landi whi (that) no one passes through iti and whi (that) no one lives therei?’ ‫א־יֹועלוּ ָה ָלכוּ׃‬ ִ ֹ ‫אַח ֵרי ל‬ ֲ ְ‫יאים נִ ְבּאוּ ַב ַבּ ַעל ו‬ ִ ‫ ‏וְ ַהנְּ ִב‬Jer 2:8 ‘the prophets prophesied by Baal and after ∅(things)i whi (that) ___i do not profit they walked’ ‫יֹועיל׃‬ ִ ‫ ‏וְ ַע ִמּי ֵה ִמיר ְכּבֹודֹו ְבּלֹוא‬Jer 2:11 ‘and my people has exchanged its glory for ∅(what)i whi (that) ___i does not profit’ ‫א־ת ַדע ְלשׁ ֹנֹו וְ לֹא ִת ְשׁ ַמע ַמה־יְ ַד ֵבּר׃‬ ֵ ֹ ‫עֹולם הוּא ֚גֹּוי ל‬ ָ ‫יתן הוּא גֹּוי ֵמ‬ ָ ‫ ‏גֹּוי ׀ ֵא‬Jer 5:15 ‘It is an enduring nation, it is an ancient nation, a nationi whi (that) you do not know itsi language and you cannot hear what it says’ ‫ ‏ ִשׁ ְמעוּ־נָ א זֹאת ַעם ָס ָכל וְ ֵאין ֵלב ֵעינַ יִ ם ָל ֶהם וְ לֹא יִ ְראוּ אָזְ נַ יִ ם ָל ֶהם וְ לֹא יִ ְשׁ ָמעוּ׃‬Jer 5:21 (3×, stacked) ‘hear this, O foolish peoplei whi (that) a heart (belongs to) ___i, whi (that) eyes belong to themi and they do not see, whi (that) ears belong to themi and they do not hear’ ‫אַר ֵתְּך׃‬ ְ ‫ן־לְך צֹאן ִתּ ְפ‬ ָ ‫ ‏אַיֵּ ה ָה ֵע ֶדר נִ ַתּ‬Jer 13:20 ‘where is the flocki whi (that) ___i was given you, your beautiful flock? ’ ‫ וְ ַה ֲע ַב ְר ִתּי ֶאת־א ֶֹיְביָך ְבּ ֶא ֶרץ לֹא יָ ָד ְע ָתּ‬Jer 15:14 ‘and I will pass your enemies into a landi whi (that) you do know ___i’ ‫וּמ ְפּנֵ י ִדּ ְב ֵרי ָק ְדשֹׁו׃‬ ִ ‫וּכגֶ ֶבר ֲע ָברֹו יָ יִ ן ִמ ְפּנֵ י יְ הוָ ה‬ ְ ‫יתי ְכּ ִאישׁ ִשׁכֹּור‬ ִ ִ‫ ָהי‬Jer 23:9 ‘I have become like a drunkard, like a mani whi (that) wine has overcome himi’ ‫וּכ ַפ ִטּישׁ יְ פ ֵֹצץ ָס ַלע׃ ס‬ ְ ‫ ֲהלֹוא ֧כֹה ְד ָב ִרי ָכּ ֵאשׁ נְ ֻאם־יְ הוָ ה‬Jer 23:29 ‘Is not my word like fire, says Yhwh, and like a hammeri whi (that) ___i shatters a rock?’ ‫ת־לבֹּו ָלגֶ ֶשׁת ֵא ַלי נְ ֻאם־יְ הוָ ה׃‬ ִ ‫ ִכּי ִמי הוּא־זֶ ה ָע ַ ֧רב ֶא‬Jer 30:21 ‘because who is thisi whi (that) ___i would pawn his heart to approach me?—says Yhwh’ ‫יכם ֶאל־נַ ֲח ֵלי ַמיִ ם ְבּ ֶד ֶרְך יָ ָשׁר לֹא יִ ָכּ ְשׁלוּ ָבּהּ‬ ֵ ‫אֹול‬ ִ Jer 31:9 ‘I will guide them to streams of water, on a straight pathi whi (that) they shall not stumble on iti’ ‫אשׁיָּ הוּ וְ ַעד ַהיֹּום ַהזֶּ ה׃‬ ִ ֹ ‫ימי י‬ ֵ ‫ ִמיֹּום ִדּ ַבּ ְר ִתּי ֵא ֶליָך ִמ‬Jer 36:2 ‘from the dayi whi (that) I spoke to you ___i, from the days of Josiah until today’ ‫אָבדוּ׃‬ ָ ‫ל־כּן יִ ְת ַרת ָע ָשׂה‬ ֵ ‫‏ע‬ ַ Jer 48:36 ‘because thus—the richesi whi (that) they gained ___i have perished’

Relative Clause Data

313

‫אַתּה הוּא‬ ָ ְ‫ר־אין ִמ ְשׁ ָפּ ָטם ִל ְשׁתֹּות ַהכֹּוס ָשׁתֹו יִ ְשׁתּוּ ו‬ ֵ ‫אָמר יְ הוָ ה ִהנֵּ ה ֲא ֶשׁ‬ ַ ‫ ִכּי־כֹה ׀‬Jer 49:12 ‫נָ קֹה ִתּנָּ ֶקה‬ ‘because thus said Yhwh: Look—those who their judgment is not to drink the cup will certainly drink it, and youi are the one whi (that) ___(you)i shall really go unpunished?’ ‫ל־אישׁ וְ לֹא־יַ ֲעבֹר ָבּ ֵהן‬ ִ ‫יה ְל ַשׁ ָמּה ֶא ֶרץ ִציָּ ה וַ ֲע ָר ָבה ֶא ֶרץ לֹא־יֵ ֵשׁב ָבּ ֵהן ָכּ‬ ָ ‫ ָהיוּ ָע ֶר‬Jer 51:43 (2×, ‫ן־אָדם׃‬ ָ ‫ ֶבּ‬stacked) ‘her cities have become an object of horror, a land of drought and a desert, a landi whi (that) no one lives in themi, and whi (that) no human passes through themi’ ‫וּמ ְשׁ ָפּ ִטים לֹא יִ ְחיוּ ָבּ ֶהם׃‬ ִ ‫טֹובים‬ ִ ‫ם־אנִ י נָ ַת ִתּי ָל ֶהם ֻח ִקּים לֹא‬ ֲ ַ‫ וְ ג‬Ezek 20:25 ‘and even I gave them statutesi whi (that) ___i were not good and (2×) ordinancesj whj (that) they could not live by themj’ ‫אַתּ ֶא ֶרץ לֹא ְמט ָֹה ָרה ִהיא לֹא גֻ ְשׁ ָמהּ ְבּיֹום זָ ַעם׃‬ ְ ‫ר־להּ‬ ָ ‫ן־אָדם ֱא ָמ‬ ָ ‫ ֶבּ‬Ezek 22:24 ‘human, say to it: You are a landi whi (that) not cleansed is iti, whi (2×, stacked) (that) itsi rain has not (fallen) on a day of indignation’ ‫וּב ִר ַיח ְוּד ָל ַתיִ ם ֵאין ָל ֶהם׃‬ ְ ‫חֹומה‬ ָ ‫ ‏אָבֹוא ַהשּׁ ְֹק ִטים י ְֹשׁ ֵבי ָל ֶב ַטח ֻכּ ָלּם י ְֹשׁ ִבים ְבּ ֵאין‬Ezek 38:11 ‘I will fall upon the quiet people who live in safety, all of themi living without walls, and whi (that) bars and gates do not belong to themi’ ‫א־יָבין יִ ָלּ ֵבט׃‬ ִ ֹ ‫ וְ ָעם ל‬Hos 4:14 ‘and a peoplei whi (that) ___i does not understand will come to ruin’ ‫אָרץ׃‬ ֶ ‫יֹורה‬ ֶ ‫ ‏וְ יָבֹוא ַכגֶּ ֶשׁם ָלנוּ ְכּ ַמ ְלקֹושׁ‬Hos 6:3 ‘and he will come to us like the showers, like the spring rainsi whi (that) ___i water the earth’ ‫יכם ַכּ ֲא ֶשׁר ָתּ ִעיק ָה ֲעגָ ָלה ַה ְמ ֵלאָה ָלהּ ָע ִמיר׃‬ ֶ ‫‏הנֵּ ה אָנ ִֹכי ֵמ ִעיק ַתּ ְח ֵתּ‬ ִ Amos 2:13 ‘look—I am creaking below you, like the full carti whi that a sheaf belongs to iti creaks’ ‫ל־אַר ְמנֹות ְבּ ֶא ֶרץ ִמ ְצ ָריִ ם‬ ְ ‫אַשׁדֹּוד וְ ַע‬ ְ ‫ל־אַר ְמנֹות ְבּ‬ ְ ‫‏ה ְשׁ ִמיעוּ ַע‬ ַ Amos 3:9 ‘proclaim to the citadels in Ashdod, and to the citadelsi whi (that) ___i are in the land of Egypt’ ‫ית‬ ָ ‫אמרוּ ֵא ָליו ַמה־זֹּאת ָע ִשׂ‬ ְ ֹ ‫דֹולה וַ יּ‬ ָ ְ‫ ‏וַ יִּ ְיראוּ ָה ֲאנָ ִשׁים יִ ְראָה ג‬Jonah 1:10 ‘and the men were even more afraid, and said to him, “What is thisi whi (that) you have done ___i!”’ ‫יֹול ָדה יָ ָל ָדה‬ ֵ ‫ד־עת‬ ֵ ‫ ָל ֵכן יִ ְתּנֵ ם ַע‬Mic 5:2 ‘therefore he shall give them up until the timei whi (that) she who is in labor has given birth ___i’ ‫יכם לֹא ַת ֲא ִמינוּ ִכּי יְ ֻס ָפּר׃‬ ֶ ‫ימ‬ ֵ ‫ ִכּי־פ ַֹעל פּ ֵֹעל ִבּ‬Hab 1:5 ‘because (I) am doing a worki in your days whi (that) you would not believe ___i if you were told’

314

Appendix A

‫י־א ֶרץ ָל ֶר ֶשׁת‬ ֶ ‫הֹולְך ְל ֶמ ְר ֲח ֵב‬ ֵ ‫ת־ה ַכּ ְשׂ ִדּים ַהגֹּוי ַה ַמּר וְ ַהנִּ ְמ ָהר ַה‬ ַ ‫י־הנְ נִ י ֵמ ִקים ֶא‬ ִ ‫ ִכּ‬Hab 1:6 ‫ִמ ְשׁ ָכּנֹות לֹּא־לֹו׃‬ ‘Indeed, I am rousing the Chaldeans, the bitter and hasty nation that marches through the breadth of the earth to seize dwellingsi whi (that) ___i are not its’ ‫וּפ ָר ָשׁיו ֵמ ָרחֹוק יָ בֹאוּ יָ ֻעפוּ ְכּנֶ ֶשׁר ָחשׁ ֶל ֱאכֹול׃‬ ָ Hab 1:8 ‘its horsemen come from far away; they fly like an eaglei whi (that) ___i hurries to eat’ ‫אָדם ִכּ ְדגֵ י ַהיָּ ם ְכּ ֶר ֶמשׂ לֹא־מ ֵֹשׁל בֹּו׃‬ ָ ‫ וַ ַתּ ֲע ֶשׂה‬Hab 1:14 ‘you have made people like the fish of the sea, like crawling thingsi whi (that) no one rules over iti’ ‫גוּדנּוּ׃‬ ֶ ְ‫אָנוּח ְליֹום ָצ ָרה ַל ֲעלֹות ְל ַעם י‬ ַ ‫ ֲא ֶשׁר‬Hab 3:16 ‘where I wait for the day of distress to come upon the peoplei whi (that) ___i attack us.’ ‫קֹושׁשׁוּ וָ קֹושּׁוּ ַהגֹּוי לֹא נִ ְכ ָסף׃‬ ְ ‫ ִה ְת‬Zeph 2:1 ‘gather together, gather, O nationi whi (that) ___i is not ashamed’ ‫ ְבּ ֶט ֶרם ֶל ֶדת חֹק ְכּמֹץ ָע ַבר יֹום‬Zeph 2:2 ‘before you are driven about* like chaffi whi (that) passes by (in) a day’ [emend: ‫;ּב ֶט ֶרם לֹא ִת ָּד ֲחקּו‬ ְ see Roberts 1991: 188] ‫יה זְ ֵא ֵבי ֶע ֶרב לֹא גָ ְרמוּ ַלבּ ֶֹקר׃‬ ָ ‫יה ְב ִק ְר ָבּהּ ֲא ָריֹות שׁ ֲֹאגִ ים שׁ ְֹפ ֶט‬ ָ ‫ ָשׂ ֶר‬Zeph 3:3 ‘its officials in its midst are roaring lions; its judges are evening wolvesi whi (that) ___i do not gnaw bones until morning’ ‫יֹושׁ ַיע‬ ִ ‫ֹלהיִ ְך ְבּ ִק ְר ֵבּך גִּ בֹּור‬ ַ ‫הו֧ה ֱא‬ ָ ְ‫ י‬Zeph 3:17 ‘Yhwh, your God, is in your midst, a warriori whi (that) ___i gives victory’ ‫ה־אישׁ ֶצ ַמח ְשׁמֹו‬ ִ ֵ‫אָמר יְ הוָ ה ְצ ָבאֹות ֵלאמֹר ִהנּ‬ ַ ‫אָמ ְר ָתּ ֵא ָליו ֵלאמֹר כֹּה‬ ַ ְ‫ ו‬Zech 6:12 ‘say to him: Thus says Yhwh of hosts: Here is a mani whi (that) hisi name is Branch’ ‫אָמר יְ הוָ ה ְצ ָבאֹות‬ ַ ‫ל־לבוּשֹׁו‬ ְ ‫ֹלהי יִ ְשׂ ָר ֵאל וְ ִכ ָסּה ָח ָמס ַע‬ ֵ ‫אָמר יְ הוָ ה ֱא‬ ַ ‫י־שׂנֵ א ַשׁ ַלּח‬ ָ ‫ ִכּ‬Mal 2:16 ‘because I hate divorce, says Yhwh, the God of Israel, and ∅(one) i whi (that) ___i covers his clothing (with) violence, says Yhwh of hosts’ ‫ירֹושׁם ָרבּוּ׃‬ ָ ‫ נָ ַת ָתּה ִשׂ ְמ ָחה ְב ִל ִבּי ֵמ ֵעת ְדּגָ נָ ם וְ ִת‬Ps 4:8 ‘you have put joy in my heart from the timei whi (that) their grain and wine have increased ___i’ ‫אָתּה לֹא יְ גֻ ְרָך ָרע׃‬ ָ ‫ל־ח ֵפץ ֶר ַשׁע ׀‬ ָ ‫ ִכּי ׀ לֹא ֵא‬Ps 5:5 ‘because a Godi whi (that) ___i delights in wickedness you are not; evil does not sojourn with you’

Relative Clause Data

315

‫וּכ ֻת ִמּי ָע ָלי׃‬ ְ ‫ יְ הוָ ה יָ ִדין ַע ִמּים ָשׁ ְפ ֵטנִ י יְ הוָ ה ְכּ ִצ ְד ִקי‬Ps 7:9 ‘Yhwh judges the peoples; judge me, O Yhwh, according to my righteousness and according to my blamelessnessi whi (that) ___i is upon me’ [alt., NP-internal PP: ‘my blamelessness upon me’; possibly read ‫ עלי‬as abbreviation for ‫עליון‬, ‘O Most High’; see Craigie 1983: 98] ‫ בֹּור ָכּ ָרה וַ יַּ ְח ְפּ ֵרהוּ וַ יִּ פֹּל ְבּ ַשׁ ַחת יִ ְפ ָעל׃‬Ps 7:16 ‘a pit he has made and dug it out and he has fallen into the holei whi (that) he made ___i’ ‫ ָט ְבעוּ גֹויִ ם ְבּ ַשׁ ַחת ָעשׂוּ ְבּ ֶר ֶשׁת־זוּ ָט ָמנוּ נִ ְל ְכּ ָדה ַרגְ ָלם׃‬Ps 9:16 ‘nations have sunk in the holei whi (that) they made ___i; in the net that they hid their foot has been caught’ ‫ֹלהים׃‬ ִ ‫אֹולה ָכּל־גֹּויִ ם ְשׁ ֵכ ֵחי ֱא‬ ָ ‫ יָ שׁוּבוּ ְר ָשׁ ִעים ִל ְשׁ‬Ps 9:18 ‘the wicked shall return to Sheol, all the nationsi whi (that) ___i are forgetful of God’ ‫אָשׁית ְבּיֵ ַשׁע יָ ִפ ַיח לֹו׃‬ ִ ‫אמר יְ הוָ ה‬ ַ ֹ ‫ ִמשּׁ ֹד ֲענִ יִּ ים ֵמאַנְ ַקת ֶא ְביֹונִ ים ַע ָתּה אָקוּם י‬Ps 12:6 ‘because of the despoiling of the poor, because of the groan of the needy, now I will rise up—says Yhwh—I will place (him) in a safe placei whi (that) he pants for iti’ ‫ל־קר ֹבֹו׃ נִ ְבזֶ ה ׀‬ ְ ‫א־ע ָשׂה ְל ֵר ֵעהוּ ָר ָעה וְ ֶח ְר ָפּה לֹא־נָ ָשׂא ַע‬ ָ ֹ ‫ל־לשׁ ֹנֹו ל‬ ְ ‫א־רגַ ל ׀ ַע‬ ָ ֹ ‫ ל‬Ps 15:3–5 (8×, ‫ ְבּ ֵעינָ יו נִ ְמאָס וְ ֶאת־יִ ְר ֵאי יְ הוָ ה יְ ַכ ֵבּד נִ ְשׁ ַבּע ְל ָה ַרע וְ לֹא יָ ִמר׃ ַכּ ְספֹּו ׀ לֹא־נָ ַתן ְבּנֶ ֶשְׁך‬stacked) ‫עֹולם׃‬ ָ ‫ה־א ֶלּה לֹא יִמֹּוט ְל‬ ֵ ‫וְ שׁ ַֹחד ַעל־נָ ִקי לֹא ָל ָקח ע ֵֹשׂ‬ ‘∅(he)i whi (that) ___i does not slander by his tongue, and whi (that) ___i does not do evil to his companion, and whi (that) a reproach ___i does not take up against his neighbors, whi (that) the despised person in hisi eyes is rejected, and whi (that) the fearers of Yhwh ___i honors, whi (that) ___i swears so that he harms but does not change, whi (that) his money ___i does not give with usury, and whi (that) a bribe against the innocent ___i does not take; the doer of these things will not ever stumble’ ‫מֹותם‬ ָ ‫ת־שׁ‬ ְ ‫ל־א ָשּׂא ֶא‬ ֶ ‫וּב‬ ַ ‫יהם ִמ ָדּם‬ ֶ ‫ל־אַסּיְך נִ ְס ֵכּ‬ ִ ‫אַחר ָמ ָהרוּ ַבּ‬ ֵ ‫בֹותם‬ ָ ‫ יִ ְרבּוּ ַע ְצּ‬Ps 16:4 ‫ל־שׂ ָפ ָתי׃‬ ְ ‫ַע‬ ‘let ∅(those)i whi (that) ___i “marry” another (god) increase their sorrows; their drink offerings of blood I will not pour out and I will not take their names upon my lips’ ‫ ְתּ ִפ ָלּה ְל ָדוִ ד ִשׁ ְמ ָעה יְ הוָ ה ׀ ֶצ ֶדק ַה ְק ִשׁ ָיבה ִרנָּ ִתי ַה ֲאזִ ינָ ה ְת ִפ ָלּ ִתי ְבּלֹא ִשׂ ְפ ֵתי‬Ps 17:1 ‫ִמ ְר ָמה׃‬ ‘a prayer of David: hear a just cause, O Yhwh; attend to my ringing cry; give ear to myi prayer whi (that) ___i am without lips of deceit’ ‫ ִמ ְפּנֵ י ְ ֭ר ָשׁ ִעים זוּ ַשׁדּוּנִ י א ַֹיְבי ְבּנֶ ֶפשׁ יַ ִקּיפוּ ָע ָלי׃‬Ps 17:9 ‘from the wicked who despoil me, my enemiesi whi (that) mortally ___i surround me’

316

Appendix A

‫וּמ ַפ ְל ִטי ֵא ִלי צוּ ִ֭רי ֶא ֱח ֶסה־בֹּו ָמגִ נִּ י וְ ֶק ֶרן־יִ ְשׁ ִעי ִמ ְשׂגַּ ִבּי׃‬ ְ ‫צוּד ִתי‬ ָ ‫וּמ‬ ְ ‫ יְ הוָ ה ׀ ַס ְל ִעי‬Ps 18:3 ‘Yhwh is my rock, my fortress, and my deliverer, my God, my rocki whi (that) I take refuge in himi, my shield, and the horn of my salvation, my stronghold’ ‫ימנִ י ְלרֹאשׁ גֹּויִ ם ַעם לֹא־יָ ַד ְע ִתּי יַ ַע ְבדוּנִ י׃‬ ֵ ‫ ְתּ ַפ ְלּ ֵטנִ י ֵמ ִר ֵיבי ָעם ְתּ ִשׂ‬Ps 18:44 ‘you delivered me from strivings with people; you made me a head of nations; a peoplei whi (that) I had not known ___i served me’ ‫עֹולם וָ ֶעד׃‬ ָ ‫ ַחיִּ ים ׀ ָשׁאַל ִ ֭מ ְמָּך נָ ַת ָתּה לֹּו א ֶֹרְך יָ ִמים‬Ps 21:5 ‘the lifei whi (that) he asked ___i from you you gave to him—length of days forever and ever’ ‫יֹורנּוּ ְבּ ֶד ֶרְך ְיִב ָחר׃‬ ֶ ‫ ִמי־זֶ ה ָ ֭ה ִאישׁ יְ ֵרא יְ הוָ ה‬Ps 25:12 (2×) ‘who is this, the mani whi (that) ___i is a fearer of Yhwh? He will teach them the wayj whj (that) he should choose ___j’ ‫אָדם לֹא יַ ְחשׁ ֹב יְ הוָ ה לֹו ָעֹון וְ ֵאין ְבּרוּחֹו ְר ִמיָּ ה׃‬ ָ ‫אַשׁ ֵרי‬ ְ Ps 32:2 (2×, ‘happy is the mani whi (that) Yhwh does not reckon to himi iniquity stacked) and whi (that) deceit is not in hisi spirit’ ‫ֹלהיו ָה ָעם ׀ ָבּ ַחר ְלנַ ֲח ָלה לֹו׃‬ ָ ‫אַשׁ ֵרי ַ ֭הגֹּוי ֲא ֶשׁר־יְ הוָ ה ֱא‬ ְ Ps 33:12 ‘happy is the nation that Yhwh is its God, the peoplei whi (that) he has chosen iti as his heritage’ ‫אַשׁ ֵרי ַהגֶּ ֶבר יֶ ֱח ֶסה־בֹּו׃‬ ְ ‫ ַט ֲעמוּ וּ ְ֭ראוּ ִכּי־טֹוב יְ הוָ ה‬Ps 34:9 ‘taste and see that Yhwh is good; happy is the mani whi (that) ___i takes refuge in him’ ‫ל־בּהּ׃‬ ָ ‫ר־ט ַמן ִתּ ְל ְכּדֹו ְבּשֹׁואָה יִ ָפּ‬ ָ ‫בֹואהוּ שֹׁואָה לֹא־יֵ ָדע וְ ִר ְשׁתֹּו ֲא ֶשׁ‬ ֵ ‫ ְתּ‬Ps 35:8 ‘let come upon him a disasteri whi (that) he does not know ___i; and let the net that he hid ensnare him; with disaster let him fall in it’ ‫ח־פּיו׃‬ ִ ‫וּכ ִא ֵלּם לֹא יִ ְפ ַתּ‬ ְ ‫ וַ ֲאנִ י ְכ ֵח ֵרשׁ לֹא ֶא ְשׁ ָמע‬Ps 38:14 ‘and I, like a deaf person I do not hear, and like a mutei whi (that) ___i does not open his mouth’ ‫ימי ֶק ֶדם׃‬ ֵ ‫יהם ִבּ‬ ֶ ‫ימ‬ ֵ ‫רוּ־לנוּ פּ ַֹעל ָפּ ַע ְל ָתּ ִב‬ ָ ‫בֹותינוּ ִס ְפּ‬ ֵ ‫ֹלהים ׀ ְבּאָזְ נֵ ינוּ ָשׁ ַמ ְענוּ ֲא‬ ִ ‫ ֱא‬Ps 44:2 ‘we have heard with our ears, O God, our ancestors have told us, the worki whi (that) you did ___i in their days, in the days of old’ ‫ֹלהים ָ ֭לנוּ ַמ ֲח ֶסה וָ עֹז ֶעזְ ָרה ְב ָצרֹות נִ ְמ ָצא ְמאֹד׃‬ ִ ‫ ֱא‬Ps 46:2 ‘God is for us a refuge and strength, a helpi whi (that) in troubles ___i is easily found’ ‫ית׃‬ ָ ‫יענִ י ָשׂשֹׂון וְ ִשׂ ְמ ָחה ָתּגֵ ְלנָ ה ֲע ָצמֹות ִדּ ִכּ‬ ֵ ‫ ַ ֭תּ ְשׁ ִמ‬Ps 51:10 ‘let me hear joy and gladness; let the bonesi whi (that) you crushed ___i rejoice’ ‫ֹלהים ָמעוּזֹּו ַ ֭ו ְיִּב ַטח ְבּר ֹב ָע ְשׁרֹו יָ עֹז ְבּ ַהוָּ תֹו׃‬ ִ ‫ ִהנֵּ ה ַהגֶּ ֶבר לֹא יָ ִשׂים ֱא‬Ps 52:9 ‘look, the mani whi (that) ___i would not make God his stronghold, but trusted in abundance of riches, and was strong in his desire!’ ‫ יֹום ִא ָירא ֲאנִ י ֵא ֶליָך ֶא ְב ָטח׃‬Ps 56:4 ‘(in) the dayi whi (that) I am afraid ___i, I put my trust in you’

Relative Clause Data

317

‫ֹלהים ִלי׃‬ ִ ‫י־א‬ ֱ ‫אֹויְבי אָ֭חֹור ְבּיֹום ֶא ְק ָרא זֶ ה־יָ ַד ְע ִתּי ִכּ‬ ַ ‫ אָז יָ שׁוּבוּ‬Ps 56:10 ‘then my enemies shall turn back, on the dayi whi (that) I call out ___i; this I know, that God is for me’ ‫מֹו־פ ֶתן ֵח ֵרשׁ יַ ְא ֵטם אָזְ נֹו׃‬ ֶ ‫ת־למֹו ִכּ ְדמוּת ֲח ַמת־נָ ָחשׁ ְכּ‬ ָ ‫ ֲח ַמ‬Ps 58:5 ‘they have venom like the venom of a serpent, like a deaf adderi whi (that) ___i stops its ear’ Ps 65:5 (3×, ‫יכ ֶלָך׃‬ ָ ‫יתָך ְקד ֹשׁ ֵה‬ ֶ ‫וּת ָק ֵרב יִ ְשׁכֹּן ֲח ֵצ ֶריָך ִנ ְ֭שׂ ְבּ ָעה ְבּטוּב ֵבּ‬ ְ ‫אַשׁ ֵרי ׀ ִתּ ְב ַחר‬ ְ ‘happy is ∅(he)i whi (that) you choose ___i and whi (that) you bring stacked) ___i near, whi (that ) ___i will live in your courts; we shall be satisfied with the goodness of your house, your holy temple’ ‫ ַה ָשּׂם ַ ֭נ ְפ ֵשׁנוּ ַבּ ַחיִּ ים וְ לֹא־נָ ַתן ַלמֹּוט ַרגְ ֵלנוּ׃‬Ps 66:9 ‘∅(he)i who places our life among the living, and whi ___i has not let our feet stumble’ ‫י־כ ֶסף ִבּזַּ ר ַע ִמּים ְק ָרבֹות‬ ָ ‫אַבּ ִירים ׀ ְבּ ֶעגְ ֵלי ַע ִמּים ִמ ְת ַר ֵפּס ְבּ ַר ֵצּ‬ ִ ‫ גְּ ַער ַחיַּ ת ָקנֶ ה ֲע ַדת‬Ps 68:31 ‫יֶ ְח ָפּצוּ׃‬ ‘rebuke the creature of the reeds, the herd of bulls, with the calves of the peoples, trampling under foot pieces of silver; scatter the peoplesi whi (that) ___i enjoy wars’ [vocalizing ‫ בזר‬as imperative] ‫רֹועָך ְלדֹור ְל ָכל־יָבֹוא גְּ ב‬ ֲ ְ‫אַל־תּ ַעזְ ֵבנִ י ַעד־אַגִּ יד ז‬ ַ ‫ֹלהים‬ ִ ‫ וְ גַ ם ַעד־זִ ְקנָ ה ׀ וְ ֵשׂ ָיבה ֱא‬Ps 71:18 ‫ָוּר ֶתָך׃‬ ‘and even up to old age and gray hair, O God, do not forsake me, until (that) I tell your might to a generation, (until I tell) your power to alli whi (that) ___i will come’ ‫ר־ציֹּון זֶ ה ׀ ָשׁ ַכנְ ָתּ בֹּו׃‬ ִ ‫אַל ָתּ ֵשׁ ֶבט נַ ֲח ָל ֶתָך ַה‬ ְ ‫ית ֶקּ ֶדם ָ ֭גּ‬ ָ ִ‫ זְ כֹר ֲע ָד ְתָך ׀ ָקנ‬Ps 74:2 (2×, stacked) ‘remember your congregationi whi (that) you acquired ___i long ago, whi (that) you redeemed to be the tribe of your inheritance; remember Mount Zion, which you came to dwell in it’ ‫יהם׃‬ ֶ ֵ‫יס ְפּרוּ ִל ְבנ‬ ַ ִ‫ ְל ַמ ַען יֵ ְדעוּ ׀ דֹּור אַ ֲ֭חרֹון ָבּנִ ים יִ וָּ ֵלדוּ יָ ֻקמוּ ו‬Ps 78:6 ‘in order that a later generation might know (them), childreni whi (that) ___i will be born, and rise up and tell (them) to their children’ ‫ת־אל רוּחֹו׃‬ ֵ ‫א־ה ִכין ִלבֹּו וְ לֹא־נֶ ֶא ְמנָ ה ֶא‬ ֵ ֹ ‫סֹורר וּמ ֶֹרה דּ֭ ֹור ל‬ ֵ ‫בֹותם דֹּור‬ ָ ‫ וְ לֹא יִ ְהיוּ ׀ ַכּ ֲא‬Ps 78:8 (2×, ‘and they will not be like their ancestors, a stubborn and rebellious stacked) generation, a generationi whi (that) ___i make firm its heart, whi (that) itsi spirit was not faithful to God’ ‫הֹולְך וְ לֹא יָ שׁוּב׃‬ ֵ ‫רוּח‬ ַ ‫י־ב ָשׂר ֵה ָמּה‬ ָ ‫ ַ ֭ויִּ זְ כֹּר ִכּ‬Ps 78:39 ‘and he remembered that they were flesh, a windi that passes and whi (that) ___i does not return’ ‫ל־א ֶרץ ִמ ְצ ָריִ ם ְשׂ ַפת לֹא־יָ ַד ְע ִתּי ֶא ְשׁ ָמע׃‬ ֶ ‫יהֹוסף ָשׂמֹו ְ ֭בּ ֵצאתֹו ַע‬ ֵ ‫ ֵעדוּת ׀ ִבּ‬Ps 81:6 ‘he made it a testimony in Joseph, when he went out over the land of Egypt; a languagei whi (that) I not known ___i I hear’

318

Appendix A

‫וּכ ֶל ָה ָבה ְתּ ַל ֵהט ָה ִרים׃‬ ְ ‫ ְכּ ֵאשׁ ִתּ ְב ַער־יָ ַער‬Ps 83:15 (2×) ‘like a firei whi (that) ___i consumes a forest, and like the flamej whj ___j burns the mountains’ ‫אָדם עֹוז־לֹו ָבְך ְמ ִסלֹּות ִבּ ְל ָב ָבם׃‬ ָ ‫אַשׁ ֵרי‬ ְ Ps 84:6 (2×, ‘happy is the mani whi (that) strength is hisi because of you, whi stacked) (that) highways are in theiri heart’ ‫ זְ ַר ְמ ָתּם ֵשׁנָ ה יִ ְהיוּ ַבּבּ ֶֹקר ֶכּ ָח ִציר יַ ֲחֹלף׃‬Ps 90:5 ‘you sweep them away; they will become (like) sleep; in the morning, (they will become) like grassi whi (that) ___i passes away’ ‫יתנוּ ְשׁנֹות ָר ִאינוּ ָר ָעה׃‬ ָ ִ‫ ַשׂ ְמּ ֵחנוּ ִכּימֹות ִענּ‬Ps 90:15 (2×) ‘make us glad acccording to the daysi whi (that) you have afflicted us ___i, and yearsj whj (that) we saw evil ___j’ ‫עוּריְ ִכי׃‬ ָ ְ‫ ַה ַמּ ְשׂ ִבּיַ ע ַבּטֹּוב ֶע ְדיֵ ְך ִתּ ְת ַח ֵדּשׁ ַכּנֶּ ֶשׁר נ‬Ps 103:5 ‘[he (v. 2)] who satisfies with goodness youri ornamentation, whi (that) your youth is renewed like the eagle’ ‫אָרץ׃‬ ֶ ‫בוּל־שׂ ְמ ָתּ ַבּל־יַ ֲעבֹרוּן ַבּל־יְ שׁוּבוּן ְל ַכסֹּות ָה‬ ַ֭ ְ‫ גּ‬Ps 104:9 ‘a boundaryi whi (that) you set ___i they may not pass; they may not again cover the earth.’ ‫אָרץ וַ ִתּ ְר ָעד יִ גַּ ע ֶבּ ָה ִרים וְ יֶ ֱע ָשׁנוּ׃‬ ֶ ‫ ַה ַמּ ִבּיט ָ ֭ל‬Ps 104:32 ‘[Godi (v. 31)], who looks on the earth and it trembles, whi (that) ___i touches the mountains and they smoke’ ‫עֹולם ְבּ ִריתֹו ָדּ ָבר ִצוָּ ה ְל ֶא ֶלף דֹּור׃‬ ָ ‫ זָ ַכר ְל‬Ps 105:8 ‘he remembers his covenant forever, the wordi whi (that) he commanded ___i for a thousand generations’ ‫א־ד ֶרְך׃‬ ָ ֹ ‫ שׁ ֵֹפְך ֭בּוּז ַעל־נְ ִד ִיבים וַ יַּ ְת ֵעם ְבּתֹהוּ ל‬Ps 107:40 ‘he pours contempt on princes and has caused them to wander in a wastelandi whi (that) ___i is not a road’ ‫מֹותת׃‬ ֵ ‫ישׁ־ענִ י וְ ֶא ְביֹון וְ נִ ְכ ֵאה ֵל ָבב ְל‬ ָ ‫ יַ ַען ֲא ֶשׁר ׀ לֹא זָ ַכר ֲעשֹׂות ָח ֶסד וַ יִּ ְרדּ ֹף ִא‬Ps 109:16 ‘because that he did not remember to do kindness and pursued a mani whi (that) ___i was poor and needy and broken of heart to kill (himself)’ ‫ ֶ֭א ֶבן ָמ ֲאסוּ ַהבֹּונִ ים ָהיְ ָתה ְלרֹאשׁ ִפּנָּ ה׃‬Ps 118:22 ‘the stonei whi (that) the builders rejected ___i has become the cornerstone’ ‫ ָשׁ ְמ ֵרנִ י ִמ ֵידי ַפח יָ ְקשׁוּ ִלי וּמ ְֹקשֹׁות פּ ֲֹע ֵלי אָוֶ ן׃‬Ps 141:9 ‘keep me from the hands of ∅(those)i whi (that) a trap ___i laid for me, and from the snares of evildoers’ ‫אָמר ה ָֹרה גָ ֶבר׃‬ ַ ‫אבד יֹום ִאוָּ ֶלד בֹּו וְ ַה ַלּיְ ָלה‬ ַ ֹ ‫ י‬Job 3:3 (2×) ‘let perish the dayi whi (that) I was born in iti, and the nightj whj (that) one said, “A man-child has been conceived”’ ‫יהם ָכּ ֶסף׃‬ ֶ ‫ם־שׂ ִרים זָ ָהב ָל ֶהם ַה ְמ ַמ ְל ִאים ָבּ ֵתּ‬ ָ ‫ אֹו ִע‬Job 3:15 ‘or with princesi whi (that) gold belongs to themi, who fill their houses (with) silver’

Relative Clause Data

319

‫א־ראוּ אֹור׃‬ ָ ֹ ‫ אֹו ְכנֵ ֶפל ָ ֭טמוּן לֹא ֶא ְהיֶ ה ְכּע ְֹל ִלים ל‬Job 3:16 ‘or (why) could I not have been like a hidden miscarriage, like sucklingsi whi (that) ___i have not seen the light?’ ‫אָבד וְ ֵאיפֹה יְ ָשׁ ִרים נִ ְכ ָחדוּ׃‬ ָ ‫ זְ ָכר־נָ א ִמי הוּא נָ ִקי‬Job 4:7 (3×, 2× stacked) ‘remember, please, who is hei whi (that) ___i was innocent (and) whi (that) ___i perished? And where are the uprightj whj (that) ___j were destroyed?’ ‫וּת ִהי עֹוד ׀ נֶ ָח ָמ ִתי וַ ֲא ַס ְלּ ָדה ְ ֭ב ִח ָילה לֹא יַ ְחמֹול ִכּי־לֹא ִכ ַח ְד ִתּי ִא ְמ ֵרי ָקדֹושׁ׃‬ ְ Job 6:10 ‘and it would still be my comfort; and I would revel in the paini whi (that) ___i does not spare, because I have not denied the words of the Holy One’ ‫קֹומם׃‬ ָ ‫ ְ ֭בּ ֵעת יְ ז ְֹרבוּ נִ ְצ ָמתוּ ְבּ ֻחמֹּו נִ ְד ֲעכוּ ִמ ְמּ‬Job 6:17 ‘at the timei whi (that) they should flow ___i, they have vanished; when it is hot, they vanish from their place’ ‫וּכ ָשׂ ִכיר יְ ַקוֶּ ה ָפ ֳעלֹו׃‬ ְ ‫אַף־צל‬ ֵ ‫ ְכּ ֶע ֶבד יִ ְשׁ‬Job 7:2 (2×) ‘like a slavei whi (that) ___i pants for a shade, and like a laborerj whj (that) ___j waits for his wage’ ‫ם־אנִ יֹּות ֵא ֶבה ְכּנֶ ֶשׁר יָ טוּשׂ ֲע ֵלי־א ֶֹכל׃‬ ֳ ‫ ָח ְלפוּ ִע‬Job 9:26 ‘they pass by like boats of reed, like an eaglei whi (that) ___i swoops upon food’ ‫ ִכּי־אַ ָ֭תּה ָע ָמל ִתּ ְשׁ ָכּח ְכּ ַמיִ ם ָע ְברוּ ִתזְ כֹּר׃‬Job 11:16 ‘because you will forget toil; like watersi whi (that) ___i have passed by, you will remember (your toil)’ ‫ וְ הוּא ְכּ ָר ָקב ְיִב ֶלה ְכּ ֶבגֶ ד ֲא ָכלֹו ָעשׁ׃‬Job 13:28 ‘and he wastes away like a rotten thing, like a garmenti whi (that) a moth has eaten iti’ ‫ימה־נָּ א ָע ְר ֵבנִ י ִע ָמְּך ִמי הוּא ְליָ ִדי יִ ָתּ ֵק ַע׃‬ ָ ‫ ִשׂ‬Job 17:3 ‘put down a pledge for me with yourself; who is hei whi (that) by my hand ___i will be struck?’ ‫ע־אל׃ ס‬ ֵ ‫אְַך־א ֶלּה ִמ ְשׁ ְכּנֹות ַעוָּ ל וְ זֶ ה ְמקֹום לֹא־יָ ַד‬ ֵ Job 18:21 ‘surely these are the dwellings of the perverse, and this is the place of ∅(those)i whi (that) ___i do not know God’ ‫רוּ־לי׃‬ ִ ‫א־תבֹשׁוּ ַתּ ְה ְכּ‬ ֵ ֹ ‫ זֶ ה ֶע ֶשׂר ְפּ ָע ִמים ַתּ ְכ ִלימוּנִ י ל‬Job 19:3 ‘these ten timesi whi (that) you have cast reproach upon me ___i; you have not been ashamed (that) you wrong me’ ‫וּמזִ מֹּות ָע ַלי ַתּ ְחמֹסוּ׃‬ ְ ‫יכם‬ ֶ ‫בֹות‬ ֵ ‫ ֵהן יָ ַד ְע ִתּי ַמ ְח ְשׁ‬Job 21:27 ‘look—I know your thoughts, and the schemesi whi (that) you devise ___i against me’ ‫אַפּי׃‬ ִ ‫לֹוהּ ְבּ‬ ַ ‫רוּח ֱא‬ ַ ְ‫י־כל־עֹוד נִ ְשׁ ָמ ִתי ִבי ו‬ ָ ‫ ִכּ‬Job 27:3 (2×, ‘because all whilei whi (that) my breath is in me ___i and whi (that) stacked) the spirit of God is in my nose ___i’

320

Appendix A

‫וּמקֹום ַלזָּ ָהב יָ זֹקּוּ׃‬ ָ ‫מֹוצא‬ ָ ‫ ִכּי יֵ שׁ ַל ֶכּ ֶסף‬Job 28:1 ‘indeed, silver has a mine, and a place exists for gold whi (that) they refine ___i’ ‫ נָ ִתיב לֹא־יְ ָדעֹו ָעיִ ט וְ לֹא ְשׁזָ ַפתּוּ ֵעין אַיָּ ה׃‬Job 28:7 (2×, ‘a pathi whi (that) bird of prey does not know iti and whi (that) the stacked) falcon’s eye has not seen iti’ ‫י־א ַמ ֵלּט ָענִ י ְמ ַשׁוֵּ ַע וְ יָ תֹום וְ לֹא־ עֹזֵ ר לֹו׃‬ ֲ ‫ ִכּ‬Job 29:12 ‘because I delivered a poor who cries, and an orphan, and ∅(one)i whi (that) no one helps himi’ ‫ אָב אָנ ִֹכי ָל ֶא ְביֹונִ ים וְ ִרב לֹא־יָ ַד ְע ִתּי ֶא ְח ְק ֵרהוּ׃‬Job 29:16 ‘I was a father to the needy, and the cause of ∅(one)i whi (that) I do not know ___i—I investigated it’ ‫בוּאָתי ְת ָשׁ ֵרשׁ׃‬ ִ ‫ל־תּ‬ ְ ‫וּב ָכ‬ ְ ‫אכל‬ ֵ ֹ ‫ד־א ַבדֹּון תּ‬ ֲ ‫ ִכּי ֵאשׁ ִהיא ַע‬Job 31:12 (2×, ‘because firei is it whi (that) ___i consumes unto Abaddon, and whi stacked) (that) ___i burns to the root all my produce’ ‫ה־ב ְטנִ י ְכּיַ יִ ן לֹא־יִ ָפּ ֵת ַח ְכּאֹבֹות ֲח ָד ִשׁים ָיִבּ ֵק ַע׃‬ ִ ֵ‫ ִהנּ‬Job 32:19 ‘look—my belly is like winei whi (that) ___i is not opened; like new wineskins, it is ready to burst’ ‫מֹותיו לֹא ֻרּאוּ׃‬ ָ ‫ יִ ֶכל ְבּ ָשׂרֹו ֵמר ִֹאי ושׁפי [וְ ֻשׁפּוּ] ַע ְצ‬Job 33:21 ‘his flesh shall waste beyond sight; and its bonesi whi (that) ___i could not be seen were laid bare [Qere]’ ‫אָרח ְ ֭ל ֶח ְב ָרה ִעם־פּ ֲֹע ֵלי אָוֶ ן וְ ָל ֶל ֶכת ִעם־‬ ַ ְ‫ה־לּ ַעג ַכּ ָמּיִ ם׃ ו‬ ַ ‫ ִמי־גֶ ֶבר ְכּ ִאיֹּוב יִ ְשׁ ֶתּ‬Job 34:7–8 ‫י־ר ַשׁע׃‬ ֶ ‫( אַנְ ֵשׁ‬2×, stacked) ‘who is a mani like Job, whi (that) ___i drinks up mockery like water, and whi (that) ___ i travels in company with evildoers and by walking with the wicked?’ ‫ ַה ֵבּט ָשׁ ַמיִ ם ְוּר ֵאה וְ שׁוּר ְשׁ ָח ִקים גָּ ְבהוּ ִמ ֶמּךָּ ׃‬Job 35:5 ‘gaze at the heavens and see; observe the cloudsi whi (that) ___i are higher than you’ ‫י־מיִ ם יָ זֹקּוּ ָמ ָטר ְל ֵאדֹו׃‬ ָ ‫ ִכּי יְ גָ ַרע נִ ְט ֵפ‬Job 36:27 ‘because he draws up the drops of wateri whi (that) one refines ___i as rain for its mist’ ‫ ֵאי־זֶ ה ַה ֶדּ ֶרְך יִ ְשׁ ָכּן־אֹור וְ ח ֶֹשְׁך ֵאי־זֶ ה ְמקֹמֹו׃‬Job 38:19 ‘where is this, the way of ∅(the place)i whi (that) light dwells ___i, and darkness—where is this, its place?’ ‫י־אָרץ׃‬ ֶ ‫ ֵאי־זֶ ה ַה ֶדּ ֶרְך יֵ ָח ֶלק אֹור יָ ֵפץ ָק ִדים ֲע ֵל‬Job 38:24 (2×) ‘where is this, the way of ∅(the place)i whi (that) light is distributed ___i, and (where is) ∅(the place)j whj (that) one scatters the east wind ___j over the earth?’ ‫[תֹּוע ַבת] נַ ְפשֹׁו׃‬ ֲ ‫שׁ־הנָּ ה ָשׂנֵ א יְ הוָ ה וְ ֶשׁ ַבע תועבות‬ ֵ ‫ ֶשׁ‬Prov 6:16 ‘six are theyi whi (that) Yhwh hates ___i, seven are the abominations [Kethiv] of his soul’

Relative Clause Data

321

‫אַשׁ ֵרי ְדּ ָר ַכי יִ ְשׁמֹרוּ׃‬ ְ ְ‫עוּ־לי ו‬ ִ ‫ וְ ַע ָתּה ָבנִ ים ִשׁ ְמ‬Prov 8:32 ‘and now, O children, listen to me: happy are ∅(those)i whi (that) my ways ___i keep’ ‫וּשׁתוּ ְבּיַ יִ ן ָמ ָס ְכ ִתּי׃‬ ְ ‫ ְלכוּ ַל ֲחמוּ ְב ַל ֲח ִמי‬Prov 9:5 ‘come, eat of my bread and drink of the winei whi (that) I have mixed ___i’ ‫אָמ ָרה לֹּו׃‬ ְ ְ‫ר־לב ו‬ ֵ ‫י־פ ִתי יָ ֻסר ֵהנָּ ה וַ ֲח ַס‬ ֶ ‫ ִמ‬Prov 9:16 ‘“whoever is simple, turn in here!” And to ∅(him)i whi (that) ___i lacks sense she says, . . .’ ‫י־מ ָל ִכים יִ ְתיַ ָצּב‬ ְ ֵ‫אכתֹּו ִל ְפנ‬ ְ ‫ית ִאישׁ ׀ ָמ ִהיר ִבּ ְמ ַל‬ ָ ִ‫ ָחז‬Prov 22:29 ‘have you seen a mani whi (that) ___i is adept in his work? He will stand before kings’ [alt., adjective-internal PP: ‘an adept-in-hiswork man’; see §1:1.3] ‫וּכ ִצ ְפעֹנִ י יַ ְפ ִרשׁ׃‬ ְ ‫אַח ִריתֹו ְכּנָ ָחשׁ יִ ָשְּׁך‬ ֲ Prov 23:32 ‘its ends is like a snakei whi (that) ___i bites and like a viperj whj (2×) (that) ___j stings’ ‫ל־ריב לֹּא־לֹו׃‬ ִ ‫י־כ ֶלב ע ֵֹבר ִמ ְת ַע ֵבּר ַע‬ ָ ֵ‫ ַמ ֲחזִ יק ְבּאָזְ נ‬Prov 26:17 ‘one that seizes a passing dog by the ears is (like) one that meddles in a quarreli whi that ___i is not his’ ‫א־אָמרוּ הֹון׃‬ ְ ֹ ‫אַר ַבּע ל‬ ְ ‫לוּקה ׀ ְשׁ ֵתּי ָבנֹות ַהב ׀ ַהב ָשׁלֹושׁ ֵהנָּ ה לֹא ִת ְשׂ ַבּ ְענָ ה‬ ָ ‫ ַל ֲע‬Prov 30:15 (2×) ‘To the leech are two daughters; “Give, give! Three are theyi whi (that) ___i are not satisfied; four are ∅(they)j whj (that) ___j do not say, “Wealth!”’ ‫לוּה ְבנֵ י־נָ ֶשׁר׃ פ‬ ָ ‫אכ‬ ְ ֹ ‫רוּה ע ְֹר ֵבי־נַ ַחל וְ י‬ ָ ‫ת־אם יִ ְקּ‬ ֵ ‫ ַעיִ ן ׀ ִתּ ְל ַעג ְלאָב וְ ָתבוּז ִל ֲיקּ ַה‬Prov 30:17 (2×, stacked) ‘the eyei whi (that) ___i mocks a father and ___i shows contempt for obedience to a mother— ravens of the valley will peck it out and vultures will eat it’ ‫אַר ָבּ ָעה] לֹא יְ ַד ְע ִתּים׃‬ ְ ְ‫ֹלשׁה ֵ ֭ה ָמּה נִ ְפ ְלאוּ ִמ ֶמּנִּ י וארבע [ו‬ ָ ‫ ְשׁ‬Prov 30:18 ‘three are theyi whi (that) ___i are too wonderful for me; four [Qere] (2×) are ∅(they)j whj (that) I do not understand themj’ ‫י־אָרץ וְ ֵה ָמּה ֲח ָכ ִמים ְמ ֻח ָכּ ִמים׃‬ ֶ ֵ‫אַר ָבּ ָעה ֵ ֭הם ְק ַטנּ‬ ְ Prov 30:24 ‘four are theyi whi (that) ___i are the small things of the land, and they are wise things (that) are taught’ ‫יט ֵבי ָל ֶכת׃‬ ִ ‫אַר ָבּ ָעה ֵמ‬ ְ ְ‫יט ֵיבי ָצ ַעד ו‬ ִ ‫ֹלשׁה ֵה ָמּה ֵמ‬ ָ ‫ ְשׁ‬Prov 30:29 ‘three are theyi whi (that) ___i make a good stride; four are ∅(they)i (2×) whi (that) ___i make a good walk’ ‫אָדם ַל ֲענֹות בֹּו׃‬ ָ ‫ֹלהים ִל ְבנֵ י ָה‬ ִ ‫ הוּא ׀ ִענְ יַ ן ָרע נָ ַתן ֱא‬Eccl 1:13 ‘it is an unfortunate taski whi (that) God has given ___i to men to be occupied with iti’ ‫יֹוסיף ַמ ְכאֹוב׃‬ ִ ‫יֹוסיף ַדּ ַעת‬ ִ ְ‫ב־כּ ַעס ו‬ ָ ‫ ִכּי ְבּר ֹב ָח ְכ ָמה ָר‬Eccl 1:18 ‘because in an abundance of wisdom exists an abundance of vexation and hei whi (that) ___i will add knowledge will add pain’

322

Appendix A

Eccl 2:1 ‫אָמ ְר ִתּי ֲאנִ י ְבּ ִל ִבּי ְל ָכה־נָּ א ֲאנַ ְסּ ָכה ְב ִשׂ ְמ ָחה ְוּר ֵאה ְבטֹוב וְ ִהנֵּ ה גַ ם־הוּא ָה ֶבל׃‬ ַ ‘I said, I with my ‫לב‬: Come, I will make you experienced in joy; look upon ∅(what)i whi (that) ___i is good. See―even this is absurd!’ ‫ֹלהים גַּ ם־זֶ ה ֶה ֶבל ְוּרעוּת‬ ִ ‫חֹוטא נָ ַתן ִענְ יָ ן ֶל ֱאסֹוף וְ ִל ְכנֹוס ָל ֵתת ְלטֹוב ִל ְפנֵ י ָה ֱא‬ ֶ ‫ וְ ַל‬Eccl 2:26 ‫רוּח׃‬ ַ ‘and to the offender he has given the business of gathering and collecting in order to give (everything) to the onei whi (that) ___i is good before God. Even this is absurd and chasing wind’ ‫יתי ַתּ ַחת ַה ָשּׁ ֶמשׁ ע ֶֹשׁר ָשׁמוּר ִל ְב ָע ָליו ְל ָר ָעתֹו׃‬ ִ ‫חֹולה ָר ִא‬ ָ ‫ יֵ שׁ ָר ָעה‬Eccl 5:12 ‘There is a sickly misfortunei whi (that) I have seen ___i under the sun: wealth is kept by its owner, to his misfortune’ ‫יתי ַתּ ַחת ַה ָשּׁ ֶמשׁ ִכּ ְשׁגָ גָ ה ֶשׁיּ ָֹצא ִמ ִלּ ְפנֵ י ַה ַשּׁ ִלּיט׃‬ ִ ‫ יֵ שׁ ָר ָעה ָר ִא‬Eccl 10:5 ‘there is a tragedyi whi (that) I saw ___i under the sun, like a mistake that proceeds from the ruler’ ‫טוּעים ַבּ ֲע ֵלי ֲא ֻספֹּות נִ ְתּנוּ ֵמר ֶֹעה ֶא ָחד׃‬ ִ ְ‫וּכ ַמ ְשׂ ְמרֹות נ‬ ְ ‫ ִדּ ְב ֵרי ֲח ָכ ִמים ַכּ ָדּ ְרבֹנֹות‬Eccl 12:11 ‘the words of the wise are like goads and like nails sunk in are wellgrouped sayingsi, whi (that) ___i are given by one shepherd’ ‫יכה ׀ יָ ְשׁ ָבה ָב ָדד ָה ִעיר ַר ָבּ ִתי ָעם‬ ָ ‫ ֵא‬Lam 1:1 ‘how lonely sits the cityi whi (that) ___i was full of people!’ ‫יתה לֹא־יָ בֹאוּ ַב ָקּ ָהל ָלְך׃ ס‬ ָ ִ‫י־ר ֲא ָתה גֹויִ ם ָבּאוּ ִמ ְק ָדּ ָשׁהּ ֲא ֶשׁר ִצוּ‬ ָ ‫ ִכּ‬Lam 1:10 ‘indeed, she has seen the nations enter her sanctuary, who you commanded (that) they should not enter the congregationi whi ___i belongs to you’ [alt., NP-internal PP: ‘congregation belonging to you’] ‫ארי ִה ְכ ִשׁיל כּ ִֹחי נְ ָתנַ נִ י ֲאד ֹנָ י ִבּ ֵידי לֹא־‬ ִ ָ‫ל־צוּ‬ ַ ‫ נִ ְשׂ ַקד עֹל ְפּ ָשׁ ַעי ְבּיָ דֹו יִ ְשׂ ָתּ ְרגוּ ָעלוּ ַע‬Lam 1:14 ‫אוּכל קוּם׃ ס‬ ַ ‘the Lord gave me into the hands of ∅(those)i whi (that) I am not able to withstand ___i’ ‫את וְ יִ ְהיוּ ָכמֹונִ י׃ ס‬ ָ ‫יֹום־ק ָר‬ ָ ‫את‬ ָ ‫ ֵה ֵב‬Lam 1:21 ‘you brought the dayi whi (that) you have announced ___i, and let them be as I am’ ‫ ֲאנִ י ַהגֶּ ֶבר ָראָה ֳענִ י ְבּ ֵשׁ ֶבט ֶע ְב ָרתֹו׃‬Lam 3:1 ‘I am the mani whi (that) ___i has seen affliction under the rod of God’s wrath’ ‫ טֹוב יְ הוָ ה ְלקוָ ֹו ְלנֶ ֶפשׁ ִתּ ְד ְר ֶשׁנּוּ׃‬Lam 3:25 ‘Yhwh is good to those who wait for him, to the souli whi (that) ___i seeks him. ’ ‫ל־אל ַבּ ָשּׁ ָמיִ ם׃‬ ֵ ‫ל־כּ ָפּיִ ם ֶא‬ ַ ‫ נִ ָשּׂא ְל ָב ֵבנוּ ֶא‬Lam 3:41 ‘let us lift up our heart beside our hands to the Godi whi (that) ___i is in heaven’ [alt., NP-internal PP: ‘God in heaven’]

Relative Clause Data

323

‫אַל־תּ ָירא׃ ס‬ ִ ‫אָמ ְר ָתּ‬ ַ ָ‫ ָק ַר ְב ָתּ ְבּיֹום ֶא ְק ָר ֶאךּ‬Lam 3:57 ‘You came near on the dayi whi (that) I called you ___i; you said, “Do not fear!”’ ‫יהם׃ ס‬ ֶ ‫יוּכלוּ יִ גְּ עוּ ִבּ ְל ֻב ֵשׁ‬ ְ ‫ נָ עוּ ִעוְ ִרים ַבּחוּצֹות נְ ג ֲֹאלוּ ַבּ ָדּם ְבּלֹא‬Lam 4:14 ‘they wandered (as) blind men through the streets, defiled with blood, by ∅(those)i whi (that) others were not able to touch theiri garments’ ‫יֹושׁ ַע׃ ס‬ ִ ‫ ְבּ ִצ ִפּיָּ ֵתנוּ ִצ ִפּינוּ ֶאל־גֹּוי לֹא‬Lam 4:17 ‘at our watch-tower we looked for a nationi whi (that) ___i would not save’ ‫ֹלהים ֶאת־ר‬ ִ ‫וּבנְ יָ ִמן וְ ַהכּ ֲֹהנִ ים וְ ַה ְלוִ יִּ ם ְלכֹל ֵה ִעיר ָה ֱא‬ ִ ‫יהוּדה‬ ָ ‫אשׁי ָהאָבֹות ִל‬ ֵ ‫ וַ יָּ קוּמוּ ָר‬Ezra 1:5 ‫ירוּשׁ ָלםִ׃‬ ָ ‫ת־בּית יְ הוָ ה ֲא ֶשׁר ִבּ‬ ֵ ‫וּחֹו ַל ֲעלֹות ִל ְבנֹות ֶא‬ ‘and the heads of the families of Judah and Benjamin, and the priests and the Levites arose, with regard to everyonei whi (that) God had stirred hisi spirit to go up and rebuild the house of Yhwh that was in Jerusalem’ ‫וּשׁתוּ ַמ ְמ ַת ִקּים וְ ִשׁ ְלחוּ ָמנֹות ְל ֵאין נָ כֹון לֹו‬ ְ ‫אמר ָל ֶ֡הם ְלכוּ ִא ְכלוּ ַמ ְשׁ ַמנִּ ים‬ ֶ ֹ ‫ וַ יּ‬Neh 8:10 ‘and he said to them, “Go your way, eat the fat and drink sweet wine and send portions for ∅(the one)i whi (that) nothing is prepared for himi”’ ‫ל־דּוִ יד ֶח ְברֹונָ ה ְל ָה ֵ֞סב ַמ ְלכוּת ָשׁאוּל‬ ָ ‫אשׁי ֶה ָחלוּץ ַל ָצּ ָבא ָבּאוּ ַע‬ ֵ ‫ ְו ֵ֠א ֶלּה ִמ ְס ְפּ ֵ ֞רי ָר‬1 Chr 12:24 ‫ֵא ָליו ְכּ ִפי יְ הוָ ה׃ ס‬ ‘and these are the numbersi of the heads of those that were equipped for war whi (that) ___i came to David, to Hebron, to turn the kingdom of Saul over to him, according to the word of Yhwh’ ‫יתם ֵאת‬ ֶ ‫יכם וְ ַה ֲע ִל‬ ֶ ‫אַתּם וַ ֲא ֵח‬ ֶ ‫אשׁי ָהאָבֹות ַל ְלוִ יִּ ם ִה ְת ַק ְדּשׁוּ‬ ֵ ‫אַתּם ָר‬ ֶ ‫אמר ָל ֶהם‬ ֶ ֹ ‫ וַ יּ‬1 Chr 15:12 ‫ינֹותי לֹו׃‬ ִ ‫ל־ה ִכ‬ ֲ ‫ֹלהי יִ ְשׂ ָר ֵאל ֶא‬ ֵ ‫ֲארֹון יְ הוָ ה ֱא‬ ‘and you shall bring up the Ark of Yhwh, the God of Israel, to ∅(the place)i whi (that) I have prepared ___i for it’ ‫עֹולם ְבּ ִריתֹו ָדּ ָבר ִצוָּ ה ְל ֶא ֶלף דֹּור׃‬ ָ ‫ זִ ְכרוּ ְל‬1 Chr 16:15 ‘remember his covenant forever, the wordi whi (that) he commanded ___i, for a thousand generations’ ‫ֹלהי ְל ַמ ְע ָלה ִמ ָכּל־‬ ַ ‫ית־א‬ ֱ ‫שׁ־לי ְסגֻ ָלּה זָ ָהב וָ ָכ ֶסף נָ ַת ִתּי ְל ֵב‬ ִ ֶ‫ֹלהי י‬ ַ ‫צֹותי ְבּ ֵבית ֱא‬ ִ ‫ וְ עֹוד ִבּ ְר‬1 Chr 29:3 ‫ינֹותי ְל ֵבית ַהקּ ֶֹדשׁ׃‬ ִ ‫( ֲה ִכ‬2×) ‘and more, in my desire for the House of my God, I have my own possession of gold and silveri whi (that) I have dedicated ___i to the House of my God, above allj whj (that) I set aside ___j for the Holy House’ ‫ֹלהים ֶה ֱע ָלה ָדוִ יד ִמ ִקּ ְריַ ת יְ ָע ִרים ַבּ ֵה ִכין לֹו ָדּוִ יד ִ ֧כּי נָ ָטה־לֹו א ֶֹהל‬ ִ ‫ ֲא ָבל ֲארֹון ָה ֱא‬2 Chr 1:4 ‫ירוּשׁ ָלםִ׃‬ ָ ‫ִבּ‬ ‘but the Ark of God David had brought up from Kiriath-jearim into ∅(the place)i whi (that) David had prepared ___i for it, because he had pitched a tent for it in Jerusalem’

324

Appendix A

‫ן־ה ָשּׁ ָלל ֵה ִביאוּ ָבּ ָקר ְשׁ ַבע ֵמאֹות וְ צֹאן ִשׁ ְב ַעת ֲא ָל ִפים׃‬ ַ ‫ וַ יִּ זְ ְבּחוּ ַליהוָ ה ַבּיֹּום ַההוּא ִמ‬2 Chr 15:11 ‘and they sacrificed to Yhwh on that day, from the bootyi whi (that) they had brought ___i, seven hundred oxen and seven thousand sheep. ’ ‫ם־ל ָב ָבם ָשׁ ֵלם ֵא ָליו‬ ְ ‫אָרץ ֠ ְל ִה ְת ַחזֵּ ק ִע‬ ֶ ‫ל־ה‬ ָ ‫ ִכּי יְ הוָ ה ֵע ָ֞יניו ְמשׁ ְֹטטֹות ְבּ ָכ‬2 Chr 16:9 ‘because Yhwh—his eyes roam throughout the whole earth, to show himself strong with ∅(those)i whi (that) theiri heart is completely for him’ ‫רוּח־‬ ַ ‫אמר ֵאי זֶ ה ַה ֶדּ ֶרְך ָע ַ ֧בר‬ ֶ ֹ ‫ל־ה ֶלּ ִחי וַ יּ‬ ַ ‫יכיְ הוּ ַע‬ ָ ‫ת־מ‬ ִ ‫ן־כּנַ ֲענָ ה וַ יַּ ְך ֶא‬ ְ ‫ וַ יִּ גַּ שׁ ִצ ְד ִקיָּ הוּ ֶב‬2 Chr 18:23 ‫יְ הוָ ה ֵמ ִא ִתּי ְל ַד ֵבּר א ָֹתְך׃‬ ‘and he said, “Where is this, the wayi whi (that) the spirit of Yhwh passed from me to speak to you?”’ ‫ר־שׂ ִעיר‬ ֵ ‫ל־בּנֵ י ַעמֹּון מֹואָ֧ב וְ ַה‬ ְ ‫אָר ִבים ַע‬ ְ ‫וּת ִה ָלּה נָ ַתן יְ הוָ ה ׀ ֠ ְמ‬ ְ ‫וּב ֵעת ֵה ֵחלּוּ ְב ִרנָּ ה‬ ְ 2 Chr 20:22 ‫יהוּדה וַ יִּ נָּ גֵ פוּ׃‬ ָ ‫ַה ָבּ ִאים ִל‬ ‘and at the timei whi (that) they began with song and praise ____i, Yhwh set an ambush against the Ammonites, Moab, and Mount Seir, who had come against Judah, so that they were routed’ ‫ם‬ ֒ ִ‫ל־פּ ֻק ַדּת ַה ֶמּ ֶלְך ְבּיַ ד ַה ְלוִ יּ‬ ְ ‫ת־האָרֹון ֶא‬ ָ ‫ וַ יְ ִ֡הי ְבּ ֵעת ִיָביא ֶא‬2 Chr 24:11 ‘and it was, at the timei whi (that) one would bring the chest to the inspection of the king by the Levites ___i’ ‫גוּ־בם ְבזַ ַעף ַעד‬ ָ ‫הוּדה נְ ָתנָ ם ְבּיֶ ְד ֶכם וַ ַתּ ַה ְר‬ ָ ְ‫יכם ַעל־י‬ ֶ ‫בֹות‬ ֵ ‫י־א‬ ֲ ‫ֹלה‬ ֵ ‫הו֧ה ֱא‬ ָ ְ‫ ֠ ִהנֵּ ה ַבּ ֲח ַמת י‬2 Chr 28:9 ‫ַל ָשּׁ ַמיִ ם ִהגִּ ַיע׃‬ ‘look—in the anger of Yhwh, the God of your ancestors, against Judah, he gave them into your hand, and you killed them in a ragei whi (that) ___i reached up to heaven’ ‫עֹולה ֵה ֵחל ִשׁיר־יְ הוָ ה וְ ַה ֲחצ ְֹצרֹות וְ ַעל־יְ ֵדי ְכּ ֵלי ָדּוִ יד ֶמ ֶלְך־יִ ְשׂ ָר ֵאל׃‬ ָ ‫וּב ֞ ֵעת ֵה ֵחל ָה‬ ְ 2 Chr 29:21 ‘and at the timei whi (that) the burnt offering began ___i, the song of Yhwh and the trumpets began, and besides the instruments of King David of Israel’ ‫בֹותיו וְ לֹא ְכּ ָט ֳה ַרת ַהקּ ֶֹדשׁ׃ ס‬ ָ ‫ֹלהי ֲא‬ ֵ ‫ֹלהים ׀ יְ הוָ ה ֱא‬ ִ ‫ל־ל ָבבֹו ֵה ִכין ִל ְדרֹושׁ ָה ֱא‬ ְ ‫ ָכּ‬2 Chr 30:19 ‘everyonei whi (that) ___i has set his heart to seek God, Yhwh, the God of their ancestors’ ‫ל־ה ְתיַ ֵחשׂ ַבּ ְלוִ יִּ ם׃‬ ִ ‫וּל ָכ‬ ְ ‫ ֲאנָ ִשׁים ֲא ֶשׁר נִ ְקּבוּ ְבּ ֵשׁמֹות ָל ֵתת ָמנֹות ְל ָכל־זָ ָכר ַבּכּ ֲֹהנִ ים‬2 Chr 31:19 ‘the people who were designated by names were to give portions to every male among the priests and to everyonei whi (that) ___i was enrolled among the Levites’

7.  Extraposed Relative Data (133×) Included in the list below are all cases in which I have determined a relative head to be positioned at a distance from its modifying relative clause. This data set of 133 examples is exhaustive for cases involving ‫ אשׁר‬but also includes 8 zero-relatives (Exod 15:17; Lev 13:20, 39; 17:11; Jer 49:12; Hab

Relative Clause Data

325

1:5; and Job 31:12 [2×]) and 6 illustrative ‫ה‬-relatives (Ps 137:7; 147:7–8 [3×]; Ruth 1:22; 2:6), though I do not make any exhaustiveness claim for extraposed ‫ה‬-relatives. Note that some, if not many, of these examples are arguably not extraposition, properly speaking—that is, the rightward movement of the relative clause—but reflect the raising of the relative head (Focus-induced raising, for example) and the subsequent relative-clause stranding or are due to the presence of an extrasentential element (e.g., vocative, parenthesis) inserted after the formation of the relative clause. However, below, I provide the full data set and leave to the reader the specific interpretation of each example. ‫אָמר ַהיֹּום ְבּ ַהר יְ הוָ ה יֵ ָר ֶאה׃‬ ֵ ֵ‫ם־ה ָמּקֹום ַההוּא יְ הוָ ה ׀ יִ ְר ֶאה ֲא ֶשׁר י‬ ַ ‫אַב ָר ָהם ֵשׁ‬ ְ ‫ וַ יִּ ְק ָ ֧רא‬Gen 22:14 ‘and Abraham called the name of that placei ___j “Yhwh will provide,” [whi that it is said to this day ___i]j, “On the mount of Yhwh it shall be provided”’ ‫ן־מ ְל ָכּה ֵא ֶשׁת‬ ִ ‫תוּאל ֶבּ‬ ֵ ‫ וַ יְ ִהי־הוּא ֶט ֶרם ִכּ ָלּה ְל ַד ֵבּר֒ וְ ִה ֵנּ֧ה ִר ְב ָקה י ֵֹצאת ֲא ֶשׁר יֻ ְלּ ָדה ִל ְב‬Gen 24:15 ‫ל־שׁ ְכ ָמהּ׃‬ ִ ‫אַב ָר ָהם וְ ַכ ָדּהּ ַע‬ ְ ‫נָ חֹור ֲא ִחי‬ ‘and it happened before he finished speaking, and—look! Rebekahi ___j was coming out [whi that ___i was born to Bethuel]j’ ‫י־ב ֶטן׃‬ ָ ‫ר־מנַ ע ִמ ֵמְּך ְפּ ִר‬ ָ ‫ֹלהים אָנ ִֹכי ֲא ֶשׁ‬ ִ ‫אמר ֲה ַת ַחת ֱא‬ ֶ ֹ ‫ וַ יִּ ַחר־אַף יַ ֲעקֹב ְבּ ָר ֵחל וַ יּ‬Gen 30:2 ‘and Jacob’s anger burned against Rachel and he said, “In the place of Godi ___j am I, [whi that ___i has withheld from you the fruit of the womb?]j” ‫ת־פֹּוטי ֶפ ַרע‬ ִ ‫אָסנַ ת ַבּ‬ ְ ‫יֹוסף יֻ ַלּד ְשׁנֵ י ָבנִ ים ְבּ ֶט ֶרם ָתּבֹוא ְשׁנַ ת ָה ָר ָעב ֲא ֶשׁר יָ ְל ָדה־לֹּו‬ ֵ ‫וּל‬ ְ Gen 41:50 ‫כּ ֵֹהן אֹון׃‬ ‘and to Joseph, two sonsi ___j were born before the two famine years came, [whi that Asenath, the daughter of Potiphera, the priest of On, bore ___i for him]j’ ‫ֹלהים ָבּזֶ ה‬ ִ ‫ן־לי ֱא‬ ִ ‫ל־אָביו ָבּנַ י ֵהם ֲא ֶשׁר־נָ ַת‬ ִ ‫יֹוסף ֶא‬ ֵ ‫אמר‬ ֶ ֹ ‫ וַ יּ‬Gen 48:9 ‘and Joseph said to his father, “My sonsi ___j are they, [whi that God has given ___i to me here”]j”’ ‫וּב ַק ְשׁ ִתּי׃ פ‬ ְ ‫ל־אַחיָך ֲא ֶשׁר ָל ַק ְח ִתּי ִמיַּ ד ָה ֱאמ ִֹרי ְבּ ַח ְר ִבּי‬ ֶ ‫אַחד ַע‬ ַ ‫ וַ ֲא ִ֞ני נָ ַ ֧ת ִתּי ְלָך ְשׁ ֶכם‬Gen 48:22 ‘and I give to you one “shoulder”i ___j over your brothers, [whi that I took ___i from the hand of the Amorites with my sword and with my bow]j’ ‫ת־יֹוסף׃‬ ֵ ‫ל־מ ְצ ָריִ ם ֲא ֶשׁר לֹא־יָ ַדע ֶא‬ ִ ‫ְך־ח ָדשׁ ַע‬ ָ ‫ וַ יָּ ָקם ֶמ ֶל‬Exod 1:8 ‘and a new kingi ___j arose over Egypt [whi that ___i did not know Joseph]j’ ‫אתֹת׃ פ‬ ֹ ‫ת־ה‬ ָ ‫ת־ה ַמּ ֶטּה ַהזֶּ ה ִתּ ַקּח ְבּיָ ֶדָך ֲא ֶשׁר ַתּ ֲע ֶשׂה־בֹּו ֶא‬ ַ ‫ וְ ֶא‬Exod 4:17 ‘this staffi ___j take in your hand, [whi you shall perform with iti the signs]j’

326

Appendix A

‫ל־א ֶרץ ִמ ְצ ָריִ ם ֲא ֶשׁר ָכּמֹהוּ לֹא נִ ְהיָ ָתה וְ ָכמֹהוּ לֹא ת ִֹסף׃‬ ֶ ‫ וְ ָהיְ ָתה ְצ ָע ָקה גְ ד ָֹלה ְבּ ָכ‬Exod 11:6 ‘and there shall be a loud cryi ___j throughout the whole land of Egypt, [whi like iti has never been and like it will never be again]j’ ‫ ָמ ֧כֹון ְל ִשׁ ְב ְתָּך ָפּ ַע ְל ָתּ יְ הוָ ה ִמ ְקּ ָדשׁ ֲאד ֹנָ י כֹּונְ נוּ יָ ֶדיָך׃‬Exod 15:17 ‘a place (that) you made for you to dwell, O Yhwh, a sanctuaryi ___j O Lord, [whi (that) your hands established ___i]j’ ‫אתיָך ֵמ ֶא ֶרץ ִמ ְצ ַריִ ם ִמ ֵבּית ֲע ָב ִדים׃‬ ִ ‫הֹוצ‬ ֵ ‫ֹלהיָך ֲא ֶ ֧שׁר‬ ֶ ‫ אָנ ִֹכי יְ הוָ ה ֱא‬Exod 20:2 ‘Ii ___j am Yhwh, your God, [whi ___(I)i brought you out of the land of Egypt, out of the house of slavery]j’ ‫ל־מֹועד ִל ְפנֵ י יְ הוָ ה ֲא ֶשׁר ִאוָּ ֵעד ָל ֶכם ָשׁ ָמּה ְל ַד ֵבּר‬ ֵ ‫יכם ֶפּ ַתח א ֶֹה‬ ֶ ‫ ע ַֹלת ָתּ ִמיד ְלדֹר ֵֹת‬Exod 29:42 ‫ֵא ֶליָך ָשׁם׃‬ ‘(it shall be) a perpetual burnt offering for your generations (at) the entrance of the tent of meetingi ___j before Yhwh, [whi that I will meet with you to speak to you therei]j’ ‫ֹלהיָך יִ ְשׂ ָר ֵאל ֲא ֶשׁר‬ ֶ ‫אמרוּ ֵא ֶלּה ֱא‬ ְ ֹ ‫ וַ יִּ ַקּח ִמיָּ ָדם וַ יָּ ַצר אֹתֹו ַבּ ֶח ֶרט וַ יַּ ֲע ֵשׂהוּ ֵעגֶ ל ַמ ֵסּ ָכה וַ יּ‬Exod 32:4 ‫ֶה ֱעלוָּך ֵמ ֶא ֶרץ ִמ ְצ ָריִ ם׃‬ ‘and he took [the gold earings (v. 3)] from their hand and bound it in the bag and he made it (into) a calf of metal and he said, “These are your godsi ___j, O Israel, [whi that ___i brought you up from the land of Egypt!]j”’ ‫ת־ה ָדּם‬ ַ ‫ת־ה ָדּם וְ זָ ְרקוּ ֶא‬ ַ ‫אַהר ֹן ַהכּ ֲֹהנִ ים ֶא‬ ֲ ‫ת־בּן ַה ָבּ ָקר ִל ְפנֵ י יְ הוָ ה וְ ִה ְק ִריבוּ ְבּנֵ י‬ ֶ ‫ וְ ָשׁ ַחט ֶא‬Lev 1:5 ‫מֹועד׃‬ ֵ ‫ר־פּ ַתח א ֶֹהל‬ ֶ ‫ל־ה ִמּזְ ֵבּ ַח ָס ִביב ֲא ֶשׁ‬ ַ ‫ַע‬ ‘and one shall slaughter the bull before Yhwh; and Aaron’s sons the priests shall offer the blood and they shall dash the blood on the altari ___j all around [whi that ___i is at the entrance of the tent of meeting] j’ ‫אשר‬ ֶ ‫וּמ ְשׁ ַחת ָבנָ יו ֵמ ִא ֵשּׁי יְ הוָ ה ְבּיֹום ִה ְק ִריב א ָֹתם ְל ַכ ֵהן ַליהוָ ה׃‬ ִ ‫אַהר ֹן‬ ֲ ‫ זֹאת ִמ ְשׁ ַחת‬Lev 7:35–36 ‫ִצוָּ ה יְ הוָ ה ָל ֵתת ָל ֶהם ְבּיֹום ָמ ְשׁחֹו א ָֹתם‬ ‘this is the sharei of Aaron and the sharei of his sons ___j from the fire-offerings to Yhwh, on the day (that) one brought them to be priests for Yhwh, [whi that Yhwh commanded to give ___i to them on the day he anointed them]j’ ‫ע־צ ַר ַעת ִהוא ַבּ ְשּׁ ִחין ָפּ ָר ָחה׃‬ ָ ַ‫ נֶ ג‬Lev 13:20 ‘a leprous diseasei ___j is it [whi (that) in boils ___i has broken out]j’ ‫עֹור־בּ ָשׂ ָרם ֶבּ ָהר ֹת ֵכּהֹות ְל ָבנֹת בּ ַֹהק הוּא ָפּ ַרח ָבּעֹור ָטהֹור‬ ְ ‫ וְ ָראָה ַהכּ ֵֹהן וְ ִה ֵנּ֧ה ְב‬Lev 13:39 ‫הוּא׃ ס‬ ‘and the priest shall look, and—look, (if) on the skin of the body are dull white spots, a lesioni ___j is it [whi (that) ___i has broken out on the skin]j; he is clean’ ‫י־ה ָדּם הוּא ַבּנֶּ ֶפשׁ יְ ַכ ֵפּר׃‬ ַ ‫ ִכּ‬Lev 17:11 ‘because the bloodi ___j is it [whi (that) by means of life ___i atones]j’

Relative Clause Data

327

‫ת־שׁם ָק ְד ִשׁי ֲא ֶשׁר‬ ֵ ‫ל־בּנָ יו וְ יִ נָּ זְ רוּ ִמ ָקּ ְד ֵשׁי ְבנֵ י־יִ ְשׂ ָר ֵאל וְ לֹא יְ ַח ְלּלוּ ֶא‬ ָ ‫ל־אַהר ֹן וְ ֶא‬ ֲ ‫ ַדּ ֵבּר ֶא‬Lev 22:2 ‫ֵ ֧הם ַמ ְק ִדּ ִשׁים ִלי ֲאנִ י יְ הוָ ה׃‬ ‘speak to Aaron and his sons and let them keep separate from the holy donationsi of the children of Israel ___j (so they do not profane my holy name) [whi that they are consecrating ___i to me]j’ ‫יִמּ ְכרוּ ִמ ְמ ֶכּ ֶרת ָע ֶבד׃‬ ָ ‫אתי א ָֹתם ֵמ ֶא ֶרץ ִמ ְצ ָריִ ם לֹא‬ ִ ‫ר־הֹוצ‬ ֵ ‫י־ע ָב ַדי ֵהם ֲא ֶשׁ‬ ֲ ‫ ִכּ‬Lev 25:42 ‘because my servantsi ____j are they, [whi that I brought themi out of the land of Egypt]j; they shall not be sold as slaves are sold’ ‫אֹותם ֵמ ֶא ֶרץ ִמ ְצ ָריִ ם ֲאנִ י יְ הוָ ה‬ ָ ‫אתי‬ ִ ‫ר־הֹוצ‬ ֵ ‫י־לי ְבנֵ י־יִ ְשׂ ָר ֵאל ֲע ָב ִדים ֲע ָב ַדי ֵהם ֲא ֶשׁ‬ ִ ‫ ִכּ‬Lev 25:55 ‫יכם׃‬ ֶ ‫ֹלה‬ ֵ ‫ֱא‬ ‘because to me the sons of Israel are servants; my servantsi ___j are they [whi I brought themi out from the land of Egypt]j; I am Yhwh your God’ ‫יט ְבָך ְבּאַ‬ ִ ‫וּל ַמ ַען נַ סּ ֶֹתָך ְל ֵה‬ ְ ‫ ַה ַמּ ֲא ִכ ְלָך ָמן ַבּ ִמּ ְד ָבּר ֲא ֶשׁר לֹא־יָ ְדעוּן ֲאב ֶֹתיָך ְל ַמ ַען ַענּ ְֹתָך‬Deut 8:16 ‫יתָך׃‬ ֶ ‫ֲח ִר‬ ‘[Yhwh (v. 14), who fed you mannai in the wilderness [whi that your ancestors did not know ___i]j, in order to humble you and in order to test you to do you good in the end’ ‫אתם ִמ ָשּׁם‬ ֶ ‫א־שׁ ָמּה ְל ִר ְשׁ ָתּהּ לֹא ְכ ֶא ֶרץ ִמ ְצ ַריִ ם ִהוא ֲא ֶשׁר יְ ָצ‬ ָ ‫אַתּה ָב‬ ָ ‫אָרץ ֲא ֶשׁר‬ ֶ ‫ ִכּי ָה‬Deut 11:10 ‫ית ְב ַרגְ ְלָך ְכּגַ ן ַהיָּ ָרק׃‬ ָ ‫( ֲא ֶשׁר ִתּזְ ַרע ֶאת־זַ ְר ֲעָך וְ ִה ְשׁ ִק‬2×) ‘because the land that you are about to enter there to possess—not like the land of Egypti is it, [whi that you have come from therei, whi that you sow your seed ___i and irrigate by foot like a vegetable garden]j’ ‫הו֧ה‬ ָ ְ‫אַה ָבה ֶאת־י‬ ֲ ‫ל־ה ִמּ ְצוָ ה ַהזֹּאת ַל ֲעשׂ ָֹתהּ ֲא ֶשׁר אָנ ִֹכי ְמ ַצוְּ ָך ַהיֹּום ְל‬ ַ ‫ת־כּ‬ ָ ‫י־ת ְשׁמֹר ֶא‬ ִ ‫ ִכּ‬Deut 19:9 ‫ל־היָּ ִמים וְ יָ ַס ְפ ָתּ ְלָך עֹוד ָשֹׁלשׁ ָע ִרים ַעל ַה ָשֹּׁלשׁ ָה ֵא ֶלּה׃‬ ַ ‫ֹלהיָך וְ ָל ֶל ֶכת ִבּ ְד ָר ָכיו ָכּ‬ ֶ ‫ֱא‬ ‘if you keep this whole commandmenti ___j by doing it [whi that I command you ___i today]j, by loving Yhwh your God and walking always in his ways, then you shall add for yourself three cities in addition to these three’ ‫ל־אד ֹנָ יו ֲא ֶשׁר־יִ נָּ ֵצל ֵא ֶליָך ֵמ ִעם ֲאד ֹנָ יו׃‬ ֲ ‫א־ת ְסגִּ יר ֶע ֶבד ֶא‬ ַ ֹ ‫ ל‬Deut 23:16 ‘you will not deliver a slavei ___j to his master [whi that ___i escaped to you from his master]j’ ‫את ֶכם ִמ ִמּ ְצ ָריִ ם׃ ֲא ֶשׁר ָק ְרָך ַבּ ֶדּ ֶרְך וַ יְ זַ נֵּ ב‬ ְ ‫ר־ע ָשׂה ְלָך ֲע ָמ ֵלק ַבּ ֶדּ ֶרְך ְבּ ֵצ‬ ָ ‫ זָ כֹור ֵאת ֲא ֶשׁ‬Deut ‫ֹלהים׃‬ ִ ‫אַתּה ָעיֵ ף וְ יָ גֵ ַע וְ לֹא יָ ֵרא ֱא‬ ָ ְ‫אַח ֶריָך ו‬ ַ ‫ל־הנֶּ ֱח ָשׁ ִלים‬ ַ ‫ ְבָּך ָכּ‬25:17–18 ‘remember (the thing) that Amaleki ___j did to you on the road when you went out from Egypt, [whi that ___i encountered you on the road and cut off among you all the stragglers behind you and you were tired and weary]j, and he did not fear God’

328

Appendix A

‫ת־ה ִעיר‬ ָ ‫וּבנָ ה ֶא‬ ָ ‫הֹושׁ ַע ָבּ ֵעת ַה ִהיא ֵלאמֹר אָרוּר ָה ִאישׁ ִל ְפנֵ י יְ הוָ ה ֲא ֶשׁר יָ קוּם‬ ֻ ְ‫ וַ יַּ ְשׁ ַבּע י‬Josh 6:26 ‫יה׃‬ ָ ‫וּב ְצ ִעירֹו יַ ִצּיב ְדּ ָל ֶת‬ ִ ‫ַהזֹּאת ֶאת־יְ ִריחֹו ִבּ ְבכֹרֹו יְ יַ ְסּ ֶדנָּ ה‬ ‘and Joshua swore at that time, “Cursed is the mani ___j before Yhwh [whi that ___i rises and rebuilds this city, Jericho!]j At the cost of his firstborn he shall lay its foundation, and at the cost of his youngest he shall set up its gates!”’ ‫אוּבן וַ ֲח ִצי ֵשׁ ֶבט ַה ְמנַ ֶשּׁה‬ ֵ ‫י־כ ֻהנַּ ת יְ הוָ ה נַ ֲח ָלתֹו וְ גָ ד ְוּר‬ ְ ‫ין־ח ֶלק ַל ְלוִ יִּ ם ְבּ ִק ְר ְבּ ֶכם ִכּ‬ ֵ ‫ ִכּי ֵא‬Josh 18:7 ‫ָל ְקחוּ נַ ֲח ָל ָתם ֵמ ֵע ֶבר ַליַּ ְר ֵדּן ִמזְ ָר ָחה ֲא ֶשׁר נָ ַתן ָל ֶהם מ ֶֹשׁה ֶע ֶבד יְ הוָ ה׃‬ ‘because a portion does not belong to the Levites among you, because the priesthood of Yhwh is their inheritance; and Gad and Reuben and the half-tribe of Manasseh have received their inheritancei ___j beyond the Jordan eastward, [whi that Moses the servant of Yhwh gave them ___i]j’ ‫יהם ֲא ֶשׁר לֹא־יָ ְדעוּ‬ ֶ ‫אַח ֵר‬ ֲ ‫אַחר‬ ֵ ‫בֹותיו וַ יָּ ָקם דֹּור‬ ָ ‫ל־א‬ ֲ ‫ל־הדֹּור ַההוּא נֶ ֶא ְספוּ ֶא‬ ַ ‫ וְ גַ ם ָכּ‬Judg 2:10 ‫ת־ה ַמּ ֲע ֶשׂה ֲא ֶשׁר ָע ָשׂה ְליִ ְשׂ ָר ֵאל׃ ס‬ ַ ‫ֶאת־יְ הוָ ה וְ גַ ם ֶא‬ ‘and moreover, all of that generation was gathered to their ancestors, and another generationi ___j rose after them, [whi that ___i did not know Yhwh or even the work that he had done for Israel]j’ ‫יתם לֹו׃ ֲא ֶשׁר־נִ ְל ַחם‬ ֶ ‫ם־כּגְ מוּל יָ ָדיו ֲע ִשׂ‬ ִ ‫ם־בּיתֹו וְ ִא‬ ֵ ‫יתם ִעם־יְ ֻר ַבּ ַעל וְ ִע‬ ֶ ‫ם־טֹובה ֲע ִשׂ‬ ָ ‫ וְ ִא‬Judg ‫יכם וַ יַּ ְשׁ ֵלְך ֶאת־נַ ְפשֹׁו ִמנֶּ גֶ ד וַ יַּ ֵצּל ֶא ְת ֶכם ִמיַּ ד ִמ ְדיָ ן׃‬ ֶ ‫אָבי ֲע ֵל‬ ִ 9:16–17 ‘and if according to the deed of his hands ∅(you)i ___j do to him [whi that my father fought for youi, and risked his life, and rescued you from the hand of Midian]j ‫י־דן׃‬ ָ ֵ‫גוּרים ְכּ ֵלי ִמ ְל ַח ְמ ָתּם נִ ָצּ ִבים ֶפּ ַתח ַה ָשּׁ ַער ֲא ֶשׁר ִמ ְבּנ‬ ִ ‫שׁ־מאֹות ִאישׁ ֲח‬ ֵ ‫ וְ ֵשׁ‬Judg 18:16 ‘and six hundred meni ___j girded (with) their weapons of war were stationed at the entrance of the gate [whi that ___i were from the Danites]j’ ‫ית־אל ִמזְ ְר ָחה ַה ֶשּׁ ֶמשׁ‬ ֵ ‫ימה ֲא ֶשׁר ִמ ְצּפֹונָ ה ְל ֵב‬ ָ ‫יָמ‬ ִ ‫אמרוּ ִהנֵּ ה ַחג־יְ הוָ ה ְבּ ִשׁלֹו ִמיָּ ִמים ׀‬ ְ ֹ ‫ וַ יּ‬Judg 21:19 ‫וּמנֶּ גֶ ב ִל ְלבֹונָ ה׃‬ ִ ‫ית־אל ְשׁ ֶכ ָמה‬ ֵ ‫ִל ְמ ִס ָלּה ָהע ָֹלה ִמ ֵבּ‬ ‘and they said, “Look, the festival of Yhwh is at Shilohi ___j, from year to year, [whi that ___i is north of Bethel, on the east of the highway that goes up from Bethel to Shechem, and to the south of Lebonah]j”’ ‫מוּאל ִה ֵנּ֧ה אָנ ִֹכי ע ֶֹשׂה ָד ָבר ְבּיִ ְשׂ ָר ֵאל ֲא ֶשׁר ָכּל־שׁ ְֹמעֹו ְתּ ִצ ֶלּינָ ה‬ ֵ ‫ל־שׁ‬ ְ ‫אמר יְ הוָ ה ֶא‬ ֶ ֹ ‫ וַ יּ‬1 Sam 3:11 ‫ְשׁ ֵתּי אָזְ נָ יו׃‬ ‘and Yhwh said to Samuel, “Look—I am doing a thingi ___j in Israel [whi that anyone who hears iti—both his ears will tingle]j’ ‫א־הגִּ יד‬ ִ ֹ ‫לוּכה ל‬ ָ ‫ת־דּ ַבר ַה ְמּ‬ ְ ‫אמר ָשׁאוּל ֶאל־דֹּודֹו ַהגֵּ ד ִהגִּ יד ָלנוּ ִכּי נִ ְמ ְצאוּ ָה ֲאתֹנֹות וְ ֶא‬ ֶ ֹ ‫ וַ יּ‬1 Sam 10:16 ‫מוּאל׃ פ‬ ֵ ‫אָמר ְשׁ‬ ַ ‫לֹו ֲא ֶשׁר‬ ‘and Saul said to his uncle, “He clearly to