Linguistic Studies on Biblical Hebrew 9789004448858, 9789004448841

Citation preview

Linguistic Studies on Biblical Hebrew

- 978-90-04-44885-8

Studies in Semitic Languages and Linguistics Editorial Board Aaron D. Rubin and Ahmad Al-Jallad

volume 102

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

- 978-90-04-44885-8

Linguistic Studies on Biblical Hebrew Edited by

Robert D. Holmstedt

leiden | boston

- 978-90-04-44885-8

Cover illustration: Fol. 107recto of the Aleppo Codex (ca. 925 c.e.) containing Isaiah 5:29–7:12. Public domain. Library of Congress Cataloging-in-Publication Data Names: Holmstedt, Robert D., editor. Title: Linguistic studies on biblical Hebrew / edited by Robert D. Holmstedt. Description: Leiden; Boston : Brill, [2021] | Series: Studies in semitic languages and linguistics, 0081-8461 ; volume 102 | Includes bibliographical references and index. Identifiers: lccn 2020054308 (print) | lccn 2020054309 (ebook) | isbn 9789004448841 (hardback) | isbn 9789004448858 (ebook) Subjects: lcsh: Bible. Old Testament–Language, style. | Hebrew language– Grammar. | Hebrew language–Grammar, Historical. | Hebrew language–Dialects–Israel–Tiberias. Classification: lcc pj4564 .l56 2021 (print) | lcc pj4564 (ebook) | ddc 492.4/5–dc23 lc record available at https://lccn.loc.gov/2020054308 lc ebook record available at https://lccn.loc.gov/2020054309

Typeface for the Latin, Greek, and Cyrillic scripts: “Brill”. See and download: brill.com/brill‑typeface. issn 0081-8461 isbn 978-90-04-44884-1 (hardback) isbn 978-90-04-44885-8 (e-book) Copyright 2021 by Robert D. Holmstedt. Published by Koninklijke Brill nv, Leiden, The Netherlands. Koninklijke Brill nv incorporates the imprints Brill, Brill Hes & De Graaf, Brill Nijhoff, Brill Rodopi, Brill Sense, Hotei Publishing, mentis Verlag, Verlag Ferdinand Schöningh and Wilhelm Fink Verlag. Koninklijke Brill nv reserves the right to protect this publication against unauthorized use. Requests for re-use and/or translations must be addressed to Koninklijke Brill nv via brill.com or copyright.com. This book is printed on acid-free paper and produced in a sustainable manner.

- 978-90-04-44885-8

For Edit (‫)ז״ל‬ and For Susan (‫)ז״ל‬

∵

- 978-90-04-44885-8

- 978-90-04-44885-8

Contents 1

Introduction: Linguistic Theory and Philology in the Study of Biblical Hebrew 1 Robert D. Holmstedt

2

Pausal vs. Context Forms in Tiberian Hebrew: A Multi-Planar Analysis of Vowel Reduction and Stress 9 Roman Himmelreich and Outi Bat-El Foux

3

Prosodic Dependency in Tiberian Hebrew Vincent DeCaen and B. Elan Dresher

4

Ordinals in Biblical Hebrew 60 Susan Rothstein and Adina Moshavi

5

Investigating Ellipsis in Biblical Hebrew Robert D. Holmstedt

6

A Unified Account of the Infinitive Absolute in Biblical Hebrew Elizabeth Cowper and Vincent DeCaen

7

The Nature of the Infinitive Absolute Galia Hatav

8

The Infinitive in Biblical Hebrew Edit Doron

9

Light Verbs in Biblical Hebrew Todd Snider

10

Argument Sharing Secondary Predicates in Biblical Hebrew E.J. Jacques Boulet

11

The Causative-Inchoative Alternation and the Semantics of Hiphil 209 Kevin Grasso

39

84

103

125

144

169

191

- 978-90-04-44885-8

viii 12

contents

Hybrid Syntactic Constructions in Biblical Hebrew Tamar Zewi

231

Index of Authors 253 Index of Scriptures 257

- 978-90-04-44885-8

chapter 1

Introduction: Linguistic Theory and Philology in the Study of Biblical Hebrew Robert D. Holmstedt

1

Introduction

The study of Biblical Hebrew grammar reaches back through the centuries to the medieval period, to at least Saadiah Gaon (882–942ce), who is typically considered to be the first Hebrew grammarian (Téné, Maman, and Barr 2007: 29–30). The medieval Jewish grammarians, from Saadiah Gaon to David Qimhi (1160–1235) and their successors for the subsequent four centuries described Hebrew grammar through the lens of the native Arabic grammatical descriptions that had already been developing for two centuries by his time. The emphasis was on description; rarely did they speculate about theoretical issues, such as the origin of language or the relationship of language to thought (Zwiep 2006: 257). Though Jewish grammatical work did not cease after the sixteenth century (see, for example, Spinoza’s Compendium grammatices linguae hebraeae [1677]), it was overshadowed by the Christian turn to Biblical Hebrew, spurred by both the humanism of the Renaissance and the Reformation zeal for scriptural sources (Téné, Maman, and Barr 2007: 54). Elijah Levita (1469–1549) served as the primary bridge between the Jewish tradition and the new Christian interest both with his own grammatical description as well as his study on the Masorah and a commentary on the grammar of David Qimhi’s brother Moses (1127–1190). Through the seventeenth and eighteenth centuries the description of Hebrew was increasingly done in concert with not only Latin and Greek, from which many grammatical categories were adopted, but alongside newly available Semitic languages, such as Arabic, Syriac, Aramaic, Ethiopic, and eventually also Akkadian in the nineteenth century. The nineteenth century also witnessed the bloom of the historical-comparative study of languages. The first generation of scholars involved in this type of study (e.g., Friedrich von Schlegel, Jacob Grimm, Rasmus Rask, Franz Bopp) cast their activity as “neue Philologie.” (Koerner 1997: 169). “Philology” used to describe a language related activity can traced back to Plato, but by the third century bce it had become “associated specifically with the study of

© Robert D. Holmstedt, 2021 | doi:10.1163/9789004448858_002

- 978-90-04-44885-8

2

holmstedt

language—with reading, rhetoric, literature, textual scholarship” (Turner 2014: 3). With the second generation of nineteenth century historical-comparative scholars (August Schleicher [1821–1868], in particular), philology as the study of language was being displaced in the world of language study: philology came to be seen as historical inquiry, using language as a vehicle in the study of culture; in contrast, linguistics was the scientific investigation of language itself (Koerner 1997: 170). The view of linguistics as a science continued to develop in the twentieth century. Saussure was both a pivotal and bridging figure—he viewed language as a system that could be described both diachronically (and so consistent with the historical focus of the nineteenth century) and synchronically (a significant new direction). Saussure continued the scientific view of language study but gave life to a new and soon dominant interest in describing synchronic systems. Moreover, Saussure’s sensitivity to the terminology, the use of data, and the relationship of linguistics to other disciplines, such as sociology and psychology (Sanders 2006: 769), prepared the scene for the methodological and theoretical advances in the twentieth century, from Bloomfield to Chomsky. Increasingly overshadowed in the twentieth-century linguistics revolution was philology. As a highly synthetic activity, attending to language within the context of literature and history, philology was soon orphaned in the increasing fragmentation of the academic world. With the study of language firmly ensconced in the realm of linguistics and the disciplines of history, literary studies, communication, and so on, hiving off much that had previously fit within the scope of philology, this venerable pursuit was marginalized, often considered odd and arcane: … no discipline in today’s university is more misunderstood, disdained, and threatened. For many, philologist is hardly more than a term of abuse, “what you call the dull boys and girls of the profession.” For others, philology has ceased to be. It is a “now defunct field,” a “protohumanistic empirical science” that “no longer exists as such,” its decline “a conspicuous and puzzling fact.” pollock 2009: 934; see also turner 2014: 381

The question that directly arises from this is “whether it still makes sense to speak institutionally and epistemologically of ‘philology.’ Does this venerable title still signify a truly coherent field, and not rather a multitude of scattered currents and competing genealogies, differing national characteristics and inconsistent methodologies?” (Bajohr et al 2014: 1; similarly, see Pollock 2015: 2–3). The answer is, arguably, no. For many, the term philology is inter-

- 978-90-04-44885-8

linguistic theory and philology in the study of biblical hebrew

3

changeable with historical-comparative linguistics (Barr 1969: 37; Sampson 1980: 243, n. 1). For others, philology concerned “those things which are peculary to specific texts” (whereas linguistics concerns language features common to all texts in a given language) (Gleason 1974: 200; Bodine 1987: 40; Turner 2014). For still others, philology is equated with textual criticism (Hendel 2016). And for some, philology seems to be catch-all term for any activity related to reading ancient texts in their original languages (for a recent example, see Najman 2017). At its best, philology should perhaps be seen as a reflection of a compilation of skills and attention to detail that result in a Nietzschean “slow reading” of texts, better opening the textual and linguistic data for analysis in whichever academic discipline is relevant. Nearly twenty years ago, Michael O’Connor anticipated the current discussion that has been rippling through Hebrew studies when he wrote (read “philology” for “reading”), Linguistics is, in fact, a science and therefore committed to a modern horizon, involving verifiability, falsifiability, or comparable criteria for proceeding; it is also, unlike biblical studies, oriented away from the unique. Reading, in contrast, is devoted to the unique. Reading, as an act or an endeavor can be modern in that sense, but it can also be pre-modern or pre-critical or it can be post-modern. It can be, as linguistics cannot be, naive or canny. 2002: 42

Seen this way, philology might be something to which any linguist or literary scholar should aspire—the careful, reflective, self-aware, linguistically competent analysis of texts. But it is imperative to recognize that there is no specific method or theory associated with this activity. As such, it is neither scientific nor identifiable as a discrete academic discipline. And so how do we process the continued recognition and gravitas of philology in biblical studies? On the one hand, lack of awareness of and/or facility with theories in related disciplines is understandable in a field requiring competence in multiple ancient languages, superficial knowledge of hundreds of literary compositions, and deep knowledge of dozens, all spanning centuries if not millennia. To become proficient in the field of biblical studies requires an immense intellectual and temporal investment (this also pertains to, for example, classics and medieval studies). In such large academic space there is room for the nontheoretical scholar to operate. On the other hand, the lack of any engagement with theory, whether linguistic or otherwise (literary, sociological, anthropological, etc.), is scientifically indefensible.

- 978-90-04-44885-8

4

holmstedt

A middle road allows that for many the case is simply lack of access to guidance through the theoretical landscape as well as with the complexities of analyzing ancient language date by means of a modern linguistic theory. Building bridges between the theoretical study of language and the larger world of Biblical Hebrew studies was the motivation two years ago for the creation of the Biblical Hebrew Linguistics and Philology Network (BHLaP).1 In the summer of 2018, the second workshop of this new network was held in Jerusalem, at the Hebrew University. The participants of the workshop were overwhelmingly theoretical linguistics, turning their attention to the analysis of Biblical Hebrew; a minority (including this editor) were Biblical Hebrew scholars who are committed to theoretically informed language research. This volume represents the fruit of that meeting, with (besides the introduction) eleven papers covering a variety of phonological, morphological, syntax, and semantic features of Biblical Hebrew.

2

The Essays

This volume includes ten essays that were delivered at the second BHLaP workshop, held at the Hebrew University of Jerusalem on June 26–29, 2018. There is also an eleventh essay that was written for this volume by an attendee of the workshopm, Galia Hatav, who had been investigating linguistic issues treated at the workshop. Notably, all but one of these essays reflect a generative linguistic approach to the study of language, while at the same time making clear connections to the history of Biblical Hebrew grammatical study. Beyond the insights afforded by each study, the overwhelmingly generative orientation of this volume makes it unique within contemporary Biblical Hebrew studies. In the first essay, Roman Himmelreich and Outi Bat-El Foux analyze the allomorphy between pausal (phrase final) and contextual (phrase medial) forms in Tiberian Hebrew, manifested by alternation in stress and vowel reduction. The challenging aspect of this allomorphy is that vowels in stressed syllables do not resist reduction, contrary to universal typology. They argue that stress and vowel reduction in Tiberian Hebrew are not directly related, and propose a multi-planar metrical structure—a separate metrical plane for each phenomenon. Foot assignment in the two planes is identical (right-aligned

1 See the brief statement at the following site: https://bhlapworkshop2018.wordpress.com/ about‑bhlap‑network/.

- 978-90-04-44885-8

linguistic theory and philology in the study of biblical hebrew

5

trochaic foot), but mora assignment is minimally distinct. This minimal contrast is phonetically and phonologically supported and fits within universal typology. In their essay on prosodic dependency, Vincent DeCaen and B. Elan Dresher investigate why there seem to be two distinct definitions of prosodic dependency in Tiberian Hebrew. One prosodic dependency is morphosyntactic: the distinction of bound (construct) nouns from their free (absolute) forms. The other prosodic dependency is musical: the distinction between accented free forms and unaccented bound forms, connected to the clitic hosts by a ‘hyphen’ (the Hebrew maqqef ). While there is great overlap between these two systems, which is expected, DeCaen and Dresher focus on the unexpected divergences, on the cases of bound forms that are accented and free forms that are deprived of an accent, as well as a whole class of monosyllabic nouns that behave with prosodic variability, such as leḇ ‘heart’. Susan Rothstein (z”l) and Adina Moshavi’s study of ordinal numerals marks the volume’s turn to primarily syntactic phenomena. As in many languages, Biblical Hebrew has specific adjectives that function as ordinal numerals (e.g., first, second, third) to express the property of having a certain position in a (partial) order relation, in contrast to cardinal numerals (e.g., one, two, three), which appear to be nouns and serve to count pluralities. While ordinals and cardinals would seem to have non-overlapping semantic and syntactic uses, Rothstein and Moshavi note and examine the unexpected use of cardinal numerals to express express ordinality in Biblical Hebrew temporal expressions. Their study describes the data and offers a formal semantic analysis for the ordinal effect of the biblical constructions. Robert Holmstedt’s study of ellipsis in Biblical Hebrew moves the discussion of the phenomenon beyond poetry to its use in prose. It has long been observed that ellipsis, especially verb-gapping, is frequent in Biblical Hebrew poetry. Holmstedt surveys the work done especially in the last fifteen years as well as the kinds of ellipsis that occur in biblical poetry before moving to demonstrate that most kinds also occur in non-poetic contexts. Moreover, Holmstedt investigates whether the biblical ellipsis data allow for the discernment of subtleties such as sloppy identity readings and locality constraints. The next three essays in the volume examine the same construction—the infinitive, especially the so-called infinitive absolute (ia)—that has long defied an analysis accounting for all the data. Elizabeth Cowper and Vincent DeCaen argue that the ia is a morphologically underspecified form, or the default spellout of a verb without any inflectional features, and that synctically it is a verbal projection (vP or VoiceP) that is adjoined to another verbal projection. They are thus able to explain the position of the ia, which may appear before or after

- 978-90-04-44885-8

6

holmstedt

the associated finite verb, as the in situ adjunction position (when it follows the finite verb) or its raising to spec,TopP (when it precedes the finite verb). Galia Hatav obverses that that the ia is able to take complements, but not modifiers, which leads her to set aside the label “infinitive” as inaccurate and instead analyze the ia as the root phrase of a verb that represents or names the eventuality depicted by the respective root. And Edit Doron (z”l) argues that what have traditionally been takes as two distinct types of infinitives in Biblical Hebrew are actually a single infinitve with free and bound variants. Doron takes the free form as either a bare infinitive that is not clausal (i.e., the oft-called adverbial infinitive) or a [-indicative] verb of a root clause with imperative force. The bound form has two manifestations, the pro-inf that has no temporal specification and the Poss-inf that is primarily temporal. Importantly, Doron takes the infinitive in any of its contexts as verbal, in contrast to some traditional analyses that take the infinitive construct as essential nominal. Todd Snider’s essay moves beyond infinitives to consider a particular construction in which two finite verbs, taken together, jointly describe a single event, e.g., ‘he rose and went’. He considers four analyses: (a) both verbs are main verbs, (b) both verbs jointly make up a serial verb construction, (c) the first verb is an auxiliary verb, and (d) the first verb is a light verb. Snider concludes that the data are best explained by a light verb analysis, in which the first verb is somewhat delexicalized and so contributes the aspectual meaning of immediacy and the non-aspectual meaning of volitionality while the second verb remains fully lexical. Jacques Boulet’s essay introduces the concept of secondary predication to Biblical Hebrew studies; he defines a secondary predicate as any non-finite predicate that occurs under the scope of a primary predication of any kind, whether by adjunction or as a complement, e.g., ‘I went out full’. For this study he considers a subset of secondary predicates in which the secondary predicate shares its subject argument with an argument of the primary predication, either the subject or the object, e.g., ‘God created them male and female’ (< God created them (and) they be male and female). Boulet finds five kinds of argument sharing secondary predicate in Biblical Hebrew: depictives, circumstantials, resultatives, explicit creation productives, and capacitives. Kevin Grasso re-examines the Hiphil binyan, typically understood to be a derivational strategy of verb formation that carries causative semantics. Focusing on the alternation in the Hiphil between inchoative (‘the food did not stink [Hiphil]’) and causative (‘you made us stink [Hiphil]’), Grasso argues that a better characterization of the Hiphil than causative is that it is resultative. Moreover, this resultative contribution may concern events or states.

- 978-90-04-44885-8

linguistic theory and philology in the study of biblical hebrew

7

Tamar Zewi’s essay (the lone non-generative study) considers constructions that both defy classification in the typical syntactic categories and show properties of more than one syntactic category. She examines five examples: the genitive šɛl (i.e., the relative š attached to the preposition l), ‘inverted adjectival annexion’ (e.g., ‘he is weak of hands’ < he is weak and his hands are weak), imperfect extrapositon (where she uses extraposition as a form of constituent fronting), the object marker ʾɛṯ on a subject noun, and the use of the third person pronoun between the subject and predicate of a verbless clause. Zewi suggests that these kinds of constructions show syntactic innovation and often fluctuate among more than one syntactic pattern as Hebrew developed.

References Barr, James. 1969. The Ancient Semitic Languages—The Conflict between Philology and Linguistics. Pp. 37–55 in Transactions of the Philological Society 1968. Oxford: Basil Blackwell. Bajohr, Hannes, Benjamin Dorvel, Vincent Hessling, Tabea Weitz. 2014. Introduction. Pp. 1–26 in The Future of Philology: Proceedings of the 11th Annual Columbia University German Graduate Student Conference, ed. H. Bajohr, B. Dorvel, V. Hessling and T. Weitz. Cambridge: Cambridge Scholars Press. Bodine, Walter R. 1987. Linguistics and Philology in the Study of Ancient Near Eastern Languages. Pp. 39–54 in “Working with No Data”: Semitic and Egyptian Studies Presented to Thomas O. Lambdin; ed. D.M. Golomb. Winona Lake, IN: Eisenbrauns. Gleason, H.A. 1974. Linguistics and Philology. Pp. 199–212 in On Language, Culture, and Religion: In Honor of Eugene A. Nida, ed. M. Black and W.A. Smalley, Jr.; The Hague: Mouton. Hendel, Ronald S. 2016. Steps to a New Edition of the Hebrew Bible. Text Critical Studies 10. Atlanta, GA: sbl Press. Koerner, Konrad. 1997. Linguistics vs Philology: Self-Definition of a Field or Rhetorical Stance? Language Sciences 19 (2): 167–175. Najman, Hindy. 2017. Ethical Reading: The Transformation of the Text and the Self. The Journal of Theological Studies 68 (2): 507–529. O’Connor, Michael P. 2002. Discourse Linguistics and the Study of Biblical Hebrew. Pp. 17–41 in Congress Volume Basel 2001, ed. André Lemaire. Leiden: Brill. Pollock, Sheldon. 2009. Future Philology? The Fate of a Soft Science in a Hard World. Critical Inquiry 35 (4): 931–961. Pollock, Sheldon. 2015. Introduction. Pp. 1–24 in World Philology, ed. S. Pollock, B.A. Elman and K.K. Chang. Cambridge, MA: Harvard University Press. Sampson, Geoffrey. 1980. Schools of Linguistics. Stanford: Stanford University Press.

- 978-90-04-44885-8

8

holmstedt

Sanders, C. 2006. Saussurean Tradition in the 20th-Century Linguistics. Pp. 769–774 in Encyclopedia of Language and Linguistics, 2nd edition, volume 10, ed. Keith Brown. Amsterdam: Elsevier. Téné, David, Aharon Maman, and James Barr 2007. Linguistic Literature, Hebrew. Pp. 29–61 in Encyclopaedia Judaica, Second Edition, volume 13. Detroit: Macmillan. Turner, James 2014. Philology: The Forgotten Origins of the Modern Humanities. Princeton: Princeton University Press. Zwiep, I. 2006. Hebrew Linguistic Tradition. Pp. 256–259 in Encyclopedia of Language and Linguistics, 2nd edition, volume 5, ed. Keith Brown. Amsterdam: Elsevier.

- 978-90-04-44885-8

chapter 2

Pausal vs. Context Forms in Tiberian Hebrew: A Multi-Planar Analysis of Vowel Reduction and Stress Roman Himmelreich and Outi Bat-El Foux

1

Introduction

Tiberian Hebrew exhibits positional allomorphy, whereby a word has different surface structures in different positions: the pausal form in phrase final position and the contextual form elsewhere (Revell 1981, 2012; Goerwitz 1993; Dresher 2009). As shown in (1), the two allomorphs may appear within the same phrase. (1) Pausal and context forms within the same phrase1 ʔɛθ-ˈzɛ toχəˈlu miˈkkol ʔaˈʃɛr bamˈmɔjim kol ʔaʃɛr-ˈlo sənaˈpir this eat 2pl from-all that in-the-water all that-has fin wəqasˈqɛsɛθ toˈχelu and-scale eat 2pl ‘These you shall eat of all that is in the waters: all that have fins and scales you shall eat’ (Deut 14:9) There are two types of alternation between the allomorphs: either in vowel quality alone (2a), or in vowel quality and stress (2b).2

1 The distribution of pausal forms is conditioned by the verse structure of the Biblical text, which is denoted by an elaborate system of cantillation marks (teʾamim). Although pausal forms do not co-occur with any specific cantillation mark, their appearance is nevertheless largely predictable on the basis of the cantillation system reflecting the underlying prosodic structure of the text (Dresher 1994; Churchyard 1999). In the majority of cases, pausal forms co-occur with the major disjunctive cantillation marks silluq and atnah, which mark the verse’s main subdivisions (DeCaen 2005). 2 Vowel alternation is oblivious to morphological structure, thus applying also within a suffix (e.g. [jirɔˈʃ-ɛχɔ]P—[jirɔʃ-əˈχɔ]C ‘inherit 3msg you’).

© Roman Himmelreich and Outi Bat-El Foux, 2021 | doi:10.1163/9789004448858_003

- 978-90-04-44885-8

10

himmelreich and bat-el foux

(2) Pausal vs. context forms a. kɔ.ˈtɔv kɔ.ˈtav ʔɔ.ˈmɔr.tɔ ʔɔ.ˈmar.tɔ b. ʃɔ.ˈmɔ.r-u ʃɔ.mə.ˈr-u ˈle.x-u lə.ˈx-u

‘wrote 3msg’ ‘said 2ms’ ‘guarded 3mp’ ‘go! 2mp’

The two types of alternation result from vowel reduction. In (2a), the output of reduction is [a] and stress persists in its position; in (2b) the output of reduction is schwa [ə] and stress shifts to the final syllable. That is, the reduced vowel in the contextual form is either stressed (2a) or corresponding to a stressed vowel (2b). Such interaction between stress and vowel reduction is typologically odd and theoretically problematic; stressed syllables are prosodically strong and vowels in strong positions tend to resist alternation, let alone reduction (Beckman 1997; Crosswhite 2001, 2004; Padgett and Tabain 2005; Barnes 2006). In this paper we solve this discrepancy with biplanar metrical structure, allocating one plane for stress assignment and another for vowel reduction. Crucially, the two planes do not interact directly, thus allowing vowel reduction to apply in a weak position in its own metrical structure. The difference between the two metrical planes stems from different schemes of syllable weight. The stress plane is sensitive to syllable structure, assigning extra weight to closed syllables; thus, the syllables that are heavy for the purpose of stress are cvc. The vowel reduction plane is sensitive to phonetic duration, assigning extra weight to vowels in domain final positions; thus, the syllables that are heavy for the purpose of vowel reduction are those which feature a vowel which is lengthened due to its final position in the word or the phrase. (3) Syllable weight in the two metrical planes Syllable structure Stress assignment

cv cvc

Light Heavy

Vowel reduction Word medial Word final Light

Heavy Light

The paper’s roadmap is as follows: We start with generalizations regarding the pause-context alternation (§2), and then introduce the theoretical problem addressed here: while cross-linguistically vowel reduction targets vowels in

- 978-90-04-44885-8

11

pausal vs. context forms in tiberian hebrew

unstressed syllables, in Tiberian Hebrew vowels in stressed syllables undergo reduction (§3). The path towards resolving this problem goes through the concept of phenomenon-specific syllable weight (§ 3.2) which is extended to the realm of vowel reduction (§3.3). Detailed expositions of the derivation of contextual and pausal forms are presented in §4. Finally, the conclusion is given in § 5.

2

Pause-Context Alternation: Data and Generalizations

We start this section with two basic assumptions, one regarding the Tiberian Hebrew vowel system and the other addressing the base of the derivation. There is an ongoing debate in the literature regarding the vowel system in Tiberian Hebrew, whether it is a 5-vowel system with length contrast, or a 7vowel system with quality contrast only (Khan 1987; Churchyard 1999; Anstey 2005). Here we adopt Khan’s (1987) 7-vowel system, attributing vowel length contrast to a late phonetic effect. Tiberian Hebrew vowel system thus includes one low vowel [a], two high vowels [i, u] and four mid vowels; two tense [e, o] and two lax [ɛ, ɔ]. The schwa, although transcribed here as [ə] is phonetically a low vowel (Khan 2013) and thus the output of vowel reduction in both (2a) and (2b) is phonetically [a].3 We will, however, maintain the [ə] symbol, which reflects the phonological behavior of the vowel. The second assumption concerns the base of derivation. Although pausal forms exhibit more vowel contrasts and are often identical to the base of derivation, pentasyllablic words (4) show that both pausal and contextual forms undergo reduction. Therefore, we assume that the base of derivation is an abstract form comprised of the non-reduced vowels in both allomorphs. (4) Pausal forms missing segmental information w.r.t to Contextual forms Pause

Context

Base

lə.hɔ.rə.ˈʁɛ.χɔ la.hă.rɔ.ʁə.ˈχɔ lə.hɔ.rɔ.ʁɛ.χɔ ‘to kill you’ lə.χal.kə.ˈlɛ.χɔ lə.χal.kɛ.lə.ˈχɔ lə.χal.kɛ.lɛ.χɔ ‘to support you’

3 The Shewa diacritic ‫ ְא‬in Tiberian Hebrew correlates with two phonological realizations: either a syllable coda (silent Shewa) or a reduced vowel (vocalic Shewa). For example, in ‫[ ִתְּשְׁמרוּ‬tiʃ.mə.ˈr-u] ‘keep 3mpl’, the first Shewa diacritic marks the /∫/ in the coda and the other indicates the reduced vowel [ə] following /m/. We employ the term ‘schwa’ to refer to the vocalic Shewa.

- 978-90-04-44885-8

12

himmelreich and bat-el foux

With these assumptions in hand, the rest of this section displays the pause– context alternation in cv (§2.1) and cvc (§2.2) syllables; as shown, the structure of syllable is a crucial factor in determining the nature of the alternation. 2.1 Open Syllables (cv) The vowel in a penultimate cv syllable alternates between a full mid vowel in the pausal allomorph, and a schwa in the contextual allomorph. As shown in (5a), the mid vowels [o, ɔ, e, ɛ]4 in the pausal allomorph correspond to [ə] in the contextual forms; the high vowels [i, u] resist alternation (5b). (5) Alternation in open syllables (cv) Pause

Context

a. tiʃ.ˈmo.r-u hɔ.ˈjɔ.θ-ɔ to.ˈχe.l-u b. jɔ.ˈmu.θ-u jag.ˈgi.ð-u

tiʃ.mə.ˈr-u hɔ.jə.ˈθ-ɔ to.χə.ˈl-u jɔ.ˈmu.θ-u jag.ˈgi.ð-u

Alternation ‘keep 2mpl’ ‘was 3fsg’ ‘eat 2pl’ ‘die 3msg’ ‘say 3msg’

o ~ ə and stress ɔ ~ ə and stress e ~ ə and stress – –

There is a clear correlation between vowel alternation and stress alternation, i.e. stress shift; neither apply in (5b) while both apply in (5a). This correlation is due to the prohibition on a stressed schwa in Tiberian Hebrew (GeseniusKautzsch-Cowley 2006; Prince 1975), as in many other languages (Flemming 2009; Becker-Kristal 2010; Gordon 2017). Thus, where a vowel is reduced to schwa stress must shift, and since stress in Tiberian Hebrew can be either final or penultimate, stress shifts to the final syllable.5 The patterns of vowel quality alternation in (5b) mirror prototypical patterns of vowel reduction in the world’s languages. First, mid vowels are the most commonly attested targets of vowel reduction due to their low contrastivity and distinctiveness in compare to corner vowels [i, a, u]; see Dispersion Theory (Lindblom 1963; Padgett and Tabain 2005). Second, schwa is 4 The vowel /ɛ/ is included in this list on the basis of the alternation found in the possessive/accusative suffix /ɛχɔ/, as in [jirɔˈʃ-ɛχɔ]P – [jirɔʃ-əˈχɔ]C ‘inherit 3msg you’. 5 Antepenultimate stress is possible in the exceptional case where phrasal stress clash (nesiga) prohibits a final stressed syllable and the penultimate syllable cannot bear stress because it contains a schwa; e.g. /hɔrɔʁ-u ʔiʃ/ ⇒ [ˈhɔ.rə.ʁu ˈʔiʃ] ‘killed 3pl a man’ (Dresher 2009).

- 978-90-04-44885-8

13

pausal vs. context forms in tiberian hebrew

the most common output of vowel reduction cross-linguistically (Crosswhite 2001; Barnes 2006), structurally represented as an empty V-slot (Anderson 1982; Clements and Kayser 1983). In Tiberian Hebrew, this featureless representation is supported by the hataf vowels which acquire their features from neighboring consonants or vowels. For example, in [ʔă.vaq.ˈqeʃ] ‫‘ ֲאַבֵקּש‬ask 1sg’ the vowel [ă] is an empty V-slot, i.e. a schwa, that gets its features from the glottal stop. 2.2 Closed Syllables (cvc) Unlike in cv syllables, where the reduced vowel in context forms surfaces as schwa [e], in cvc syllables the output of reduction is [a]. (6) Alternation in closed syllables (cvc) Pause a. jo.ˈχel hɔ.ˈrɔʁ b. ʔɔ.ˈmɔr.-ti mɔ.ˈrɔð.-nu c. jɔ.ˈmuθ jag.ˈgið

Context jo.ˈχal hɔ.ˈraʁ ʔɔ.ˈmar.-ti mɔ.ˈrað.-nu jɔ.ˈmuθ jag.ˈgið

Alternation ‘eat 3msg’ ‘killed 3msg’ ‘said 1msg’ ‘rebelled 1pl’ ‘die 3msg’ ‘say 3msg’

e~a ɔ~a ɔ~a ɔ~a – –

As in cv syllables, the high vowels resist alternation (6c), and the mid vowels [ɔ] and [e], whether in the final (6a) or the penultimate (6b) syllable, alternate with [a]. Thus, contextual allomorphs exhibit less structural complexity, and it is in this sense that they can be considered reduced (Bosch and Wiltshire 1993).6 Similar pattern of vowel reduction is found in Belarusian (7), where high vowels do not alternate, and mid vowels alternate with [a].

6 Unlike in cv syllables, the alternation in cvc syllables is not exception-free; that is, mid vowels do not always alternate in cvc syllables and consequently the pausal and contextual forms can be identical, e.g. [tir.ˈdof]C/P ‘chase 2msg’, [ʔă.vaq.ˈqeʃ]C/P ‘ask 1sg’. Such exceptions are commonly attributed to diachronic phenomena (Qimron 1986, 2006; Khan 1994).

- 978-90-04-44885-8

14

himmelreich and bat-el foux

(7) Vowel reduction in Belarusian (data from Crosswhite 2004) Stressed

Unstressed

a. ˈno.ɣi ‘legs’ ˈre.ki

‘rivers’

b. ˈru.ki ‘hands’ ˈspi.nɨ ‘backs’

na.ˈɣa naʐ.ˈnoj ra.ˈka ratʂ.ˈnoj ru.ˈka spi.ˈna

Alternation

‘leg’ ‘of legs adj’ ‘river’ ‘of rivers adj’ ‘hand’ ‘back’

o~a o~a e~a e~a – –

The same reduction pattern is also attested in certain southern dialects of Russian (Crosswhite 2000) and in Shimakonde (Barnes 2006).

3

A Conflict in Prominence

The following table presents the prosodic positions where pause–context alternation occurs. The generalization is that the position where the two allomorphs differ is the stressed syllable of the pausal allomorph. However, this position varies as a function of word and syllable structure. (8) Positions of vowel reduction in pause-context allomorphy

a. b. c. d.

Position

Structure Pause

Penultimate Penultimate Final Final

cvc cv cvc cv

Context

ʔɔ.ˈmɔr.tɔ ʔɔ.ˈmar.tɔ ‘said 2msg’ ʃɔ.ˈmɔ.ru ʃɔ.mə.ˈru ‘kept 3mpl’ kɔ.ˈθɔv kɔ.ˈθav ‘wrote 2msg’ not attested

The data in (8) exhibit a typologically a-typical reduction pattern and thus poses a theoretical problem. Cross-linguistically, vowel reduction tends to affect unstressed syllables, as stressed syllables are prosodically strong positions, thus exempt from reduction (Beckman 1997; Crosswhite 2004; Barnes 2006). Here, contrary to cross-linguistic tendencies, vowels in stressed syllable are reduced. However, it is not the case that stress is isomorphic with prosodic strength, nor that vowel reduction always targets unstressed syllables. Indeed, vowel

- 978-90-04-44885-8

pausal vs. context forms in tiberian hebrew

15

reduction does not affect strong positions, but stressed syllables are only one of several strong positions in the word (Bosch and Wiltshire 1993; Bosch 1996; Barnes 2006). Thus, vowel reduction in Tiberian Hebrew could as well be independent of stress, as it is the case in French and Northern Welsh. French lacks word-level stress, but it does exhibit alternating pattern of vowel reduction in casual speech, attributed to a phrase-level foot structure (Garcia et al. 2017). In Northern Welsh, as in Biblical Hebrew, vowel reduction applies in stressed syllables (9). All vowels, except the final one, are reduced to [ə], including the vowels in the stressed syllables, which are usually penultimate (Hannahs 2007). (9) Reduction of stressed vowels in Northern Welsh (data from Ball and Williams 2001) Stem forms

Suffixed forms

ˈkuχ ‘boat’ ˈbrɨn ‘hill’ ˈmə.nið ‘mountain’

ˈkə.χɔd ‘boats’ ˈbrə.nja ‘hills’ mə.ˈnə.ðɔɨð ‘mountains’

3.1 Multiple Prominent Positions On the phonetic level, stress is not a homogenous phenomenon; its phonetic correlates are typically pitch contour, increased intensity, and/or prolonged duration (Hayes 1995). Different languages employ varying subsets of these acoustic characterizations to mark stress (Gordon and Roettger 2017). For example, Modern Hebrew marks stress mainly by duration (Silber-Varod et al. 2016, Cohen et al. 2018) while Welsh marks it by intensity. However, these same languages employ the other phonetic correlates of stress in different positions: in Modern Hebrew a peak of high pitch appears in the first pre-tonic syllable (Becker 2003), and in Welsh duration and pitch rise are associated with the word-final syllable (Ball and Williams 2001; Hannahs 2013). Bosch (1996) proposes that prosodic words in Northern Welsh have two distinct prominent positions, one relevant to stress assignment and another determining the position of vowel reduction. The prominent position for stress assignment is determined by the metrical system—a right-aligned trochaic foot. The prominent position for vowel reduction is determined by the phonetic properties of the syllable—the final syllable is the longest in terms of duration (Ball and Williams 2001), thus resisting vowel reduction; all other syllables are reduced. Bosch’s analysis is in line with recent studies on conflicting syllable weight criteria and prosodic prominence within the same language (Ryan

- 978-90-04-44885-8

16

himmelreich and bat-el foux

2016, 2019). In fact, it has been shown that such apparent discrepancies are typologically not uncommon (Gordon 2006). So, if languages employ distinct syllable weight criteria for stress assignment vs. tone licensing, as shown for Lhasa Tibetan (see §3.2), why not for stress assignment vs. vowel reduction? Dismantling stress to its phonetic correlates does not impeach its phonological prominence. However, it does show that on the phonetic level, there are multiple positions of prominence which correspond to different phonetic phenomena. These phonetically prominent positions coincide in some cases, while in others they do not. Therefore, given a phonological phenomenon other than stress, such as vowel reduction, which correlates with phonetic vowel duration—the relevant position(s) of prominence may be different to the ones which are relevant for the purposes of stress. Earlier proposals for the resolution of the conflict between stress and vowel reduction in Tiberian Hebrew have invoked multi-planar metrical structure. One such proposal includes an independent plane of vowel reduction feet (vrfeet) in parallel to stress feet, where vr-feet are trochaic and stress feet are iambic (Rappaport 1984). The superimposition of the two metrical planes, illustrated in (10), achieves the sought duality, where a single syllable occupies a strong position for stress and a weak position for reduction. Notice that this analysis assumes a 5-vowel quantity-sensitive vowel inventory, while in this study we assume a 7-vowel inventory. (10) Analysis: Context forms (Rappaport 1984) a. ʃɔː.mə.ˈru: ‘guarded 3mpl’ b. kɔː.ˈθav ‘wrote 3msg’ Stress plane Underlying base Reduction plane Derived form Surface form

[* _] ʃaː . ma . ru [*][_ *] ʃaː . mV . rú ʃɔː . mV . rúː

[*] kaː . tav [*][*] kaː . táv kɔː . θáv

The penultimate syllable in (10a) exemplifies the main idea of a multi-planar metrical system, with the co-occurrence of prosodic strength on the stress plane (marked with *) and prosodic weakness on the vowel reduction plane (marked with _). Since the vowel is reduced, stress must shift to the final syllable because a reduced vowel cannot be stressed in Tiberian Hebrew. Additional rules are applied to derive the surface representation, including wordfinal vowel lengthening, rounding of [aː] to [ɔː], and post-vocalic spirantization

- 978-90-04-44885-8

pausal vs. context forms in tiberian hebrew

17

(Rappaport 1984). The surface form [ʃɔː.mV.rúː] contains the empty V-slot for the reduced vowel; the surface segmental content (or lack thereof) may be further conditioned by the adjacency of guttural consonants or set by default (see § 2.1). In the case of (10b), there is no environment for vowel reduction to occur, i.e. no weak position. Rappaport’s (1984) vr-feet are not compatible with the 7-vowel system assumed in the present study, where the underlying base contains the full set of vowel qualities exhibited by both the pausal and the contextual allomorphs (see §2). In a 7-vowel system, the underlying base of (10b) is not /kaːtav/ but rather /kɔtɔv/. With /kɔtɔv/ as the underlying base, the derivation /ɔ/ ⇒ [a] in the final syllable of the contextual form [kɔ.ˈtav] is not expected with vr-feet. If vr-feet are quantity-sensitive iambs, i.e. treating the final syllable [tɔv] as heavy, reduction is not expected to apply. Conversely, if vr-feet are quantityinsensitive iambs, i.e. treating [tɔv] as light, reduction of the penultimate syllable [kɔ] is expected, but not attested. This problem may lead one to the conclusion that the 7-vowel system for Tiberian Hebrew assumed in the present study is simply wrong. However, Dresher (2009) points out that vr-feet cannot fully account for the data, even with a 5-vowel quantity-sensitive system. The admission of pausal lengthening into the account predicts that all stressed vowels in pausal forms will surface as long. Per contra, pausal forms affixed with the accusative clitic [-eχɔː], such as [lə.hɔː.rə.ˈʁe.χɔː] ‘to kill inf you’, feature a short [e] in the stressed penultima. This is a case where the rule of pausal lengthening predicts an incorrect form. vr-feet also predict that this penultimate [e] should be reduced, as per the iambic reduction foot at the prosodic word’s right edge: lə.hɔː.rə.[ˈʁe.χɔː]f. Had vowel reduction taken place, the expected output would be a schwa in the penultimate syllable—in effect, yielding the contextual form of this word. Thus, neither pausal lengthening nor vowel reduction apply to the penultimate [e] in this case, and so the case of affixed forms with [-eχɔː] cannot be accounted for under vr-feet. Alternatively, under the assumption of a 7-vowel system, the underlying quality of the penultimate vowel is /ɛ/. The surface form features a straightforward preservation of the underlying quality. Thus, no derivation process is required to account for /-ɛχɔ/ affixed words under the 7-vowel system. According to our analysis, vowel reduction does not apply to this vowel due to the effects of phrase final lengthening, which renders this vowel immune to vowel reduction (see §3.3). Rappaport’s (1984) analysis with multiple metrical planes has been reinforced in recent studies of phenomenon-specific prominence (Gordon 2006; Ryan 2019). The present study adopts and follows the general concept of this

- 978-90-04-44885-8

18

himmelreich and bat-el foux

analysis, but proposes a novel architecture to the vowel reduction plane—one which is tightly grounded upon the cross-linguistically attested phenomenon of final lengthening (see §3.4). 3.2 Phenomenon-Specific Weight Criteria Cross-linguistic typology shows numerous cases of phenomenon-specific weight criteria, whereby languages employ different syllable weight schemes as a function of the phenomenon at hand. For example, Lhasa Tibetan exhibits three different syllable weight schemes corresponding to three phenomena: stress, compensatory lengthening, and contour tone (Dawson 1980; Gordon 2006). cvc syllables are light for the purposes of stress assignment, but heavy for the purposes of compensatory lengthening. As for contour tones, cvr syllables (where R stand for a sonorant) are heavy but cvo syllables (where O stands for an obstruent) are light. (11) Phenomenon specific prominence in Lhasa Tibetan (data from Dawson 1980) Stress

Contour tone

Compensatory lengthening

a. ˈtýː.tṹː ‘shirt’ lɔ̂ ː ‘electricity’ tsík ~ tsîː ‘one’ b. khá.ˈpáː ‘school gen’ kâː ‘stop’ kə̀ pkí ~ kə̀ ːki ‘will do’ c. láp.ˈʈéː ‘telephone’ khâm ‘Kham’ tʃúrkú ~ tʃúːkú ‘nineteen’

In the stress system, cvv syllables are heavy and cvc are light; in (11a) the initial cvv syllable is stressed but in (11c) the initial cvc is not stressed. In the tone system, both cvv and cvr are heavy as they license contour tone; there are no cvc syllables with contour tone where the coda is an obstruent. Finally, for compensatory lengthening, all cvc are heavy since any deleted consonant coda position is compensated via lengthening of the preceding vowel. Gordon (2006) provides a detailed analysis grounding this behavior in the phonetic manifestation of the different phenomena. In brief, the realization of a contour tone requires a long sequence of sonority, where a sonorous segment is a vowel or a sonorant consonant; therefore, cvr is heavy like cvv. Compensatory lengthening, on the other hand, is sensitive to prosodic structure and applies whenever a deleted consonant leaves an empty coda position. Stress as a metrical system is different from both, giving no weight priority to coda consonants. Thus, Lhasa Tibetan exhibits three different syllable weight schemes for different phonological phenomena.

- 978-90-04-44885-8

pausal vs. context forms in tiberian hebrew

19

A similar case is reported for Early and Classical Greek, where different weight criteria are required for the pitch accent system as opposed to the systems of stress, poetic meter and minimal root requirement (Steriade 1991). The solution proposed in recent literature is to redefine weight distinction as phenomenon-driven rather than language-driven (Gordon 2006). In such a system, different phenomena can utilize distinct metrical planes within the same language, where each phenomenon determines its prominent (and weak) positions. The proposed system is restricted in the sense that all syllable weight schemes must adhere to the universal scale of syllable weight (Hyman 1984, Hayes 1989): cvv> cvr> cvc> cv. However, different schemes may place the borderline between light and heavy syllables in different positions along this universal scale. The different syllable weight schemes employed in Lhasa Tibetan are as follows (grayed cells represent bimoraic syllable types). (12) Phenomenon specific syllable weight schemes in Lhasa Tibetan Stress cvv σµµ > cvr > cvo > cv Tone cvv σµµ > cvr σµµ > cvo > cv Compensatory Lengthening cvv σµµ > cvr σµµ > cvo σµµ > cv

Under such an analysis, each phenomenon adheres to a strict dichotomy of light vs. heavy, where heavy syllables are bimoraic. 3.3 Syllable Weight Criteria for Vowel Reduction Introducing new phenomenon-specific weight criteria must be rigorously restricted because it is a powerful theoretical device that may easily over-generate; therefore, any phenomenon-specific syllable weight scheme must be well-supported. Following typological surveys (Gordon 2006; Ryan 2019), the following phenomena may have independent syllable weight schemes: stress, tone, minimal word restrictions, compensatory lengthening, syllabic template and poetic meter. Previous studies propose three elements that are required to support a phenomenon-specific weight scheme (Gordon 2006; Ryan 2016, 2019): i. A phonological phenomenon with a binary contrast; ii. The phonetic manifestation of the phenomenon; and iii. Typological evidence supporting the correlation between (i) and (ii). The following exposition uses the case of contour tone licensing in Lhasa Tibetan as an example. On the phonetic level, contour tone is implemented by modulation of pitch (raising then falling) over a continuous sonorous sig-

- 978-90-04-44885-8

20

himmelreich and bat-el foux

nal. In order to achieve a perceptible contrast between the two parts, “sufficient” duration of the sonorous signal is required. Notice that in phonetic terms, “sufficiency” is determined by assuming some critical threshold over the temporal dimension. However, from a phonological point of view, licensing is binary—contour tone is either licensed or not. In Lhasa Tibetan, only cvv and cvr syllables boast such “sufficiently” long sonorous sequences. Thus, the relation between the phonetic implementation and the phonological contrast is established. Within the moraic theory, such binary contrasts are represented by using different number of morae (Hyman 1984; Hayes 1989). By definition, morae encode the number of timing positions as weight distinctions (ibid.), making them a fitting means of representation when the relevant categorical contrast is determined by the duration of the underlying phonetic signal. Therefore, for the purposes of contour tone licensing in Lhasa Tibetan, cvv and cvr syllables are considered bimoraic, while cvc and cv are monomoraic. Thus, the three required elements for postulating a proprietary weight scheme for contour tone are as follows: i. Categorical phenomenon: Licensing of contour tone ii. Phonetic manifestation: Pitch modulation over “sufficient” sonorous sequence iii. Typological evidence: Lhasa Tibetan and other languages (see survey in Gordon 2006) Now, we turn to the argument supporting a phenomenon-specific weight scheme for vowel reduction. On the phonetic level, the main correlate of vowel reduction is vowel duration (Lindblom 1963; Moon and Lindblom 1994; Flemming 2005). Like in the case of contour tone, some “sufficient” phonetic duration is required for the accurate production and perception of vowel quality. Vowel reduction starts manifesting when duration falls below this “sufficient” threshold. This effect is best seen in the cases of gradient reduction systems such as non-first-pretonic syllables in Russian, where vowel quality is gradually altered in direct correlation with phonetic duration (Crosswhite 2000; Barnes 2007). Conversely, vowels with durations longer than the aforementioned “sufficient” threshold suffer no quality degradation. Therefore, vowels having durations above “sufficient” can be considered resistant to vowel reduction, while vowels with durations below “sufficient” are reducible. The exact phonetic duration that comprises “sufficiency” is language specific (Barnes 2006) and also vowel specific, as some vowels are inherently longer than others (Becker-Kristal 2010). Therefore, we intentionally leave the term “sufficient” unspecified, as it is not strictly relevant for purposes of the argument. Rephrasing the reducible vs. non-reducible contrast to fit the terminology of the moraic theory results in a binary contrast between reduction-resistant vowels which are bimoraic and reducible vowels are monomoraic. - 978-90-04-44885-8

pausal vs. context forms in tiberian hebrew

21

The actual application of vowel reduction is further conditioned by metrical structure, which is again, language specific. Just as different languages exhibit different stress patterns, be they rhythmically alternating or not, left or right aligned, trochaic or iambic, the positions where vowel reduction applies are too determined by the parsing and grouping of morae attributed to vowels or syllables. Therefore, a monomoraic vowel is not necessarily reduced, it is just reducible. In this study we propose a new syllable weight scheme specific to vowel reduction, whereby moraic structure is assigned as a function of the vowel’s phonetic duration: a short vowel is mono-moraic and a long vowel is bimoraic. This contrast distinguishes between reducible and non-reducible vowels respectively. Crucially, coda consonants do not contribute weight for the purpose of vowel reduction. Finalizing the argument, here are the three elements required to support the existence of a proprietary vowel reduction weight scheme: i. Categorical phenomenon: The reducibility vs. reduction-resistance of a vowel. ii. Phonetic manifestation: Vowel production over “sufficient” phonetic duration. iii. Typological evidence: Non reducible vowels under stress or final lengthening in various languages (see survey in Barnes 2006). This proposal finds both theoretical support and parallels in the literature. In general, the notion of bimoraic non-reducible vowels is a case of inalterability (Hayes 1986). Reformulated in the terminology of moraic and prosodic theories, it follows the same line of thought regarding the inalterability of segments which are associated with multiple prosodic slots, such as Tiberian Hebrew geminate stops, which do not undergo spirantization due to the one-to-many representation, one segment to two prosodic positions. With regard to the phenomenon of vowel reduction, the current proposal is very similar to Bosch’s (1996) notion of phonetic-level vs. word-level prominence types. In both proposals, phonetic duration determines prominence, which in turn conditions the applicability of vowel reduction. In parallel, a separate prominence level, called “word-level” in her terms, determines the position of word stress. In our proposal, “word-level” prominence is simply called stress. 3.4 Final Lengthening Commonly attested vowel reduction systems target prosodically weak positions. The most widespread case being the licensing of a large vowel inventory in stressed syllables as opposed to a small vowel inventory in unstressed syllables. In such languages, stress correlates with increased phonetic duration

- 978-90-04-44885-8

22

himmelreich and bat-el foux

of the stressed vowel (Hayes 1995; Gordon and Roettger 2017). In other words, the phonetic lengthening that is incurred by stress renders vowels in stressed syllables non-reducible. In such “simple” cases, there is a clear position which is prominent both in terms of stress and in terms of vowel non-reducibility, while other positions are prosodically weak and prone to vowel reduction. In languages with secondary stress, like English, this scheme is repeated in an alternating pattern throughout the entire word (e.g. [səˈrɛnəɾi] ‘serenity’; Hammond 1997). Resistance to vowel reduction is incurred by prolonged phonetic duration, regardless of the phenomenon that caused duration to increase. Acoustic studies exploring the durational effects of prosodic boundaries have shown that final lengthening manifests at all levels of the prosodic hierarchy; starting from the prosodic word and climbing up to the utterance. The amount of lengthening increases with higher prosodic boundaries (Cambier-Langeveld 1997; Byrd and Saltzman 1998; Cho 2006; Tabain 2003; Tabain and Perrier 2005). Final lengthening at the level of the prosodic word affects the final syllable, prolonging the duration of its vowel (Beckman and Edwards 1987; Wightman et al. 1992). However, phrase-final lengthening targets two distinct positions—the final syllable and the rightmost stressed syllable in the phrase (Berkovits 1994; Turk 1999; Turk and Shattuck-Hufnagel 2007). In complete accordance to these observations, various languages exhibit the blocking of vowel reduction due to the increased phonetic duration of vowels in domain-final positions. This behavior is attested at the word-level, in languages such as Northern Welsh, Belarusian, Ukrainian, Central Eastern Catalan, English and Bonggi, and at the phrase-level, in Russian, Brazilian Portuguese, Yakan, Nawuri, Shimakonde, Murut (Barnes 2006).

4

A Multi-Planar Analysis of Tiberian Hebrew Pause-Context Allomorphy

In light of the above discussion, we present in this section a detailed analysis of the context (§4.1) and pausal (§4.2) allomorphs, employing phenomenonspecific syllable weight scheme. We argue that vowel reduction in Tiberian Hebrew is independent of stress, by showing that stress does not block, license or determine the site of vowel reduction. Rather, the application of vowel reduction is determined by prosodic factors, which include: (i) the position of the word in the phrase, (ii) the position of the syllable in the word, and (iii) the syllable structure. The apparent correlation between stress and vowel reduction is attributed to the shared prosodic elements.

- 978-90-04-44885-8

23

pausal vs. context forms in tiberian hebrew

Additional support for the existence of separate metrical planes comes from the phenomenon of minor pause, which shows that stress and vowel reduction are independent (Revell 1981, 2012). For example, the word /ʔɔttɔ/ ‘you 2msg’ has three surface forms: [ˈʔɔt.tɔ] in major pause, [ˈʔat.tɔ] in minor pause, and [ʔat.ˈtɔ] in context (DeCaen 2005). Notice that the reduced vowel, [a], appears in a stressed syllable in minor pause and an unstressed syllable in context. This phenomenon suggests that stress and vowel reduction are independent of each other, as vowel reduction (/ɔ/ ⇒ [a]) may apply in a stressed or unstressed syllable, depending on the phrase-level prosodic environment. 4.1 Vowel Reduction in the Contextual Allomorph This section begins the exposition of the core proposal of this study. In order to employ the metrical plane of vowel reduction developed above in § 3.3, the distribution of phonetic vowel duration must be determined. Following Khan (1987), phonetic duration in Tiberian Hebrew is conditioned by stress and syllable structure: (13) Tiberian Hebrew phonetic vowel duration – Phonetically long: – Vowels in stressed syllables – Vowels in open syllables (cv) – Phonetically short: – Vowels in unstressed closed syllables (cvc) – Reduced vowels (hataf vowels and [ə]; always in open syllable) Stressed vowels (cv́(c)]σ) are long regardless of syllable structure. When unstressed, vowels in open syllables (cv]σ) are long while vowels in closed syllables (cvc]σ) are short. Finally, reduced vowels (cv̌]σ) are always short and never stressed. This yields the following vowel duration scale (a comma indicates an equal position in the hierarchy): (14) Vowel duration hierarchy (first version) cv́(c)]σ , cv]σ > cvc]σ > cv̌]σ Stressed Open Closed Reduced However, this scale does not prove useful for predicting the alternation involved in pause-context allomorphy, because vowels in all syllable types can be reduced. As shown in (15) below (repeated from (8)), contextual forms display reduction in closed and open syllables, stressed or unstressed.

- 978-90-04-44885-8

24

himmelreich and bat-el foux

(15) Positions of vowel reduction in pause-context allomorphy Position a. b. c. d.

Pause

Context

Penultimate cvc ʔɔ.ˈmɔr.tɔ ʔɔ.ˈmar.tɔ ‘said 2msg’ Penultimate cv ʃɔ.ˈmɔ.ru ʃɔ.mə.ˈru ‘kept 3mpl’ Final cvc kɔ.ˈθɔv kɔ.ˈθav ‘wrote 2msg’ Final cv not attested

The only position that never allows reduction is word-final open syllable (15d). We propose that word-final vowels in Tiberian Hebrew are phonetically lengthened due to final lengthening on the word-level (see § 3.4). The lengthened state of the final vowel renders it resistant to vowel reduction (see § 3.3). Given the position-based behavior difference in cv syllables, i.e. final vs. non-final position, the word-final open syllable, cv]ω, is added to the scale of phonetic vowel duration. (16) Vowel duration hierarchy (revised version; cf. (14)) cv]ω > cv́(c)]σ , cv]σ > cvc]σ > cv̌]σ Open word-final Stressed Open Closed Reduced cv]ω is the only syllable type that should be considered heavy (i.e. bimoraic) for the purposes of vowel reduction, since its vowel is never reduced (15d). The claim that word final vowels in Tiberian Hebrew are long has already been made in Balcaen (1995) and Dresher (2009). However, we argue that this length is relevant only for the vowel reduction system, not for the stress system. The resulting state of affairs in similar to Northern Welsh, where the vowel in the final syllable is longer than the stressed vowel and never reduced (Ball and Williams 2001; Bosch 1996). The following table presents the proposed phenomenon-specific syllable weight for stress and vowel reduction, where monomoraic syllables are light and bimoraic (shaded) are heavy.

- 978-90-04-44885-8

25

pausal vs. context forms in tiberian hebrew

(17) Phenomenon-specific weight for stress and vowel reduction Syllable structure Stress assignment

cvµ cvµcµ

cv cvc

Vowel reduction Word medial Word final cvµ

cvµµ cvµc

For the purposes of stress, cv syllables are monomoraic while cvc syllables are bimoraic. For the purposes of vowel reduction, cvc syllables are monomoraic, while cv syllables vary—monomoraic when word-medial but bimoraic when word-final. That is, there are two types of heavy syllables, one for each phenomenon: cvc for stress and final cv for vowel reduction; all other syllables in all other positions are light. These differences in moraic structure yield different definitions of prominence (i.e. bimoracity): stress is sensitive to the complexity of syllable structure (cv vs. cvc), where coda consonants get an extra mora, while vowel reduction is sensitive to phonetic vowel duration (cv in word final position), where lengthened vowels receive an extra mora. These difference lead to a multi-planar metrical structure, illustrated below for the context forms [hɔ.ˈraʁ] ‘killed 3msg’ and [ʔɔ.ˈmar.tɔ] ‘said 2msg’. (18) Multi-planar analysis for Context forms a. Stress assignment Base

Weight assignment

hɔ.rɔʁ hɔµ.rɔµʁµ ʔɔ.mɔr.tɔ ʔɔµ.mɔµrµ.tɔµ

Trochaic footing

Stress

hɔµ.|rɔµʁµ|F ʔɔµ.|mɔµrµ.tɔµ|F

hɔ.|ˈrɔʁ|F ʔɔ.|ˈmɔr.tɔ|F

Trochaic footing

Reduction

|hɔµ.rɔµʁ|F |ʔɔµ.mɔµr|F.|tɔµµ|F

|hɔ.raʁ|F |ʔɔ.mar|F.|tɔ|F

b. Vowel reduction Base

Weight assignment

hɔ.rɔʁ hɔµ.rɔµʁ ʔɔ.mɔr.tɔ ʔɔµ.mɔµr.tɔµµ

- 978-90-04-44885-8

26

himmelreich and bat-el foux

As emphasized above, the only difference between these two metrical systems is the assignment of an additional mora to coda consonants in (18a) vs. word-final vowels in (18b). Otherwise, the derivation follows a straightforward right-to-left construction of quantity-sensitive trochaic feet. The different weight assignment affects the footing, which in turn affects the output (rightmost columns): in the case of stress (18a), the strong position of the trochaic foot (in bold) is assigned with stress, whereas in (18b), the strong position resists reduction, and thus the vowel in the weak position of the trochaic foot (in bold) undergoes reduction (ɔ ⇒ a). The forms in (18) are cases where stress and vowel reduction do not conflict, and perhaps do not interact at all. However, there are forms in which the two metrical planes interact in a manner that causes alternation in the resulting stress pattern. An example for such a form is [ʃɔ.mə.ˈru] ‘kept 3mpl’, which lacks cvc syllables, so its metrical parsing for the purpose of stress assignment results in one foot which spans the final and penultimate syllables—[ʃɔµ.|mɔµ.ruµ|f]. Therefore, it is expected that the surface form should be stressed at the penultima—*[ʃɔ.ˈmɔ.ru]. In parallel, vowel reduction metrics single out the same penultimate syllable as weak and thus targeted for reduction— [|ʃɔµ.mɔµ|f.|ruµµ|f]. The actual surface form [ʃɔ.mə.ˈru] suggests that reduction wins; consequently, stress shifts to the final syllable since a syllable with a reduced vowel cannot bear stress. Note that this is not likely to be an independent effect of stress assignment, as parallel forms featuring high vowels in the penultimate syllable do not surface with final stress; e.g. [jɔ.ˈmu.θu] ‘will die 3mpl’. The following scheme presents the multi-planar metrics for the context form [ʃɔ.mə.ˈru] ‘kept 3mpl’. (19) Multi-planar metrical systems (s=strong, w=weak) Stress µ Base

s [µ

w µ]

ʃɔ . mɔ . ru

[µ Vowel Reduction s

µ] w

⇒ [ʃɔ.mə.ˈru]

[µµ] s

- 978-90-04-44885-8

pausal vs. context forms in tiberian hebrew

27

The alternation in stress position is explained by the generalization that schwa is never stressed in Tiberian Hebrew. In the cases presented above in (18), the surface quality of reduced vowels is [a], providing no motivation for stress shift. However, in the case where vowel reduction results in a schwa, stress cannot remain in its designated position, and it thus shifts rightward within its foot. This is an important theoretical difference between the present analysis and the vr-feet analysis (Rappaport 1984; see §2.1). In the multi-planar architecture of the vr-feet analysis, the vowel reduction plane takes precedence by incapacitating weak prosodic positions from bearing stress. Consequently, feet parsing on the stress plane is affected directly, and in some cases, re-parsing is imposed. The current proposal eliminates this additional complexity because it does not suppose any direct interaction between the different metrical planes; the alternation in stress position is motivated solely by the inability of schwa [ə] to bear stress. This behavior is easily accountable in parallel derivation frameworks like the Optimality Theory (see Himmelreich 2019), by posing a constraint which bans stress from schwa (McCarthy and Cohn 1998). 4.2 Vowel Reduction in the Pausal Allomorph In earlier studies, the pausal form is often assumed to be similar to the base (Prince 1975; Rappaport 1984; Revell 1981, 2012; Dresher 2009; Qimron 2008; inter alia). In this study we show that pausal forms resemble the base because they undergo less vowel reduction relative to context forms. The reason they undergo less reduction and are thus more “faithful” to the underlying base, is phrase-final lengthening. In the previous section, the analysis of vowel reduction in context forms has employed word-level final lengthening to account for reduction resistance in word-final open syllables. Pausal allomorphs appear at the right edge of the intonational phrase (Dresher 1994) and are thus affected both by word-level and phrase-level final lengthening (see §3.4). The effects of final lengthening at different levels are not identical in their domain of application. Lengthening at both word and phrase level affects final vowels (of each domain respectively). However, phrase-final lengthening targets another position—the stressed syllable of the phrase’s last word (Berkovits 1994; Turk and ShattuckHufnagel 2007). This is the crucial phenomenon that distinguishes phrase-final vs. phrase-medial words. Ultimately, this is also the origin of pausal forms. Recall that the position of alternation between pause and context forms is always the stressed vowel of the pausal form. Inversely phrased, the data in (5), (6) and (8) suggest that the pausal form’s stressed vowel never undergoes reduction—it is non-reducible. Conversely, the stressed vowel of a phrase-

- 978-90-04-44885-8

28

himmelreich and bat-el foux

medial word is not protected by additional lengthening, and so it does undergo reduction, yielding the attested contextual forms. In complete parallel to the analysis of word-final vowels as reduction resistant, in this study, we propose to analyze the pausal form’s stressed vowel as reduction resistant, i.e. V́ ]ip and cv]ω are analyzed as bimoraic for the metrical scheme of vowel reduction. Thus, the growing vowel duration scale can now be extended with its final member: V́ ]ip—the last stressed vowel in an intonational phrase. (20) Vowel duration hierarchy: final version V́ ]ip , cv]ω > cv́(C)]σ , cv]σ > cvc > cv̌ Stressed Open Stressed Open Closed Reduced Phrase-Final Word-Final The following table presents the complete set of proposed phenomenon-specific syllable weight schemes for stress and reduction: (21) Phenomenon-specific weight schemes for stress and reduction Syllable structure cv cvc

Stress Vowel reduction assignment Word medial Word final Last stressed in ip cvµ cvµcµ

cvµ cvµC

cvµµ

cvµµ

For the purposes of stress assignment, phrase-level position is irrelevant, thus no change is required by the added reference to the phrase-final position. For the purposes of vowel reduction, the stressed vowel of the last word in the phrase is lengthened and thus it is bimoraic. The following tables present the application of the proposed multi-planar metrical structure for pausal forms: [ʃɔ.ˈmɔr.ti] ‘kept 1MSG’ and [hɔ.ˈrɔʁ] ‘killed 3MSG’. The data from (18a) is repeated below in (22a) for convenience, the metrical account for stress assignment in both context and pausal forms is identical.

- 978-90-04-44885-8

29

pausal vs. context forms in tiberian hebrew

(22) Multi-planar analysis for Pausal forms a. Stress assignment Base i. hɔ.rɔʁ ii. ʃɔ.mɔr.ti

Weight assignment

Trochaic footing

Stress

hɔµ.rɔµʁµ ʃɔµ.mɔµrµ.tiµ

hɔµ.[rɔµʁµ] ʃɔµ.[mɔµrµ.tiµ]

hɔ.[ˈrɔʁ] ʃɔ.[ˈmɔr.ti]

Weight assignment

Trochaic footing

Reduction

hɔµ.rɔµµʁ ʃɔµ.mɔµr.tiµµ

hɔµ.[rɔµµʁ] ʃɔµ.[mɔµµr].[tiµµ]

hɔ.[rɔʁ] ʃɔ.[mɔr].[ti]

b. Vowel reduction Base i. hɔ.rɔʁ ii. ʃɔ.mɔr.ti

The resulting metrical structure in both items in (22b) contains only bimoraic feet. There are no metrically weak positions, and therefore there is no reduction (see inalterability in §3.3). This is the common state of affairs in pausal forms which span up to 4 syllables. However, given a longer word such as /lə.hɔ.rɔ.ʁɛ.χɔ/ ‘to kill you’, which is long enough to host three feet, vowel reduction will manifest in a pausal form. The metrical parsing will result in [ləµ.|hɔµ.rɔµ|f.|ʁɛµµ|f.|χɔµµ|f], where the leftmost trochaic foot is disyllabic, thus targeting the antepenultimate syllable [rɔ] for vowel reduction. The derivation of both the contextual and the pausal allomorphs of /lə.hɔ.rɔ.ʁɛ.χɔ/ are given in (23) below: (23) Derivation of long Context and Pausal forms Context ‘to kill you’

Pause ‘to kill you’

Base

lə . hɔ . rɔ . gɛ . χɔ

lə . hɔ . rɔ . gɛ . χɔ

vr weight assignment trochaic footing trochaic reduction

ləµ.hɔµ.rɔµ.ʁɛµ.χɔµµ [ləµ.hɔµ].[rɔµ.ʁɛµ].[χɔµµ] [la.hă].[rɔ.ʁə].[χɔ]

ləµ.hɔµ.rɔµ.ʁɛµµ.χɔµµ ləµ.[hɔµ.rɔµ].[ʁɛµµ].[χɔµµ] lə.[hɔ.rə].[ʁɛ].[χɔ]

- 978-90-04-44885-8

30

himmelreich and bat-el foux (cont.)

Base

lə . hɔ . rɔ . gɛ . χɔ

lə . hɔ . rɔ . gɛ . χɔ

Stress weight assignment trochaic footing stress assignment

laµ.hăµ.rɔµ.ʁəµ.χɔµ laµ.hăµ.rɔµ.[ʁəµ.χɔµ] la.hă.rɔ.ʁə.ˈχɔ

ləµ.hɔµ.rəµ.ʁɛµ.χɔµ ləµ.hɔµ.rəµ.[ʁɛµ.χɔµ] lə.hɔ.rə.ˈʁɛ.χɔ

The crucial point of divergence in the derivation processes of the two allomorphs is the stage of weight assignment in vr. The assignment of an extra mora to the penultimate syllable in the pausal form renders its vowel nonreducible. The rest of the derivation process follows from standard parsing to right-aligned trochaic feet and application of vowel reduction at the weak positions. The same derivation process is illustrated in the following diagram: (24) Derivation of long Context and Pausal forms Context

Reduction feet

[ʃɔµ.mɔµ].[ruµµ] Vowel Reduction

ʃɔ . mə . ˈru

Pause ʃɔµ.[mɔµµ].[ruµµ] ◆

◆

◆

◆

Bimoraic vowels resist reduction

ʃɔ . ˈmɔ . ru

4.3 Moraic Projection To account for the complex pattern of the moraic projection required by phenomenon-specific syllable weight schemes, the grammar of weight assignment must be allowed to refer to other levels in the prosodic hierarchy. For example, contour tone licensing in Lhasa Tibetan is licensed on cvv and cvr syllables (R = Sonorant), but not cvc syllables (C=Stop). This suggests that contour tone licensing requires a structure where two consecutive prosodic slots are associated with the segmental feature [+son]. In other words, weight assignment for the purposes of contour tone must refer to the segmental tier, in order to be able to distinguish between sonorant and non-sonorant consonants.

- 978-90-04-44885-8

31

pausal vs. context forms in tiberian hebrew

(25) Moraic projection in different structures of a cvc syllable a. cvv µ µ | | C V V \ / [+son]

b. cvr µ µ | | C V R | | [+son] [+son]

c. cvc µ µ | C V C | [+son]

In the classic case, the moraic projection for each C and V slot is determined by its position within the syllable. However, if the projection of a mora can be conditioned by its association with segmental features, the grammar of weight assignment must refer to more complex structures. In the case of contour tone licensing, it is necessary to refer to both the cv tier and the segmental tier to determine whether a mora is to be projected. In (25), the conditions for the projection of two morae are met only in (a) and (b), thus contour tone is licensed only in these two structures. This type of analysis can be used to formalize the grammars of various phenomenon-specific weight assignment schemes (Gordon 2006; Rayan 2019). For the purposes of weight assignment on the metrical plane of vowel reduction, we assume that the projection of additional morae (beyond the basic one-mora-per-vowel) requires a structure where a V slot is associated with a prosodic position which is phonologically prominent due to final lengthening. This analysis follows the Structural Prominence approach to the phoneticsphonology interface (Beckman 1997), according to which prosodic structure is assumed to include abstract prominence features such as [strong]. Prominence features are assumed to exist at different levels of the prosodic hierarchy. For example, at the foot level, the feature [strong] distinguishes between trochaic and iambic feet. At the word level, [strong] marks the foot which bears primary stress. At the phrase level, it marks the word which bears phrasal stress/focus. The tree representation in (26) presents the multi-planar prosodic parsing of an intonational phrase in Tiberian Hebrew. The structure that extends upwards corresponds to the stress metrical plane; the structure that extends downwards represents the vowel reduction metrical plane. Strong prosodic positions are marked by a subscript “s” (e.g. φs or Fs).

- 978-90-04-44885-8

32

himmelreich and bat-el foux

(26) Prosodic structure of an intonational phrase

wə.ka.ʔă.ˈʃɛr ʔɔ.ˈvað.ti ʔɔ.ˈvɔð.ti and if perish 1sg perish 1sg … and if I perish, I perish. (Esth 4:16) Note that the word /ʔɔvɔðti/ ‘perished 1sg’ appears twice; the first occurrence is a contextual form [ʔɔ.ˈvað.ti] and the second is a pausal form [ʔɔ.ˈvɔð.ti]. In both occurrences of this word, an additional mora is projected by the word-final vowel. This occurs because the word-final vowel is affected by final lengthening (see §3.4), and thus, it is assigned with the [strong] feature. Moreover, the pausal form’s stressed vowel also projects an additional mora. This occurs because this vowel is the rightmost stressed vowel in the intonational phrase, and thus it is also affected by final lengthening (at the ip level). Thus, the grammar of weight assignment for the vowel reduction plane depends on higher prosodic levels, namely, the word-level and the ip-level.

- 978-90-04-44885-8

pausal vs. context forms in tiberian hebrew

33

This dependency must be embodied by the rules or constraints employed in any formal account of these phenomena. Rules targeting a word-final vowel are straightforward, while for singling out the “rightmost stressed vowel” one can refer to phrase-level stress models such as the nsr (Chomsky and Halle 1968) and its later developments (Liberman and Prince 1977; Gussenhoven 1992; Cinque 1993; see Truckenbrodt 2006 for survey). In an Optimality Theoretic (Prince and Smolensky 1993) account, one would use constraints such as *Vµ]ω and *ˈVµ]ip which propagate that a vowel associated with a word-final position or the rightmost stressed position in the ip—must not be monomoraic (see Himmelreich 2019 for ot analysis).

5

Conclusion

In this study, we analyzed the allomorphy between pausal and contextual forms in Tiberian Hebrew, where the alternation is conditioned by the word’s position in a phrase. As there is no semantic or morphological difference between pausal and contextual forms, this case presented an opportunity to examine a phenomenon which is purely phonological and provide a glimpse into the phonological grammar of Tiberian Hebrew and the nature of the interaction between stress, vowel reduction and phrase final lengthening in general. Regarding the study of Tiberian Hebrew, the proposed analysis improves upon its predecessors by incorporating universally attested phenomena into the account. First, vowel reduction patterns which are cross-linguistically common are shown to match the segmental alternation exhibited by pausal allomorphy (§2). Second, phrase final lengthening provides a simplified account to the reduction-resistant nature of pausal forms (§ 4.2), which finds parallels in many other languages (Barnes 2006). Third, the employment of phenomenonspecific prominence (Gordon 2006) to account for the metrical conflict of reduction in stressed syllables reinforces the conceptual core of multi-planar metrical systems (Rappaport 1984). We argued that vowel reduction in Tiberian Hebrew operates independently of stress, where the positions it targets and the vowel alternation are determined by prosodic factors such as the position of the word in the phrase, the position of the syllable in the word, and syllable structure. Regarding linguistic theory, we introduced the phenomenon-specific syllable weight scheme for vowel reduction (§3.3). The strong link between vowel reduction and phonetic vowel duration has been demonstrated and widely accepted for some time now (Lindblom 1963; Moon and Lindblom 1994; Flem-

- 978-90-04-44885-8

34

himmelreich and bat-el foux

ming 2005). In the common case, the metrical systems of stress and vowel reduction are harmonically interweaved, creating an alternating pattern of stressed and reduced syllables which does not justify two separate metrical planes. However, the cases where these metrical systems do not align, such as Northern Welsh (Bosch 1996) and hereby Tiberian Hebrew, suggest that these mechanisms can operate independently. In fact, languages with vowel reduction but no secondary stress (e.g. Russian; Crosswhite 2000) are clear cases of such metrical misalignment. In this type of languages, vowel reduction typically targets all non-stressed positions, resulting in two fundamentally different metrical domains. While stress assignment metrics employ a single foot, the domain of vowel reduction spans throughout the entire remainder of the prosodic word. While the segmental and phonetic realms of vowel reduction have been widely studied (Crosswhite 2001; Beckman 1997; Flemming 2005; Padgett and Tabain 2005; inter alia), the nature of metrical alignment between stress and vowel reduction seems like a promising endeavor for future research. Natural candidates for such research would be languages where the main phonetic correlate of stress is not phonetic vowel duration.

References Anderson, Stephen R. 1982. Where’s morphology? Linguistic Inquiry 13 (4): 571–612. Anstey, Matthew P. 2005. Towards a typological presentation of Tiberian Hebrew. Hebrew Studies 46(1): 71–128. Balcaen, M. Jean. 1995. The prosody of Tiberian Hebrew. ma thesis, University of Saskatchewan. Ball, Martin J., and Briony J. Williams. 2001. Welsh phonetics. Lampeter, Wales: Edwin Mellen. Barnes, Jonathan. 2006. Strength and weakness at the interface: Positional neutralization in phonetics and phonology. Berlin: Walter de Gruyter. Barnes, Jonathan. 2007. Phonetics and phonology in Russian unstressed vowel reduction: A study in hyperarticulation. Unpublished ms. Becker, Michael. 2003. Hebrew Stress: Can’t you hear those trochees? University of Pennsylvania Working Papers in Linguistics 9(1): 45–57. Becker-Kristal, Roy. 2010. Acoustic typology of vowel inventories and Dispersion Theory: Insights from a large cross-linguistic corpus. PhD dissertation, University of California, Los Angeles. Beckman, Jill N. 1997. Positional faithfulness, positional neutralization and Shona vowel harmony. Phonology 14(1): 1–46. Beckman, Mary E., and Jan Edwards. 1992. Intonational categories and the articula-

- 978-90-04-44885-8

pausal vs. context forms in tiberian hebrew

35

tory control of duration. Pp. 359–375 in Speech perception, production, and linguistic structure, ed. Y. Tohkura, E. Vatikiotis-Bateson, and Y. Sagisaka. Amsterdam: ios Press. Berkovits, Rochele. 1994. Durational effects in final lengthening, gapping, and contrastive stress. Language and Speech 37(3): 237–250. Bosch, Anna. 1996. Prominence at two levels: Stress versus pitch prominence in North Welsh. Journal of Celtic Linguistics 5: 121–165. Bosch, Anna and Wiltshire, Caroline. 1993. The licensing of prosodic prominence in Tamil. Pp. 1–15 in Proceedings of the Third Annual Meeting of the Formal Linguistics Society of Midamerica, ed. L. Smith Stvan. Bloomington, IN: Indiana University Linguistics Club. Byrd, Dani, and Elliot Saltzman. 1998. Intragestural dynamics of multiple prosodic boundaries. Journal of Phonetics 26(2): 173–199. Cambier-Langeveld, Tina, Marina Nespor and Vincent J. van Heuven. 1997. The domain of final lengthening in production and perception in Dutch. Pp. 931–934 in eurospeech 1997, Fifth European Conference on Speech Communication and Technology, ed. G. Kokkinakis, N. Fakotakis, and E. Dermatas. isca Archive. Cho, Taehong. 2006. Manifestation of prosodic structure in articulatory variation: Evidence from lip kinematics in English. Laboratory phonology 8: 519–548. Chomsky, Noam, and Morris Halle. 1968. The sound pattern of English. New York: Harper and Row. Churchyard, Henry. 1999. Topics in Tiberian Biblical Hebrew metrical phonology and phonetics. PhD dissertation, University of Texas, Austin, TX. Cinque, Guglielmo. 1993. A null theory of phrase and compound stress. Linguistic inquiry 24(2): 239–297. Clements, Nick and Kayser, Samuel. 1983. CV Phonology: A Generative Theory. Cambridge, MA: The mit Press. Cohen, Evan-Gary, Vered Silber-Varod, and Noam Amir. 2018. The acoustics of primary and secondary stress in Modern Hebrew. Brill’s Annual of Afroasiatic Languages and Linguistics 10: 1–19. Crosswhite, Katherine M. 2000. Vowel reduction in Russian: a unified account of standard, dialectal, and “dissimilative” patterns. University of Rochester working papers in the language sciences 1: 107–172. Crosswhite, Katherine M. 2001. Vowel reduction in optimality theory. New York: Routledge. Crosswhite, Katherine M. 2004. Vowel reduction. Pp. 191–231 in Phonetically based phonology, ed. Bruce Hayes, Robert Kirchner, and Donca Steriade. Cambridge: Cambridge University Press. Dawson, Willa. 1980. Tibetan phonology. PhD dissertation, University of Washington, Seattle, WA.

- 978-90-04-44885-8

36

himmelreich and bat-el foux

DeCaen, Vincent. 2005. On the distribution of major and minor pause in Tiberian Hebrew in the light of the variants of the second person independent pronouns. Journal of Semitic Studies 50(2): 321–327. Dresher, Bezalel Elan. 1994. The prosodic basis of the Tiberian Hebrew system of accents. Language 70(1): 1–52. Dresher, Bezalel Elan. 2009. The Word in Tiberian Hebrew. Pp. 95–111 in The nature of the word: Essays in honor of Paul Kiparsky, ed Kristin Hanson and Sharon Inkelas. Cambridge, MA: mit Press. Flemming, Edward. 2005. A phonetically-based model of phonological vowel reduction. Unpublished ms. Flemming, Edward. 2009. The phonetics of schwa vowels. Pp. 78–98 in Phonological weakness in English: from old to present-day English, ed. Donka Minkova. New York: Palgrave Macmillan. Garcia, Guilherme D., Heather Goad, and Natália Brambatti Guzzo 2017. Footing is not always about stress: Formalizing variable high vowel deletion in Québec French. Pp. 1–10 in Proceedings of the Annual Meetings on Phonology, vol. 4, ed. Karen Jesney, Charlie O’Hara, Caitlin Smith and Rachel Walker. Washington, dc: Linguistic Society of America. Gesenius, Wilhelm. 1909. Gesenius’ Hebrew Grammar, rev. E. Kautzsch, trans. A.E. Cowley. Oxford: Oxford University Press. Goerwitz, Richard L. 1993. Tiberian Hebrew Pausal Forms. PhD dissertation, University of Chicago, Chicago, IL. Gordon, Matthew. 2006. Syllable weight: phonetics, phonology, typology. New York: Routledge. Gordon, Matthew, and Timo Roettger. 2017. Acoustic correlates of word stress: A crosslinguistic survey. Linguistics Vanguard 3(1). Gussenhoven, Carlos. 1992. Sentence accents and argument structure. Pp. 79–106 in Thematic structure: Its role in grammar, ed. I.M. Roca. Berlin: Foris. Hammond, Michael. 1997. Vowel quantity and syllabification in English. Language 73:1– 17. Hannahs, Stephen J. 2007. Constraining Welsh vowel mutation. Journal of Linguistics 43(2): 341–363. Hannahs, Stephen J. 2013. The phonology of Welsh. Oxford University Press. Hayes, Bruce. 1986. Inalterability in cv phonology. Language 62: 321–351. Hayes, Bruce. 1989. Compensatory lengthening in moraic phonology. Linguistic inquiry 20(2): 253–306. Hayes, Bruce. 1995. Metrical stress theory: Principles and case studies. University of Chicago Press. Himmelreich, Roman. 2019. Positional Allomorphy: Pausal vs. Context forms in Tiberian Hebrew. ma thesis, Tel-Aviv University.

- 978-90-04-44885-8

pausal vs. context forms in tiberian hebrew

37

Hyman, Larry M. 1984. On the weightlessness of syllable onsets. Pp. 1–14 in Annual Meeting of the Berkeley Linguistics Society 10, ed. Claudia Brugman and Monica Macaulay. Berkeley, CA: Berkeley Linguistics Society. Khan, Geoffrey. 1987. Vowel length and syllable structure in the Tiberian tradition of Biblical Hebrew. Journal of Semitic Studies 32(1): 23–82. Khan, Geoffrey. 1994. The historical background of the vowel ṣere in some Hebrew verbal and nominal forms. Bulletin of the School of Oriental and African Studies 57(1): 133–144. Khan, Geoffrey. 2013. Shewa: pre-modern Hebrew. Pp. 543–554 in Encyclopedia of Hebrew Language and Linguistics, vol. 3, ed. Geoffrey Khan. Leiden: Brill. Liberman, Mark, and Alan Prince. 1977. On stress and linguistic rhythm. Linguistic inquiry 8(2): 249–336. Lindblom, Björn. 1963. Spectrographic study of vowel reduction. The journal of the Acoustical society of America 35(11): 1773–1781. McCarthy, John J., and Abigail Cohn. 1998. Alignment and parallelism in Indonesian phonology. Working Papers of the Cornell Phonetics Laboratory 12, 53–137. Moon, Seung-Jae, and Björn Lindblom. 1994. Interaction between duration, context, and speaking style in English stressed vowels. The Journal of the Acoustical society of America 96(1): 40–55. Padgett, Jaye, and Marija Tabain. 2005. Adaptive dispersion theory and phonological vowel reduction in Russian. Phonetica 62(1): 14–54. Prince, Alan. 1975. The phonology and morphology of Tiberian Hebrew. PhD dissertation, Massachusetts Institute of Technology, Cambridge, MA. Prince, Alan and Paul Smolensky. 2004. Optimality Theory: Constraint Interaction in Generative Grammar. Pp. 1–71 in Optimality Theory in Phonology: A Reader, ed. John J. McCarthy. Malden, MA: Wiley. Qimron, Elisha. 1986. Interchanges of e and a vowels in accented closed syllables in Biblical Hebrew. Leshonen, 50: 77–102. Qimron, Elisha. 2006. The Pausal Patah in Biblical Hebrew. Pp. 305–314 in Biblical Hebrew in its Northwest Semitic Setting, ed. S. Fassberg and A. Hurvitz. Jerusalem: Hebrew University Magnes Press. Qimron, Elisha. 2008. Pausal forms—what is their nature? Pp. 92–106 in Shaʾarei Lashon: Studies in Hebrew, Aramaic, and Jewish Languages Presented to Moshe BarAsher, vol. 1: Biblical Hebrew, Masorah, and Medieval Hebrew, ed. A. Maman, S.E. Fassberg, and Y. Breuer. Jerusalem: Bialik Institute. Rappaport, Malka. 1984. Issues in the phonology of Tiberian Hebrew. PhD dissertation, Massachusetts Institute of Technology, Cambridge, MA. Revell, Ernest J. 1981. Pausal forms and the structure of biblical poetry. Vetus Testamentum 31(2): 186–199. Revell, Ernest J. 2012. The Occurrence of Pausal Forms. Journal of Semitic Studies 57(2): 213–230. - 978-90-04-44885-8

38

himmelreich and bat-el foux

Ryan, Kevin M. 2016. Phonological weight. Language and Linguistics Compass 10(12): 720–733. Ryan, Kevin M. 2019. Prosodic end-weight reflects phrasal stress. Natural Language and Linguistic Theory 37(1): 315–356. Silber-Varod, Vered, Hagit Sagi, and Noam Amir. 2016. The acoustic correlates of lexical stress in Israeli Hebrew. Journal of Phonetics 56:1–14. Smolensky, Paul, and Alan Prince. 1993. Constraint interaction in generative grammar. Oxford: Blackwell. Steriade, Donca. 1991. Moras and other slots. Pp. 254–228o in Proceedings of the Formal Linguistics Society of Midamerica 1, ed. D. Meyer, S. Tomioka, L. Zidani-Eroglu. Madison, WI: Linguistics Student Organization. Tabain, Marija. 2003. Effects of prosodic boundary on/aC/sequences: articulatory results. The Journal of the Acoustical Society of America, 113(5): 2834–2849. Tabain, Marija, and Pascal Perrier. 2005. Articulation and acoustics of /i/ in preboundary position in French. Journal of Phonetics 33(1): 77–100. Truckenbrodt, Hubert. 2006. Phrasal stress. Pp. 572–579 in The encyclopedia of languages and linguistics, vol. 9, ed. Keith Brown. Boston: Elsevier Science. Turk, Alice E. 1999. Structural influences on boundary-related lengthening in English. Pp. 1117–1120 in Proceedings of the 14th International Congress of Phonetic Sciences, vol. 2, ed. John J. Ohala, Yoko Hasegawa, Manjari Ohala, Daniel Granville, and Ashlee C. Bailey. Berkeley, CA: University of California Press. Turk, Alice E., and Stefanie Shattuck-Hufnagel 2007. Multiple targets of phrase-final lengthening in American English words. Journal of Phonetics 35(4): 445–472. Wightman, Colin W., Stefanie Shattuck-Hufnagel, Mari Ostendorf, and Patti J. Price. 1992. Segmental durations in the vicinity of prosodic phrase boundaries. The Journal of the Acoustical Society of America 91(3): 1707–1717.

- 978-90-04-44885-8

chapter 3

Prosodic Dependency in Tiberian Hebrew Vincent DeCaen and B. Elan Dresher

Ernest John Revell, 1934–2017 In Memoriam

∵ 1

Introduction

We examine two related but different notions of prosodic freedom and dependency in Tiberian Hebrew (th). One is a morphosyntactic notion, whereby absolute forms are free and construct forms are dependent. Examples of absolute forms are dɔːvɔ́ ːʀ ‘word’ and davɔːʀíːm ‘words’; in the construct, these become davà(ː)ʀ, as in davàːʀ hammɛ́ːlɛχ ‘the king’s word’, and divʀèː, as in divʀèː hammɛ́ːlɛχ ‘the king’s words’.1 The other is a prosodic notion whereby free forms receive an accent (in the musical interpretation of the prosody), and dependent forms are clitics bound to a free form by a maqqef ‘hyphen’ (marked throughout by the equal-sign ‘=’ to distinguish it from a morpheme boundary, marked by a hyphen ‘-’ where required). An example is the accusative particle, which appears as ʔéːθ when it is prosodically free and receives a musical accent, and as ʔɛθ= when it is prosodically bound by maqqef.2 One would reasonably expect there to be isomorphism between the prosodic dependency of the construct and the system of accents: that is, we might expect that construct words would be unaccented and marked with maqqef, and that absolute words would always be prosodically free and accented. This is not always the case, however: a construct word is sometimes accented (daváːʀ)

1 Our phonetic transcriptions of th forms follow Khan (1987, 2013). We indicate the lower degree of stress that construct forms receive by a grave accent (`); see (12) below. We assume that this stress is lost when the construct is cliticized with maqqef, and promoted to a full word stress (´) when the construct is a prosodic word with a musical accent. 2 Joüon & Muraoka (2006: 54n1) note three exceptions where ʔɛ́θ occurs accented and uncliticized: Ps 47:5, 60:2, Prov 3:12.

© Vincent DeCaen and B. Elan Dresher, 2021 | doi:10.1163/9789004448858_004

- 978-90-04-44885-8

40

decaen and dresher

and sometimes deprived of the accent (davaʀ=). Conversely, absolute state words are sometimes deprived of an accent according to the rules governing phrasing. It appears, therefore, that there are two distinct definitions of prosodic dependency: morphosyntactic versus accentual. That is not the end of the matter, however: “small” nouns (Breuer 1982: 167)— that is monosyllabic stems such as léːv ‘heart’, šéːm ‘name’, ḥóːq ‘ordinance’, róːv ‘multitude’—are caught up in a conflict between these two notions. There is considerable variability in how these nouns behave in this respect. In this article we will consider the reasons for the development of two different notions of prosodic dependency.

2

The Prosodic Dependency of the Construct

There are various phonological differences between the absolute and construct forms. Some of these involve differences that are morphological or morphophonological (Prince 1975; Joüon & Muraoka 2006). The masculine plural suffix is -iːm in the absolute form, as in davɔːʀ-íːm (1a); this suffix does not appear in the construct, divʀèː (1b). Instead we find an ending -èː, which may be connected to the augment found in suffixed forms such as davɔːʀ-éː-nuː, divʀeː-hɛ́m (1c). (1) Absolute and construct forms of masculine words a. Absolute Singular Plural dɔːvɔ́ ːʀ davɔːʀ-íːm word.ms.abs word-mp.abs ‘word’ ‘words’ b. Construct Singular davàːʀ ham-mɛ́ːlɛχ word.ms.cstr the-king.ms.abs ‘the king’s word’

Plural divʀ-èː ham-mɛ́ːlɛχ word.mp.cstr the-king.ms.abs ‘the king’s words’

c. Suffixed Singular davɔːʀ-éː-nuː word.ms-aug-1p ‘our word’

Plural divʀ-eː-hɛ́ːm word-mp.aug-3mp ‘their words’

- 978-90-04-44885-8

41

prosodic dependency in tiberian hebrew

In the feminine, the absolute singular form ends in -ɔ́ ː (2a), whereas the construct singular ends in -àːθ (2b). Pretonically, the feminine singular morpheme appears as -ɔːθ-, thus ṣiðq-ɔːθ-íː ‘my righteousness’ (2c). Therefore, the construct morphology of both the masculine and feminine nouns has some connection to morphology found in the suffixed forms, suggesting there is some underlying form from which both the absolute and construct forms of a noun can be derived. For example, the unspirantized underlying representation of the feminine singular must be /-ɔt/ < *at. (2) Absolute and construct forms of feminine words a. Absolute Singular Plural ṣaðɔːq-ɔ́ ː ṣaðɔːq-óːθ righteousness-fs.abs righteousness-fp.abs ‘righteousness’ ‘acts of righteousness’ b. Construct Singular ṣiðq-àːθ haṣ-ṣaddíːq righteousness-fs.cstr the-righteous.ms.abs ‘the righteousness of the righteous’

‘the gracious acts of the Lord’

c. Suffixed Singular ṣiðq-ɔːθ-íː righteousness-fs-1s ‘my righteousness’

Plural ṣiðq-oːθ-éː-nuː righteousness-fp-aug-1p ‘our virtues’

Plural ṣiðq-òːθ ʔaðo:nɔ́ :j righteousness.cstr-fp lord.ms

Here, we will focus on differences between absolute and construct forms that can be attributed to differences of stress. These are of particular interest because they support the idea that construct forms are prosodically dependent on the word that stands at the end of a construct chain. According to Joüon & Muraoka (2006: 253), “The two nouns form a phonetic unit … The first noun is said to be in the construct state because it rests phonetically on the second … [it] always loses something of its stress.” The vowel deletions and reductions observed in the construct all follow from the assumption that a word in the construct lacks the full word stress that words in the absolute form receive. Consider the derivations of the absolute and construct singular of dɔːvɔ́ ːʀ, based on Prince (1975) as modified by Dresher (2009a). We assume that their

- 978-90-04-44885-8

42

decaen and dresher

lexical representations are the same, except that the construct and the word that follows it “form a phonetic unit”, formalized by Prince (1975) as a single word boundary (3b), in contrast to the double word boundary that follows a word in the absolute (3a). (3) Lexical representations of the absolute and construct of ‘word’ a. Absolute b. Construct x x x x Line 0 da baʀ## da baʀ# Dresher (2009a) proposes a new analysis of Biblical Hebrew stress in the framework of the Simplified Bracketed Grid (sbg) metrical theory (Idsardi 1992; Halle & Idsardi 1995). This theory builds on the metrical theory of Liberman & Prince (1977), Halle & Vergnaud (1987), and Hayes (1995). On this approach, stress is computed by projecting elements (grid marks and brackets) onto a metrical grid. The examples in (3) show the first line of the grid, conventionally called line 0. Every potential stress-bearing unit receives a grid mark on line 0; in Biblical Hebrew, these units are vowels. It appears that Biblical Hebrew main stress must apply early in the derivation to account for Pretonic Lengthening and the phonology of pausal and contextual forms. This creates a problem, in that later rules require that the early metrical feet must be over-written by conflicting feet that govern vowel reduction and deletion as well as secondary stress (Blake 1951; Prince 1975; Rappaport 1984; Malone 1993; Balcaen 1995). Dresher (2009a) proposes that rather than an early rule assigning main stress or stress feet, there is an early rule of Left Bracket Insertion (lbi), given in (4): (4) Left Bracket Insertion (lbi) Insert a left bracket to the left of the last vowel of the word that is not absolutely word-final. What we are proposing here is that lbi does not operate in the domain of a single #. Therefore, a left bracket is inserted in the absolute form (5a), but not in the construct (5b). (5) Left Bracket Insertion (lbi) in the absolute of ‘word’ a. Absolute b. Construct x (x x x Line 0 da baʀ## da baʀ#

- 978-90-04-44885-8

prosodic dependency in tiberian hebrew

43

The next rule that applies is Pretonic Lengthening (ptl), given in (6).3 (6) Pretonic Lengthening (ptl) Lengthen a vowel in an open syllable immediately to the left of a left bracket. ptl applies in the absolute state (7a) but not in the construct (7b), because the latter does not have a left bracket. (7) Pretonic Lengthening (ptl) in the absolute of ‘word’ a. Absolute b. Construct x (x x x Line 0 daa baʀ## da baʀ# Next, heavy syllables (syllables with long vowels or that are closed by a consonant), which are the heads of feet, receive a right bracket. In the absolute form of ‘word’ (8a), both syllables receive a right bracket because they are both heavy: the first because of its long vowel, the second because it is closed by a consonant. In the construct (8b), only the final syllable receives a right bracket. (8) Heavy syllables receive a right bracket a. Absolute b. Construct x) (x) x x) Line 0 daa baʀ## da baʀ# Two more bracket insertion rules apply on line 0, as described in (9) and (10).4 Edge marking puts a left bracket at the edge of every word; Iterative Constituent Construction (icc) limits the size of line 0 metrical constituents (feet) to be maximally binary. These rules yield the line 0 representations in (11) (in (11b), the effects of (9) and (10) coincide to place a left bracket at the left edge). (9) Edge marking on line 0 In every word, insert a left bracket to the left of the leftmost element on line 0.

3 This formulation is inspired by Balcaen (2000). 4 This part of the analysis departs from that in Dresher (2009a) but adheres to the proposals of Halle & Idsardi (1995).

- 978-90-04-44885-8

44

decaen and dresher

(10) Iterative Constituent Construction (icc) Starting from the right on line 0, insert a left bracket after every pair of grid marks in which no bracket intervenes. (11) Representations showing edge marking (and icc) a. Absolute b. Construct (x) (x) (x x) Line 0 daa baʀ## da baʀ# Iambic (right-headed) feet are assigned by projecting the rightmost element in a line 0 foot to the next line (line 1, the level of heads of feet). The rightmost foot head is projected to receive the word-level stress on line 2. Then words that are the heads of their prosodic word (clitic group) receive a further level of stress on line 3. This level of stress is typically assigned to words in the absolute state (12a), but not to words in the construct state (12b), for they are not the heads of a prosodic word. (12) Projection of foot-level, word-level, and phrase-level stress: masculine singular nouns a. Absolute b. Construct x Line 3 (prosodic word stress) x x Line 2 (word stress) x x) x) Line 1 (heads of feet) (x) (x) (x x) Line 0 (stressable units) daa baʀ## da baʀ# At this point the metrical grids are in place, and further rules shown in (13) apply to derive the phonetic forms of the absolute (14a) and construct (14b) forms.5 (13) Some segmental rules a. Tone Lengthening: A vowel with prosodic word stress is lengthened under certain conditions.

5 These derivations can be understood either as derivations from underlying to surface forms or as a compact summary of diachronic developments. Since the Masoretic manuscripts date from around 500 years after Hebrew had ceased to be spoken as a first language (Blau 2010: 11), we might expect that a ‘synchronic’ grammar of Tiberian Hebrew would have some unusual properties; see Edzard (2013) and Rendsburg (2013).

- 978-90-04-44885-8

prosodic dependency in tiberian hebrew

45

b. Vowel reduction: A vowel in the weak position of a foot loses its grid mark (is reduced or deleted). c. Spirantization of non-geminate, non-emphatic, stop consonants: / b, g, d, k, p, t/ become [v, ʁ, ð, χ, f, θ], respectively, following a vowel. d. Rounding: [aː] becomes [ɔː]. e. Phonetic Lengthening: Vowels with word stress and vowels in open syllables are lengthened. f. Schwa Realization: /ə/ is pronounced as [a] in most environments. (14) Application of segmental rules to the forms in (13) a. Absolute b. Construct Output of (13) (daː) (báʀ)## (da bàʀ)# Tone Lengthening (daː) (báːʀ)## — Vowel reduction — (də bàʀ)# Spirantization (daː) (váːʀ)## (də vàʀ)# Rounding (dɔː) (vɔ́ ːʀ)## — Phonetic Lengthening — (də vàːʀ)# Schwa Realization — (da vàːʀ)# Phonetic forms [dɔːvɔ́ ːʀ] [davàːʀ] Similarly, it can be shown (without dwelling on the steps) that the differences between the masculine plural absolute and construct forms can be derived from /dabaʀ+iːm/ and /dabaʀ+ay/, respectively. The rules that construct metrical grids and ptl give the representations in (15). (15) Projection of foot-level, word-level, and phrase-level stress: masculine plural nouns a. Absolute b. Construct x) Line 2 (prosodic word stress) x) x Line 2 (word stress) x x) x x) Line 1 (heads of feet) (x x) (x) (x (x x) Line 0 (stressable units) da baaʀ + iim## da ba ʀ + ay# Then, the rules of vowel reduction/deletion and other segmental rules apply as in (16) to give the phonetic forms [davɔːr-íːm] and [divr-èː].

- 978-90-04-44885-8

46

decaen and dresher

(16) Application of segmental rule to the forms in (15) a. Absolute b. Construct Output of (15) (da baːʀ)+(íːm)## (da (ba ʀ+ày)# Tone Lengthening — — Vowel reduction (də baːʀ)+(íːm)## (da (bə ʀ+ày)# Other rules — (dib (ʀ+èː)# Spirantization (də vaːʀ)+(íːm)## (div (ʀ+èː)# Rounding (də vɔːʀ)+(íːm)## — Schwa Realization (da vɔːʀ)+(íːm)## — Phonetic forms [davɔːʀíːm] [divʀèː] We have established that almost all differences between the absolute and construct forms of a noun derive from the fact that the latter do not have the same level of stress as the former. Another way to put this is that a construct forms a single prosodic word with a following word. If we did not have a vocalized text of the Bible, then from the phonology alone we would be inclined to indicate that a construct is part of the same prosodic word as the word that follows it. We might, for example, leave spaces around a prosodic word and connect a construct to a following word with a hyphen, as in (17). (17) Indication of prosodic dependency by a hyphen a. Absolute b. Construct dɔːvɔ́ ːʀ ṭóːv davaʀ=ham-mɛ́ːlɛχ word.ms.abs good.ms.abs word.ms.cstr=the-king.ms.abs ‘good word’ ‘the king’s word’ We actually do have such a text, but, as we mentioned at the outset, the maqqef does not always appear where we would put it, based on the phonology of the construct. This sets up a tension between the prosody we expect based on morphosyntax, and the prosody indicated in the Tiberian transcription. We will show that even if the intention of the Masoretes was to indicate every construct form with a maqqef, the constraints of their system would have prevented this outcome. In order to understand why this is, we need to review some basics of the Tiberian prosodic hierarchy and the rules of cliticization that are tightly bound up with the phrasing.

- 978-90-04-44885-8

47

prosodic dependency in tiberian hebrew

3

The Tiberian Prosodic Hierarchy

Prosodic representation mediates the relationship between phonology and syntax. On this view, a prosodic hierarchy organizes domains in which phonological rules operate (Selkirk 1984, 1986; Nespor & Vogel 1986; Hayes 1989). From the word level up, the units of the prosodic hierarchy are commonly supposed to have at least the levels in (18a): (18) The prosodic hierarchy a. Contemporary theory Utterance

U

Intonational Phrase

I

Phonological Phrase

P

Prosodic Word (w/clitics) pw

b. The Tiberian hierarchy Verse Hierarchy of phonological phrases

V D0 D1 D2 D3

Prosodic Word (w/clitics) pw

The Tiberian transcription marks the bottom and top of the hierarchy very systematically (18b). Between the Utterance and the Word, however, the Tiberian transcription departs from the contemporary understanding of the prosodic hierarchy. Rather than a Phonological Phrase and an Intonational Phrase, the Tiberian transcription parses each verse into a hierarchy of phrases, D0–D3, where D0 is the highest level and D3 is the lowest. The Tiberian notation distinguishes two types of accents: conjunctive and disjunctive. A conjunctive accent C on a word indicates that the word is part of the same phrase as the word that follows it. A disjunctive accent Di indicates that a word is final in its phrase. A phrase that ends in a disjunctive accent and which contains no other disjunctive accents is a Minimal Phrase (mp; Strauss 2009). We can identify this mp with the Phonological Phrase, P. In the example in (19), the first word vajjillɔːḥamúː ‘fought’ has a conjunctive accent and forms a minimal phrase with vaneː=jahuːðɔ́ : ‘the men of Judah’. The third prosodic word, biːʀuːšɔːláːjim ‘against Jerusalem’, makes a second phrase.

- 978-90-04-44885-8

48

decaen and dresher

(19) Conjunctive and disjunctive accents C D2 D1 (va-j-jillɔːḥam-úː vaneː=jahuːðɔ́ :) (b-iːʀuːšɔːláːjim) and-then-3m.npst-fight-p the.men.mp.cstr= against-Jerusalem Judah.ms.abs ‘The men of Judah fought against Jerusalem’ (Judg 1.8) The mp forms the domain for three phonological rules: spirantization, gemination, and rhythmic stress shift (nasiʁɔː). In the first phrase in (19), the initial consonant of the second word, vaneː, is spirantized from underlying /b/ because it follows a vowel that ends the preceding word in the same mp. By contrast, the initial /b/ of biːʀuːšɔːláːjim is not spirantized, though it also follows a word-final vowel, because the preceding word is not in the same mp. Recall that the disjunctive accents form a hierarchy with four levels. The hierarchy of disjunctives indicates that Tiberian phonological phrases are nested, so that a phrase with accent of level Di is divided by a phrase ending in accent Di+1. In example (19), the second disjunctive, D1, terminates a non-minimal phrase comprising all three words. This non-minimal phrase is divided by accent D2. The prosodic structure can be represented as a tree, where a phrase ending in a disjunctive Di is itself labelled Di. Here, the inner phrase is labelled D2, and the entire phrase is a D1. (20) Tree representation of the partial verse in (19) D1 D2 C D2 (va-j-jillɔːḥam-úː vaneː=jahuːðɔ́ :) and-then-3m.npst-fight-p the.men.mp.cstr= Judah.ms.abs

D1 (b-iːʀuːšɔːláːjim) against-Jerusalem

Why does the phrase in (20) end in D1? Recall that the top of the hierarchy is labelled D0. The three prosodic words in (20) form the beginning of a verse; the phrasing of the complete verse is shown in (21). The verse has ten prosodic words, labelled pw1–pw10. There are seven mp s, indicated by parentheses (). These mp s can be equated with the Phonological Phrase and serve as the domain of the three phonological rules mentioned above. The higher-level phrases are not associated with phonological rules but indicate how the P-phrases are organized. The verse is divided into two parts by

- 978-90-04-44885-8

49

prosodic dependency in tiberian hebrew

D0 accents. The largest break comes after pw7, which ends the first half-verse. There is a maximum of two D0 accents in a verse, and every verse ends in a D0 accent. Short verses may lack a second D0. The first half-verse consists of five mp s. The main division comes after the second mp (pw3). Since the whole half-verse ends in D0, it is divided by a D1 accent on pw3. This D1 phrase is in turn divided by a D2 accent on pw2. This is the three-word phrase we looked at earlier in (20). This organization is important in governing a series of phrasal simplification rules: in certain prosodic conditions, two or more mps may be combined into one. It also crucially interacts with cliticization, which is sensitive to position in the prosodic tree. (21) Tree representation of Judg 1.8

and- the.men= against- and. om. andfought of.Judah J’lem cap’d it put.it

to=the. and-om= they. onsword the-city set fire

‘The men of Judah fought against Jerusalem and captured it and put it to the sword; and they set the city on fire.’6

4

The Tiberian Prosodic Hierarchy and the Rules of Cliticization

In this section we review some principles governing cliticization in Tiberian Hebrew, as set out by Breuer (1982), Dresher (2009b), and Holmstedt & Dresher (2013). We can begin with the accusative particle, which takes the form ʔɛθ= when it is a clitic, and ʔéːθ when it stands as an independent prosodic word. The lengthened vowel is due to the aforementioned rule of Tone Lengthening

6 We give our own translation, since neither jps nor niv follow the mt in putting the major break after pw7.

- 978-90-04-44885-8

50

decaen and dresher

under the main stress of a prosodic word. This particle is one of the most easily cliticizable morphemes, and in the majority of cases it is attached by maqqef to the following word. However, there are various cases where cliticization does not occur. First, there is a very strong constraint that the half-verse, which ends with a D0 accent, should consist of at least two phrases. In some verses, the main division is such that one of the half-verses contains only two words, one of which is a small cliticizable word. In such a case, the small word almost always remains an independent word in its own phrase, marked with a disjunctive accent, as in the example in (22). (22) Suspension of cliticization of om ʔéːθ in a two-word D0 D0 D1 D0 pw pw (va-ʔéːθ) (baθu:ʔéːl) and-om Bethuel.ms.abs ‘and Bethuel’ (Gen 22:22) Another constraint is that a long word does not easily coexist with another word in a mp governed by D0.7 Thus, a small word is generally not cliticized to a long word in a D0 phrase but again is placed in its own phrase with a disjunctive accent. This phenomenon is illustrated by the verses in (23): ʔéːθ is normally cliticized before a long word, for example, haggiʀgɔːší: in (23a), but it remains prosodically independent before the same word in a D0 phrase (23b). (23) Cliticization of om ʔéːθ in D1 and suspension of cliticization in D0 a. ʔéːθ cliticized before a long word D1

pw (va-ʔɛːθ=hag-girgɔːší:) and-om=the-Girgashites.ms.abs ‘and the Girgashites’ (Gen 15:21) 7 A long word is one which has at least two vowels before the stressed vowel, not counting reduced vowels, or else contains a long vowel in a closed syllable or before a shewa (often marked with metheg); see Wickes (1887: 62n4), Breuer (1982: xvi), and Dresher (2013).

- 978-90-04-44885-8

51

prosodic dependency in tiberian hebrew

b. Cliticization suspended in a D0 phrase D0 D1 D0 pw pw (va-ʔéːθ) (hag-girgɔːší:) and-om the-Girgashites.ms.abs ‘and the Girgashites’ (Gen 10:16) In general, the conditions on cliticization are very restrictive in prominent prosodic positions (in the domain of D0) and become more liberal as one proceeds down the prosodic hierarchy. Thus, returning to construct forms, it follows that though we may want to cliticize them all the time, in many positions cliticization would violate the phrasing rules. We can illustrate this point with some verses from the book of Esther, though similar examples occur all through the Bible. In Esth 1:17 (24), the construct ðavaʀ is in the domain of D2, that is, not in a prominent prosodic position. Therefore, cliticization can proceed even though it creates a long prosodic word, and the preceding pw is assigned a conjunctive accent to complete the two-word mp. (24) Cliticization of construct ðavaʀ in the domain of a D2 accent D2 C D2 pw pw (kiː-yeːṣéː ðavaʀ=ham-malk-ɔ́ ː) for-3ms.npst-go.out the.conduct.ms.cstr=the-queen-fs.abs ‘for the queen’s conduct will become known’ (Esth 1:17 niv) In Esth 1:12 (25), the construct bi-ðváːʀ is in the domain of a D1 accent. If it were cliticized with the following word it would create a pw that is too long, so the phrasing of the previous verse is not allowed here.

- 978-90-04-44885-8

52

decaen and dresher

(25) Suspension of cliticization of construct bi-ðəvar in the domain of a D1 accent

D1 D2 C pw pw (lɔː-vóː) bi-ðváːʀ to-come.inf at-command.ms.cstr ‘to come at the king’s command’ (Esth 1:12 jps)

D1 pw ham-mɛ́ːlɛχ) the-king-ms.abs

Esth 1:19 (26) exemplifies a different complication: the combination ‘Persia and Media’ is typically kept together as a single pw, leaving the construct ba-ðɔːθ-eː to form a second pw on its own. Note the retracted secondary stress in fɔ̀ ːʀas-, the result of treating fɔ̀ ːʀas=u-mɔːðáːj as a single pw, even though fɔ̀ ːʀas is in the absolute state. The initial [f] is due to spirantization from /p/ following a vowel-final word in the same mp. (26) Suspension of the cliticization of a construct before an already cliticized phrase D0 D1 D0 pw pw (ba-ðɔːθ-éː fɔ̀ ːʀas=u-mɔːðáːj) in-laws-cstr.mp Persia.ms.abs=and-Media.ms.abs ‘into the laws of Persia and Media’ (Esth 1:19 jps)

5

Small Nouns

We will start our survey of small nouns with dɔːm ‘blood’ and jɔːð ‘hand’. They behave just like we would expect, based on what we observed from nouns like dɔːvɔ́ ːʀ. The absolute forms are always free and have the long vowel [ɔː]. The construct forms tend to be cliticized, except when prevented by the phrasing rules; their vowel is [a] when cliticized and unaccented, or [aː] when free and accented.

- 978-90-04-44885-8

53

prosodic dependency in tiberian hebrew

(27) Absolute and construct forms of dɔːm ‘blood’ a. Absolute Free Bound dɔ́ ːm — b. Construct Free (dáːm hɔː-ʔɔːšɔ́ ːm) D0 blood.ms.cstr the-guilt.offering.ms.ab ‘the blood of the guilt offering’ (Lev 14:17)

Bound dam=ziːvḥ-íː) D0 blood.ms.cstr=sacrifice.ms.abs-1s ‘the blood of my sacrifice’ (Exod 23:18)

(28) Absolute and construct forms of yɔːð ‘hand’ a. Absolute Free Bound yɔ́ ːð — b. Construct Free (ʕal=jáːð joːséːf ) D1 on-hand.fs.cstr Joseph.ms.abs ‘on Joseph’s hand’ (Gen 41:42)

Bound jað=paʀʕóː) D1 hand.fs.cstr=Pharaoh.ms.abs ‘Pharaoh’s authority’ (Gen 41:35)

Other small nouns do not all behave like this; we will continue with nouns with stem vowel [oː] whose suffixed forms have [u]. Consider kóːl ‘all’, which occurs over 5,000 times.8 We assume that the underlying form of the vowel is short /o/.9 Parallel to /a/ in (14), we expect it to be lengthened to [oː] by Tone Lengthening when absolute. Assuming that Tone Lengthening does not apply in the construct, the underlying vowel should remain short until very late in the derivation; thus, it is expected to surface as [ɔ], the surface reflex of short /o/, or as [ɔ́ ː], by late (phonetic) lengthening. That is, we expect the vowel in /kol/ to parallel the [ɔː] ~ [a(ː)] alternation in (27) and (28), appearing as [kóːl] when absolute and as [kɔl=] or [kɔ́ ːl] in construct, independent of maqqef.

8 There are 5,194 instances according to Even-Shoshan (1990), plus another 100 or so Aramaic instances, which pattern in the same way. A search of morph yields 5,201 instances. 9 This o is itself derived from /u/ which is still visible in suffixed forms, e.g. kullɔ́ ːm ‘all of them’ (Rendsburg 2103).

- 978-90-04-44885-8

54

decaen and dresher

In fact, the absolute forms are always free and have a long vowel [oː] as expected (29a), but the construct forms depend on the phrasing: with five exceptions, they are short [ɔ] when cliticized with maqqef, but long [oː] when prosodically free (accented) (29b). That is, koːl behaves like the accusative particle ʔeːθ in being sensitive to phrasing. Of course, morphosyntactically ʔeːθ is always the same. Perhaps koːl is itself becoming a grammatical particle. (29) Absolute and construct forms of Roːl ‘all’ a. Absolute Free Bound kóːl 11810 — 0 b. Construct Free kóːl 460 kɔ́ ːl 412

Bound kol= 1 (Jer 33:8)11 kɔl= 4,344

Another frequent small noun in this class is ʀoːv ‘multitude’. In the absolute state it is as expected, prosodically free and with a long vowel [ó:]. When in construct and prosodically free, the vowel is always long, following the prosody like koːl does. When bound by maqqef, the construct is usually short [ɔ], again like kɔl-. But there are relatively more (3/15) [o] vowels with maqqef. (30) Absolute and construct forms of Roːv ‘multitude’ a. Absolute Free Bound ʀóːv 5913 — 0

10

11 12

13

This is the number of absolute forms listed by Even-Shoshan (1990). morph lists 396 instances coded as absolute, a very significant discrepancy that appears to point to a major disagreement in how to classify forms as absolute or construct. However, the two sources agree that we find [kóːl] when free and [kɔl=] when cliticized, leaving aside the 5 exceptions. This form is doubly anomalous in being written plene with a vav. According to Joüon & Muraoka (2006: 54n2), the 4 instances of construct kɔ́ ːl without maqqef occur in Ps 35:10, 87:7 (but bhs has maqqef ), Isa 40:12, and Prov 19:7. morph and bhs have [laχɔl] in Ps 119:96 with no maqqef, but also no accent. This is the number we find in Even-Shoshan (1990). Once again morph classifies more of the free forms as absolute, with 67 absolute and 69 construct. How these forms are classified does not affect our conclusion that prosody is the main determinant of the vowel in this word.

- 978-90-04-44885-8

55

prosodic dependency in tiberian hebrew

b. Construct Free ʀóːv 76 ʀɔ́ ːv 0

Bound ʀov= 3 (Isa 1:11, Job 37:23, Lam 1:5) ʀɔv= 12

ħóːq ‘statute’ in absolute is mostly free and has a long [ó:], with two exceptions: Job 26:10 has ħoq=ħɔ́ ːʁ ‘horizon=mark. pst.3ms’ (‘He marks out the horizon’ niv), with maqqef in stress clash; and Ps. 148:6 has ħɔq=nɔːθáːn ‘statute=give .pst.3ms’ (‘he issued a decree’ (niv)).14 The construct is mostly bound with maqqef and has a short vowel [ɔ]. There is also one free construct form with [ó:]. (31) Absolute and construct forms of ḥoːq ‘statute’ a. Absolute Free Bound ħóːq 21 ħoq= 1 (Job 26:10) ħɔ́ ːq 0 ħɔq= 1 (Ps 148:6) b. Construct Free ħóːq 1 (Ezek 45:14) ħɔ́ ːq 0

Bound ħoq= 0 ħɔq= 14

Finally, we will consider some small nouns with stem vowel [eː]. The normal absolute of léːv ‘heart’ is free with a long [éː]. Surprisingly, there are 3 bound absolute forms with a short [ɛ], all in stress clash: lɛv=ṭóːv, lɛv=rɔ́ ːʕ. In the construct, there are 13 forms with short [ɛ], all in clash (lɛv=ʔíːš, lɛv=mɛ́ːlɛːχ), and 2 forms with long vowels when not in clash. For this word, then, stress clash is the best predictor of vowel length. (32) Absolute and construct forms of leːv ‘heart’ a. Absolute Free Bound léːv 142 lev= 1 (Ps 51:19) lɛ́ːv 0 lɛv= 2 (Prov 17:16, 26:23) (stress clash)

14

This form is classified by both Even-Shoshan (1990) and morph as a construct, but it is not clear what the rationale for this is (apart from having the vowel ɔ).

- 978-90-04-44885-8

56

decaen and dresher

b. Construct Free léːv 86 lɛ́ːv 0

Bound lev= 11 (not in stress clash) lɛv= 14 (all in stress clash)

Similarly, šéːm ‘name’ has a long vowel [eː] except for 6 cases with [ɛ] where the construct is bound and in a stress clash. (33) Absolute and construct forms of šeːm ‘name’ a. Absolute Free Bound šéːm 53 šeːm= 0 b. Construct Free šéːm 233 šɛ́ːm 0

Bound šeːm= 5415 šɛm= 6 (all in stress clash)16

Finally, beːn ‘son’ has [eː] in the absolute and [ɛ] in the construct, whether these forms are free or bound. An example of a bound absolute form is béːn=šiššíː ‘a sixth son’ (Gen 30:19). It has a long [eː] despite being bound with maqqef. (34) Absolute and construct forms of beːn ‘son’ a. Absolute Free Bound béːn beːn= b. Construct Free bɛ́ːn

Bound bɛn=

To sum up, some small nouns follow the morphosyntax, some follow the prosody, and some have more complicated behaviours. This is an area where we might expect to find that manuscripts differ, because the lack of clear principles would make these forms particularly prone to copying errors.

15 16

Some forms have a metheg indicating that the vowel is long and has some stress; thus, some of these forms should be represented as šéːm=. Gen 16:15, 21:3, 1 Sam 8:2, 1 Kgs 16:24, Ezek 39:16, Prov 30:4.

- 978-90-04-44885-8

prosodic dependency in tiberian hebrew

6

57

Conclusion

To conclude, we have argued that the reason for the divergence between construct phonology and the prosody is to be found in basic principles of Tiberian phrasing, which force many construct forms to be independent prosodic words. In earlier work (Dresher 1994; DeCaen 2005, 2009) we have also argued that the Tiberian phrasing is not a made-up system, but has properties found in other prosodic systems, and appears to reflect an actual reading tradition grounded in natural speech (Revell 1980). If something like Tiberian phrasing was operative at the time when construct phonology emerged, could it be that all construct forms at one time followed the prosody the way some small nouns do in the Tiberian text? That is, the construct of dɔːvɔ́ ːʀ would have been davaʀ when actually prosodically dependent on a following word, but it would have been dɔːvɔ́ ːʀ when the phrasing prevented its cliticization. If this line of thinking is correct, it would suggest that the behaviour of some of the small nouns is not simply an innovation that is a reaction to a chaotic system but might point back to a time when all construct forms alternated depending on their prosodic representation.

References Balcaen, M. Jean. 1995. The prosody of Tiberian Hebrew. Master’s thesis, University of Saskatchewan, Saskatoon. Balcaen, M. Jean. 2000. Prosody and syntax in Tiberian Hebrew. Ms., University of Toronto. bhs = K. Elliger & W. Rudolph (eds.) 1977. Biblia Hebraica Stuttgartensia. Stuttgart: Deutsche Bibelgesellschaft. Blake, Frank. 1951. Pretonic vowels in Hebrew. Journal of Near Eastern Studies 10: 243– 255. Blau, Joshua. 2010. Phonology and morphology of Biblical Hebrew. Winona Lake, IN: Eisenbrauns. Breuer, Mordecai. 1982. ‫[ טעמי המקרא בכ״א ספרים וספרי אמ״ת‬The Biblical accents in the twenty-one books and in the three books]. Jerusalem: Mikhlala. DeCaen, Vincent. 2005. On the distribution of major and minor pause in Tiberian Hebrew in the light of the variants of the second person independent pronouns. Journal of Semitic Studies 50: 321–327. DeCaen, Vincent. 2009. Theme and variation in Psalm 111: Phrase and foot in generativemetrical perspective. Journal of Semitic Studies 54: 81–109.

- 978-90-04-44885-8

58

decaen and dresher

Dresher, B. Elan. 1994. The prosodic basis of the Tiberian Hebrew system of accents. Language 70(1): 1–52. Dresher, B. Elan. 2009a. Stress assignment in Tiberian Hebrew. In Contemporary views on architecture and representations in phonology, ed. by Charles Cairns & Eric Raimy, 213–224. Cambridge, MA: mit Press. Dresher, B. Elan. 2009b. The word in Tiberian Hebrew. In The nature of the word: Essays in honor of Paul Kiparsky, ed. by Kristen Hanson & Sharon Inkelas, 95–111. Cambridge, MA: mit Press. Dresher, B. Elan. 2013. Biblical accents: Prosody. In Khan (2013a). Edzard, Lutz. 2013. Phonology, Generative. In Khan (2013a). Even-Shoshan, Abraham (ed.). 1990. ‫קונקורדנציה חדשׁה לתורה נביאים וכתובים׃ אוצר לשׁון‬ ‫ שׁמות פרתיים צרופים ונרדפים‬,‫ מלים‬,‫[ המקרא ־ עברית וארמית שׁרשׁים‬A new concordance of the Bible: thesaurus of the language of the Bible—Hebrew and Aramaic roots, words, proper names, phrases and synonyms]. Jerusalem: Kiryat Sefer. Halle, Morris & William J. Idsardi. 1995. General properties of stress and metrical structure. In The handbook of phonological theory, ed. by John Goldsmith, 403–443. Cambridge, MA: Blackwell. Halle, Morris & Jean-Roger Vergnaud. 1987. An essay on stress. Cambridge, MA: mit Press. Hayes, Bruce. 1989. The prosodic hierarchy in meter. In Rhythm and meter, ed. by Paul Kiparsky & Gilbert Youmans, 201–260. Orlando, FL: Academic Press. Hayes, Bruce. 1995. Metrical stress theory: Principles and case studies. Chicago: University of Chicago Press. Holmstedt, Robert D. & B. Elan Dresher. 2013. Clitics: Pre-Modern Hebrew. In Khan (2013a). Idsardi, William J. 1992. The computation of prosody. Doctoral dissertation, mit, Cambridge, MA. Joüon, Paul & Takamitsu Muraoka. 2006. A grammar of Biblical Hebrew. Rev. ed. Rome: Pontifical Biblical Institute Press. jps = Tanakh: A new translation of the Holy Scriptures according to the traditional Hebrew text. 1988. Philadelphia: Jewish Publication Society. Khan, Geoffrey. 1987. Vowel length and syllable structure in the Tiberian tradition of Biblical Hebrew. Journal of Semitic Studies 32(1): 23–82. Khan, Geoffrey. 2013a. (ed.) Encyclopedia of Hebrew language and linguistics. BrillOnline Reference Works. https://referenceworks.brillonline.com/browse/encyclopedia ‑of‑hebrew‑language‑and‑linguistics. Khan, Geoffrey. 2013b. Tiberian reading tradition. In Khan (2013a). Liberman, Mark & Alan Prince. 1977. On stress and linguistic rhythm. Linguistic Inquiry 8(2): 249–336. Malone, Joseph. 1993. Tiberian Hebrew phonology. Winona Lake, IN: Eisenbrauns.

- 978-90-04-44885-8

prosodic dependency in tiberian hebrew

59

morph = The Westminster Hebrew Morphology Database (morph 4.20). 2016. Glenside, PA: The J. Alan Groves Center for Advanced Biblical Research. Nespor, Marina & Irene Vogel. 1986. Prosodic phonology. Dordrecht: Foris. niv = The Holy Bible, New International Version. 2011. Grand Rapids: Zondervan House. Prince, Alan S. 1975. The phonology and morphology of Tiberian Hebrew. Doctoral dissertation, mit, Cambridge, MA. Rappaport, Malka. 1984. Issues in the phonology of Tiberian Hebrew. Doctoral dissertation, mit, Cambridge, MA. Rendsburg, Gary A. 2013. Phonology: Biblical Hebrew. In Khan (2013a). Revell, E.J. 1980. Pausal forms in Biblical Hebrew: Their function, origin and significance. Journal of Semitic Studies 25(2): 165–179. Selkirk, Elisabeth O. 1984. Phonology and syntax: The relation between sound and structure. Cambridge, MA: mit Press. Selkirk, Elisabeth O. 1986. On derived domains in sentence phonology. Phonology Yearbook 3: 371–405. Strauss, Tobie. 2009. ‫[ השפעת גורמים פרוזודיים וגורמים אחרים על חלוקת טעמי כ״א ספרים‬The effects of prosodic and other factors on the parsing of the biblical text by the accents of the 21 books]. Doctoral dissertation, The Hebrew University of Jerusalem. Wickes, William. 1887. Two treatises on the accentuation of the Old Testament. Reprinted with a prolegomenon by Aron Dotan. New York, NY: Ktav, 1970.

- 978-90-04-44885-8

chapter 4

Ordinals in Biblical Hebrew Susan Rothstein and Adina Moshavi

1

Introduction*

Ordinal numerals (e.g., first, second, third) are adjectives which express the property of having a certain position in a (partial) order relation, in contrast to cardinal numerals (e.g., one, two, three), which are used to count pluralities. Ordinals and cardinals would seem to have non-overlapping semantic and syntactic uses. In Biblical Hebrew, however, cardinals are used to express ordinal meaning in a variety of syntactic constructions. In this paper we examine the ways in which cardinal constructions express ordinality in bh temporal expressions. Section 2 presents a description of the biblical data, and section 3 presents formal semantic analyses which account for the ordinal effect of the biblical constructions, while respecting their respective syntax. Conclusions are presented in section 4.

2

The Biblical Data

2.1 Hebrew Lexical Ordinals as Compared to Cardinals The Hebrew lexical ordinals exist only up to 10, while cardinals can be combined to express a number of any size. The ordinals are adjectives, while the cardinals appear to be nouns. The ordinals are derived from cardinals, with the added -i ending and a variety of morphophonemic changes, as illustrated in (1). Ordinals exhibit standard gender marking: masculine ordinals have zeromarking, and feminine ordinals have the characteristically-feminine ending -t. * This research was funded in part by Israel Science Foundation award 1345/13 to Susan Rothstein. The translations in the biblical citations below are our own, based primarily on consultation of nrsv and jps. Susan Rothstein tragically passed away before the final version of this paper was prepared. Adina Moshavi is extraordinarily grateful to Galia Hatav, who gave generously of her time in composing explanations and elaborations of the technical formulas in the second section of the paper. Publication of this paper would not have been possible without her help. I am grateful to the participants of the workshop for their helpful comments. Thanks also go to Elisheva Jeffay who constructed and coded the data base used to prepare this paper.

© Susan Rothstein and Adina Moshavi, 2021 | doi:10.1163/9789004448858_005

- 978-90-04-44885-8

61

ordinals in biblical hebrew

This pattern is in stark contrast with the “reversed” gender marking of the cardinals (from 2 to 19), in which the feminine forms have zero-marking and the masculine forms are marked by the -a ending. (1) Ordinals a. šəliši (m)/ šəlišit (f) ‘third’ b. šišši (m)/ šiššit (f) ‘sixth’ c. təšiʿi (m)/ təšiʿit (f) ‘ninth’

Cardinals šəloša (m)/ šaloš (f) ‘three’ šišša (m)/ šeš (f) ‘six’ tišʿa (m)/ tešaʿ (f) ‘nine’

The syntactic properties of ordinal and cardinal numerals in bh are in sharp contrast in several respects, as detailed in the subsections below. 2.1.1 Word Order Ordinal numerals occur after the modified noun, like other attributive adjectives (2), while cardinal numerals usually precede the noun in appositional structures (3). (2) ben šišši son.ms sixth.ms ‘a sixth son’ (Gen 30:19) (3) šišša ḇan-im six.m son-mp ‘six sons’ (Gen 30:20) 2.1.2 Syntactic Number of the Modified Noun Ordinals nearly always modify a singular noun, as in (2), above, while cardinals over 1 normally modify plural nouns (3). This rule applies to cardinal numerals over ten as well, as shown in (4). When certain common nouns, such as ʾiš ‘man’, yom ‘day’ and šana ‘year’ are involved (jm §142e), cardinals over ten modify singular rather than plural nouns (5). (4) meʾa ṣimmuq-im hundred.f cluster.of.raisins-mp ‘one hundred clusters of raisins’ (1Sam 25:18) (5) meʾa ʾiš hundred.f man-ms ‘one hundred men’ (Judg 20:35)

- 978-90-04-44885-8

62

rothstein and moshavi

2.1.3 Definiteness of the Numeral Like other attributive adjectives, ordinals agree with the noun in definiteness, as illustrated by the contrast between the indefinite and definite np s in (6)– (7). Cardinals, in contrast, do not take the definite article, even when the noun is definite. Cardinals appear in the absolute state before an indefinite noun (8) and in the construct state before a definite noun (9). (6) paʿam ḥamišiṯ time.fs fifth.fs ‘a fifth time’ (Neh 6:5) (7) bap=paʿam haš=šəḇiʿiṯ in.the=time.fs the=seventh.fs ‘at the seventh time’ (Josh 6:16) (8) ʾarbaʿa məlaḵ-im four.m king-mp ‘four kings’ (Gen 14:9) (9) ḥămešɛṯ ham=məlaḵ-im five.m.cstr the-king-mp ‘the five kings’ (Josh 10:16) 2.2 Ordinal and Cardinal Temporal Phrases Ordinal temporal phrases in bh most often express the temporal location of an event or state. A temporal location is an interval occupying a position on a timeline counted off from an initial reference point; for example, on the third day of the week locates an event during the third 24-hour period counted off from the beginning of the week. The reference point can be evoked by the preceding context (10) or expressed by a lə ‘to’ phrase, as in ‘to the king Rehoboam(’s reign)’ (11): (10) u=ḇaš=šana ha=rəḇiʿiṯ yihyɛ and=in.the=year.fs the=fourth.fs be.impf.3ms kol=piry=o qoḏɛš hillul-im la-Y’ all.cstr=fruit.cstr=his holy.object.ms praise-mp to.the-lord ‘and in the fourth year (the fruit of the tree) shall be holy for jubilation in the Lord’ (Lev 19:24)

- 978-90-04-44885-8

63

ordinals in biblical hebrew

(11) wayhi baš=šana ha=ḥămišiṯ lam=mɛlɛḵ and.be.impf.3ms in.the=year.fs the=fifth.fs to.the=king.ms rəḥaḇʿam Rehoboam ‘And it came about in the fifth year to (the reign of) the king Rehoboam’ (1Kgs 14:25) In (10) the initial reference point is the end of a three-year duration invoked in the previous clause, starting from the time the tree is planted, during which time eating the fruit is forbidden. Example (11) contains a common formula in which the relevant interval begins at the time of the ascent of the king to his throne. Temporal phrases with cardinals, in contrast, usually express duration, that is, a multiple of intervals such as days or years: (12) ki šešɛṯ yam-im ʿaśa Y’ ʾɛt haš=šamay-im for six.ms.cstr day-mp make.perf.3ms lord om the-heavens-mp wə=ʾɛt ha=ʾarɛṣ and=om the=earth.ms ‘for (in) six days the lord made the heavens and the earth’ (Exod 31:17) (13) wayaʿǎḇoḏ ʿod šɛḇaʿ šan-im ʾăḥer-oṯ and.serve.impf.3ms more seven.f year-fp other-fp ‘And (he) served him further (for) another seven years’ (Gen 29:30) The cardinal expression in (12) denotes the time interval over which God created the world, namely, six days;1 similarly, the cardinal expression in (13) denotes the duration of Jacob’s service to Laban: seven years. Ordinal and cardinal temporal phrases exhibit the same syntactic contrasts which characterize ordinal and cardinal phrases in general; i.e., in the definite ordinal temporal phrase the numeral follows a singular noun and has the definite article, while in the definite cardinal temporal phrase the numeral precedes a plural noun and is in the construct form, without the article, as in the near-minimal pair in (14) and (15). Although the ordinal temporal phrase ordinarily exhibits definiteness agreement (14), the noun yom ‘day’ often lacks the article in definite ordinal phrases (jm §138b), in what Borg (2000) terms the 1 This phrase violates a syntactic rule that cardinals appear in the construct form only before definite np s. As discussed in Moshavi and Rothstein (2019), this construction is common in temporal adverbials, as well as in measure expressions.

- 978-90-04-44885-8

64

rothstein and moshavi

“pseudo-construct” construction, as in (16) and (17). The possibility of using of the object marker ʾɛṯ with this construction (17), shows that the phrase is nonetheless definite, as the object marker only appears with syntactically definite noun phrases.2 (14) bay=yom haš=šəliši in.the=day.ms the=third.ms ‘on the third day’ (Gen 22:4) (15) šəlošɛt hay=yam-im three.m.cstr the=day-mp ‘the three days’ (1Sam 9:20) (16) ʿad yom haš=šəliši until day.ms the=third.ms ‘until the third day’ (Lev. 19:6) (17) wayḇarɛḵ ʾɛ̌lohim ʾɛṯ yom haš=šəḇiʿi and.bless.impf.3ms God om day.ms the=seventh.ms ‘And God blessed the seventh day’ (Gen 2:3) In addition to these contrasts, additional syntactic generalizations can be made that distinguish most ordinal and cardinal temporal phrases. Ordinal temporal phrases are nearly always definite, while cardinal temporal phrases are overwhelmingly indefinite, as illustrated by the near-minimal pair in (18)–(19); the definite cardinal phrase in (15), above, is unusual.3 Ordinal temporal phrases are most frequently governed by a preposition (usually bə- ‘in’, but also min ‘from’, ʿad ‘until’, etc.), while cardinal temporal phrases are usually bare np s, as (18)–(19) illustrate.4

2 According to Borg, phrases like yom haš=šəḇiʿi and other phrases with an indefinite noun modified by a definite adjective have undergone a certain degree of lexicalization and are becoming compounds. Although the pseudo-construct looks like a construct phrase, Borg points out that the initial noun is usually in absolute rather than construct form (e.g., 1Kgs 7:12). 3 Nonetheless, definite cardinal temporal phrases are possible, as shown by the li=šəloš haš= šan-im ‘for the three years’ (Lev 25:21), with a definite article preceding the noun šan-im ‘years’. 4 In li=šəloš haš=šan-im ‘for the three years’ (Lev 25:21) the cardinal phrase is governed by a preposition (see also note 3, above).

- 978-90-04-44885-8

ordinals in biblical hebrew

65

(18) ba=ḥoḏɛš haš=šəliši in.the=month.ms the=third.ms ‘in the third month’ (Exod 19:1) (19) šəloša ḥŏḏaš-im three.ms month-mp ‘three months’ (2Sam 6:11) To sum up, ordinal and cardinal phrases in general, and ordinal and cardinal temporal phrases in particular, seem to have mostly non-overlapping semantic interpretations, internal syntax, and syntactic contexts of occurrence. In spite these differences, however, cardinals are used to express ordinal meaning in a number of syntactic constructions. 2.2

Cardinal Constructions Expressing Ordinality in Temporal Phrases in bh As mentioned above, lexical ordinals exist in Hebrew, both Biblical and modern, only up to the number 10, making it necessary to find a different way of expressing larger ordinals. In modern Hebrew, ordinalities over 10 are expressed by substituting the corresponding cardinal numeral in the standard ordinal construction, with the usual definiteness agreement (compare ha=yom ha=šiši ‘the sixth day’), as shown in (20). This construction, however, is extremely rare in bh. The only indisputable occurrence, either temporal or non-temporal, appears to be (21). Example (22) is another possible occurrence, although this is not the only possible syntactic interpretation of this verse. (20) ba=yom ha=šiša=asar on.the=day.ms the=six.m=teen.m ‘on the 16th day’ (Modern Hebrew) (21) u=ḇaš=šana ha=ʾaḥaṯ ʿɛśre and=in.the=year.fs the=one.f teen.f ‘and in the eleventh year’ (1Kgs 6:38) (22) aḏ yom ha=ʾɛḥaḏ wə=ʿɛśr-im la=ḥoḏɛš until day.ms the=one.m and=twenty-cp to.the=month.ms ‘until the twenty-first day of the month’ (Exod 12:18)

- 978-90-04-44885-8

66

rothstein and moshavi

The numerical phrase in (22) is taken here as a pseudo-construct phrase, in the pattern of ʿaḏ yom haš=šəḇiʿi ‘the seventh day’ (16, above), with the definite numeral ha=ʾɛḥaḏ wə=ʿɛśr-im ‘twenty-one’ standing in for an ordinal.5 In every other case of an ordinality over ten in bh, ordinality is expressed by one of several constructions involving cardinal numerals, as pointed out by Steiner (1997) Some of these constructions are also used to express ordinalities under 10. None appear to be used for non-temporal ordinality.6 In section 3 we discuss the special properties of the temporal realm which enable it to take advantage of cardinals for the purpose of expressing ordinality. In this paper we analyze the three most common cardinal constructions expressing temporal ordinality, focusing on occurrences with yom ‘day’ and šana ‘year’, the two most common temporal lexemes. The three constructions will be referred to as the calendrical, reduplicative, and pseudo-durational constructions, as defined below. 2.2.1 The Calendrical Construction In the calendrical construction, as shown in (23)–(26), the temporal noun is the head of a construct phrase and has a bare cardinal is its annex; nearly all occurrences end with a lə phrase. The cardinal in this construction appears to designate a unique abstract numeric entity. When used in this way, numerals are similar to proper names, which denote a particular individual (Rothstein 2017: 25). The calendrical construction designates a temporal location that is calculated by counting off a particular number of units, such as years, from the beginning of the duration designated by the lə phrase, e.g., the beginning of a king’s reign. šənaṯ here acts as a counter that combines with a number to produce the equivalent of English Year 18. This construction seems to presuppose a calendar and a systematic method of dating. Perhaps for this reason the calendrical construction is attested with numerals under ten as well, as in (26), despite the availability of a lexical ordinal construction.

5 An alternative interpretation is to take yom as a real construct (the construct and absolute forms of this noun are identical). In this case the phrase would be an example of an extremely rare construction expressing ordinality with a construct noun followed by a definite cardinal numeral, exemplified by Num 33:38 bi=šnaṯ ha=ʾarbaʿim ‘in the fortieth year’. 6 In fact, almost all ordinality over ten in the Bible is temporal. There are non-temporal ordinals above 10 in 1Chr (12:14, 24:13–18, 25:18–31, 27:14–15), but in none of these examples is the ordinal used explicitly as an adjectival modifier.

- 978-90-04-44885-8

ordinals in biblical hebrew

67

bə-[noun.cstr cardinal] (lə- np) (23) u=ḇi=šnaṯ šəmonɛ ʿɛśre lam=mɛlɛḵ yaroḇʿam and=in=year.fs.cstr eight.f teen.f to.the=king.ms Jeroboam bɛn nəḇaṭ son.ms.cstr Nebat ‘and in the 18th year of the king Jeroboam son of Nebat’ (1 Kgs 15:1) (24) gam bə=yom ʾarbaʿa ʿaśar lə=ḥoḏɛš ʾăḏar also in=day.ms.cstr four.m teen.m to=month.ms.cstr Adar ‘also on the 14th day of the month Adar’ (Esth 9:15) (25) u=ḇi=šnaṯ šəmonɛ ʿɛśre lə=molḵ=o in=year.fs.cstr eight.f teen.f to=reign.inf.cstr=his ‘and in the 18th year of his reign’ (2Chr 34:8) (26) bi=šnaṯ šaloš lə=ʾasa mɛlɛḵ yəhuḏa in=year.fs.cstr three.f to=Asa king.ms.cstr Judah ‘in the third year of (the reign of) Asa king of Judah’ (1 Kgs 15:28) Comparing the calendrical construction to the corresponding ordinal construction bay=yom ha-ʿăśiri ‘on the third day’, we see that the two constructions have matching word order, with the temporal noun preceding the numeral. Rather than consisting of a noun phrase followed by adjective phrase, however, the calendrical construction is a construct phrase. Interestingly, the numeral in the calendrical construction always has the same gender as the temporal noun, even though there is normally no rule of gender agreement between construct and annex in a construct phrase. Assuming that the numeral in the calendrical construction designates an abstract number, an intrinsically definite entity, the calendrical construction should be considered syntactically definite, since the annex normally determines the definiteness of the entire phrase. If this is the case, the calendrical construction matches the corresponding ordinal construction in definiteness as well. 2.2.2 The Reduplicative Construction In the reduplicative construction, as shown in (27) and (28), below, the preposition bə- modifies a construct phrase which has a temporal noun as the construct and an appositional phrase of the form ‘cardinal—noun’ (featuring the same noun) as the annex. The temporal noun occurs in the construct form before the cardinal numeral, and again in the absolute form, following the numeral. A prepositional phrase with lə-, designating the reference point from

- 978-90-04-44885-8

68

rothstein and moshavi

which the temporal intervals are to be counted off, normally ends the phrase, although it may be absent in phrases with yom ‘day’. It is unclear whether this final pp modifies the entire construct phrase or only the appositional phrase. The lə- phrase supplies the initial reference point for the duration to be counted off from. (27), for example, has the literal rendering ‘in the year of 26 years to Asa king of Judah’, conveying the ordinal meaning ‘the 26th year of Asa’s reign’. Another example of the reduplicative construction, shown in (28), has the literal rendering ‘the year of 600 years to Noah’, conveying the ordinal meaning ‘the 600th year of Noah’s life’. bə=[noun.cstr [cardinal noun]] (lə= np) (27) bi=šnaṯ ʿɛśr-im wa=šeš šana lə=ʾasa mɛlɛḵ in=year.fs.cstr twenty-cp and=six year.fs to=Asa king.ms.cstr yəhuḏa Judah ‘in the 26th year of Asa king of Judah(’s reign)’ (1 Kgs 16:8) (28) bi=šnaṯ šeš meʾoṯ šana lə=ḥayye noaḥ in=year.fs.cstr six.f hundred.fp year.fs to=life.mp.cstr Noah ‘in the 600th year of Noah’s life’ (Gen 7:11) The reduplicative construction in (27) can be compared to the semantically similar phrase with a lexical ordinal in (11), repeated here as (29). The semantic equivalence of the reduplicative construction and the standard lexical ordinal construction is also clear from the excerpted numerical phrases from Num 7, shown in (30)–(32). The chiefs’ offerings on the 1st–10th days are expressed using ordinals, exemplified by (30), switching to the reduplicative cardinal construction for the 11th day (31) and the 12th day (32): (29) wayhi bas̆=s̆ana ha=ḥămišiṯ lam=mɛlɛḵ Rehoboam and.be.impf.3ms in.the=year.fs the=fifth.fs to.the=king.ms Rehoboam ‘And it came about in the 5th year of King Rehoboam(’s rule)’ (1 Kgs 14:25) (30) bay=yom ha=ʿăśiri in.the-day.ms the=tenth.ms ‘on the tenth day’ (Num 7:66) (31) bə=yom ʿašte ʿaśar yom on-day.ms.cstr one.m teen.m day.ms ‘on the 11th day’ (Num 7:72)

- 978-90-04-44885-8

ordinals in biblical hebrew

69

(32) bə=yom šənem ʿaśar yom on=day.ms.cstr two.m teen.m day.ms ‘on the 12th day’ (Num 7:78) As compared to the ordinal temporal construction bay=yom ha-ʿăśiri ‘on the third day’, the reduplicative cardinal construction displays both syntactic similarities and contrasts. The most obvious contrast is that the ordinal construction contains a noun phrase and modifying adjectival phrase, while the cardinal construction is a construct phrase. The cardinal construction, furthermore, is indefinite, like other cardinal durative phrases, contrasting with the normally-definite ordinal temporal construction. Nevertheless, there is a point of similarity: in both the ordinal and the reduplicative construction a temporal noun precedes the numeral (in the reduplicative construction, as we have seen the noun is repeated after the numeral as well.) 2.3.3 The Pseudo-Durational Construction In the pseudo-durational construction, as illustrated in (33–35), the preposition bə- governs what appears to be a durational phrase. Nevertheless, the construction expresses ordinal meaning, designating the last interval in the specified duration; thus (35) denotes the 14th year of Hezekiah’s reign. Pseudo-durational structures with yom nearly all include a lə- phrase, while occurrences with šana often do not, with the initial reference point derived contextually. The ordinal effect of this construction is clear not only from the contexts in which it occurs, but also from the fact that duration phrases are not ordinarily modified by bə-, nor followed by a lə- phrase designating a reference point.7 bə= [cardinal noun] lə=np yom la=ḥoḏɛš ʿaśar (33) bə=šiḇʿa in=seven.m teen.ms day.ms to.the=month.ms ‘on the 17th day of the month’ (Gen 8:4) šana meʾoṯ (34) bə=ʾaḥaṯ wə=šeš in=one.fs and=six.fs hundred.fp year.fs ‘in the 601st year (of Noah’s life)’ (Gen 8:13)

7 An exception to the first generalization is the durative bə=šɛḇaʿ šəney haś=śaḇaʿ ‘in the seven years of plenty’ (Gen 41:34), with the meaning ‘throughout the seven-year duration of the plenty’.

- 978-90-04-44885-8

70

rothstein and moshavi

(35) u=ḇə=ʾarbaʿ ʿɛśre šana lam=mɛlɛḵ ḥizqiyya and=in=four.f teen.fs year.fs to.the=king.ms Hezekiah ‘in the 14th year of the king Hezekiah(’s reign)’ (2 Kgs 18:13) The pseudo-durational differs from both the reduplicative and the bi=šnaṯ X constructions in that the temporal noun follows the numeral, just as in the corresponding cardinal construction. It is clearly indefinite, like the reduplicative construction and unlike the bi=šnaṯ X construction.

3

Semantics for Cardinal Constructions Expressing Ordinality in bh

3.1 A Semantic Account of Ordinal Numerals The following semantic account of ordinal numerals is a (very basic) version of the one given in Rothstein (2017a). Explanations follow each formula. Ordinal numerals such as first, second etc. are adjectives which express the property of having a certain position in a contextually relevant order. The ordinal numeral expresses how many elements precede it in the order. By way of illustration, let us say there were three people who gave a talk in a certain conference in the following order: John, then Mary, and then Lucy. These three people form the set {John, Mary, Lucy} and the order relation between them is as follows: ‘John spoke before Mary and Lucy’ and ‘Mary spoke before Lucy’. We attribute the property ‘first’ to John because no one preceded him, the property ‘second’ to Mary because one person preceded her and the property ‘third’ to Lucy because two people preceded her. More generally, we can say that: (36) Given a set and an order relation, a. first denotes the property x has if nothing precedes it in the order. b. second denotes the property x has if 2–1 (i.e. 1) entities precede it in the order. c. third denotes the property x has if 3–1 (i.e. 2) entities precede it in the order etc. We can express this formally as follows: (37) nth: λnλPλRλx.P(x) ∧ |{y: P(y) ∧ R(y,x) }| = n – 1

- 978-90-04-44885-8

71

ordinals in biblical hebrew

This formula gives the denotation of any ordinal numeral: take a property P (e.g. speaker), an individual x and a set of entities y, all of which have the property P, and a relation of order R, such that R says that every y precedes x with respect to property P. For x to have the property of being nth it has to be the case that the cardinality of the set that has y(s) as its member(s), i.e., the number of the individual(s) represented by ‘y’ that have the property P and that precede(s) x (with respect to that property) should be n-1. Let us substitute the number 3 for ‘n’ in the formula. In such case, we will get the property ‘third’, whose denotation is as follows: (38) third: λPλRλx.P(x) ∧ |{y: P(y) ∧ R(y,x) }| = 3 – 1 Since we substituted ‘3’ for n in formula (37), we eliminate the λn expression. Formula (38) says the following: take any property P and any individual x that has that property and any individual(s) y that has/have that property, such that the order relation R between x and y determines that y precedes x. The number of individuals represented by y is 3 – 1 = 2. Let us now substitute ‘(be a) speaker’ for the property P in formula (38): (39) third speaker: λx.λR.speaker(x) ∧ |{y: speaker(y) ∧ R(y,x) }| = 3 – 1 Since we substituted ‘speaker’ for P, we eliminate the λP expression. The new formula says: take any individual x and a set of individual(s) y, such that all individuals x and y have the property of being a speaker. In addition, the relation R determines that the y(s) spoke before x. The number of the individuals y that spoke before x is 2. Ordering relations often but not always involve temporality. An example which does not is: third prize = the prize awarded for third highest achievement (i.e., there are two higher achievers). Thus it is possible to say The third prize will be awarded first, where first, but not third, refers to temporal order. 3.2 A Semantic Account of Lexical Ordinals in Biblical Hebrew We assume that the above semantics is applicable to lexical ordinals in bh. Take (20) as an example, referring to a row of stones on the high priest’s breastplate:

- 978-90-04-44885-8

72

rothstein and moshavi

(40) wə=haṭ=ṭur haš=šəliši and=the=row.ms the=third.ms ‘and the third row’ (Exod 28:19) Applying formula (37) above, where property P is ‘(be a) row’ and the ordering relation is ‘spatial ordering’, we derive the following denotation: (41) ṭur šəliši ‘third row’: λx.row(x) ∧ | {y: row(y) ∧ R (y,x)} | = 3 – 1 R(y,x): y is spatially ordered above x In other words, if we take a row x and a set of rows y and relation R specifying that y is spatially ordered above x, the third row is the row that has two rows before it in the spatial ordering of rows on the breast plate. An additional example is (42), referring to a son born by Leah to Jacob: (42) ben šišši son.ms sixth.ms ‘a sixth son’ (Gen 30:19) Applying formula (37) again and specifying property P as ‘(be a) son’, the denotation of (42) is as follows: (43) ben šišši ‘a sixth son’: λx.son(x,jacob) ∧ | {y: son(y,jacob) ∧ R(y,x)} | = 6–1 R(y,x): y is born before x In other words, taking the relation R specifying that y is born before x, ben šišši ‘a sixth son’ is a son that came sixth in the order of birth. Ordinality in the temporal domain involves the relation of temporal precedence. In (44), for example, the ordinal expression functions as a temporalframe adverbial locating the event of the gift offering of the prince of the Gadites. (44) bay=y0m haš=šišši in.the=day.ms the=sixth.ms ‘on the sixth day’ (Num 7:42) We assume that nouns expressing units of time denote sets of contextuallyrestricted non-overlapping intervals. For example, day denotes a set of intervals that do not overlap; this is expressed formally as {i: i ∈ 1 day}. In other words,

- 978-90-04-44885-8

73

ordinals in biblical hebrew

the entities of this set are intervals i that are members of the set (= unit of time) ‘1 day’. These are naturally ordered by temporal precedence, i.e. interval i’ precedes interval i’’, and i’’ precedes i’’’, etc. These intervals can but need not form a continuous stretch of time. We can now derive the denotation of the temporal expression hay=yom haš=šišši ‘the sixth day’: (45) hay=yom haš=šišši ‘the sixth day’: σday(i) ∧ |{i’: day(i’) ∧ i’ < i}| = 6–1 σ symbolizes the sum of the subsets of a certain set. In our case, the set day has intervals i as its subsets. σday(i) therefore means the sum of the intervals that compose together the countable temporal unit day. The formula in (45) specifies that, given the set day consisting of the collection of the intervals represented by i and the set of intervals i’ such that each i’ is a day that precedes i, the number of the members in the latter set is 6–1 = 5. In other words, hayyōm haš-šiššī ‘the sixth day’ is the day which is sixth in the order of temporal precedence. The temporal noun phrase hay=yom haš=šišši ‘the sixth day’ is governed by the preposition bə- ‘in’ in (44), above. Temporal phrases governed by bə- function as adverbials which locate the time of the event, corresponding to pp s with at or in in English, e.g., at five o’clock, in February. We treat prepositions like these as functions from nps denoting intervals into the events whose running times are parts of those intervals. When a preposition applies to an ordinal temporal noun phrase, it yields an expression that denotes an event that takes place sometime during the interval designated by the temporal phrase. The complete denotation of (44) can therefore be given as follows: (46) bay=yom haš=šišši ‘on the sixth day’: λe.τ(e) ⊆ σday(i) ∧ |{i’: day(i’) ∧ i’ < i}| = 6–1 In other words, the ordinal expression denotes the set of events whose running time is included in a day such that there are 6–1 = 5 days preceding that day. 3.3 Ordinality Above Ten We have seen that lexical ordinal constructions express precedence in a contextually-relevant ordering relation, with the relation introduced by the ordinal adjective. Since Hebrew has no ordinal adjectives above ten, larger ordinal numbers must be expressed in some other way. As discussed in section 2, in Modern Hebrew, a cardinal greater than ten, such as šiša=asar ‘six=teen’, can be used as a post-nominal ordinal adjective, with ordinary ordinal semantics (repeated from [20]):

- 978-90-04-44885-8

74

rothstein and moshavi

(47) ba=yom ha=šiša=asar on.the=day.ms the=six=teen.m ‘on the 16th day’ In bh, in contrast, ordinality above ten is most frequently expressed periphrastically by one of the three constructions discussed in section 2. Since almost all ordinality using numerals above 10 is temporal ordinality, these constructions do not need to specify an ordinal relation, but can use the implicit temporal precedence relation which the use of time language, including words like šana and yom, presupposes. In deriving a semantics for these three constructions, the challenge is to achieve the ordinal effect without explicitly introducing an ordinal relation, and also respect the syntactic structure of the construction. Both of these goals can be achieved making use of the semantic concept of the scale, as explained in the next subsection. 3.4 A (Very Basic) Semantics for Temporal Scales A scale is a triple ⟨D, ℝ, U⟩, involving the following parameters: D is a continuous dimension, ℝ is the set of real numbers, and U is a set of units. Scales in the temporal domain involve the temporal dimension (or time-line), a set of units such as day, year, or month, which divides the time line into units, and the set of numerals used to number the units. This gives us the temporal scale ⟨time, ℝ, unit⟩. In the formulas below, the parameter time is represented by the property ‘dur’ (for duration), the parameter ℝ by the number ‘n’, and the parameter unit by ‘i’. Duration expressions like six days denote functions from numbers into sets of intervals with a particular duration; e.g. six days denotes periods of six days. This can be represented formally as: (48) six days: λnλi.dur = ⟨n, day⟩ (6) = λi.dur = ⟨6, day⟩ The formula λnλi.dur takes any number n and any interval i, and yields a duration ⟨n, i⟩. If we substitute day for i, we get the expression ⟨n, day⟩. If we substitute 6 for n, we get ⟨6, day⟩, an interval i that has a duration of six days. This enables us to derive the semantics for sentences like (29), where six days denotes an interval of time which expresses the duration of the event that the sentence is about, i.e. a property of its running time. (49) Mary studied ( for) six days: ∃e[study(e) ∧ ag(e) = mary ∧ τ(e) ∈ six days]

- 978-90-04-44885-8

ordinals in biblical hebrew

75

∃e means that there was some event e. The text within the square brackets is a description of that event: it is an event of studying such that the agent (ag) of the event is mary and the duration of that specific event, represented as τ(e), was six days, which represents the expression ⟨6, day⟩. In other words, “There was an event of Mary’s studying, and its running time was an interval of six days.” We can derive a similar interpretation for the sentence in (50). The interpretation is shown in (51):

(50) u=ḇən-e=ḵɛm yihyu roʿ-im bam=miḏbar and=child-mp=yours be.impf.3mp shepherd-mp in.the=desert.ms ʾarbaʿim šana forty.cp year.fs ‘And your children will be shepherds in the desert forty years’ (Num 14:33) (51) ∃e[be-shepherd-in-the-desert(e) ∧ exp(e) = your children ∧ τ(e) ∈ 40 years] This formula means that there will be a state of your children being shepherds in the desert, and the duration of that state will be an interval of 40 years. 3.5 Back to Cardinal Constructions Expressing Ordinality The invented expressions in (52)–(54) illustrate three ways to express the meaning ‘in the 20th year of Hezekiah’s rule’, based on the definitions of the calendrical, reduplicative and pseudo-durative constructions presented in section 2. The semantics presented below are based on these expressions, which contain the absolute form šana ‘year’ and/or the construct šənaṯ ‘year of’. The semantics for cardinal temporal phrases with yom ‘day’, which does not have a distinct construct form, can be derived in a similar fashion. (52) Calendrical construction: bə- [noun.cstr cardinal] (lə- np) bi=šnaṯ ʿɛśr-im lam=mɛlɛḵ ḥizqiyya in=year.fs.cstr twenty-cp to.the=king.ms Hezekiah lit. ‘in the year twenty to the king Hezekiah’ (53) Reduplicative construction: bə- [noun.cstr [cardinal year]] (lə- np) bi=šnaṯ ʿɛśr-im šana lam=mɛlɛḵ ḥizqiyya in=year.fs.cstr twenty-cp year.fs to.the=king.ms Hezekiah lit. ‘in the year of twenty years to the king Hezekiah’

- 978-90-04-44885-8

76

rothstein and moshavi

(54) Pseudo-durational construction: bə- [cardinal noun] lə- np ḇə=ʿɛśr-im šana lam=mɛlɛḵ ḥizqiyya in=twenty-cp year.fs to.the=king.ms Hezekiah lit. ‘in twenty years to the king Hezekiah’ There are two ways to group these structures: the calendrical and reduplicative constructions can be categorized together because they both feature the construct form šənaṯ, while the pseudo-durational construction involves the absolute form šana. Alternatively, the reduplicative and pseudo-durational constructions can be categorized together because they both appear to involve durational phases (ʿɛśrim šana ‘20 years’), while the calendrical construction does not. We will show that both groupings are relevant. We assume, following Dobrovie-Sorin (2000), Heller (2002), Rothstein (2012, 2017b) and Doron (2014), that construct morphology expresses a shift from a sortal to a relational meaning. A sortal noun has a meaning that does not involve another entity (e.g., woman), while the meaning of a relational meaning involves another entity (e.g., wife, whose meaning necessarily involves a spouse). For example, while absolute bayiṯ ‘house’ denotes a set of houses, the construct bet in the construct phrase beṯ ham=mɛlɛḵ ‘house of the king’ denotes a set of houses that stand in the relation possessed-by to the king. In this account the denotation of the absolute form of a noun is distinguished from those of the construct form, which in turn has a multiplicity of potential denotations depending on the relation between construct and annex in the construct phrase. For example, bet1 ‘house belonging to’ occurs in phrases like beṯ ham=mɛlɛḵ ‘the house of the king’, bet2 ‘house made of’ occurs in phrases like beṯ ha=ʿeṣ ‘the house of wood’; etc.8 Applying this approach to the cardinal temporal constructions, we treat the construct form šənaṯ in the reduplicative and calendrical constructions as a relational noun whose denotation is distinct from absolute šana and depends on its semantic relation with the annex. Since the annex is a durational phrase in the reduplicative construction and a numeral in the calendrical construction, we do not expect šənaṯ to denote the same relation in each phrase. We shall assume that the reduplicative and pseudo-durational constructions both build the ordinal meaning on the durative phrase, for reasons explained below.

8 An advantage of ascribing the relational meaning to the construct noun rather than the annex noun is that it is the construct noun which exhibits morphological change as compared to the absolute noun forms.

- 978-90-04-44885-8

ordinals in biblical hebrew

77

3.5.1 Semantics for the Calendrical Construction We start with the calendrical construction, repeated in (55). šənaṯ here is an operator that combines directly with a numeral annex (ʿɛśr-im ‘twenty’), and with a pp (lam=mɛlɛḵ ḥizqiyya ‘to the king Hezekiah’). The lə- phrase maps the event of the reign of the king onto the interval which is its running time, i.e., the years of his reign. The result is an expression that denotes the 20th year in the interval of the king’s reign. (55) ḇi=šənaṯ ʿɛśr-im lam=mɛlɛḵ ḥizqiyya in=year.fs.cstr twenty-cp to.the=king.ms Hezekiah lit. ‘in the year twenty to the king Hezekiah’ We call the construct noun in this construction šənaṯc (c for calendrical). šənaṯc is a relational noun at type ⟨n, ⟨i, ⟨i, t⟩⟩⟩, which means that it is a function that combines with an interval i’ (e.g., the duration of the king’s reign) and a numeral n, and yields an expression which denotes year n in the interval i’, in our case, the 20th year of the interval which is the running time of the reign of King Hezekiah. The formal derivation of šənaṯc is given in (56): (56) šənaṯc: λnλi’’λi’λi.dur(i) ≤ 1 year ∧ i ⊆ i’ ∧ dur(max (i’’ : i’’ ⊆ i’ ∧ i’’< i)) = n-1 years This formula defines the interval denoted by an expression with šənaṯc: take any number n and any interval i, such that the duration of i is up to one year long (i.e. one year or less). Take also i’ and i’’, such that i is a subinterval of i’ and i’’ is a subinterval of i’ and precedes i. The maximum duration of i’’ is then n-1 years. To illustrate how this works, let us take the phrase šənaṯcʿɛśr-im ‘year twenty’: (57) šənaṯc ʿɛśr-im: λi’λi’’λi.dur(i) ≤ 1 year ∧ i ⊆ i’ ∧ dur(max (i’’ : i’’ ⊆ i’ ∧ i’’< i)) = 20-1 years This formula says if we take the number 20 and the set of intervals i such that the duration of i is up to one year long (i.e. one year or less), and take also two intervals i’ and i’’, such that i is a subinterval of i’ and i’’ is a subinterval of i’ and precedes i, the maximum duration of i’’ is 19 years. Let us now consider the complete calendrical construction, taking i’ to be the time of the reign of King Hezekiah:

- 978-90-04-44885-8

78

rothstein and moshavi

(58) ḇi=šnaṯ ʿɛśr-im l=am=mɛlɛḵ ḥizqiyya: λi λi’’.dur(i) ≤ 1 year ∧ i ⊆ τ(king hezekiah’s reign) ∧ dur(max (i’’ : i’’ ⊆ τ(king hezekiah’s reign) ∧ i’’< i)) = 20-1 years This specifies that if we take the number 20 and the set of intervals i which are up to a year long and which are part of the reign of King Hezekiah, the maximal interval which precedes i during the reign of King Hezekiah is 20-1, i.e. 19 years. This is a singleton set, i.e. the construct phrase identifies a unique year period. bə- maps this unique interval onto the set of events that took place during this interval, and thus the whole phrase has the semantics of a temporal frame adverbial. 3.5.2 Semantics for the Reduplicative Construction The reduplicative construction is repeated in (58): (59) ḇi=šnaṯ ʿɛśr-im šana lam=mɛlɛḵ ḥizqiyya in=year.fs.cstr twenty-cp year.fs to.the=king.ms Hezekiah lit. ‘in the year of twenty years to the king Hezekiah’ = ‘in the twentieth year of King Hezekiah’s rule’ The construct šənaṯ here takes what appears to be a durational phrase as its complement; we term this šənaṯr (for reduplicative). There is a parallel between the reduplicative construction and superlative construct phrases, in which the same noun appears as the construct and as the annex: qoḏaš-im (60) qoḏɛš holiness.ms.cstr holiness-mp ‘most holy of holy objects’ (Exod 29:37) (61) mɛlɛḵ məlaḵ-im king.ms.cstr king-mp ‘most kingly of kings’ (Ezek 26:7) (62) šir haš-šir-im song.ms.cstr the-song-mp ‘the most outstanding of songs’ (Song 1.1) This construction expresses superlativity in terms of salience, i.e., the highest position on a scale with respect to a class or set of like entities. For example, a holy object even among holy objects is the highest on the scale of holiness; an

- 978-90-04-44885-8

79

ordinals in biblical hebrew

outstanding king among kings (Nebuchadnezzar) is the highest on the scale of kings (in terms of power, greatness, etc.) We propose that the reduplicative construction expresses ordinality in the same way. We take as a presupposition that the salient year in a continuing period is the last one, as it is highest on the temporal scale, i.e. furthest from the starting point. Therefore, ḇi=šnaṯ ʿɛśr-im šana lam=mɛlɛḵ ḥizqiyya lit. ‘in the year of twenty years of King Hezekiah’s rule’ expresses the year that is salient, i.e. highest on the scale of years, in a period of 20 years counting from the beginning of the reign of King Hezekiah. As above, lə- maps (the event of) King Hezekiah’s reign onto its running time. In the reduplicative construction šənaṯr is a relational noun applying to an interval np (the latter denoted by the duration phrase n šana) and a pp. It applies to a period of n years which begins at the same point as the interval denoted by the pp and yields the most salient (= last) year in the period. We can define šənaṯr formally as follows: (63) šənaṯr: λi’λi’’λi.dur(i) = 1 year ∧ initial(i’)=initial(i’’) ∧ i ⊆salient i’ According to this formula šənaṯr is a relation between three intervals which yields the salient interval year in an interval of a specified time denoted by the annex whose starting point is fixed by the starting point of the interval given by the pp. Let us now substitute 20 for n: (64) šənaṯ ʿɛśr-im šana: λi’’λi.∃i’[dur(i) = 1 year ∧ i’ ∈ 20 years ∧ initial(i’)= initial(i’’) ∧ i ⊆salient i’]9 Adding the pp we get the following: (65) šənaṯ ʿɛśr-im šana l=am=mɛlɛḵ ḥizqiyya: λi.∃i’[dur(i) = 1 year ∧ i’ ∈ 20 years ∧ initial(i’)=initial(τ(king hezekiah’S reign)) ∧ i ⊆salient i’] This formula specifies a year interval which is the salient (= last) part of a 26year interval which has the same starting point as the interval of the reign of King Hezekiah. Again, this is a singleton set, i.e. a unique interval. The preposition bə- maps this interval onto the set of events that took place dur-

9 The annex ʿɛśr-im šana is interpreted via semantic incorporation; for details see Rothstein (2012).

- 978-90-04-44885-8

80

rothstein and moshavi

ing this interval, yielding a frame adverbial. Temporal ordinality (in the nth year/day) is expressed by picking out a specific, unique interval in the temporal order. 3.5.3 Semantics for the Pseudo-Durational Construction The pseudo-durational construction is repeated in (66): (66) ḇə=ʿɛśr-im šana lam=mɛlɛḵ ḥizqiyya in=twenty-cp year.fs to.the=king.ms Hezekiah lit. ‘in the twenty years to the king Hezekiah’ Although the syntactic structure of ʿɛśr-im šana ‘twenty years’ matches that of a duration phrase, it clearly has ordinal meaning in this construction, i.e., it refers to the 20th year of the king’s reign, rather than to a 20-year interval. How is ordinal meaning derived from what appears to be a durational construction? There are two possibilities: a) ʿɛśr-im šana isn’t a durational phrase, but has the same semantics as the calendrical construction, i.e., it is equivalent to šənaṯ ʿɛśr-im. b) The pseudo-durational construction is essentially similar to the reduplicative construction, with a null operator rather than the explicit construct šənaṯ picking out the last year in a designated interval. Against the first alternative, we note that the construction cardinal šana has a durational interpretation in all its other occurrences. Furthermore, this possibility involves interpreting šana as a relational noun equivalent to šənaṯc despite the fact that it does not have construct morphology; there is no precedent for such a shift.10 We therefore assume that Construction C involves an implicit operator equivalent to šənaṯc in the reduplicative construction, which maps an interval onto the salient subpart of that interval. A formal definition of the null operator is given in (67): (67) operator: λi’λi’’λi. initial(i’)=initial(i’’) ∧ i ⊆salient i’

10

An additional argument against the first possibility is an occurrence of the pseudodurative construction, which involves a repetition of the temporal noun: ḇi-šmonim šana wə-ʾarbaʿ meʾoṯ šana lit. ‘in 80 years and 400 years’ = ‘in the 480th year’ (1Kgs 6:1). If šana had the semantics of šənaṯc we would expect it to combine directly with the numeral šəmonim wə-ʾarbaʿ meʾoṯ ‘480’.

- 978-90-04-44885-8

81

ordinals in biblical hebrew

We can now derive the same results as for the reduplicative construction: (68) [operator] + ʿɛśr-im šana: λi’’λi. ∃i’[i’ ∈ 20 years ∧ initial(i’)=initial(i’’) ∧ i ⊆salient i’ Adding the pp yields: (69) [operator]+ ʿɛśr-im šana l=am=mɛlɛḵ ḥizqiyya: λi. ∃i’[i’ ∈ 20 years ∧ initial(i’)=initial(τ(king hezekiah’S reign) ∧ i ⊆salient i’] The formula in (69) defines the meaning of the pseudo-durational construction as ‘the salient (final) interval in the period of 20 years which began with the beginning of Hezekiah’s reign’.

4

Conclusions

In this paper we have made the following claims: A. Ordinality at 10 and below is expressed through lexical ordinals which introduce a contextually determined order and assign an individual position in the order. B. Ordinal effects in the temporal domain can be expressed indirectly via a temporal scale. Three constructions are frequently used for this purpose: 1. The calendrical construction: the construct head denotes an operator which maps from a numeral to a numbered unit in a period. 2. The reduplicative construction: ordinality is expressed by highlighting a salient = final unit in an interval. The construct is headed by an operator which maps from an interval to a salient, final subinterval. 3. The pseudo-durational construction has the same semantics as the reduplicative construction, but the operator is implicit and does not make reference to the length of the salient final subinterval. C. We noted the role of the construct form in expressing ordinality. In the calendrical and the reduplicative constructions, the temporal nouns šənaṯ and yom shift to relational interpretations, a recognised semantic correlate of construct morphology. D. The literature often relates ordinals to superlatives (Stateva and Sharvit 2002, Bylinina et al 2015). We note that bh has no grammaticalized superlatives (or comparatives). In general, superlativity is expressed in terms of salience, e.g. by using a definite + adjective.

- 978-90-04-44885-8

82

rothstein and moshavi

(70) wə-ḏawiḏ hu haq-qaṭan and-David he the-young.ms ‘And David was the young one = the youngest’ (1 Sam 17:14) We have seen that an alternative syntactic technique to express superlativity is to use the construct, e.g. mɛlɛḵ məlaḵ-im ‘king of kings’, which expresses superlativity by focusing on an individual which is salient with respect to a comparison class of similar individuals, since it has the relevant property to the highest degree. We suggest that ordinality can be expressed in a similar fashion, whether explicitly, as in the reduplicative construction, or implicitly, as in the pseudo-durational construction.

References Borg, A. 2000. Some observations on the ‘yom ha-šišši’ syndrome in the Hebrew of the Dead Sea Scrolls. Pp. 26–39 in Diggers at the Well; Proceedings of a Third International Symposium on the Hebrew of the Dead Sea Scrolls and Ben Sira, ed. T. Muraoka & J.F. Elwolde. Leiden: Brill. Bylinina, L., Y. Sudo, N. Ivlieva, and A. Podobryaev. 2015. An in situ semantics for ordinals. Pp. 135–145 in nels 45: Proceedings of the Forty-Fifth Annual Meeting of the North East Linguistic Society, vol. 1, ed. Thuy Bui and Deniz Özyildiz. Amherst, MA: glsa. Dobrovie-Sorin, C. 2000. (In)definiteness spread: From Romanian genitives to Hebrew construct state nominals. Pp. 177–226 in Comparative studies in Romanian syntax, ed. V. Motapanyane. Amsterdam: Elsevier. Doron, E. 2014. The interpretation of construct-state morphology. Pp. 361–374 in The form of structure, the structure of form: Essays in honor of Jean Lowenstamm, ed. S. Bendjaballah, N. Faust, M. Lahrouchi and N. Lampitelli. Amsterdam: John Benjamins. Heller, D.. 2002. Possession as a lexical relation: Evidence from the Hebrew construct state. Pp. 127–140 in Proceedings of wccfl 21, ed. L. Mikkelsen and C. Potts. Somerville, MA: Cascadilla Press. jm = Jouon, P. and T. Muraoka. 2009. A Grammar of Biblical Hebrew, 2nd ed. Rome: Pontifical Biblical Institute. Moshavi, A. and S. Rothstein. 2018. Indefinite numerical construct phrases in Biblical Hebrew. jss 63(1): 99–123. Rothstein, S. 2012. Reconsidering the construct state in Modern Hebrew. Italian Journal of Linguistics 24(2): 227–266. Rothstein, S. 2017a. Semantics for counting and measuring. Cambridge: Cambridge University Press. - 978-90-04-44885-8

ordinals in biblical hebrew

83

Rothstein, S. 2017b. Proper names in Modern Hebrew construct phrases. Folia Linguistica 51 (2): 419–451. Sharvit, Y. and P. Stateva. 2002. Superlative expressions, context, and focus. Linguistics and Philosophy 25: 453–505. Screnock, J. 2018. The syntax of cardinal numbers in Judges, Amos Esther and 1qm. jss 63(1): 125–154. Steiner, R.C. 1997. Ancient Hebrew. Pp. 145–173 in The Semitic languages, ed. R. Hetzron. London: Routledge. Waltke, B. and M. O’Connor. 1990. An Introduction to Biblical Hebrew Syntax. Winona Lake, IN: Eisenbrauns.

- 978-90-04-44885-8

chapter 5

Investigating Ellipsis in Biblical Hebrew Robert D. Holmstedt

1

Introduction

In a recent survey of analytical approaches to the phenomenon of ellipsis, Merchant suggests that the ongoing interest in elliptical constructions is due to the break down or absence of the typical mapping of sounds and gestures onto their corresponding meanings: “in ellipsis, there is meaning without form,” he writes (2018: 19; see also Aelbrecht 2010: 1–2). For ellipsis in Biblical Hebrew (bh), there is often an additional motivation: the quest to interpret biblical poetry, which regularly employs ellipsis. Though a great deal of insightful work has been done on the features of ellipsis in bh poetry, some gaps remain (pun intended). This study will survey what we know about bh ellipsis, propose a few types that have not yet been explicitly identified, investigate whether bh ellipsis is constrained to local contexts, and look for the occurrence of ellipsis beyond poetry. Ellipsis in bh has long been noticed by biblical scholars, particularly as a regular feature of poetic verse, though not of course referred to as “ellipsis” or “gapping” until fairly recently.1 Previous to this—from the medieval grammarians to the twentieth century—the phenomenon was variously described as one item “standing for” or “serving for” a missing item (Miller 2003: 252, n. 6; 2007a: 165, n. 2) or a feature of “incomplete parallelism” (Gray 1915: 72–83). Thus, Gray, in his 1915 The Forms of Hebrew Poetry, adduced the poetic verse in (1) as an example of what he characterized as a.b.c.//b’.c’ “incomplete parallelism,” in which the a element, the verb ʾašiḇa, is not repeated. (1) wǝ=ʾašiḇa šop̱ṭ-ay=iḵ kǝ=ḇ-a=riʾšona and-restore.juss.1cs judges-mp=your like=at-the=first wǝ=yoʿăṣ-ay=iḵ kǝ=ḇ-at=tǝḥilla and-counsellors-mp=your like=at-the=beginning ‘I will restore thy judges as at the first, and thy counsellors as at the beginning’ (Isa 1:26a; Gray 1915: 75). 1 For studies that mention ellipsis in bh before Miller 2003, 2005, 2007a,b,c, 2008, 2013, see Geller 1979: 299–317; O’Connor 1980: 122–129,401–407; Watson 1984: 303–306.

© Robert D. Holmstedt, 2021 | doi:10.1163/9789004448858_006

- 978-90-04-44885-8

investigating ellipsis in biblical hebrew

85

The ellipsis of the verb in (1) is one of a number of phenomena that have been categorized under the rubric of ellipsis (McShane 2005: 6–7; Merchant 2018: 19–20). The type in (1) is typically referred to as “gapping,” that is, ellipsis in which a constituent (often a finite verb) is elided from the coordinate clause, leaving a “gap” (McShane 2005: 136–142). Common types of ellipsis, with English examples, are gapping (2), vp-ellipsis (3), np-ellipsis (4), sluicing (5), stripping (or bare argument ellipsis) (6), and answer ellipsis (7). (2) John can play the guitar, and Kathy ____ [can play] the piano. (3) John can play the guitar and Kathy can ____ [play the guitar], too. (4) John reads five/these ancient languages, and Kathy reads six/those ____ [ancient languages]. (5) John knows something, but I have no idea what ____ [John knows] (6) John plays the guitar, and Kathy ____ [plays the guitar], too. (7) Who can read six languages? Kathy ____ [can read six languages]. In vp-ellipsis (3), a non-finite vp is elided and, in English at least, the ellipsis is introduced by an auxiliary like can or the particle to (McShane 2005: 146– 152). In np-ellipsis (4) a noun and its modifiers are elided, often leaving behind an overt determiner or quantifier (McShane 2005: 128–134). Sluicing (5) is the elision of everything following a wh-word (McShane 2005: 143–145). Stripping (6) is similar to gapping, except that only one main constituent of the coordinate clause is left after the elision of the relevant material (McShane 2005: 143). And finally, in answer ellipsis (7) all the material in the answer clause is elided except for the constituent that corresponds to the wh-word in the question. The precise linguistic features of bh ellipsis were left largely unaddressed until a series of studies by Miller in the last fifteen years (2003, 2005, 2007a,b,c, 2008, 2013). Miller has focused primarily on verb gapping, like the example in (1) as well as the case in (8), which demonstrates backwards verb gapping, and the example in (9), which illustrates backwards pp gapping: (8) ʾellɛ ḇ=a=rɛḵɛḇ wǝ=ʾellɛ ḇ=as=susim wa=ʾănaḥnu these on=the=chariotry and=these on=the=horses and=we

- 978-90-04-44885-8

86

holmstedt

bǝ=šem yhwh ʾǝlohe=nu nazkir on=name.of Yhwh god=our call.impf.1cp ‘Some on chariot(s) ____, and some on horses ____, but we on the name of the lord our God will call.’ (Ps 20:8; Miller 2003: 263) b=aš=šamayim yiśḥaq ʾăḏonay (9) yošeḇ sitting.ms.ptcp in=the=heavens mock.impf.3ms lord yilʿaḡ̱ la=mo ridicule.impf.3ms to=them ‘The one who sits in the heavens will mock ____; the Lord will ridicule them.’ (Ps 2:4; Miller 2008: 105)2 Miller argues that backwards ellipsis like that in (8) and (9), together with a constraint that backwards ellipsis is also from the final position in the second clause, and ellipsis with chiastic word order—v-pp-s, s-pp-[V], as in (10), are distinctive features of bh poetic grammar (2003: 262–265). (10) ki niḇqǝʿu ḇ=am=miḏbar mayim because break.forth.pass.perf.3cp in=the=wilderness water u=nǝḥal-im b=a=ʿăraḇa and=rivers-mp in=the=desert ‘For will break forth in the wilderness waters and rivers in the desert ____.’ (Isa 35:6; Miller 2003: 265) Beyond verb gapping, Miller (2013) also mentions that bh allows the gapping of the existential ʾen ‘there is not’ (11) and yeš ‘there is’ (12) (11) ʾen ʾɛškol lɛ=ʾɛ̆ḵol bikkura ʾiwwǝṯa nap̱̄š=i exst.neg cluster to=eat.inf early.fig crave.perf.3fs soul=my ‘A cluster doesn’t exist to eat, the early figs that I crave ____’ (Mic 7:1)

2 Miller asserts that both Ps 2:4 (backwards) and Ps 132:8 (forwards) exhibit pp ellipsis (2013). However, only Ps 2:4 clearly contains ellipsis (she acknowledges that Ps 132:8 is ambiguous and the non-elliptical analysis is just as likely; 2008: 106). Critically, in Ps 2:4 the pp is likely not an optional adjunct (as Miller describes it; 2013), but a complement of the typically bivalent verb śḥq. Thus, verbal valency is crucial in licensing ellipsis (“since [adjuncts] are not grammatically necessary they are not subject to syntactic ellipsis”; McShane 2005: 15).

- 978-90-04-44885-8

investigating ellipsis in biblical hebrew

87

(12) ki yeš l=ak=kɛsɛp̱̄ moṣaʾ u=maqom l=az=zahaḇ excl exst for=the=silver mine and=place for=the=gold yazoqqu refine.impf.3mp ‘Indeed, a mine exists for silver and a place ____ for gold that they refine’ (Job 28:1) In other studies, Miller covers not-stripping (2005), as in (13), and comparative deletion (2007c), as in (14), (13) ki ḥɛsɛḏ ḥap̱̄aṣti wǝ=loʾ zaḇaḥ because loyalty desire.perf.1cs and=neg sacrifice ‘For loyalty I desire, and not ____ sacrifice’ (Hos 6:6; Miller 2005: 47; 2013)3 (14) nǝg̱ašša ḵa=ʿiwr-im qir grope.juss.1cp like=blind-mp wall ‘we shall grope a wall like blind men ____’ (Isa 59:10) though she has not yet discussed the type of comparative ellipsis that is more typically covered under this category and illustrated by the example in (15). (15) han=nɛḥɛ̆maḏ-im miz=zahaḇ u=mip=paz raḇ the=desirable-mp from=gold and=from=fine.gold much.adj.ms u=mǝṯuq-im mid=dǝḇaš wǝ=nop̱̄ɛṯ ṣup̱̄-im and=sweet-mp from=honey and=honey.of comb-mp ‘[the ordinances of Yhwh, v. 10] are what are desirable more than gold ____ and than much fine gold ____; and they are sweeter than honey ____ and honey of the comb ____’ (Ps 19:11)4 The existence of as-of-yet unaddressed types of ellipsis in bh motivates this study of bh ellipsis and provides the guiding questions: 1) what types of ellipsis does bh exhibit that have not yet been describes?; 2) does bh license non-local 3 See also Gen 24:21 (laḏaʿaṯ hahiṣliăḥ yhvh darko ʾim loʾ); 27:12 (vǝheḇeʾṯi ʿalay qǝlala vǝloʾ ḇǝraḵa); as well as Deut 28:13; Exod 16:4; Num 11:23; Deut 8:2; Judg 2:22. Note that there is not a single bh example similar to the Modern Hebrew pattern of not-stripping, in which the negative follows the subject remnant: dina heḵina ʾet haharṣaʾa šɛlah, ʾaḇal dani loʾ ‘Dina prepared her lecture but not Dani’ (Doron 1999: 135). 4 See also Ps 76:5; 139:6. In prose this type of ellipsis is very common; see, e.g., Gen 3:1 vǝhannaḥaš haya ʿarum mikkol ḥayyaṯ haśśaḏe ʾăšer ʿaśa yhvh ʾǝ̆lohim ‘the serpent was craftier than any field creature that Yhwh God had created ____ (it)’.

- 978-90-04-44885-8

88

holmstedt

ellipsis, and if so, what does it look like?; and 3) is ellipsis a nearly exclusive feature of bh poetry, as is often claimed?

2

Further Diversity of Ellipsis in Biblical Hebrew

What additional kinds of ellipsis occur in bh? Using the list of six common ellipsis types in (2)–(7) as a baseline, we have already seen examples of verbgapping and stripping, but what about, for instance, np-ellipsis. As it turns out, np-ellipsis is difficult to find, for two reasons. First, bh is a pro-drop language, which makes the frequent cases of null subjects better candidates for null anaphora than ellipsis.5 Second, bh uses bare adjectives in abundance, which makes it unclear how to to determine whether this is a type of head ellipsis or the adjective is simply functioning as a noun. Currently the best candidates for np-ellipsis occur with numerals, such as in (16) and (17).6 (16) ko ʾamar yhwh ʿal šǝloša pišʿ-e ḏammɛśɛq thus say.perf.3ms Yhwh because three sin-mp.of Damascus wǝ=ʿal ʾarbaʿa loʾ ʾăšiḇ=ɛnnu and=because four neg turn.caus.impf.1cs=3ms ‘Thus has said Yhwh: Because of three sins of Damascus, because of four ____ I will not turn it back’ (Amos 1:3) (17) ʾulay yaḥsǝrun ḥămišš-im haṣ=ṣaddiq-im ḥămišša perhaps lack.impf.3mp fifty-mp the=righteous-mp five hă=ṯašḥiṯ ba=ḥămišša ʾeṯ kol ha=ʿir interr=destroy.impf.2ms on=five om all.of the=city ‘Perhaps the fifty righteous men lack five ____. Would you destroy the whole city because of five ____?’ (Gen 18:28) Similarly elusive are cases of vp-ellipsis, in contrast to the frequent verbgapping we have already seen. The examples in (18) and (19) seem to be legitimate candidates.

5 Thus, for example, the covert subject in the last clause of Amos 5:18 is more likely to be a case of null anaphora than ellipsis (pace Miller 2013): huʾ ḥošɛḵ vǝloʾ ʾor ‘it is darkness, [it] is not light’. 6 Miller (2013) asserts that bh poetry exhibits np ellipsis, both with objects and adjuncts, but I do not follow how the examples she cites (Ps 97:6, 100:2) exhibit ellipsis.

- 978-90-04-44885-8

investigating ellipsis in biblical hebrew

89

(18) ṣɛḏɛq yalin b=ah wǝ=ʿatta mǝraṣṣǝḥ-im righteousness dwell.impf.3ms in=her and=now murderer-mp ‘Righteousness was dwelling in her / but now murderers ____.’ (Isa 1:21; Miller 2007a: 170) (19) wa=y=yaḥăzeq dawiḏ bi=ḇg̱aḏ-a=w (Qr) and=narr=seize.pret.3ms David in=clothes-mp=3ms wa=y=yiqraʿ=em wǝ=g̱am kol ha=ʾănaš-im ʾăšɛr and=tore.pret.3ms=3mp and=also all.of the=men-mp rel ʾit=to with=3ms ‘David seized his clothes and tore them and also all the men who were with him [tore their own clothes]’ (2Sam 1:11)7 The antecedent in (18) appears to be both the bivalent verb and its pp complement, with the remnant in the ellipsis clause the contrastive subject. In (19), the elided constituent is a conjoined V’ consisting of two verbs and their complements: ‘seized his/their clothes and tore them’. Example (19) highlights the greater complexity sometimes exhibited in ellipsis outside biblical poetry. Given the limited cross-linguistic distribution of vp-ellipsis (Aelbrecht 2010: 13), it is not particularly unusual that bh does not productively utilize it. In fact, since Hebrew lacks auxiliaries to license the vp-ellipsis (McShane 2005: 146–148), more surprising than the lack of vp-ellipsis in bh is the licensing of it in Modern Hebrew (Doron 1999). Similarly difficult to find are clear examples of sluicing in bh. There are zero cases of the interrogative lamma ‘why’ without the full overt content of the interrogative clause or the deictic zɛ ‘this’ following. In fact, in the entire Hebrew Bible I found only two possible examples, both with the interrogative ma ‘what’, given in (20) and (21). (20) wa=ʾăni ʾăḏabber bǝ=ḵa ʾel ʾaḇ=i and=I speak.impf.1cs about=you to father=my wǝ=raʾiṯi ma and=see.perf.irreal.1cs what ‘I will speak about you to my father and I will see what ____ [he speaks about you to me] …’ (1Sam 19:3) 7 See also Ps 15:1–5; 24:3–4. Note that the ellipsis in (19) allows a sloppy identity reading: the men in the second clause could be interpreted as tearing their own clothes (which is likely what was intended) or tearing David’s clothes along with him (unlikely).

- 978-90-04-44885-8

90

holmstedt

(21) raʾiṯi hɛ=hamon hag=gaḏol li=šloăḥ ʾɛṯ see.perf.1cs the=commotion the=great.adj.ms to=send.inf om ʿɛḇɛḏ ham=mɛlɛḵ yoʾaḇ wǝ=ʾɛṯ ʿaḇd=ɛḵa wǝ=loʾ servant.of the=king Joab and=om servant=your and=neg yaḏaʿti ma know.perf.1cs what ‘I saw the great commotion by Joab’s sending off the servant of your king, your servant, but I do not know what ____ [I saw]’ (2 Sam 18:29) On the other hand, stripping like (22) is fairly common (even apart from the specific not-stripping type discussed in Miller 2005). (22) wǝ=ʿatta loʾ ʾattɛm šǝlaḥtɛm ʾoṯ=i henna ki and=now neg you.mp send.perf.2mp om=1cs here but ha=ʾɛ̆lohim the=god ‘Now, you have not sent me here, but God ____’ (Gen. 45:8)8

3

Non-local Ellipsis in Biblical Hebrew

The question about constraints on ellipsis has generated a great deal of discussion in the relevant literature. In his survey of approaches to ellipsis, Merchant addresses constraints on the various types of ellipsis within the question about what licenses ellipsis and what locality conditions might be in effect for the licensor-ellipsis relationship (2018: 22, 26–28, 32–34; see also, among many others, Murguía 2004, Goldberg 2005). Much the evidence examined concerns how wh-words, relative operators, and topicalization create islands interfering with ellipsis. And the different types of ellipsis exhibit different constraints, with gapping perhaps the most restrictive. Thus, while the vp-ellipsis in (23) is acceptable, the gapping from a similarly embedded structure in (24) is not. (23) Susan prepared lunch, and I believe John did too. (24) *Susan prepared lunch, and I think John dinner.

8 See also Gen 32:29; 39:6,9; 42:15; Josh 14:4; 17:3; 1 Sam 21:7; 30:22 (?); 2Sam 21:2; 1Kgs 17:1; 22:18; 2Kgs 4:2; 5:15, 17; 7:10.

- 978-90-04-44885-8

investigating ellipsis in biblical hebrew

91

The issue of locality in bh ellipsis is raised by Miller: she asserts that “in general, an ellipsis site must find its antecedent in the nearest conjunct which meets the other conditions for ellipsis” (2007: 175). She presents only a couple examples as non-local, though she leaves the reader with the idea that nonlocal gapping is grammatical acceptable in bh (2007a: 176–177), as in (25) and (26):9 (25) yǝšawwǝʿû wǝ=ʾen mošiăʿ ʿal yhwh wǝ=loʾ cry.out.impf.3mp and=exst.neg rescue.ptcp.ms to Yhwh and=neg ʿana=m answer.perf.3ms=3mp ‘(a) They cried out /(b) and there was no rescuer /(c) ____ to the lord /(d) and he did not answer them’ (Ps 18:42; Miller 2007a: 176) (26) loʾ ʾiš ʾel w=iḵazzeḇ uḇɛn ʾaḏam neg man god and=lie.impf.3ms and=son.of man wǝ=yiṯnɛḥam and=repent.impf.3ms ‘(a) God is not a man (b) so that he should lie and (c) [not] a human [is God] (d) that he should repent’ (Num 23:19a; Miller 2007a: 176, n. 39) For (25), Miller suggest that “ellipsis takes place across a clause boundary, but the antecedent is found in the nearest clause that meets the other conditions on ellipsis” (2007: 176). That is, the gapped verb in the (c) clause finds its antecedent in the (a) clause, which implies that the antecedent-ellipsis relationship crosses over the (b) clause. And she analyzes (26) as the ellipsis of both the subject and negative of the (a) clause from the (c) clause, again crossing the intervening (b) clause. However, neither of these two particular examples is a straightforward case of non-local ellipsis. In (26), the only way to make contextual sense of the (b) line is to analyze it as a case of successive negative gapping (see Miller 2005: 49–52), i.e., ‘God is not man, (he) does [not] lie’, and then again into the (c) and (d) clauses, ‘(he) is [not] human; (he) does [not] repent’.10 As for (25), it is similar to a number of examples, like (27).

9

10

Miller also lists Isa 45:13 and Zech 9:5, though neither example requires non-local ellipsis; and in Miller 2013, she also adds Isa 59:9, nǝqawwe laʾor wǝhinne ḥošɛḵ ling̱ohoṯ bāʾăp̱eloṯ nǝhalleḵ ‘We wait for light, and behold darkness! / ____ for brightness, (but) in gloom we walk’. A similar example also appears in Deut 32:7, šǝʾal ʾaḇiḵa wǝyaggeḏḵa zǝqenɛḵa wǝyoʾmǝru laḵ. Gapping the negative in Num 23:19 does not match both conditions that Miller argues for

- 978-90-04-44885-8

92

holmstedt

(27) 29wa=y=yoʾmɛr ʾɛ̆lohim hinne naṯatti la=ḵɛm ʾɛṯ and=narr=say.pret.3ms god excl give.perf.1cs to=you om kol ʿeśɛḇ … la=ḵɛm yihyɛ lǝ=ʾoḵla 30u=l=ḵol ḥayyaṯ all.of plant for=you be.impf.3ms for=food and=to=all.of animal.of ha=ʾarɛṣ u=lǝ=ḵol ʿop̄ haš=šamayim … ʾɛṯ kol yɛrɛq the=earth and=to=all.of bird.of skies om all.of foiliage.of ʿeśɛḇ lǝ=ʾoḵla plant for=food ‘God said, “Hey—I am giving to you every plant … (it will be food for you) and to all the land creatures and birds … ____ every foliage of plants as food.”’ (Gen 1:29–30) With (27), the verb in v. 29, natatti ‘I am giving’ (performative perfective), must be elided from the next verse (v. 30) even though there is an imperfective verbal clause, laḵɛm yihyɛ lǝʾoḵla, distancing the ellipsis from its antecedent. But here, as my translation indicates, the nature of intervening material may be the clue to sorting such examples out. As with (27), so also the intervening clause in (25) as well as the one-word clause between the verbal antecedent and gapped verb in (28) can be analyzed as parenthetical. (28) wǝ=hoṣeʾṯɛm ʾɛṯ šǝne=hɛm ʾel šaʿar ha=ʿir and=take.perf.irrealis.2mp om both=them to gate.of the=city ha=hiʾ u=sǝqaltɛm ʾoṯ=am ba=ʾăḇanim the=that and=stone.perf.irrealis.2mp om=3mp with=stone-mp wa=meṯu ʾɛṯ han=naʿăra (Qr) ʿal dǝḇar and=die.perf.irrealis.3cp om the=girl because fact.of ʾăšɛr loʾ ṣoʿăqa ḇ=a=ʿir wǝ=ʾɛṯ ha=ʾiš ʿal rel neg cry.perf.3fs in=the=city and=om the=man because dǝḇar ʾăšɛr ʿinna ʾɛṯ ʾešɛṯ reʿ=ehu fact.of rel humiliate.perf.3ms om wife.of neighbor=his u=ḇiʿarta ha=raʿ miq=qirb=ɛḵa and=burn.perf.irrealis.2ms the=evil from=midst=your ‘You shall take both of them to that city gate and execute them with stones (and they shall die); ____ (om) the girl because she did not cry out negative ellipsis, that the negative must be appear at the front of the first conjunct, and that the constituent structures of both conjuncts must be similar (e.g., sv // sv). Clearly, the second condition is not met in the first two clauses in Num 23:19. And yet, the logic of the clause fails if the negative is not gapped (which perhaps is why Miller translates it as an irrealis result clause).

- 978-90-04-44885-8

investigating ellipsis in biblical hebrew

93

in the city and ____ (om) the man because he humiliated the wife of his neighbor, and you shall purge the evil from your midst’ (Deut 22:24) Why does it matter that the intervening material in parenthetical? It matters because parenthesis, though linearly integrated into its host, is not structurally integrated, and so it creates no barrier to the locality relationship between an ellipsis and its antecedent. To account for the linear but no structural integration of parenthesis, de Vries has made the case in a number of studies for a second Merge operation (2007, 2009, 2012a,b; Griffiths and de Vries 2015), what he calls ‘par-Merge’, which is … a syntactic concatenation operation whose output does not ‘dominate’ its input (in a technical sense) and which is triggered only when one of its inputs is the functional head Par (De Vries 2012a); in effect this creates a new c-command domain, since c-command is defined over dominance relations created by regular merge. The semantic effect of Par can be compared to Potts’ (2005) ‘comma operator’. In [Griffiths and de Vries 2015], Par is used transitively: its complement is an arc and its specifier is the arc’s anchor, as illustrated in (30). (30) [ParP Jill [Par′ Par [arc who is my neighbour]]] Because Par and the arc in (30) are concatenated by par-merge, neither the antecedent nor any other constituent in the host clause can ccommand into the arc. This generates the required opacity effects. (Griffiths and de Vries, 2019: 620). Whether it is the differently merged status of parenthesis or some aspect of their information status accounts for their behavior, there is little doubt that parenthesis interacts with elliptical processes in fascinating ways, illustrated in (29). (29) 2wa=y=yiqqaḥ yiṯro ḥoṯen mošɛ ʾeṯ and=narr=take.pret.3ms Jethro father-in-law.of Moses om ṣippora ʾešɛṯ mošɛ ʾaḥar šilluḥ=ɛha 3wǝ=ʾeṯ šǝne Zipporah wife.of Moses after send.perf.3ms=3fs and=om two.of ḇan-ɛ=ha ʾăšɛr šem ha=ʾɛḥaḏ geršom ki ʾamar son-mp=her rel name.of the=one Gershom because say.perf.3ms ger hayiṯi bǝ=ʾɛrɛṣ noḵriyya 4wǝ=šem ha=ʾɛḥaḏ alien be.perf.1cs in=land foreign.adj.fs and=name.of the=one

- 978-90-04-44885-8

94

holmstedt

ʾɛ̆liʿɛzɛr ki ʾɛ̆lohe ʾaḇ=i bǝ=ʿɛzri Eliezer because god.of father=my in=help wa=y=yaṣṣil=eni me=ḥɛrɛḇ parʿo and=narr=deliver.pret.3ms=1cs from=sword.of Pharaoh ‘Jethro, Moses’ father-in-law, took Zipporah, Moses’ wife, after her dismissal, and ____ her two sons, who the name of one was Gershom, because he said, “I was a alien in a foreign land,” and the name of the other was Eliezer, because ____, “The God of my father was in the capacity of my help and delivered me from the sword of Pharaoh”.’ (Exod 18:2– 4) While the first case of ellipsis in (29) is straightforward verb gapping between coordinate clauses, the second case of verb gapping occurs between two subordinate clauses, separated by a higher clause. And yet, the relationship of the ki causal clauses to the ‘at-issue’ proposition of the higher, null copula clauses (‘the name of the one/the other was Gershom/Eliezer’) should give us pause. These ki clauses do not motivate the predication of the null copula clauses but seem to enrich the interpretation of the host clause, either by contextualizing the host’s information or contributing a comment about the proposition communicated by the host (Blakemore 2009; see also Potts 2012). These ki clauses present non-at-issue information; as such, they are conventional implicatures, otherwise often known as parentheses. It may be that the par-Merged status of parentheses allows them to engage in a non-local relationship; or more likely, the opacity of parenthesis removes it locality consideration. Simply put, parentheses do not count when determining locality. Whether in examples like (29) or double-gapping examples like (30): yiśraʾel ʾɛṯ yǝmin=o (30) wa=y=yišlaḥ and=narr=send.pret.3ms Israel om right.hand=his wa=y=yašɛṯ ʿal roʾš ʾɛp̱rayim wǝ=huʾ haṣṣaʿir and=narr=pret.3ms on head.of Ephraim and=he the=young wǝ=ʾɛṯ śǝmoʾl=o ʿal roʾš mǝnaššɛ and=om left.hand=his on head.of Manasseh ‘Israel stretched out1 his right hand and set [it]2 on the head of Ephraim (he was the younger) and ____1 his left hand ____2 on the head of Manasseh’ (Gen 48:14) There is more to the relationship between parenthesis and ellipsis that we must consider. I have recently proposed that the “synonymous parallelism” between poetic lines is actually non-restrictive apposition, most commonly clausal in

- 978-90-04-44885-8

investigating ellipsis in biblical hebrew

95

nature (Holmstedt 2019). Thus, if we look back at the first example in (1), presented again in (31), (31) wǝ=ʾašiḇa šop̱ṭ=ay=iḵ kǝ=ḇ=a=riʾšona and-restore.juss.1cs judges-mp=your like=at-the=first wǝ=yoʿăṣ=ay=iḵ kǝ=ḇ=at=tǝḥilla and-counsellors-mp=your like=at-the=beginning ‘I will restore your judges as at the first, and ____ your counsellors as at the beginning’ (Isa 1:26a). the relationship between the two poetic lines is easily accounted for by the reformulative semantics of nonrestrictive apposition. In other words, Proposition A, that is, Proposition A’. Most relevant to the issue of ellipsis is this question: since nonrestrictive apposition is a type of parenthesis, how is it accessible for ellipsis? I suggest that parenthesis is not a barrier for ellipsis occurring around or through it. And relatedly I also suggest that while the par-Merge creates the opacity (also referred to as “invisibility”; de Vries 2007) of parenthesis,11 that is, parentheses are unable to maintain c-command relations with the constituents in the parenthetical host (de Vries 2007: 207–218), they do not block an antecedent-ellipsis relationship between the host and parenthesis. While

11

De Vries’ ten “invisibility effects” (de Vries 2007: 207–218) are: a. No movement: there cannot be movement from ParP into the host. b. No idiom chunks: no idiom can be split across a paratactic boundary. c. No Q-binding: a pronoun in ParP cannot be bound by a quantified expression in the host. d. No A-binding: a reflexive in ParP cannot be bound by an antecedent in the host. e. No Condition B effects: a pronoun in ParP does not cause Condition B effects with respect to a coreferent expression in the host. f. No Condition C effects: an R-expression in ParP does not cause Condition C effects with respect to a coreferent expression in the host. g. No npi s: no negative polarity item in ParP can be licensed by an operator in the host. h. No ppi effects: no positive polarity item in ParP can be disqualified by an operator in the host. i. No dependent Force: ParP’s illocutionary Force is independent of the host’s. j. No dependent Mood: ParP’s Mood is independent of the host’s. Due to the inability to create and test counter-examples for bh, the no-movement (a), noidiom chunks (b), Q-binding (c), and A-binding (d) tests are not applicable. Similarly, the bh data provide no examples that fail the no-Condition B (e) and no-Condition C (f) tests or the no-npi (g) and no-ppi (h) tests, but this is true across-the-board, so none of these distinguish parentheses from any other construction. The two effects for which the bh corpus provides testable data are no-dependent Force (i) and no-dependent Mood (j).

- 978-90-04-44885-8

96

holmstedt

there may be no c-command between the host and a par-merged parenthetical, examples like (32) clearly indicate that the antecedent can bind the ellipsis. (32) John—and Bill ____ too—loves Mary Poppins (Kluck, Ott, de Vries 2014: 13) Let us look at one more example (33) to see how ellipsis and parenthesis interact in poetic verse. (33) yaḏaʿ šor qone=hu wa=ḥămor ʾeḇus bǝʿal=aw know.perf.3ms ox owner=its and=ass trough.of master=its yiśraʾel loʾ yaḏaʿ ʿamm=i loʾ hiṯbonan Israel neg know.perf.3ms people=my neg consider.perf.3ms ‘the ox knows its owner2, the ass ____ [knows] the trough of its master; Israel does not know ____ [?], my people does not consider ____ [?]!’ (Isa 1:3) The first case of ellipsis occurs in the second poetic line and as a clear case of verb gapping. Additionally, the relationship between the two lines fits the nature of clausal apposition: the second line reformulates—and thus reinforces—the first line by switching to another popular work animal and narrowing from ‘owner’ to ‘master’s trough’. The effect is the idea that stock animals remember who feeds them. But it is the third and fourth lines that present some difficulty. The lack of overt complements in the third and fourth poetic lines in (33), for the verbs yadaʿ and hitbonan (both of which exhibit patterns of bivalency elsewhere)12 raises a good question: what are the antecedents for the apparent ellipses? Both lines have moved beyond the animal metaphor and so the only constituents in the clause that make semantic sense are the closer bǝʿalaw in the second line and the more distant qonehu in the first line. But the closer bǝʿalaw is embedded within a np (it is the clitic host in a bound phrase), and the fact that there are absolutely no other cases of bh ellipsis with an embed-

12

Previous scholarship appears at odds on whether the verbs yadaʿ in 1:3c and hitbonan in 1:3d are bivalent (i.e., requiring a complement; see, e.g., Gray 1912: 6–10) or monovalent (i.e., intransitive; see, e.g., Wildberger 1991: 15–16; Blenkinsopp 2000: 177; cf. also halot, s.v., dch, s.v.). The overwhelming use of yadaʿ as a bivalent verb and the clear cases of hitbonan as bivalent (see, e.g., Isa 43:18; 52:15) place the burden of proof on those who take these specific cases as monovalent/intransitive.

- 978-90-04-44885-8

investigating ellipsis in biblical hebrew

97

ded antecedent suggests that it is inaccessible for ellipsis. This seems to leave qonehu as the antecedent for both ellipses in the third and fourth lines. And that brings us again to the interaction of ellipsis with parenthesis, since the ellipsis in (c) and (d) are separated from their antecedent in (a) by the appositive clause in (b). In fact, I would argue that (b), (c), and (d) are all appositives. Line (b) reformulates (a) by extending it slightly, line (c) interprets the metaphor in (ab), and line (d) reformulating (c) for clarity. Returning to the question of this section—does bh license non-local ellipsis—the answer depends on how we consider parenthesis to contribute to locality. If an intervening parenthesis is counted in the locality determination, then the data are clear that bh licenses non-local ellipsis. However, if the nature of par-Merge is such that it allows ellipsis-antecedent relationships but also renders parentheses irrelevant to the locality determination, then we can affirm with Miller that bh does not license non-local ellipsis.

4

Verb Gapping in Prose versus Poetry

Now let us turn to the final issue: ellipsis in bh poetry and narrative prose. In her studies of bh ellipsis, Miller states more than once that verb gapping is constrained to coordinate structures, so that the example in (34) is licit but the unattested example in (35) is not. (34) ṣiyyon b=ǝmišpaṭ tippaḏɛ wǝ=šaḇ-ɛ=ha Zion in=justice redeem.pass.impf.3fs and=return.ptcp-mp=her bi=ṣḏaqah in=righteousness ‘Zion by justice shall be redeemed, and her repentant ones by righteousness’ (Isa 1:27; Miller 2007a: 166). (35) ṣiyyon bǝ=mišpaṭ tippaḏɛ lǝ=maʿan Zion in=justice redeem.pass.impf.3fs for=sake.of šaḇ-ɛ=ha bi=ṣḏaqah return.ptcp-mp=her in=righteousness ‘Zion by justice shall be redeemed, so that her repentant ones by righteousness [shall be redeemed]’ (unattested, Miller 2007a: 166) Miller also asserts that verb gapping is extremely rare in biblical prose and when it does occur, “remnants and their corresponding constituents form a disjoint set” that is contrastive (2007a: 169), as in (36):

- 978-90-04-44885-8

98

holmstedt

(36) hinne ʾanoḵi maṣṣig̱ ʾɛṯ gizzaṯ haṣ=ṣɛmɛr excl I lay.ptcp.ms om fleece.of the=wool b=ag=gorɛn ʾim ṭal yihyɛ ʿal hag=gizza on=the=threshing.floor if dew be.impf.3ms on the=fleece lǝ=ḇadd=ah wǝ=ʿal kol ha=ʾarɛṣ ḥorɛḇ for=alone=it and=upon all.of the=earth dryness wǝ=yaḏaʿti ki ṯošiăʿ and=know.perf.irrealis.1cs that deliver.caus.impf.2ms bǝ=yaḏ=i ʾɛṯ yiśraʾel ka=ʾăšɛr dibbarta by=hand=my om Israel like=rel speak.perf.2ms ‘Behold, I am going to lay a fleece of wool on the threshing floor. If dew will be upon the fleece alone / and upon all the ground ____ dryness, then I will know that you will deliver Israel by my hand, as you have said.’ (Judg 6:37; Miller 2007: 169). The contrastive sets in (36) are {dew, dryness} and {upon the fleece, upon the ground}. According to Miller, ellipsis in biblical poetry differs in the vast majority of examples, in that the paired constituents are coreferential, as in (37): (37) min ʾăram yanḥe=ni ḇalaq mɛlɛḵ moʾaḇ from Aram lead.caus.impf.3ms=1cs Balaq king.of Moab me=harr-e qɛḏɛm from.mountain-mp.of east ‘From Aram, Balaq led me, / the king of Moab ____ from the eastern mountains.’ (Num 23:7; Miller 2007: 169) In (37), ‘Balaq’ and ‘the king of Moab’ are coreferential terms, as are ‘Aram’ and ‘the eastern mountains’ in this context. While there are contrastive poetic examples, like (38), Miller describes these as a rarity. (38) ʾellɛ ḇ=a=rɛḵɛḇ wǝ=ʾellɛ ḇ=as=susim wa=ʾănaḥnu these on=the=chariot and=these on=the=horse-mp and=we bǝ=šem yhwh ʾɛ̆lohe=nu nazkir on=name.of Yhwh god=our call.impf.1cs ‘Some on chariotry ____, and some on horses ____, but we on the name of the lord our God we will call’ (Ps 20:8; Miller 2007: 170) There are three salient points to make concerning ellipsis in prose versus poetry. The first is that, while ellipsis occurs much more frequently in poetry, it

- 978-90-04-44885-8

investigating ellipsis in biblical hebrew

99

is not altogether rare in prose. From Genesis to 2 Samuel I identified more than 75 prose examples, like those in (39) and (40).13 (39) wa=y=yitten tɛḇɛn u=mispoʾ l=ag=gǝmall-im and=narr=give.pret.3ms straw and=fodder for=the=camel-mp u=mayim li=rḥoṣ rag̱l-a=w wǝ=rag̱l-e ha=ʾănaš-im and=water for=wash.inf foot-mp=his and=foot-mp.of the=man-mp ʾăšɛr ʾitt=o rel with=him ‘He gave straw and fodder for the camels and ____ water for washing his feet and the feet of the men who were with him’ (Gen 24:32) šaʾul miṣ=ṣaḏ ha=har miz=zɛ (40) wa=y=yelɛḵ and=narr=go.pret.3ms Saul from=side.of the=mountain from=this wǝ=ḏawiḏ wa=ʾănaš-a=w miṣ=ṣaḏ ha=har miz=zɛ and=David and=man-mp=his from=side.of the=mountain from=this ‘Saul went from the side of the mountain, from one way, and David and his men ____ from the side of the mountain, from the other way’ (1 Sam 23:26) The second point concerns whether it is accurate that verb gapping is constrained to coordinate structures. While this seems accurate for ellipsis in bh narrative prose, it is not accurate if the relationship of “synonymously parallel” lines, which are the most common context of verb gapping in poetry, is syntactically clausal apposition. Clausal apposition is inherently nonrestrictive and so is a type of parenthesis. Thus, verb gapping may occur in coordinate structures, but may also occur in appositive (parenthetical) structures. In light of this, I suggest that verb gapping in bh relates not to coordinate structures but to the larger category of non-subordination, under which coordination and subordination are the two subtypes, as in the schema in (41).14

13

14

See Gen 1:21, 25, 29; 7:2–3; 24:31, 32, 47; 27:12; 31:40; 32:29; 33:2; 39:6; 41:35; 43:32; 45:8; 47:18; 48:14; Exod 3:9, 19; 5:13; 12:9, 38; 16:8; 18:6; 21:23–24; 22:23; 24:4; 25:12, 20; 26:3, 10, 35; 28:21; 36:10, 13, 16; Lev 16:5; 23:9; Num 1:52; 6:19; 11:19–20; 15:24; 16:17; 21:29; 26:65; 32:16; 35:33; Deut 5:3; 11:29; 13:10; 15:22; 17:7; 28:23; 29:13–14 31:27; Josh 20:7; Judg 4:14; 10:8; 1Sam 14:13; 18:6; 1 Sam 23:26; 26:2; 30:17; 2 Sam 1:11. The following are poetic examples in Genesis–2Samuel: See Gen 4:23, 24; 49:10, 11, 13, 17; Num 21:28; 23:7; 24:5, 6; Deut 32:25; 33:10; 18, 28. The first distinction, between subordination and non-subordination, centers on syntactic hierarchy and dependency: subordinate clauses are constituents within the hierarchy of

- 978-90-04-44885-8

100

holmstedt

(41) An essential typology of clause linkage Clause Linking subordination

non-subordination coordination

parenthesis

appositives

parentheticals

Related to this is the third and final salient point about ellipsis in bh narrative prose, and it does not center on ellipsis itself but the types of clause relationships characteristic of the two discourse types. In bh prose, phrases and clauses are often linked by conjunction-coordination (which has additive semantics) or subordination. In narrative coordinated clauses there may be overlap in actors or actions/events (including time setting), but whether there is, for instance, continuity of the subject or the time of the two events is simultaneous, some major component in these linked narrative clauses must differ to achieve the purpose of narrative progression. In contrast, the convention behind bh verse is built on the frequent use of nonsubordinative nonrestrictive apposition, in which the second conjunct is a logical subset of the first. This differs from the coordination and subordination typical of narrative precisely in that the actors and events of the poetic anchor and appositive overlap (in the case of repetitive apposition, the overlap is complete).

5

Conclusion

Ellipsis is a common feature of both poetic and non-poetic texts in the Hebrew Bible. Understanding the types of ellipsis used in the biblical data allow us to interpret better these cases where “there is meaning without form.” Building on considerable recent work by Miller, I have filled in a few of the technical gaps: I have surveyed a few additional types of ellipsis that occur in bh, clarified how

another clause and are dependent on the superordinate clause; non-subordinate clauses are neither dependent nor hierarchically contained. The second distinction, between coordination and parenthesis, involves the symmetry (or lack thereof) between the host or anchor clause and the non-subordinate clause. Coordination is essentially a symmetrical linking, meaning that the coordinate constituent is of the same grammatical type as the anchor. In contrast, parenthesis exhibits no symmetry requirement. See Holmstedt 2020.

- 978-90-04-44885-8

investigating ellipsis in biblical hebrew

101

we might still assert that bh ellipsis must be a local relationship in light of the complicating features of parenthesis, and illustrated that ellipsis also occurs in non-poetic biblical texts.15

References Aelbrecht, Lobke. 2010. The Syntactic Licensing of Ellipsis. Linguistik Aktuell/Linguistics Today 149. Amsterdam: John Benjamins. Doron, Edit. 1999. V-Movement and vp Ellipsis. Pp. 124–140 in Fragments: Studies in Ellipsis and Gapping, ed. S. Lappin & E. Benmamoun. Oxford: Oxford University Press. Geller, Stephen A. 1979. Parallelism in Early Biblical Poetry. Missoula, mt: Scholars Press. Goldberg, Lotus M. 2005. Verb-Stranding vp Ellipsis: A Cross-Linguistic Study. Unpublished doctoral thesis, McGill University, Montreal, qc. Gray, George Buchanan. 1915. The Forms of Hebrew Poetry. London: Hodder and Stoughton. Griffiths, James and Mark de Vries. 2013. The Syntactic Integration of Appositives: Evidence from Fragments and Ellipsis. Linguistic Inquiry 44 (2): 332–344. Griffiths, James and Mark de Vries. 2019. Parenthesis: Syntactic Integration or Orphanage? (A Rejoinder to Ott 2016). Linguistic Inquiry 50 (3): 609–629. Holmstedt, Robert D. 2019. Hebrew Poetry and the Appositive Style: ‘Parallelism’, requiescat in pace. Vetus Testamentum 69: 617–648. Holmstedt, Robert D. 2020. Parenthesis in Biblical Hebrew as Noncoordinative Nonsubordination. Brill’s Journal of Afroasiatic Languages and Linguistics 12: 99–118. McShane, Marjorie J. 2005. A Theory of Ellipsis. Oxford: Oxford University Press. Merchant, Jason. 2019. Ellipsis: A Survey of Analytical Approaches. Pp. 19–45 in The Oxford Handbook of Ellipsis, ed. J. van Craenenbroeck and T. Temmerman. Oxford: Oxford University Press. Miller-Naudé, Cynthia L. 2003. A Linguistic Approach to Ellipsis in Biblical Poetry (Or, What to Do When Exegesis of What is There Depends on What Isn’t). bbr 13(2): 251– 270. Miller-Naudé, Cynthia L. 2005. Ellipsis Involving Negation in Biblical Poetry. Pp. 37–52 in Seeking Out the Wisdom of the Ancients: Essays Offered to Honor Michael V. Fox on the Occasion of His Sixty-Fifth Birthday, ed. R.L. Troxel, K.G. Friebel, and D.R. Magary. Winona Lake, IN: Eisenbrauns. 15

I am grateful for much insightful feedback from Edit Doron during the writing of this paper. I am also grateful to Todd Snider and Malka Rapport Hovav for input on my data and analysis.

- 978-90-04-44885-8

102

holmstedt

Miller-Naudé, Cynthia L. 2007a. Constraints on Ellipsis in Biblical Hebrew. Pp. 165– 180 in Studies in Comparative Semitic and Afroasiatic Linguistics Presented to Gene B. Gragg, ed. Cynthia L. Miller. Chicago: The Oriental Institute of the University of Chicago. Miller-Naudé, Cynthia L. 2007b. The Relation of Coordination to Verb Gapping in Biblical Poetry. jsot 32 (1): 41–60. Miller-Naudé, Cynthia L. 2007c. The Syntax of Elliptical Comparative Constructions. zah 18: 136–149. Miller-Naudé, Cynthia L. 2008. A Reconsideration of “Double-Duty” Prepositions in Biblical Poetry. janes 31: 99–110. Miller-Naudé, Cynthia L. 2013. Ellipsis: Biblical Hebrew. Pp. 807–812 in Encyclopedia of Hebrew Language and Linguistics, Volume 1: A–F, ed. Geoffrey Khan. Leiden: Brill. Murguía, Elixabete. 2004. Syntactic Identity and Locality Restrictions on Verbal Ellipsis. Unpublished doctoral thesis, University of Maryland, College Park, md. O’Connor, Michael P. 1980 [1997]. Hebrew Verse Structure. Winona Lake, IN: Eisenbrauns. Reprinted 1997 with afterword. Ross, John R. 1967. Constraints on Variables in Syntax. Unpublished doctoral thesis, Massachusetts Institute of Technology, Cambridge, MA. Vries, Mark de. 2007. Invisible Constituents? Parentheses as B-merged Adverbial Phrases. Pp. 203–234 in Parentheticals, eds. Nicole Dehé, and Yordanka Kavalova. Amsterdam: John Benjamins. Vries, Mark de. 2009. Specifying coordination: An Investigation into the Syntax of Dislocation, Extraposition, and Parenthesis. Pp. 37–98 in Language and Linguistics: Emerging Trends, ed. C.R. Dreyer. New York: Nova. Vries, Mark de. 2012a. Parenthetical main clauses—or not? On appositives and quasirelatives. Pp. 177–201 in Main Clause Phenomena: New Horizons, ed. by Lobke Aelbrecht, Liliane Haegeman & Rachel Nye. Amsterdam: John Benjamins. Vries, Mark de. 2012b. Unconventional mergers. Pp. 143–166 in Ways of Structure Building, ed. by Myriam Uribe-Etxebarria and Vidal Valmala. Oxford: Oxford University Press. Watson, Wilfred G.E. 1984. Classical Hebrew Poetry: A Guide to its Techniques. Journal for the Study of the Old Testament Supplement 26. Sheffield: Sheffield Academic Press.

- 978-90-04-44885-8

chapter 6

A Unified Account of the Infinitive Absolute in Biblical Hebrew Elizabeth Cowper and Vincent DeCaen

1

Introduction

The Biblical Hebrew (bh) infinitive absolute form (ia), illustrated in (1), has received considerable attention in the literature. Most work (Waltke and O’Connor 1990, Harbour 1999, Hatav 2017) has concentrated on constructions like (1), where the same verb root appears in both the infinitive absolute form and the main verb of the clause.1 (1) mik=kol ʿeṣ pro2 hag=gan ʾaḵol to-ʾḵel from=all.of tree.of the=garden eat.ia 2ms.npst-may.eat pro ‘Of every tree of the garden thou mayest freely eat’3 (Gen 2.16) In this paper, based on earlier work on bh clause structure (DeCaen 1995, 2014; Cowper and DeCaen 2017), we argue for a broader view of the ia, including cases where the two verbal forms have different roots. We propose a unified syntactic account of this broader range of ia forms, in which the infinitival phrase is a verbal projection (vP or VoiceP), adjoined to another verbal projection. Contra Harbour (1999, 2007), we argue that the ia is a full xp, rather than simply a head, and that the surface position of the ia depends on independently motivated movements of either the main verb or the ia phrase argued for by Cowper and DeCaen (2017). As for the morphological form of the ia, we hypothesize that the ia form is the default spellout for a verb bearing no

1 Hatav (2017) calls this construction the tautological infinitive absolute. 2 The Hebrew text follows the Leningrad codex and its Tiberian vocalization, a standard version sufficient for our purposes here (on vocalization, see Khan 2020). Where the word stress does not fall on the final syllable, an acute accent is added for clarity. The ia form under discussion, glossed with ia, is given in boldface; the main verb of the clause is underlined. Detailed morphological glossing is omitted except where relevant to the matter at hand. 3 All translations here are from the King James Version (kjv).

© Elizabeth Cowper and Vincent DeCaen, 2021 | doi:10.1163/9789004448858_007

- 978-90-04-44885-8

104

cowper and decaen

inflectional features, predicting that it might appear in other constructions as well. This prediction is borne out: we show that the same form indeed appears in certain coordinate constructions that exhibit a pattern parallel to first-conjunct agreement. We propose that these constructions involve VoiceP conjunction, where only the first conjunct moves to the relevant inflectional head, leaving the second conjunct to be spelled out with the default form.

2

The Phenomenon

We constructed a database of approximately 875 infinitives in total, extracted from BibleWorks (4.0.034d (1998) with morph 3.0), supplemented by morph 4.20, and cleaned so as to exclude misparsed examples. Approximately 420 of these were from Standard Biblical Hebrew (sbh); that is, from Genesis through 2 Kings. We added to these data tokens gleaned from the literature.4 Traditionally, the ia construction is defined as involving reduplication of the root, and possibly also the binyan, of the main verb of the clause, as illustrated in (2). When the two verbs appear in different binyanim, the infinitive overwhelmingly takes the paʿal form, as shown in (3) (Harbour 1999). (2) Same root, same binyan: wə=ʾaḇi=ha yaroq yaraq pro bə=p̄ anɛ=ha and=father=3fs spit.ia spit.pst.3ms pro in=face=3fs ‘If her father but spit in her face’ (Num 12.14) (3) Same root, different binyan (ia as plain Qal/Paʿal): loʾ ṯi-ggaʿ b=o yaḏ ki=saqol not 2ms.npst-touch on=3ms hand but=stone.ia yis-saqel 3ms.npst.pass-stone ‘There shall not a hand touch it, but he shall surely be stoned’ (Exod 19.13) While the infinitive phrase frequently consists only of a single word, it can also include arguments or modifiers, or a conjoined structure, as in (4)–(7).

4 Hatav (2017) counts 266 tokens of tautological infinitive absolute in sbh.

- 978-90-04-44885-8

a unified account of the infinitive absolute in biblical hebrew 105

(4) pp modifier and dp direct object: pro ʾɛ-ʿɛ̆ḇor bə=ḵol=ṣoʾn=əḵa hay=yom [haser pro 1s.npst-pass in=all.of=flock=2ms the=day [remove.ia miš=šam kol śɛ]5 from=there every sheep] ‘I will pass through all thy flock today, removing from thence all the sheep’ (Gen 30.32) (5) Pronominal direct object and pp modifier: pro ba=ḥɛḇɛl wa=yǝ-maddəḏ=em and.then=3ms.npst-measure=3mp pro with.the=cord [ha-škeḇ ʾoṯ=am ʾarṣ=a] [caus-lie.down om=3mp ground=to] ‘and measured them with a line, casting them down to the ground’ (2 Sam 8.2) (6) pp modifier, conjoined infinitives: wə=šiḇʿa hak=kohănim nośəʾim … holəḵim [[haloḵ] and=seven.of the=priests carry.prog … go.prog [[go.ia] [wə=ṯaqəʿu baš=šop̄ aroṯ]]6 [and=sound.ia on.the=trumpets]] ‘And seven priests bearing … went on continually, and blew with the trumpets’ (Josh 6.13) (7) Conjoined infinitives: pro haloḵ wə=ṭap̄op̄ t-elaḵ-na go.ia and=trip.ia 3f.npst-go-pl pro ‘walking and mincing as they go’ (Isa 3.16) The ia form can also appear conjoined with an inflected infinitive, as in (8), or even conjoined with a finite verb, as in (9).

5 The object here has been abbreviated, and the inverted order of the object and the modifier is the result of heavy xp shift. We use the notation pro throughout to indicate a phonologically null pronominal, and for simplicity do not distinguish between pro (Chomsky’s 1981 pronominal anaphor) and pro (Chomsky’s 1981 null pronoun). 6 We read the mt’s ‫ ותקעו‬wətaquʿ as the ia ‫ ותקוע‬wətaqoaʿ.

- 978-90-04-44885-8

106

cowper and decaen

(8) Infinitive absolute conjoined with an inflected infinitive: way=y-ešɛḇ ha=ʿam lɛ=ʾɛ̆ḵol wə=šaṯo and.then=3ms.npst-sit the=people to=eat and=drink.ia ‘and the people sat down to eat and to drink’ (Exod 32.6) (9) Infinitive absolute conjoined with a finite verb: way=y-iṯqəʿ-u pro baš=šop̄ aroṯ wə=nap̄oṣ and.then=3m.npst-blow-pl pro on.the=trumpets and=break.ia hak=kaddim ʾăšɛr bə=yaḏ-am the=jars that in=hand=3mp ‘and they blew the trumpets, and brake the pitchers that were in their hands’ (Judg 7.19) (both activities at the same time) Descriptively, there are two characteristics of the ia that must be explained in any unified account. First, the infinitival phrase never contains an overt subject; we assume that this is not an accident.7 Second, it can occupy a variety of positions in the main clause, as shown in (10)–(13). (10) At or near the end of the clause: pro ʿal=enu gam hiśtarer a. ki ṯi-śtarer for 2ms.npst-lord.over pro over=1pl even lord.over.ia ‘except thou make thyself altogether a prince over us?’ (Num 16.13) b. ʾel=ay y-eṣeʾ pro yaṣoʾ to=1s 3ms.npst-come.out pro come.out.ia ‘He will surely come out to me’ (2Kgs 5.11)

7 We are grateful to the participants at the 2018 Halbert workshop, who drew our attention to some possible cases of ia with overt subjects. We identified five such instances, perhaps not coincidentally in the Pentateuch: Gen 17:10, Exod 12:48, Lev 6:7, Num 15:35, Deut 15:2 (for additional candidates, see further Gesenius (1910: § 113gg) and Joüon (1923: §123u)). These five instances have minimal pairs in which the ia actually appears with a matching finite verb form. Thus, himmol laḵɛm kol zaḵar ‘Every man child among you shall be circumcised’ (Gen 17:10; cf. (13) below), contrasts with a clause with the finite verb yimmol ‘must be circumcised’, himmol yimmol yalid betǝḵa umiqnaṯ kaspɛḵa: ‘He that is born in thy house, and he that is bought with thy money must needs be circumcised’ (Gen 17:13). Space does not permit a full discussion of these cases here.

- 978-90-04-44885-8

a unified account of the infinitive absolute in biblical hebrew 107

pro h-eṭeb c. wə=ʾɛṯ ṣəlamay=w šibbər-u and=om idols=3ms smash.pst-3mp pro caus-good.ia ‘and his images brake they in pieces thoroughly’ (2 Kgs 11.18) d. way=yi-šṭəḥ-u la=hɛm pro šaṭoaḥ səḇiḇoṯ and.then=3m.npst-spread-pl to=3mp pro spread.ia around ham=maḥănɛ the=camp ‘and they spread them all abroad for themselves round about the camp’ (Num 11.32) (11) Directly after the main verb:8 pro haroḡ ʾim=maṣaʾ-ṯi ḥen bə=ʿenɛ=ḵa a. horḡ=eni naʾ kill.imp=1s please pro kill.ia if=found.pst-1s grace in=eyes=2ms ‘kill me, I pray thee, out of hand, if I have found favour in thy sight’ (Num 11.15) b. šimʿ-u pro šamoaʿ millaṯ=i listen.imp-mp pro listen.ia word=1s ‘Hear diligently my speech’ (Job 13.17 = 21.2) pro ha-ʿăḇir ʾɛṯ ha=ʿam c. lama he-ʿăḇar-ta why caus-cross.pst-2ms pro caus-cross.ia om the=people haz=zɛ ʾɛṯ hay=yarden the=this om the=Jordan ‘wherefore hast thou at all brought this people over Jordan’ (Josh 7.7) (12) Directly before the main verb: a. šamor ti-šmər-un pro ʾɛṯ miṣwoṯ ʾăḏonay keep.ia 3m.npst-keep-pl pro om commandments.of lord ʾɛ̆lohe=ḵɛm god=2mp ‘Ye shall diligently keep the commandments of the Lord your God’ (Deut 6.17)

8 We assume that naʾ in (11a) originates in sentence-initial position and moves to a position after horgeni for prosodic reasons.

- 978-90-04-44885-8

108

cowper and decaen

tizkor pro ʾeṯ ʾăšɛr ʿaśa b. zaḵor remember.ia 2ms.npst-remember pro om that do.pst.3ms ʾăḏonay ʾɛ̆lohe=ḵa lə=p̄ arʿo lord god=2ms to=Pharaoh ‘but shalt well remember what the Lord thy God did’ (Deut 7.18) pro c. haloḵ wə=ṭap̄op̄ t-elaḵ-na go.ia and=trip.ia 3f.npst-go-pl pro ‘walking and mincing as they go’ (Isa 3.16) (13) As the only verbal element in the clause: a. šamor pro ʾɛṯ yom haš=šabbaṯ lə=qaddəš=o keep.ia pro om day.of the=Sabbath to=make.holy=3ms ‘Keep the sabbath day to sanctify it’ (Deut 5.12; cf. Deut 16.1, 27.1) b. zaḵor pro ʾeṯ ʾăšɛr ʿaśa ʾăḏonay ʾɛ̆lohe=ḵa remember.ia pro om that do.pst.3ms lord god=2ms lə=miryam to=Miriam ‘Remember what the Lord thy God did unto Miriam’ (Deut 24.9; cf. Exod 13.3, 20.8, 25.17; Josh 1.13)

3

Prior Accounts

As mentioned above, existing discussions of the ia consider only those socalled tautological cases where the infinitive and the main verb of the clause share the same root. Traditional descriptions, such as that by Waltke and O’Connor (1990), simply list the various ia constructions, attributing different placement of the ia within the clause to factors like emphasis. Generative approaches, in contrast, provide syntactic (Harbour 1999, 2007, 2008) or information-structural (Hatav 2017) analyses, but still focus exclusively on the tautological ia. The most clearly articulated syntactic treatments of the ia are those of Harbour (1999, 2007, 2008). He analyses them as predicate clefts, with the infinitive absolute spelling out the trace of a moved verb. If the copy includes v0, which for Harbour determines the binyan, then the two verbs appear with the same binyan. If only the verb root appears in the copy, then the default Paʿal form appears due to a morphological repair operation.

- 978-90-04-44885-8

a unified account of the infinitive absolute in biblical hebrew 109

For us, given our broader definition of the phenomenon, there are two fatal problems with Harbour’s movement-based account. First, such an account inherently cannot be extended to include forms with different verb roots. Second, since the construction is derived specifically by head movement, it cannot accommodate instances such as (14), repeated from (4) above, where both the main verb of the clause and the infinitive are fully phrasal and have their own arguments or modifiers. (14) pp modifier and dp direct object: pro ʾɛ-ʿɛ̆ḇor bə=ḵol ṣoʾn-əḵa hay=yom [haser pro 1s.npst-pass in=all.of flock-2ms the=day [remove.ia miš=šam kol śɛ]9 from=there every sheep] ‘I will pass through all thy flock to day, removing from thence all the sheep’ (Gen 30.32) In the next section, we provide the theoretical framework and assumptions that serve as a point of departure for our analysis of the more broadly construed infinitive absolute construction.

4

Theoretical Framework

Following Cowper & DeCaen (2017), we assume that bh is a null-subject, verbsecond (V2) language. In that paper, we argued for the syntactic projections shown in (15), not all of which will be relevant to our discussion of the ia.

9 The object here has been abbreviated, and the inverted order of the object and the modifier is the result of heavy xp shift.

- 978-90-04-44885-8

110

cowper and decaen

(15)

In an ordinary declarative clause, the verb moves, via syntactic head movement, to the head of the Topic projection (Top), and some other constituent, frequently the subject, moves to the specifier position in TopP, giving surface subject-verb-object (svo) order as in (16). (16) huʾ hesir ʾɛṯ hab=bamoṯ he remove.pst.3ms om the=high.places ‘He removed the high places’ (2Kgs 18.4) Since bh is a null-subject language (i.e., subject pronouns are frequently not pronounced), sometimes the subject in [spec,TopP] is phonologically null, giving rise to apparent V1 order as in (17). As Cowper and DeCaen (2017) show, these sentences, while often assumed to be verb-initial, are in fact verb-second from a syntactic perspective, just like (16). The syntactic structure of (17a) is given in (18). (17) a. wə=pro šibbar ʾɛṯ ham=maṣṣeḇoṯ and=pro smash.pst.3ms om the=sacred.stones ‘and brake the images’ (2Kgs 18.4)

- 978-90-04-44885-8

a unified account of the infinitive absolute in biblical hebrew

111

b. wə=pro ḵaraṯ ʾɛṯ ha=ʾăšera and=pro cut.pst.3ms om the=Ashera ‘and cut down the groves.’ (2Kgs 18.4) (18)

As in any V2 language, sometimes a constituent other than the subject moves to [spec,TopP]. When this happens, the subject remains in [spec,FinP] while the verb still moves to Top, giving verb-subject order, as illustrated in (19). (19) bə=reʾšiṯ baraʾ ʾɛ̆lohim ʾeṯ haš=šamayim wə=ʾeṯ in=beginning create.pst.3ms God om the=heavens and=om ha=ʾareṣ the=earth ‘In the beginning God created the heaven and the earth.’ (Gen 1.1) Also, phonologically overt material can appear in one or more projections above TopP. For example, in (20), the clause-initial bracketed constituents are hanging topics, which we assume occupy the specifier position of &P. In (20a), this constituent is followed by the subject in [spec,TopP], which is in turn followed by the verb in Top0. In (20b) the nonsubject constituent ləkā ‘to you’ appears in [spec,Top], followed by the verb in Top0 and the phonologically null pro subject in [spec,FinP].

- 978-90-04-44885-8

112

cowper and decaen

(20) a. [ha=ʾiš ʾăšɛr nimṣaʾ hag=gaḇiaʿ bə=yaḏ=o] huʾ [the=man that be.found.pst.3ms the=cup in=hand=3ms] he yi-hyɛ ll=i ʿɛḇɛḏ 3ms.npst-be to=1s servant ‘but the man in whose hand the cup is found, he shall be my servant’ (Gen 44.17) b. [ki ʾɛṯ kol ha=ʾarɛṣ ʾăšɛr ʾatta roʾɛ] lə=ḵa [for om all.of the=earth that you.ms see.prog] to=2ms ʾɛ-ttənɛn=na pro 1s.npst-give=3fs pro ‘For all the land which thou seest, to thee will I give it.’ (Gen 13.15) Finally, it sometimes happens that the verb moves past Top0 to a higher head. In jussive clauses like those in (21), for example, Cowper and DeCaen (2017) argue that the verb moves through Top0 to the jussive Force head, giving a verb-initial sentence.10 (21) a. y-a-ʾer ʾăḏonay yaʾer panay=w ʾelɛ=ka face=3ms to=2ms 3ms.npst-caus-shine lord ‘the Lord make his face shine upon you.’ (Num 6.25) b. wə=ya-śem pro ya-śem lə=ka šalom to=2ms peace and=3ms.npst-give pro ‘and [the Lord] give you peace.’ (Num 6.26)

5

Our Proposal

5.1 The Infinitive Absolute Is a Verbal Projection (VoiceP or vP) We claim that the ia is a phrasal, verbal projection, namely VoiceP.11 In this we differ from Harbour (1999), who takes it as the realization of a moved verbal head.12 As mentioned earlier and repeated here in (22), the ia can contain both

10 11

12

In the examples in (21), the position from which the verb moves is indicated in the transliteration by a struck-through copy of the verb. Following Kratzer (1996) and much work since then, we assume that the external argument originates in [spec,VoiceP]. We also assume that in a transitive clause, the case that licenses the direct object is provided by the Voice head (Legate 2014). Recall that Harbour deals only with the tautological infinitive absolute construction,

- 978-90-04-44885-8

a unified account of the infinitive absolute in biblical hebrew

113

arguments and modifiers of its own, which play no role in the clause in which the ia is embedded. (22) a. pro ʾɛ-ʿɛ̆ḇor bə=ḵol ṣoʾn=əḵa hay=yom [pro haser pro 1s.npst-pass in=all.of flock=2ms the=day [pro remove.ia miš=šam kol śɛ13] from=there every sheep] ‘I will pass through all thy flock to day, removing from thence all the sheep’ (Gen 30.32) b. way=ǝ-maddəḏ=em pro ba=ḥɛḇɛl [pro and.then=3ms.npst-measure=3mp pro with.the=cord [pro ha-škeḇ ʾ oṯ=am ʾarṣ=a] caus-lie.down om=3mp ground=to] ‘and measured them with a line, casting them down to the ground’ (2Sam 8.2) Having seen from transitive examples like (22a) that the ia must include at least VoiceP, we also argue that it includes no more than VoiceP, thus lacking the inflectional structure that characterizes a full clause. In particular, it lacks the structural machinery required to license case on its subject. While we take no particular position on exactly where structural subject case originates, it is generally held to be provided by an inflectional head above VoiceP such as Tense or Finiteness. If the ia consists only of a VoiceP, there is no mechanism internal to the infinitival clause that would case-license a subject, and it is unsurprising that its subject is virtually always covert; i.e., pro. A second argument that the ia has no inflectional structure above VoiceP is that the infinitival verb is always initial in the infinitival phrase. The fact that the subject is covert, together with the absence of any higher structural positions, such as [spec,TopP] or [spec,ForceP], to which another constituent could move, explains the consistently verb-initial word order. A third, particularly compelling argument that the ia lacks inflectional heads above Voice comes from the fact that it can never bear clitics—even object clitics—despite the fact that it can take overt non-subject arguments, and be modified. This is unsurprising if, as is generally assumed, such clitics are

13

where the verbal root in the infinitive absolute is identical to that in the clause in which the ia appears. Our analysis covers a wider range of data. The object here is abbreviated to make the structure clearer; the inverted order is the result of heavy xp shift.

- 978-90-04-44885-8

114

cowper and decaen

hosted by an inflectional head above Voice. A bare VoiceP can accommodate full overt dp non-subject arguments, but not clitics—even clitics representing non-subject arguments. Full infinitives can bear clitics, as shown in (23). (23) a. ʿaḏ boʾ=i until come.inf=1s ‘until I come’ (2Kgs 18.32) b. lə=ʿozr=eni to=help.inf=1s ‘to help me’ (1Chr 12.18) We therefore assume henceforth that the ia is a VoiceP. The next question to be dealt with is where it appears in the main clause, and how its various surface positions are to be accounted for. 5.2 The Infinitive Absolute Modifies a Verbal Projection (VoiceP or vP) We claim that the ia VoiceP modifies, and is thus adjoined to, a verbal projection, potentially either VoiceP or vP,14 for two reasons. First, when the ia clause remains in situ, it always surfaces below grammatical aspect, as can be seen in (24). (24) a. wə=šiḇʿa hak=kohănim nośəʾim … [AspP holəḵim [ia haloḵ and=seven.of the=priests carry.prog … [AspP go.prog [ia go.ia wə=ṯaqəʿu baš=šop̄ aroṯ]]15 and=sound.ia on.the=trumpets]] ‘And seven priests bearing … went on continually, and blew with the trumpets’ (Josh 6.13) b. hemma [AspP holəḵim [ia haloḵ wə=ḏabber]] they.mp [AspP go.prog [ia go.ia and=talk.ia]] ‘as they still went on, and talked’ (2Kgs 2.11) Following DeCaen (1995), we assume that the participle (hōlǝkîm in (24)) spells out progressive aspect, as with walking in the English example in (25). However,

14 15

We take no position here on whether unaccusative intransitive clauses contain a VoiceP in the verbal domain, or only a vP. The matter has no bearing on our analysis. We read the Masoretic Text’s ‫ ותקעו‬wətaquʕ as the ia ‫ ותקוע‬wətāqôaʕ.

- 978-90-04-44885-8

a unified account of the infinitive absolute in biblical hebrew

115

since the ia clause in (24), unlike the sentence in (25), has no marked tense feature to be spelled out, the ia clause contains no finite auxiliary corresponding to were in (25). (25) They were walking along the street. The second reason to believe that the ia is adjoined to vP or VoiceP is the fact that, when it remains in situ, it surfaces at or near the end of the clause, as in (22) and (24), followed only by other adjuncts, as in (26). This position is exactly what would be expected if the ia clause were adjoined to either vP or VoiceP. (26) wə=ham=mayim hay-u haloḵ wə=ḥasor ʿaḏ ha=ḥoḏɛš and=the=water be.pst-3mp go.ia and=recede.ia to the=month ha=ʿăśiri the=tenth ‘And the waters decreased continually until the tenth month’ (Gen 8.5) 5.3 The Infinitive Absolute, like Other xp s, Can Move Our third core claim is that the infinitive absolute can undergo the same movements as other phrasal projections in the language. For example, it can move to [spec,TopP], participating in the verb-second construction just as other phrases do. This is what has happened in (27), where the ia is the first element in the clause. It is followed immediately by the main verb, which has, as is normal, moved to the Top0 head. (27) zaḵor t-izkor pro ʾeṯ ʾăšɛr ʿaśa ʾăḏonay remember.ia 2ms.npst-remember pro om that do.pst.3ms lord ʾɛ̆lohe=ḵa lə=p̄ arʿo god=2ms to=Pharaoh ‘but shalt well remember what the Lord thy God did’ (Deut 7.18) It can also happen that the verb can move past the ia in [spec,TopP] to occupy a higher functional head. This gives rise to examples like (28). In this instance, the main verb hēʕăbártā has moved to Force, above TopP and thus before the ia. The wh-word lā ́mâ has moved to [Spec,ForceP], and thus precedes the verb. The subject remains in [spec,FinP] below the Topic projection, and thus follows the ia. (28) lama he-ʿăḇar-ta pro ha-ʿăḇir ʾɛṯ ha=ʿam why caus-cross.pst-2ms pro caus-cross.ia om the=people

- 978-90-04-44885-8

116

cowper and decaen

haz=zɛ ʾɛṯ hay=yarden the=this om the=Jordan ‘wherefore hast thou at all brought this people over Jordan’ (Josh 7.7) Finally, a hanging topic can appear in the highest specifier position, [spec, &P]. This can derive a variety of orders, depending on which constituent has moved to [spec,TopP], and on how far the main verb has moved. In (29), kol ʿoməsɛha is a hanging topic, the infinitive absolute śaroṭ appears in [spec,TopP], and the main verb appears in Top0. (29) kol ʿoməs-ɛ=ha śaroṭ yiś-śareṭ-u all.of burden.prog-pl=3fs cut.ia 3m.npst.pass-cut-pl ‘All that burden themselves with it shall be cut in pieces.’ (Zech 12.3, cited by Harbour 1999: 169) The structure in (30) shows all the possible positions of the finite verb (marked as verb1, verb2, and verb3) and those of the infinitive absolute (marked as ia1, ia2, and ia3). (30)

- 978-90-04-44885-8

a unified account of the infinitive absolute in biblical hebrew

117

This structure derives all the observed word orders of the infinitive absolute construction, if we assume the four possible movements listed in (31), and take into account the fact that the subject may be either an overt dp or pronoun, or the phonologically null element pro. (31) a. The finite verb moves to Top0. This is the core of the verb-second pattern in Biblical Hebrew and happens in ordinary declarative clauses. b. The finite verb moves from Top0 to Force0 or another head in the left periphery. This movement is triggered by a particular marked feature in the higher head, and thus appears in questions and exclamatives. c. The finite verb may move to &0. This movement derives the narrative inversion pattern. d. The infinitive absolute moves to [spec,TopP]. This movement is one of the various possibilities available to derive verb-second order within TopP. Other adjuncts and arguments within the clause can undergo this movement; the behaviour of the ia is in this respect entirely unsurprising. The first three movements listed in (31) are argued for by Cowper and DeCaen (2017) independently of the ia construction. They thus come at no cost to the present analysis; rather, the ia construction adds to the evidence previously adduced for them. The fourth is also independently motivated for constituents other than the ia, and given our claim that the ia is a full verbal projection rather than a head, it would be surprising if it could not also move to [spec,TopP]. As to the question of which of these operations takes place in any given sentence containing an ia, we assume that the answer lies in informationstructural factors like the topicality of the ia itself (see Hatav 2017), and in the presence or absence of marked features of heads above the Topic projection.

6

Interim Summary

We have argued that the syntax of the ia construction can be understood quite straightforwardly in terms of bh clausal syntax, and that no constructionspecific syntactic processes or structures are required. We have also accounted for a wider range of data than are handled by previous accounts of the ia. However, we have not so far said anything about either the morphological shape of

- 978-90-04-44885-8

118

cowper and decaen

the ia verb, or the mechanism deriving the tautological infinitive, where the ia reduplicates the main verb root of the clause, and carries meanings often described as emphasis (see Waltke and O’Connor’s 1990: § 35.3.1b “intensifying infinitive”). Before turning to these, we first discuss another kind of construction in which the ia form appears.

7

Coordinate Structures and the ia

In addition to appearing as single phrasal adjuncts to a verbal projection, the ia form is found in a variety of conjoined structures. The first of these is illustrated in (32), where two ias together modify the main clause. pro (32) haloḵ wə=ṭap̄op̄ t-elaḵ-na go.ia and=trip.ia 3f.npst-go-pl pro ‘walking and mincing as they go’ (Isa 3.16) This example is straightforwardly accounted for by the analysis already presented. The ia consists of two conjoined VoicePs, which as a unit are adjoined to the matrix VoiceP, and have moved as a constituent to [spec,TopP]. Not all conjoined ia constructions are as easily accounted for, however. Consider the example in (33), where the infinitive absolute šātô is conjoined, not with another ia, but with the inflected infinitive lɛʾɛ̆kol. (33) way=y-ešɛḇ ha=ʿam lɛ=ʾɛ̆ḵol wə=šaṯo and.then=3ms.npst-sit the=people to=eat and=drink.ia ‘and the people sat down to eat and to drink’ (Exod 32.6) Here, the interpretation is of a conjoined structure, as is clear from the translation. The puzzle, then, is why the two verb forms are different. Given the meaning, one might expect both ‘to eat’ and ‘to drink’ to be expressed with the regular infinitive, as indeed they are in (34). (34) wa=ʾăni ʾa-ḇoʾ ʾɛl beṯ=i lɛ=ʾɛ̆ḵol wə=li=štoṯ and=I 1s.npst-go to house=1s to=eat.inf and=to=drink.inf wə=li=škaḇ ʿim ʾišt=i and=lie.inf with wife=1s ‘shall I then go into mine house, to eat and to drink, and to lie with my wife?’ (1Sam 11.11)

- 978-90-04-44885-8

a unified account of the infinitive absolute in biblical hebrew

119

A similarly puzzling example is shown in (35), where the ia phrase is conjoined with a finite verb phrase. As with (33), the interpretation is clearly one of conjoined phrases, with both activities taking place at the same time, despite the morphological difference between the two verbs. (35) way=yiṯqəʿ-u pro baš=šop̄ aroṯ wə=nap̄oṣ and.then=3m.npst-blow.pl pro on.the=trumpets and=break.ia hak=kaddim ʾăšɛr bə=yaḏ-am the=jars that in=hand=3mp ‘and they blew the trumpets, and brake the pitchers that were in their hands’ (Judg 7.19) Again, one might expect that in such a construction both verbs would take the finite form, as can be seen in (36). (36) way=ya-ḇoʾ ʾuriyya ʾelay=w way=y-išʾal and.then=3ms.npst-come Uriah to=3ms and.then=3ms.npst-ask dawiḏ li=šlom yoʾaḇ David to=peace.of Joab ‘And when Uriah was come unto him, David demanded of him how Joab did’ (2Sam 11.7) In order to explain why the verb forms in (33) and (35) differ, we turn to a parallel phenomenon that has been observed in a variety of languages: firstconjunct agreement. As illustrated in the Moroccan Arabic example in (37), coordinate structures sometimes display asymmetric agreement patterns (see also McCloskey 1986, Munn 1999, Doron 2000, Bošković 2009, and many others). Example (37) shows a typical form of asymmetric agreement, in which the verb agrees only with the first conjunct of a postverbal subject. (37) mšat kull mra w xu-ha left.f.sg each woman and brother-her ‘Each woman and her brother left.’ (Munn 1999: 653) Current syntactic accounts of coordinate structures treat them as asymmetric, with the first conjunct occupying a higher position than the second. Various versions have been proposed, two of which are illustrated in (38).

- 978-90-04-44885-8

120

cowper and decaen

(38) a.

b.

Standard X-bar view: first conjunct in [spec,&P]

Munn’s (1993) Boolean phrase, adjoined to np

In either of these structures, the first of which we adopt here, first-conjunct agreement can happen with postverbal subjects because the first conjunct is structurally closer to the head hosting the preposed verb, as illustrated schematically in (39). Assuming that in a sentence like (37), the verb has moved past the subject to a higher head position (here F), agreement then takes place between the moved verb and the higher of two conjoined dp s, as shown. (39)

We propose that a similar mechanism is responsible for the asymmetric inflectional patterns in (33) and (35), and that the relevant structure is as shown in (40). The only difference between (33) and (35) is the specification of the head F, which is either infinitival, as in (33), or finite, as in (35). The derivations are otherwise identical.

- 978-90-04-44885-8

121

a unified account of the infinitive absolute in biblical hebrew

(40)

First, each verb moves to its own Voice head, as shown by the arrows in (40). That is as far as the verb in the second conjunct can move, there being no clausal inflectional heads directly above it in the structure. However, the verb in the first conjunct is now in a position to move to the higher inflectional head, here indicated by F, and potentially higher, if other inflectional heads are present. The first conjunct then seems to behave like a full clause, whose subject is case-licensed and thus can be overt, as is the case in (41). (41) way=y-arʾ parʿo ki hayə-ṯa ha=rəwaḥa and.then=3ms.npst-see Pharaoh that be.pst-3fs the=relief wə=ha-ḵbed ʾɛṯ libb=o and=caus-hard.ia om heart=3ms ‘But when Pharaoh saw that there was respite, he hardened his heart’ (Exod 8.11) The second verb thus remains in Voice, and is spelled out with whichever binyan Voice provides. Its subject is not case-licensed, on the assumption that structural subject case originates in a position above VoiceP; the subject of the second conjunct must therefore be covert. Interestingly, however, it can be referentially distinct from the subject of the first conjunct, as in (42). (42) way=yiqrəʾ-u pro lə=p̄ anay=w ʾa-ḇreḵ and.then=3m.npst-shout-pl pro to=face=3ms caus-kneel.imp.ms

- 978-90-04-44885-8

122

cowper and decaen

wə=pro naṯon oṯ=o ʿal kol=ʾɛrɛṣ miṣrayim and= pro give.ia om=3ms over all.of=land.of Egypt ‘and they [people] cried before him, Bow the knee: and he [Pharaoh] made him [Joseph] ruler over all the land of Egypt’ (Gen 41.43 kjv) What then of sentences like (34) and (36)? In those sentences the verbs in both conjuncts have the same inflected form: a fully inflected infinitive in (34), and a full finite verb in (36). We propose, uncontroversially, that sentences like these involve coordination of constituents larger than VoiceP, and that each conjunct thus contains its own inflectional material. The verb in each conjunct moves to the relevant inflectional head, and bears the appropriate inflectional morphology. This account predicts that the subject of the second conjunct can be phonologically overt only if the structures conjoined include higher, inflectional categories like FinP or AspP. Example (38) above has such a structure. Under the account proposed here, it is unsurprising that ia forms are also found in disjunctions with ‘or’ as well as in conjunctions with ‘and’. Examples are given in (43). (43) a. wə=ḵi ṯi-mkər-u mimkar la=ʿamiṯɛ=ḵa ʾo qano and=if 2m.npst-sell-pl sale to=neighbour=2ms or buy.ia miy=yaḏ ʿamiṯɛ=ḵa from=hand neighbour=2ms ‘And if thou sell aught unto thy neighbour, or buyest ought of thy neighbour’s hand’ (Lev 25.14) b. ʾiš ki yi-ddor nɛḏɛr la=ʾḏonay ʾo hiššaḇaʿ šəḇuʿa man that 3ms.npst-vow vow to=lord or swear.ia oath ‘If a man vow a vow unto the Lord, or swear an oath’ (Num 30.3[2] kjv) It remains to be explained why the second-conjunct verb, in cases of firstconjunct inflection, is spelled out with the same form as the ia. For us, the crucial property of all of the cases we have examined is that there are no clausal inflectional heads above VoiceP in the structure, and the verb thus remains in Voice. We propose that the ia form is a so-called default inflectional form, which surfaces when a verb cannot move to, or form an Agree relation with, an inflectional head such as Aspect, Finiteness, or Tense. This account is possible under a realizational approach such as Distributed Morphology (Halle and Marantz 1993 et seq.), where morphological form is determined by a set of rules realizing specific morphosyntactic features. The ia form, we propose,

- 978-90-04-44885-8

a unified account of the infinitive absolute in biblical hebrew

123

arises when no other verbal spellout rule applies. This is parallel to Harbour’s (1999, 2007) proposal, with which we concur, that paʿal is the default binyan that appears whenever a verb has not been able to move to, or form an agree relation with, a Voice head.

8

Conclusion and Remaining Questions

We have argued that the ia construction consists of a verbal projection (vP or VoiceP), modifying another verbal projection. Syntactically, the ia behaves just like other phrasal modifiers, in that it either remains in situ, or moves to [spec,TopP]. For us, the tautological infinitive construction is a special case of this more general construction. We assume that in those cases, the root is copied from the main clause to the ia constituent. The derivation is otherwise identical to that of the hetero-radical constructions we have focussed on. The morphological form of the ia is simply the default inflectional form which appears when a verb has no access to an inflectional head. It is found, not only in VoiceP/vP modifiers, but also as the second of two conjoined VoicePs, where only the first verb moves to the inflectional head position. We have taken no position here on the mechanism by which the verb root is copied in the tautological ia, though we have ruled out Harbour’s account, in which the tautological ia is the spellout of the trace of a moved verb. Finally, we have nothing new to say about cases like (13), where the infinitive absolute is the only verb in the clause. It can be noted that these examples are highly formulaic, and that for each of them, a close counterpart can be found that also contains a finite verb. One obvious avenue to pursue is that these are fragments, with the context providing an unpronounced higher finite clause that could provide a host for a VoiceP adjunct. Another avenue is suggested by the five minimal pairs in note 7 (exceptionally overt subject) with and without the finite verb. On this view, the finite verb is optionally spelled out in the presence of a matching ia.

References Bošković, Željko. 2009. Unifying first- and last-conjunct agreement. Natural Language and Linguistic Theory 27(3): 455–496. doi 10.1007/s11049-009-9072-6. Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris Publications. Cowper, Elizabeth, and Vincent DeCaen. 2017. Biblical Hebrew: A formal perspective on the left periphery. Toronto Working Papers in Linguistics 38: 33 pp. - 978-90-04-44885-8

124

cowper and decaen

DeCaen, Vincent. 1995. On the placement and interpretation of the verb in Standard Biblical Hebrew prose. Doctoral dissertation, University of Toronto. DeCaen, Vincent. 2014. On the syntax and semantics of the Biblical Hebrew Infinitive Absolute. Presented at the Society of Biblical Literature, San Diego. Doron, Edit. 2000. vso and left-conjunct agreement: Biblical Hebrew vs. Modern Hebrew. In The syntax of verb-initial languages, ed. Andrew Carnie and Eithne Guilfoyle, 75–96. Oxford: Oxford University Press. Gesenius, Wilhelm. 1910. Gesenius’ Hebrew Grammar. 2nd ed. Edited and enlarged by E. Kautzsch, revised by A.E. Cowley. Oxford: Clarendon. Halle, Morris, and Alec Marantz. 1993. Distributed morphology and the pieces of inflection. In The view from Building 20: Essays in linguistics in honor of Sylvain Bromberger, ed. Kenneth Hale and Samuel Jay Keyser, 111–176. Cambridge, MA: mit Press. Harbour, Daniel. 1999. The two types of predicate clefts: Classical Hebrew and beyond. in Papers on morphology and syntax, cycle two, ed. Vivian Lin, Cornelia Krause, Benjamin Bruening, and Karlos Arregi, 159–175. Cambridge, MA: mit Working Papers in Linguistics. Harbour, Daniel. 2007. Against PersonP. Syntax 10(3): 223–242. doi 10.1111/j.1467–9612 .2007.00107.x. Harbour, Daniel. 2008. Klivaj predika, or predicate clefts in Haitian. Lingua 118(7): 853– 871. doi 10.1016/j.lingua.2007.11.010. Hatav, Galia. 2017. The Infinitive Absolute and topicalization of events in Biblical Hebrew. 185–207 in Advances in Biblical Hebrew linguistics: Data, method, and analyses, ed. Adina Moshavi and Tania Notarius. Winona Lake, IN: Eisenbrauns. Joüon, Paul. 1965 [1923]. Grammaire de l’ hébreu biblique. Corr. 3rd ed. Rome: Pontifical Biblical Institute. Khan, Geoffrey. 2020. The Tiberian pronunciation tradition of Biblical Hebrew. 2 vols. Online at https://www.openbookpublishers.com/10.11647/OBP.0163.pdf; https:// www.openbookpublishers.com/10.11647/OBP.0194.pdf. Kratzer, Angelika. 1996. Severing the external argument from its verb. In Phrase structure and the lexicon, ed. Johan Rooryck and Laurie Zaring, 109–137. Dordrecht: Kluwer. Legate, Julie Anne. 2014. Voice and v: Lessons from Acehnese. Cambridge, MA: mit Press. McCloskey, James. 1986. Inflection and conjunction in Modern Irish. Natural language and Linguistic Theory 4(2): 245–281. Munn, Alan S. 1993. Topics in the syntax and semantics of coordinate structures. Doctoral dissertation, University of Maryland, College Park. Munn, Alan S. 1999. First conjunct agreement: Against a clausal analysis. Linguistic Inquiry 30(4): 643–668. doi 10.1162/002438999554246. Waltke, Bruce K., and Michael O’Connor. 1990. An introduction to Biblical Hebrew syntax. Winona Lake, IN: Eisenbrauns.

- 978-90-04-44885-8

chapter 7

The Nature of the Infinitive Absolute Galia Hatav

1

Introduction

Traditional Hebraists such as Waltke & O’Connor (1990: 582) consider the Infinitive Absolute (ia) in Biblical Hebrew (bh) to be a (verbal) noun. Following studies within the Generative Grammar approach such as Harbour (1999, etc.), Cowper and DeCaen (2017, this volume), Doron (this volume) and others, I consider it to be a non-finite verb form in all of its occurrences, deprived of temporal and agreement features as well as clitics. Scholars list a number of uses of the ia in the Hebrew Bible: As an adverbial, modifying situations (Waltke & O’Connor 1990, a. o.), as a volitive, performing commands (Waltke & O’Connor 1990; Doron this volume, a. o.), as a focus or topic marker (Harbour 1999; Hatav 2017), etc. My aim in this chapter is to determine the common denominator of all those uses. As many of them cannot be accounted for if the ia is considered an infinitive, I believe that the title ‘infinitive’ given to this form is a misnomer. My contention is that the ia is a root phrase whose function is to name the eventuality of the verb in question. This explains not only why it is deprived of temporal and agreement features, but also why the ia cannot have subject or object clitics attached to it and why it cannot be negated.1 To support my claim that the ia is a root phrase, I devote Section 2 to discussing roots. Sub-sections 2.1 and 2.2 deal, respectively, with the notion of root in language in general and in Hebrew in particular. As the Hebrew root can only form verbs when casted in one of the patterns (= binyanim), Section 3 discusses the Hebrew patterns. Section 4, which is the crux of this chapter, deals with the characteristics of the ia as a root phrase. I first illustrate the different uses of the ia observed in the literature and analyze them according to my thesis (Subsection 4.1). Then, I suggest a morpho-syntactic analysis for the generation of the ia, arguing that it is a root phrase (Sub-section 4.2). Section 5 summarizes and concludes the findings and claims from the previous sections. 1 As listed by Gesenius (1910, § 113v), there are only three occurrences of the ia in the Hebrew Bible that might be considered to be negated, as they are preceded by the neg lo’: Gen 3:4, Amos 9:8, and Ps 49:8. Gesenius considers those cases to be exceptions.

© Galia Hatav, 2021 | doi:10.1163/9789004448858_008

- 978-90-04-44885-8

126 2

hatav

Root

2.1 The Nature of Root in Language Since the introduction of Distributed Morphology by Halle and Marantz (1993) it has been widely acknowledged that verbs (as well as nouns and adjectives) in language derive from roots. However, the notion of root and its syntax and morphology have been controversial. The most intuitive approach to the notion of root seems to be along the lines of studies such as Doron (2003, 2008), Arad (2005) and Borer’s earlier work (e.g., Borer 2005), where roots are understood to carry meaning. E.g., Arad (2005: 6) suggests that roots are ‘those atomic, non-decomposable lexical elements’ and Borer (2005: 30) defines roots (listemes, in her terminology) as a list of ‘bundles of conceptual meaning’ (paired with arbitrary sounds), constituting the encyclopedia. However, this line of analysis has been challenged by studies such as Harley (2014) and the later work of Borer (Borer 2014, 2015, etc.), pointing to a number of problems (which I will not reiterate here). To account for those problems, those studies suggest what I will call a ‘non-intuitive approach’, where roots do not carry meaning (but indexes).2 Notwithstanding the problems observed for the intuitive approach, I will still assume it in this study, claiming that roots have (encyclopedic) meaning, or content.3 The morphology and syntax of roots have been controversial as well. According to one approach, represented mainly by Borer’s studies (2014, 2015, etc.), roots are category-less, with neither internal grammatical nor syntactic properties, lacking argument structure. In contrast, studies such as Marantz (1997), Doron (2003) and Harley (2014) argue that roots are categories that can have complements and form with them phrases. I adopt the latter approach, suggesting that ia s in bh are root phrases, built from roots and their complements— see Section 4.2 for more details. 2.2 Roots in Hebrew Being a Semitic language, roots in Hebrew only consist of radicals—strings of (usually) three consonants. To form verbs, those radicals must be casted into one of the verb patterns, called binyanim in Hebrew (singular: binyan).4 The

2 I will not elaborate on the “non-intuitive” approach, as that will necessitate a journey into the intricacies of the different studies within it. 3 See Labelle (2014) for a criticism of the “non-intuitive” approach and arguments in favor for the “intuitive” one. 4 To form nouns, they must be casted in one of the possible noun patterns called mishqalim (mishqal in singular).

- 978-90-04-44885-8

the nature of the infinitive absolute

127

specific meaning of a certain verb depends on the binyan in which the root is mapped, but all the verbs that derive from the same root would have something in common with respect to meaning. In Table (1) below I offer two examples.5 (1) Hebrew Roots: Examples Root

Meaning

Examples

√ktb

‘related to writing’ katab ‘write’ as in ‘John wrote a letter’ hiktib ‘dictate’ (~ make somebody write) hitkatteb ‘correspond’ (~ write to each other) √lmd ‘has to do with lamad ‘learn; study’ learning’ limmed ‘teach’ (~ make someone learn)

The root √ktb, composed of the consonants /k/, /t/ and /b/ (in this order), is used to form words that have to do with writing. When it is casted in binyan i (see Section 3 below for a discussion on the binyanim and their labels), this root forms the verb kaṯaḇ, which may best be translated as ‘write’ in English. When it is casted in binyan v it forms the verb hiḵtiḇ ‘dictate’, which roughly means ‘make [someone] write’, when it is casted in binyan vii it creates the verb hiṯkatteḇ ‘correspond’, etc. Similarly, the root √lmd, composed of the consonants /l/, /m/ and /d/, may form different verbs with the binyanim. With binyan I it forms the verb lamaḏ ‘study; learn’, with binyan iii it forms limmeḏ ‘teach’, which roughly means ‘make [someone] study/learn’, etc.

3

Binyanim

As it is composed of consonants only, the root in Hebrew (and in Semitic languages in general) cannot be pronounced as is and can form a verb and get a specific meaning only when it is casted in one of the binyanim, as illustrated in Table (1) above. Traditionally, it has been claimed that Hebrew has seven binyanim, conventionally referred to by Roman numerals or via the root √pʿl, as listed in Table (2) below.6 5 Following the convention in recent studies, I use the mathematical symbol √ to represent a root. 6 Note that when using the root √pʿl to name a binyan, the labels are according to the form

- 978-90-04-44885-8

128

hatav

(2) The Hebrew Binyanim (as traditionally claimed by Hebraists) Binyan

Vocalic-Template7 Root+Binyan

Examples

i paʿal ii nip̄ʿal iii piʿel iv puʿal v hip̄ʿil vi hup̄ ʿal vii hiṯpaʿel

{a-a} {ø-a} {i-e} {u-a} {i-i} {u-a} {a-e}

lamaḏ ‘learn; study’ niḵnas ‘enter’ biššel ‘cook’ buššal ‘cook/pass’ hip̄ šir ‘defrost’ hup̄ šar ‘defrost/pass’ hiṯlabbeš ‘got dressed’

C1aC2aC3 niC1C2aC3 C1iC2C2eC3 C1uC2C2aC3 hiC1C2iC3 huC1C2aC3 hitC1aC2C2eC3

In binyan i (paʿal) the first root radical (i.e. the first consonant represented as C1 in Table 2) appears at the beginning of the verb, followed by the vowel /a/, which is followed by the second radical (i.e. the second consonant C2), which, in turn, is followed by another vowel /a/ and the third radical (C3). Binyan ii (nip̄ ʿal) consists of the prefix /ni-/, followed by the first and second radicals of the root, where the second radical is followed by the vowel /a/, which, in turn, is followed by the third radical (C3). Binyan iii (piʿel) has the first radical at the beginning, followed by the vowel /i/, which is followed by two occurrences of the second radical (i.e. the second radical is geminated), where the second occurrence is followed by the vowel /e/, etc. Studies within the tradition of Generative Grammar such as Arad (2005, etc.) and Borer (2005, etc.) argue convincingly that Hebrew only has four binyanim.8 First, templates iv (puʿal) and vi (hup̄ ʿal) are typically considered inflectional, passive variants of templates iii (piʿel) and v (hip̄ʿil), respectively, rather than independent templates (Borer 2005). As noted by Arad (2005), iii and iv as well as v and vi share the same prosodic template, and differ only in their inner vowels. So a verb in iii like biššel ‘cook’ will be inflected into its passive equivalent buššal ‘be cooked’ in iv, and the active verb hip̄ šir ‘defrost’ in V will be inflected into its passive equivalent hup̄ šar ‘be defrosted’ in vi.

qatal, conjugated in third person singular masculine. The vocalic templates may be different in other forms or conjugations. 7 This presentation is adapted from Borer (2015). Note, however, that Borer does not include Binyan i, for reasons that will be discussed below. 8 Grouping the binyanim according to the parameters simple, causative or intensive, Doron (2003, 2008) claims that there are actually only three binyanim in Hebrew.

- 978-90-04-44885-8

129

the nature of the infinitive absolute

Second, crucially for the current chapter, Binyan i (paʾal), which is also called Qal ‘light’, is actually not a “real” binyan. As Arad (2005: 237) puts it, this pattern ‘serves as in all root-derived Hebrew forms and for this reason it is a completely unmarked form—in some sense, as many linguists suggest, this is the default binyan.’ Borer (2015) observes that verbs in Qal are rarely productive and have many irregularities in forms other than the qatal ( yiqtol, the infinitive or the imperative). Accordingly, Borer (2015, n. 2) concludes that ‘for ii–vii, vocalization is binyan selected, but for Qal it is root-selected.’ As will be argued in Section 4 below, this fact is crucial for characterizing the ia in bh as a root phrase.

4

The Nature of the ia

Sub-sections 4.1 and 4.2 below deal, respectively, with the semantics of the ia in bh and its uses in the Hebrew Bible. Sub-section 4.3 suggests a morphosyntactic analysis for its generation. 4.1 The Semantics of the ia Though it usually comes in the same binyan as its finite equivalent, Harbour (1999) shows that an ia may come in the default “non-binyan” Qal even when the parallel finite verb is in another binyan. Accordingly, Harbour claims that the binyan of the ia does not contribute semantic content. Support to his observation he finds in minimal pairs such as (3)–(4): (3) wǝ=ʾatta huʾ naqo and=you.ms cop.3ms go.unpunished.ia-i tinnaqɛ loʾ ṯinnaqɛ go.unpunished.yiqtol-ii.2ms neg unpunished.yiqtol-ii.2ms ‘Are you the one to go unpunished? You shall not go unpunished.’ (Jer 49:12) (4) wǝʾ=attɛm hinnaqe ṯinnaqu and=you.mp go.unpunished.ia-ii go.unpunished.yiqtol-ii.2mp loʾ ṯinnaqu neg go.unpunished.yiqtol-ii.2mp ‘Are you to get unpunished? You shall not go unpunished!’ (Jer 25:29) While both the ia and the finite verb ‘go unpunished’ in (4) are in the same Binyan ii (nip̄ ʾal), in (3) only the finite verb is in Binyan ii, while the ia is in Binyan i (Qal), with no difference in meaning. Accordingly, Harbour (1999) con-

- 978-90-04-44885-8

130

hatav

tends that an ia represents the bare eventuality of the verb in question. Similar claims about the ia’s nature were made by other scholars. For instance, Gesenius (1910, §113a) considers the ia to be ‘the idea of the verb in the abstract, i.e. it speaks of an action (or state) without any regard to the agent or to the circumstances of time and mood under which it takes place’. Similarly, according to Doron (this volume), the ia serves as the citation of the verb and Hatav (2017) considers it to be representing the lexical concept of the verb. This semantic character of the ia suggests that it is not an infinitive, but rather a phrase that represents the root of the verb in question, providing the bundle of the lexical properties of that verb. The characteristics of the ia as a pure lexical can explain its uses observed in the literature, to which I turn now. 4.2 The Uses of the ia in the Hebrew Bible We can group the ia constructions and their uses into three main groups, as illustrated and analyzed in Sub-sections 4.2.1–4.2.3 below. 4.2.1 Group A The ia s I put in this group appear in constructions with a finite verb of the same root and (usually) the same binyan. I recognize four functions of the ia in such constructions: secondary predication (Sub-section 4.2.1.1), adverbial (4.2.1.2), focus (4.2.1.3) and topic (4.2.1.4). 4.2.1.1 Secondary Predication A number of syntactic and semantic analyses have been suggested to account for secondary predication in language. I follow the approach of studies such as Rothstein (2004, etc.), Kratzer (2005) and Hatav (2020), considering secondary predication to be an operation that results in a complex predicate created by putting together the primary- and secondary predicates. Consider the English examples in (5)–(6) below, where ‘drinking’ or ‘drunk’ is the secondary predicate of the sentence: (5) John drove his car drinking/drunk. (6) He was arrested, as driving and drinking/driving drunk is illegal. The sentence in (5) is true if and only if the driving and drinking/being drunk were simultaneous, interpreted as one (complex) event. John would not have been arrested if he only had driven his car or had been just drinking/drunk. John was arrested for engaging in the activity of driving-drinking or drivingdrunk.

- 978-90-04-44885-8

the nature of the infinitive absolute

131

Hatav (2020) shows that one of the devices in depicting such complex predicates in bh is via a certain ia construction she calls double infinitive-absolute construction. In such construction, the sentence contains a finite verb and two conjoined occurrences of an ia, where the first conjunct is of the same root and (usually) binyan as the finite verb. Consider one of her examples, illustrated in (7) below:9 (7) Secondary predication wayyelɛḵ ʾit=tah ʾišš=ah haloḵ u=ḇaḵo go.wayyiqtol.3ms with=her man=hers go.ia and=weep.ia ʾaḥăre=ha ʿaḏ baḥurim after=her till Bahurim ‘Her husband went with her till Bahurim, weeping after her.’ (2 Sam 3:16a) The verb ‘go’ or ‘walk’ appears twice. Its first occurrence is in the finite form wayyiqtol. Its second occurrence is in ia, within a construction where it is followed by the conjunction ‘and’ and the verb ‘cry; weep’ also in ia. What is important to note for the points made here is that the first part of the verse with the finite verb just reports that Michal’s husband walked with her (after she was taken from him by the king, until he was ordered to go back home). To report that he cried while walking, the finite verb of walking is duplicated, without its temporal and agreement features, using the ia. The verb depicting the crying is also in ia. Since each ia is deprived of temporal and agreement features, they seem to form together a new complex verb, namely the lexical walking-crying. 4.2.1.2 Adverbials Gesenius (1910, §113u), Waltke and O’Connor (1990: 589) and Van der Merwe et al (2017: 182), among others, observe that a conjunction of two ia s where the first is the ia of the verb ‘go’ and the second is a copy of the finite verb (sans its temporal and agreement features) may be used for modifying the eventuality depicted by the finite verb. The example in (8) below illustrates:10

9 10

For more examples Hatav refers the reader to Judg 14:9, 1Sam 6:12, 2Sam 15:3, 1Kgs 20:37b, Isa 19:22; Jer 12:17b. For more examples see: Gen 8:5, 12:9, 26:13, Judg 4:24, 1Sam 14:19, 2Sam 5:10, 1Chr 11:9.

- 978-90-04-44885-8

132

hatav

(8) Adverbial wayyašuḇu ham=mayim me=ʿal ha=ʾarɛṣ haloḵ return.wayyiqtol.3mp the=water from=above the=earth go.ia wa=šoḇ and=return.ia ‘The water gradually/continuously receded from the earth.’ (Gen 8:3a) The verb ‘return’ appears twice. Its first occurrence is in the finite form wayyiqtol. Its second occurrence is in ia, within a construction where it is preceded by the conjunction ‘and’ and the verb ‘go’ in ia, crucially, where the verb ‘go’ does not have a semantic content (i.e. does not report that the protagonist walked or went—as opposed to its appearance in examples like [7] above). A construction of two conjoined ias where one of them is the verb ‘go’ (with no semantic content) is usually understood to modify the event reported by the finite clause. In example (8), the construction is understood as the adverbial ‘gradually, continuously’, modifying the event of the water receding (lit. returning) depicted by the finite clause. 4.2.1.3 Focus As noted by Van der Merwe et al (2017: 178), most occurrences of the ia in the Hebrew Bible are within a construction referred to as the ‘tautological infinitive’ (Goldenberg 1971; Kim 2009; Hatav 2017) or the ‘paronomastic infinitive’ (Callaham 2010). In this construction, there appears a finite verb and an ia of the same root and (usually) binyan as the finite verb (but without another ia following it as in the case of secondary predication or adverbial modification, discussed above). The ia in this construction is usually understood to be the focus of its respective clause. Example (9) below illustrates:11 (9) Focus ham=mɛlɛḵ ʾɛl ʾărawna loʾ ki qano wayyoʾmɛr say.wayyiqtol.3ms the=king to Araunah no but buy.ia bi=mḥir meʾot=ḵa ʾɛqnɛ buy.yiqtol.1cs from=you.ms in=price ‘[But] the king replied to Araunah, “No, I will buy [it] from you at a price”.’ (2Sam 24:24)

11

As listed by Hatav (2017), other examples include: Gen 2:16, 31:30, 32:12, 37:8, 10; 44:5; Exod 2:19; Num 13:30, Deut 13: 10, 15:5, 7–8; 21:14, 22:1,4; 24: 12–13; 1Sam 8:9; 26: 25.

- 978-90-04-44885-8

the nature of the infinitive absolute

133

The verb ‘buy’ appears twice. Its first occurrence is in ia and its second occurrence is in the finite form yiqtol. The ia in this construction is understood by Goldenberg (1971), Harbour (1999) and Hatav (2017) to focus the eventuality in question (the buying, in our verse). 4.2.1.4 Topic Two kinds of topics have been discussed in the linguistic literature. One kind involves picking out an item from the previous context and topicalizing it (i.e. moving it to the front of its respective clause). This use of topicalization has been labeled ‘continuative topic’ in the linguistic literature (see Moshavi 2010 for elaboration). The second type of topic is a topicalized expression whose function is to set the temporal (or location) setting. (See de Swart 1999 for an analysis of temporal topic.) Hatav (2017) shows that both kinds of topics can be marked in bh by an ia via the construction discussed in the previous section (4.2.1.3). Consider one example of each kind discussed in Hatav: (10) Continuative Topic12 wayyoʾmɛr doḏ šaʾul haggiḏa nnaʾ l=i ma say.wayyiqtol.3ms uncle Saul tell.imp.ms now to=me what ʾamar la=ḵɛm šǝmuʾel wayyoʾmɛr šaʾul ʾɛl say.qatal.3ms to=you.ms Samuel say.wayyiqtol.3ms Saul to doḏ=o haggeḏ higgiḏ la=nu ki nimṣǝʾu uncle=his say.ia say.qatal.3ms to=us that find.pass.qatal.3cp ha=ʾăṯonoṯ the=asses.fp ‘Saul’s uncle asked [him], “Tell me, now, what Samuel said to you.” Saul said to his uncle: “What he said to us was that the asses have been found.”’ (1Sam 10:15–16) Saul’s response in verse 16 contains a verb with the root √ngd ‘say’ twice. Its first occurrence is in ia and its second occurrence is in the finite form qatal. Hatav (2017) recognizes the ia in this case to be the eventuality topic of its clause, as it picks out one item from Saul’s uncle’s request in the previous verse 15, namely the event of saying, using another root, though (√ʾmr).

12

Hatav only found one more case where the ia is interpreted as an example of continuative topic: 2Sam 5: 19.

- 978-90-04-44885-8

134

hatav

(11) Temporal Topic13 haloḵ halǝḵu ha=ʿeṣim li=mšowaḥ ʿăle-hɛm mɛlɛḵ go.ia go.qatal.3cp the-trees.mp to=anoint on=them.mp king ‘Once the trees went to anoint a king over them.’ (Judg 9:8) The root of the verb ‘go’ appears twice. Its first occurrence is in ia and its second occurrence is in the finite form qatal. Hatav (2017) recognizes the ia in this construction as marking the temporal topic of its clause, suggesting to translate it by the temporal adverb ‘once’. 4.2.2 Group B The ia s I put together in this group appear with finite verbs but with different roots, functioning as (quasi) adverbials (Waltke and O’Connor 1990, among others). I only found one kind of such use, namely where the ia is used to elaborate on the situation depicted by the matrix finite clause. The example in (12) illustrates. (12) Elaboration14 qǝḇuraṯ ḥămor yiqqaḇer saḥoḇ wǝ=hašleḵ burial.of donkey bury.pass.yiqtol.3ms drag.ia and=toss.ia me=halǝʾa lǝ=šaʿăre yǝrušalaim from=there to=gates.of Jerusalem ‘A donkey’s burial shall he be buried, dragged and tossed beyond the gates of Jerusalem.’ (Jer 22:19) Jeremiah the prophet cites the Lord about Jehoiakim’s fate, who is to be buried a donkey’s burial when he dies. The root √qbr casted in binyan ii forms the finite verb ‘be buried’. As it depicts a future event, it is conjugated in the finite form yiqtol, as expected. What is interesting for our discussion is that the finite verb is followed by a construction of two connected ia s: √sḥb ‘drag’ and √šlk ‘toss’. The conjunction of the two ia s is understood to specify what a donkey’s burial means, elaborating on the situation depicted by the finite clause. What is important to note for the sake of the current chapter is that representing the roots of their respective verbs, both ia s are to be understood as elaborating

13

14

Hatav (2017) refers the reader to other examples, including: Gen 15:13; 17:13; 18:10, 18; 20:18; 22:17; 24:5; 27:30; 37:33; 40:15; 43:20; 50:24, 35; Exod 18:18; Num 21:2; Judg 9:8; 1Sam 24:20; 25:28; 2Sam 1:6; 2 Kgs 11:11. Other examples include: Num 6:23; 15:35; Josh 6:3; 1 Sam 3:12; 2Sam 8:2; Isa 7:11.

- 978-90-04-44885-8

the nature of the infinitive absolute

135

on the root of the main clause, not on the temporal setting, the mood or the protagonist (Jehoiakim) of the main clause. 4.2.3 Group C As observed by Hebraists (Gesenius 1910, Waltke and O’Connor 1990, a. o.), while most of its appearances are within some construction where it is attached to a finite clause, the ia may also appear on its own. The appearance of the ia on its own seems to be challenging for the analysis defended here. In what follows, I will illustrate each of those cases. To account for them within the analysis suggested here, I can only offer intuitive hypotheses that are still to be investigated. 4.2.3.1 Commands It has been claimed that the ia can be used to perform a command, rather than the expected imperative (Waltke and O’Connor 1990: 609; Gesenius 1910, § 113b; Doron, this volume, to name a few). Consider one example analyzed in Doron (current volume). (13) Commands15 zaḵor ʾɛṯ yom haš=šabbaṯ lǝ=qaddǝš=o remember.ia om day the=Sabbath to=sanctify=it.ms ‘Remember the Sabbath day, to keep it holy.’ (Exod 20:8; see also Deut 5:12) This verse is one of the ten commandments, and like the other commandments, it is usually understood to be a command without a specific interlocutor, to be carried out always by everybody. Cross-linguistically, it has been shown that a non-finite or gerundive clause may be used to perform a speech act of command in the absence of an interlocutor (Palmer 2001: 114; Portner et al 2019: 5; among others). Consider the Italian, German and English examples in (14), (15) and (16), respectively, analyzed in Portner et al (p. 5). (14) Italian Negli armadi o negli scaffali disporre in basso i materiali in.the closets or in.the shelves put.inf in low the materials

15

Other examples include Exod 13:3; 20:8, 12; Lev 6:7; Deut 5:16; 16:1; 24:9; 25:17; 27:1; Josh 1:13; 2Sam 24:12.

- 978-90-04-44885-8

136

hatav

più pesanti more heavy ‘In closets and shelves, place the heavier materials in the lower areas.’ (15) German Bitte von der Bahnsteigkante zurücktreten please from the edge.of.the.track step.back.inf ‘Please step back from the edge of the track!’ (16) English No feeding the monkeys! The examples in (14) and (15) show that Italian and German may use the infinitive to perform a directive with the absence of an interlocutor. Example (16) shows that English allows for a nonfinite form, namely the gerund, for that purpose (though it may also use the imperative). As the ia is a non-finite form, it is not surprising that it can be used for what is understood by many to be commands. But this still begs the question of why it is the ia that is used for such purpose and not the infinitive construct or the Benoni (which is also a non-finite form and can be used as a gerundive).16 Doron (2018, this volume) argues that the ia is a kind of infinitive (contrary to what is suggested in this chapter), which differs from the infinitive construct in not carrying temporal features but only Mood. This explains, according to Doron, why it can be used to perform commands and, moreover, why such commands are (usually?) general (rather than addressed to certain people). Doron’s analysis seems to account elegantly for the ia when it is used to perform what looks like commands. However, she does not try to apply her analysis to the other uses of the ia. I believe that my thesis that ia s are root phrases

16

Note that in Modern Hebrew, which has lost the ia as a productive form, both the infinitive (construct) and the Benoni are used for performing directives such as commands and instructions. The examples in (i) and (ii) below illustrate: (i) A sign: lo’ lidroḵ ʿăl ha=deše neg to=step.inf on the=grass ‘Do not step on the grass!’ (ii) Mother to son: oḵlim be=pe sagur eat.benoni.mp in=mouth closed.ms ‘One should eat with a closed mouth.’

- 978-90-04-44885-8

the nature of the infinitive absolute

137

(rather than infinitives) may account more generally for the ia’s uses, including its appearance in the ten commandments. Though the English title ‘the ten commandments’ suggests that what is included in that list are commands, it seems to me that they are not. Note that the word used for them in Hebrew derives from the root √dbr ‘speak’ (not √pqd ‘command’). Instead, I believe, the ten items referred to as ‘commandments’ compose a kind of a to-do and not-to-do list. The ia clauses enumerate the todo items on this list, where the ia depicts the pure eventuality that is to be performed by every person at all times. However, this intuitive hypothesis may be challenged by examples such as 2Sam. 24:12, where God instructs David’s prophet to go talk to him, using the ia form of the verb ‘go’. I will have to leave this use of the ia for further investigation. 4.2.3.2 Adverbs It has been observed that some ia s are used as adverbs (Waltke and O’Connor 1990: 591, a. o.). The example in (17) below illustrates. (17) Adverbs17 wǝdarašta wǝḥaqarta investigate.weqatal.2ms inquire.weqatal.2ms wǝšaʾalta heṭeḇ interrogate.weqatal.2ms do.well.ia ‘[You shall] investigate, inquire and interrogate thoroughly.’ (Deut 13:15a) The ia heṭeḇ ‘do.well’ in (17) is understood to modify the finite verbs preceding it and is usually interpreted as the adverb ‘thoroughly; well’. Ylikaski (2003) deals with a non-finite verb form in languages like Hungarian which has been labeled in the literature ‘converb’. As those languages have also infinitives, Ylikaski’s concern is mainly to determine how the two nonfinite verb forms differ. His contention is that they differ in their function. While the infinitives function as complements, converbs function as verbal adverbs.18 What is crucial for our discussion is his conclusion that being ‘verbal adverbs’, converbs are expected to undergo grammaticalisation and be lexicalized adverbs (p. 220). He shows this expectation to be born out in languages 17 18

Other examples include: 2 Kgs 11:18; Isa 32:17; Prov 25:27a. Interestingly, most (if not all) uses of converbs I found in the literature are performed by the ia in bh. If my thesis is on the right track and ia s are root phrases, we can determine that converbs in languages like Hungarian are root phrases. I leave this thought for a future investigation.

- 978-90-04-44885-8

138

hatav

like Finnish, Russian and Hungarian. What is relevant for our chapter is that since ia s are used as (quasi) adverbials, they, too, should be expected to be lexicalized as adverbs. This expectation is borne out by cases like (17) above. 4.2.3.3 Subjects Though it is rare, as noted by studies such as Waltke and O’Connor (1990: 591), we find verses with an ia functioning as the subject of its respective clause. The verse cited in example (18) below illustrates: (18) Subjects—Rare19 ʾaḵol dǝḇaš harboṯ loʾ ṭoḇ Eat.ia honey excess.ia neg good ‘Eating honey to excess is not good.’ (Prov 25:27a) The second ia (harboṯ ‘excess’) in this verse is another example of an ia behaving like an adverb (as the ia heṭeḇ ‘do.well’ in example 17). The first ia (ʾaḵol ‘eat’), which is much rarer, functions as the subject of this clause (with its complement, namely the object ‘honey’). The function of a non-finite verb form in a subject position is a well-known phenomenon cross-linguistically. The English examples in (19)–(20) below illustrate. (19) To workout is healthy. (20) Working out is healthy. The subject in (19) is the infinitive ‘to workout’ and in (20) the gerundive ‘working out’. This can explain the appearance of the ia in examples like (18) above. 4.2.2 Summary To summarize, the ia mostly appears in constructions that are understood to function as (quasi) adverbials. But it can also appear by itself as an adverb, in what looks like commands and as the subject of its respective clause. For the sake of the discussion in this chapter, it is important to also determine where it cannot appear.

19

As a matter of fact, this is the only example I could tell for sure that an ia functions as a subject. Gesenius (1910, § 113b) lists five more verses he claims include an ia as a subject, but I am not sure his claim is adequate.

- 978-90-04-44885-8

the nature of the infinitive absolute

139

First, as noted by Gesenius (1910, §113e), the ia ‘is never used in immediate connection with prepositions.’ The only case found in the Hebrew Bible where an ia is a complement of a preposition is in the first chapter of 1 Samuel: (21) Preposition’s Complement wattaqom ḥanna ʾaḥăre ʾoḵl=a ḇǝ=šilo raise.wayyiqtol.3fs Hannah after eat.inf=hers in-Shiloh wǝ=ʾaḥăre šaṯo and=after drink.ia ‘Hannah arose after eating and after drinking.’ (1 Sam 1:9a) The first preposition ‘after’ has an infinitive construct as its complement (‘eat’ + a subject clitic), which is expected. However, the second occurrence of ‘after’ has an ia as its complement (šaṯo ‘drink’). If an ia can only be used like a (quasi) adverb, it would not be expected to function as a complement. Indeed, Gesenius (1910, §113, n. 3) claims that this occurrence is ‘impossible Hebrew, and as the lxx shows, a late addition.’ Second, the ia is not attested with modifiers in the Hebrew Bible.20 This could be just a coincidence, but I believe it is not. If my thesis that the ia is a root phrase that is built from a root, it cannot have modifiers. 4.3 The Morphology and Syntax of the ia As mentioned in Section 2 above, not only the semantics of the root is controversial but also its morphological and syntactic status. One approach, represented mainly by Borer’s work (Borer 2014, 2015 etc.), takes the root to be a-categorical. For instance, Borer (2014: 356) concludes that roots ‘have no syntactic properties—they have no category, they do not take complements, and there is no evidence that they project’. Borer shows this

20

The only example I found where it might be argued that the ia is modified by an adverb is in the following verse: (i) waʾɛkkoṯ ʾoṯo ṭaḥon heṭeḇ ʿaḏ ʾăšer daq lǝ=ʿap̄ ar crush.wayyiqtol.1cs om=it grind.ia do.well.ia until that fine to=dust ‘Grinding [the calf], [I] crushed it well until [it] was as fine as dust.’ (Deut 9:21) First note that the ia ‘grind’ is an example of an ia used to elaborate on the situation depicted in the finite clause (see discussion on Example 12 above for the use of the ia to elaborate on the finite clause). In particular, it explains how the crushing was done. Right after this ia, there appear two adverbials: the ia heṭeḇ ‘do.well’ and the pp ‘until it was as fine as dust’. Doron (this volume) understands these adverbials to modify the ia ‘grind’. However, they may also be understood to modify rather the main verb, as reflected in my English translation above.

- 978-90-04-44885-8

140

hatav

line of analysis to be able to account for a number of important morphological and syntactic problems (which will not be iterated here). Harley (2014: 249) admits that Borer’s proposals have conceptual appeal, but they also ‘face several empirical hurdles, in that there are phenomena whose analysis requires as a precondition that roots behave like normal syntactic elements, participating in Merge …, even to the point of having arguments as sisters and projecting to the √P category’. Harley’s approach is already found in Marantz (1997), and is assumed in the Hebrew studies of Doron (2003) and Arad (2005). Adopting the approach that a root is a syntactic category that can take complements and form with them a root phrase, I will argue in this section that the ia is a root phrase, which consists of a root and the internal arguments of that root. Within the framework of distributed morphology (Halle and Marantz 1993), Arad (2005) suggests that all verbs in language are made from roots, combined with verbal features and voice features. In the case of Hebrew, where the root is consonantal and cannot be pronounced on its own, ‘the verbal pattern serves as the device for syllabifying’ (Arad 2005:45). Accordingly, she suggests the tree diagram under (22) below for the root √šmr ‘guard’ as an illustration of how the root and patterns interact to generate a verb: (22) Verb Generation (Adapted from Arad’s tree 28 on p. 45) VoiceP Voice

vP v

← {a-a}

/šamar/

√rootP ← cvcvc /šVmVr/ √root ← √šmr

As explained by Arad, the root consonants are inserted at the bottom under the root node. In this example, the three consonants of the root √šmr ‘guard’ are inserted at the lowest node. Binyan i (cvcvc) is inserted above the root, at the root phrase; in our example, it forms the √P /šVmVr/. Voice exponents are inserted under Voice at the next level of the derivation, after binyan insertion takes place, providing us with the verb šamar, which is still to be assigned temporal and agreement features at the next stage of the derivation (not shown here). Now where is the ia generated? For being a pure lexical, as shown in the previous Sub-section 4.2, should we assume that the ia is generated at the root node? This assumption is problematic, as the ia is not a pure consonantal form. It is more plausible to assume

- 978-90-04-44885-8

the nature of the infinitive absolute

141

that the ia is the root phrase of the verb in question, forming, e.g., /šVmVr/. But how do we get the actual vowels at the root phrase node?21 Since binyan Qal is the default “non-binyan” (see Section 3 for discussion), I believe that its vocalic melody is what the ia gets. This explains how we get examples like (3) where the ia ‘go unpunished’ is in Qal even though its equivalent finite verb is in binyan ii. But how do we get examples like (4) where the ia is in the same binyan ii as its equivalent finite verb? It seems to me that it is due to a process similar to progressive assimilation known from phonology, but I have not tried to figure out the mechanism yet.

5

Conclusions

With Harbour (1999, etc.), I have argued that the ia can only be used to represent pure lexical elements, or roots. Since the root in Hebrew consists of consonants only, to represent it one has to cast its radicals into one of the binyanim. Accordingly, I have concluded that the ia is not the root itself, syntactically speaking, but rather the root phrase of the verb in question, used to represent, or name the eventuality depicted by the respective root. With Marantz (1997) and Harley (2014), I have assumed that roots in language may take complements and form root phrases with them. Being a (root) phrase, may explain how the ia is able to take complements and why it cannot take modifiers. Some support for my claim that an ia in bh is used to name the root of the verb in question may be found in the way the root is referred to in Modern Hebrew. Hebraists and Hebrew teachers/students make use of two ways in representing roots. One way is to indicate the letters that correspond to each radical, putting a dot between them (in writing) and naming the letters (in speaking). E.g., š.m.r. ‘guard’ (read: šin, mem, reš), l.m.d. ‘study, learn’ (lamed, mem, dalet), etc. The other way is by indicating the letters corresponding to the radicals as if the root was an acronym, e.g., šm”r, lm”d. Crucially, though Modern Hebrew has lost the ia as a productive form (cf. n. 16 above), an acronym representing a root is read as an ia, usually in binyan Qal, e.g., šamor and lamod, respectively.

21

Note that Arad argues that we get the vowels only at the next level of the derivation, at the little v Phrase vP. I believe that this is, indeed, the case of the finite verbs and the infinitive (construct), but not of the ia.

- 978-90-04-44885-8

142

hatav

References Arad, Maya. 2005. Roots and Patterns: Hebrew Morphosyntax. Dordrecht: Springer. Borer, Hagit. 2005. In Name Only. Structuring Sense, Vol. i. Oxford: Oxford university Press. Borer, Hagit. 2014. Wherefore Roots? Theoretical Linguistics 40(3/4): 343–359 Borer, Hagit. 2015. Beyond Roots. A handout for a talk delivered at Roots iv. New York: New York University. Callaham, Scott N. 2010. Modality and the Biblical Hebrew Infinitive Absolute. (Deutsche Morgenländische Gesellschaft) Wisbaden: Harrasowitz Verlag. Cowper, Elizabeth and Vincent DeCaen. 2017. Biblical Hebrew: A formal perspective on the left periphery. Toronto Working Papers in Linguistics 38: 1–33. De Swart, Henriette. 1999. Position and meaning: Time adverbials in context. Pp. 336– 361 in Focus; Linguistic, Cognitive and Computational Perspectives, ed. P. Bosch and R. van der Sandt. Cambridge: Cambridge University Press. Doron, Edit. 2003. Agency and voice: The semantics of the Semitic templates. Natural Language Semantics 11: 1–67. Doron, Edit. 2008. The binyan’s contribution to the verb’s meaning. Pp. 57–88 in Theoretical Hebrew Linguistics, ed. G. Hatav. Jerusalem: The Hebrew University Magnes Press. [In Hebrew] Doron, Edit. 2018. The infinitive in Biblical Hebrew. 2nd Workshop on Biblical Hebrew Linguistics and Philology. Hebrew University of Jerusalem, June 27. Gesenius, Wilhelm. 1910. Hebrew Grammar. Trans. A.E. Cowley. Oxford: Clarendon Press. Goldenberg, Gideon. 1971. Tautological infinitive. Israel Oriental Studies 1: 36–85. Reprinted in 1998. Studies in Semitic Linguistics: Selected Papers by Gideon Goldenberg, 66–115. Jerusalem: The Hebrew University Magnes Press. Halle, Morris and Alec Marantz. 1993. Distributive morphology and the pieces of inflection. Pp. 111–176 in The View from Building 20, ed. Ken Hale and Samuel J. Keyser. Cambridge, MA: mit Press. Harbour, Daniel. 1999. The two types of predicate clefts: Classical Hebrew and beyond. Pp. 159–175 in Papers on Morphology and Syntax, Cycle Two, ed. Vivian Lin, Cornelia Krause, Benjamin Bruening, and Karlos Arregi. Cambridge, MA: mit Working Papers in Linguistics. Harbour, Daniel. 2007. Against PersonP. Syntax 10: 223–242. Harbour, Daniel. 2008. Klivaj predika, or Predicate cleft in Haitian. Lingua 118: 853–871. Harley, Heidi. 2014. On the identity of roots. Theoretical Linguistics 40 (3/4): 225–276. Hatav, Galia. 2017. The infinitive absolute and topicalization of events in Biblical Hebrew. Pp. 185–207 in Advances in Biblical Hebrew Linguistics: Data, Method, and Analyses, ed. Adina Moshavi and Tania Notarius. Winona Lake, IN: Eisenbrauns.

- 978-90-04-44885-8

the nature of the infinitive absolute

143

Hatav, Galia. 2020. Verb phrase secondary predication: Biblical Hebrew as a case study. Linguistics 58(2): 363–378. Kim, Yoo-Ki. 2009. The Function of the Tautological Infinitive in Classical Biblical Hebrew. Winona Lake, IN: Eisenbrauns. Kratzer, Angelika. 2005. Building resultatives. Pp. 177–212 in Event Arguments: Foundations and Applications, ed. Claudia Maienborn and Angelika Wöllstein. Tübingen: Niemeyer. Labelle, Marie. 2014. Roots in models of grammar. Theoretical Linguists 40(3/4): 401– 414. Marantz, Alec. 1997. No escape from syntax: Don’t try morphological analysis in the privacy of your own lexicon. University of Pennsylvania Working Papers in Linguistics. 4(2), article 14. Moshavi, Adina. 2010. Word order in the Biblical Hebrew Finite Clause. Winona Lake, IN: Eisenbrauns. Portner, Paul, Miok Pak and Raffaella Zanuttini. 2019. The speaker-addressee relation at the syntax-semantics interface. Language 95(1): 1–39. Rothstein, Susan. 2004. Structuring Events. Oxford: Blackwell Publishing. Van der Merwe, Christo H.J., Jackie A. Naudé and Jan H. Kroeze. 2017. Biblical Hebrew Reference Grammar. 2nd ed. London: Bloomsbury. Waltke, Bruce K. and Michael O’Connor. 1990. An Introduction to Biblical Hebrew Syntax. Winona Lake, IN: Eisenbrauns. Ylikaski, Jussi. 2003. Defining non-finites: Action nominals, converbs and infinitives. Journal of Linguistics 16: 185–237.

- 978-90-04-44885-8

chapter 8

The Infinitive in Biblical Hebrew Edit Doron

1

Introduction

Biblical Hebrew (bh) verbal forms manifest rich inflection within the finite (Fin) clause, encoding the functional categories of temporality (T), mood (Mood), grammatical aspect (Asp), and modality (Mod). These categories have been widely discussed in the literature, and their relative role is still under debate (recently Hatav 1997, 2008, Joosten 2002, Cook 2006, 2012 and others). In particular, Asp and Mod have proven hard to disentangle in the morphology of the bh verb. The present work will reflect this by assuming that these two categories are composed together as Asp/Mod (am) in the inflection of the verb. The objectives of this essay are to show that: i. The same functional categories which determine the inflection of the bh finite verb also determine the feature specification of the bh infinitive. (In particular, the functional categories of the bh infinitive are clausal rather than nominal (section 4).) ii. bh has a single infinitive combined with different inflectional categories, yielding the so-called Infinitive Absolute and Infinitive Construct, which, together with the finite (Fin) verb, gives rise to 4 clause types: Fin, Possinf, pro-inf, and Nom-inf. iii. These clause types are classified by their highest functional projection TFin, T, am, Mood, which accounts for their distribution. iv. There is a concomitant 4-way alternation of attachment options of subject and object clitics to the verb: [+Scl+Ocl], [+Scl−Ocl], [−Scl+Ocl], [−Scl−Ocl]. The examples in (1)–(5) illustrate, using the same verb remember, the Fin and infinitival clause types in their typical functions. The Fin construction is a clause in the indicative mood, or in a variety of irrealis moods (imperative/ jussive/ cohortative), and Nom-inf is an irrealis root clause. Irrealis mood endows the clause with illocutionary force. Poss-inf and pro-inf are embedded clauses lacking force, and their distribution will be discussed in detail below.1

1 The distinction between the bh Poss-inf and pro-inf, which serves the base of the distinc-

© Edit Doron, 2021 | doi:10.1163/9789004448858_009

- 978-90-04-44885-8

145

the infinitive in biblical hebrew

Poss-inf often functions as a temporal adverbial, and pro-inf as a purpose adverbial: (1) Fin Indicative way=yizkor yosep̄ ʾeṯ ha=ḥălomoṯ ʾăšɛr and=remember.pret.3ms Joseph om the=dreams rel ḥalam la=hɛm dream.perf.3ms to=3mp ‘Then Joseph remembered the dreams which he had dreamed about them’ (Gen. 42:9) (2) Fin Imperative lə=ʾaḇraham lə=yiṣḥaq la=ʿăḇaḏɛ=ḵa zəḵor remember.impv.2ms to-servants=poss.2mp to-Abraham to-Isaac u=lə=yaʿăqoḇ and-to-Jacob ‘Remember Your servants, Abraham, Isaac, and Jacob’ (Deut 9:27) (3) Nom-inf zaḵor ʾɛṯ yom haš=šabbaṯ lə=qaddəš=o remember.ia om day.of the-sabbath to-sanctify.inf-acc.3ms ‘Remember the Sabbath day, to keep it holy’ (Exod 20:8) (4) Poss-inf gam baḵinu šam yašaḇnu ʿal nahăroṯ baḇɛl by rivers.of Babylon there sit.perf.1cp also weep.perf.1cp bə=zoḵr=enu ʾɛṯ ṣiyyon when-remember.inf-poss.1cp om Zion ‘By the rivers of Babylon, there we sat down and wept when we remembered Zion’ (Ps 137:1) (5) pro-inf2 wə=hayəṯa haq-qɛšɛṯ b=ɛ=ʿanan and=be.perf.mod.3fs the=rainbow.fs in=the-cloud

tion between the Modern Hebrew Gerund and Infinitive, is already found in Doron 2016, 2019. 2 It should be clear that the contrast between the overt vs. covert subject in Poss-inf vs. pro-inf is grammatical and has nothing to do with the pragmatic contrast between overt and null

- 978-90-04-44885-8

146

doron

u=rəʾiṯi=ha li=zkor bəriṯ ʿolam and=see.impf.1cs=acc.3fs to=remember.inf covenant.of eternity ben ʾɛ̆lohim u=ben kol nɛp̄ ɛš ḥayya between God and=between all soul living ‘The rainbow shall be in the cloud, and I will look on it to remember the everlasting covenant between God and every living creature’ (Gen 9:16) The Poss-inf subject is in the possessive case, overtly marked for pronominal subjects, in particular the 1st person singular, where the possessive marking differs from accusative marking of the corresponding object clitics in the proinf construction. Thus, the 1st person object clitic -eni in (6) differs in form from the 1st person subject clitic -i in (7): (6) pro-inf ha=lə=[horḡ=eni pro] ʾatta ʾomer ka=ʾăšɛr Q=to=[kill.inf=acc.1cs pro] you say.ptcp.ms as=rel haraḡta ʾɛṯ ham=miṣri killed.perf.2ms om the=Egyptian ‘Do you intend to kill me as you killed the Egyptian?’ (Exod 2:14) (7) Poss-inf wə=lo yihyɛ ḇa=ḵɛm nɛḡɛp̄ lə=mašḥiṯ and=neg be.mod at=2mp plague to=destroy.ptcp.ms bə=[hakkoṯ=i bə=ʾɛrɛṣ miṣrayim] when=[strike.inf=poss.1s at=land.of Egypt] ‘And the plague shall not be on you to destroy you when I strike the land of Egypt’ (Exod 12:13)

pronominal subjects in finite clauses like (i), where the overt/covert choice has to do with information structure: (i) 36u=ḇǝ=heʿaloṯ hɛ=ʿanan me=ʿal ham=miškan yisʿu bǝne and=in=lift.pass.inf the=cloud from=over the=tabernacle travel.impf.3mp sons.of yiśraʾel bǝ=ḵol masʿe=hɛm 37wǝ=ʾim loʾ yeʿalɛ hɛ=ʿanan Israel in=all.of travels=3mp and=cond neg lift.pass.impf.3ms the=cloud wǝ=loʾ yisʿu ʿaḏ yom heʿaloṯ=o and=neg travel.impf.3mp pro until day.of lift.pass.inf=3ms ‘and when the cloud was lifting from upon the tabernacle, the sons of Israel would travel in all their travels, but if the cloud was not lifted up, pro (they) would not travel until it had lifted’ (Exod 40:36–37).

- 978-90-04-44885-8

the infinitive in biblical hebrew

147

(8) Fin [+Scl+Ocl] ʿăśiṯi=w made.perf.1s=acc.3ms (9) Nom-inf [−Scl−Ocl] ʿaśo make.ia (10) Poss-inf [+Scl−Ocl] ba=ʿăśoṯ=i when=make.inf=poss.1s (11) pro-inf [−Scl+Ocl] la=ʿăśoṯ=eni to=make.inf=acc.1s (12) Fin hag=goyim ʾăšɛr hoṣeṯi=m lə=ʿene=hɛm the=peoples rel bring.out.perf.1cs=acc.2mp to-eyes.of=poss.3mp ‘the peoples in whose sight I had brought them out’ (Ezek 20:14) (13) pro-inf lə=hoṣiʾ=am me=ʾɛrɛṣ miṣrayim to=bring.out.inf=acc.3mp from-land.of Egypt ‘for bringing them out of the land of Egypt’ (Exod 12:42) (14) Poss-inf bə=hoṣiʾ=i ʾoṯ=am me=ʾɛrɛṣ miṣrayim when=bring.out.inf=poss.1cs om=3mp from-land.of Egypt ‘when I brought them out of the land of Egypt’ (Lev 23:43) *bə=hoṣiʾ=i=m me=ʾɛrɛṣ miṣrayim when-bring.out.inf=poss.1cs=acc.3mp from-land.of Egypt The ungrammaticality in the second example in (14) above is not due to “heaviness” of two combined clitics, since even if the subject is not a pronominal clitic but a full lexical item, even then an object clitic is impossible in the Possinf construction:

- 978-90-04-44885-8

148

doron

(15) Poss-inf way=yaśɛm yhwh lə=qayin ʾoṯ lə=ḇilti hakkoṯ ʾoṯ=o and=put.pret.3ms Yhwh to=Cain mark to=neg kill.inf om=3ms kol moṣʾ=o any find.ptc.ms=poss.3ms ‘And the Lord set a mark on Cain, lest anyone finding him should kill him’ (Gen 4:15) *lə=ḇilti hakkoṯ=o kol moṣʾ=o To=neg kill.inf=acc.3ms any find.ptc.ms=poss.3ms (16) bɛn ʾarbaʿim šana ʾanoḵi bi=šəloăḥ mošɛ ʿɛḇɛḏ yhwh son.of forty year I when=send.inf Moses servant.of Yhwh ʾoṯ=i om=1cs ‘I was forty years old when Moses the servant of the Lord sent me …’ (Josh 14:7) *bə=šolḥ=eni mošɛ when-send.inf-acc.1cs Moses The four clausal types are distinguished by what Wurmbrand 2001, 2014 has called their restructuring signature: how much of the hierarchy of clausal functional categories is projected in the clause.

(17)

Temporal Fin

Poss-inf

Non-temporal pro-inf

Nom-inf

- 978-90-04-44885-8

149

the infinitive in biblical hebrew

(18) Highest inflection +T + verbal − verbal

Fin [+Scl+Ocl] Poss-inf [+Scl−Ocl]

−T pro-inf [−Scl+Ocl] Nom-inf [−Scl−Ocl]

This corresponds to what has often been remarked in the literature: object clitics attach to inflection which is characteristically verbal (e.g. in Romance, Cardinaletti and Shlonsky 2004, Cardinaletti 2008).3

2

One Infinitive, Different Inflectional Categories

Historically, the Infinitive Absolute (ia) is the original infinitive, also found in Akkadian (Blau 1979: §30), while the Infinitive Construct (ic) has been claimed to originate in a different Proto-Semitic form, related to the imperfective (Bauer and Leander 1922: §43). Yet, synchronically in bh, I would like to propose that the two are actually two inflectional forms of a single infinitive. The derivations are shown in the following table:4

3 Indeed, the participle, which is inflected as a noun, mostly takes genitive marked object clitics: (i) mōṣʾ=i məp̄ alləṭ=i məśanʾ=i find.ptcp.ms=poss.1cs deliver.ptcp.ms=poss.1cs hate.ptcp.ms=poss.1cs ‘anyone who finds me’ ‘He delivers me’ ‘he who hates me’ (Gen 4:14) (Ps 18:49) (Job 31:29) šōlḥ=i send.ptc.ms=poss.1cs ‘He who sent me’ (2Sam 24:13) Yet the participle exhibits noun/verb duality, and there are also a few cases where it heads a finite clause with accusative object clitics: (ii) hā=ʾel ha=məʾazzər=eni ḥāyil the=God rel=arm.ptc.ms=acc.1cs strength ‘It is God who arms me with strength’ (Ps 18:33[32]). 4 The Infinitive Absolute of some verbs in derived templates also has exponents constructed by analogy to the Simple Active template, e.g. nilḥōm ‘fight’, yassōr ‘chasten’. Note passive infinitives are extremely rare.

- 978-90-04-44885-8

150

doron

(19) Agency → Voice ↓ Active Middle Passive

Simple

šamor observe.ia hiššaḇaʿ vow.ia

šəmor observe.ic hiššaḇaʿ vow.ic

Intensive

kabbed honor.ia hitnappel attack.ia gunnoḇ be-stolen.ia

kabbed honor.ic hitnappel attack.ic hukkabbes be-laundered.ic

Causative

haqreḇ offer.ia

haqriḇ offer.ic

huggeḏ hullɛḏɛṯ be-told.ia be-givenbirth.ic

The Infinitive Absolute is the citation form of the verb and has adverbial uses (typically bare of arguments). The adverbial infinitive either directly modifies the inflected verb (20)–(21), as described in Callaham 2014, Hatav 2017, and references therein, or it modifies the vp (22)–(23): (20) raʾo raʾiṯi ʾɛṯ ʿŏni ʿamm=i ʾăšɛr bə=miṣrayim see.ia saw.1cs om oppression.of people=poss.1cs rel in=Egypt ‘I have surely seen the oppression of My people who are in Egypt’ (Exod 3:7) (21) šoḇ ʾašuḇ ʾel=ɛḵa k=a=ʿeṯ ḥayya wə=hinne return.ia return.impf.1cs to=2ms as=the=season living and=behold ḇen lə=śara ʾišt=ɛḵa son to=Sarah wife=your ‘I will surely return to you next year, and behold, your wife Sarah will have a son!’ (Gen 18:10) (22) way=yašuḇu ham=mayim me=ʿal ha=ʾarɛṣ haloḵ and=recede.pret.3mp the=waters from=upon the=earth go.ia wa=šoḇ and=recede.ia ‘And the waters receded continually from the earth’ (Gen 8:3)

- 978-90-04-44885-8

the infinitive in biblical hebrew

151

(23) b=ay=yom ha=hu ʾaqim ʾɛl ʿeli ʾeṯ kol ʾăšɛr in=the=day the=that perform.impf.1cs to Eli om all rel dibbarti ʾɛl beṯ=o haḥel wə=ḵalle speak.perf.1cs to house=poss.3ms begin.ia and=end.ia ‘In that day I will perform against Eli all that I have spoken concerning his house, from beginning to end’ (1Sam 3:12) For the purposes of the present article, I will mostly ignore the adverbial use (20)–(23), where the infinitive is “bare” of any functional category, and hence is not clausal and does not introduce a subject.5 I will only be interested in the uses of the infinitive which involve clausal constructions with functional categories, and hence a subject. The present work shows that there are two types of such constructions, one classified together with finite clauses as having conversational force (through being specified for Irrealis Mood), and the other as lacking such force.

3

Two Types of Infinitival Clauses

3.1 [−Indicative] Infinitival Clauses The first type is a clause with imperative force (including jussive and cohortative). The inflectional class of the infinitive in this clause type is the Infinitive Absolute.6 According to the analysis proposed here, this is due to the fact that the only functional category specified in this construction is Mood, with a [−Indicative] value interpreted as imperative force.7 Since the tam categories in the clause are unspecified, there is no inflection to alter the citation form of the infinitive, nor to provide an attachment site for subject and object clitics. And as there is no temporal anchoring of the verb to the speech act, these

5 I consider adverbial also the “sequential use”, where the Infinitive Absolute, together with its internal arguments, is conjoined to a previous clause and interpreted within the scope of the latter’s inflection and subject, as in (i): (i) zoṯ naʿăśɛ la=hɛm wə=haḥăye ʾoṯ=am This do.impf.mod.1cp to=3mp and-let.live.ia om=3mp ‘This we will do to them: (we will) let them live’ (Josh 9:20). 6 The same is true in Arabic, where the qatāli form which corresponds to the Infinitive Absolute also serves as an imperative (Wright 1874, 1: 62). 7 As is known from the literature (Portner 1997 and references therein), Mood is the category which determines the conversational force of a root clause (Indicative, Imperative, etc.).

- 978-90-04-44885-8

152

doron

sentences tend to be generic in interpretation unlike the discourse-bound interpretation of the finite imperative.8 I call this type Nom-inf, since it includes a nominative subject, either a null pro (an addressee-oriented logophoric pronoun), as in (24), or a lexical dp, as in (25). As is to be expected of imperative clauses, they are typically root clauses (Palmer 2001). (24) Nom-inf šamor ʾɛṯ yom haš=šabbaṯ lə=qaddəš=o observe.ia om day.of the=sabbath to=sanctify.ic=acc.3ms ‘Observe the Sabbath day, to keep it holy’ (Deut 5:12) (25) Nom-inf wə=zoṯ toraṯ ham=minḥa haqreḇ ʾoṯ=ah bəne and=this.f law.f.of the=grain.offering.f offer.ia om=3fs sons.of ʾahăron lip̄ ne yhwh ʾɛl pəne ham=mizbeăḥ Aaron before Yhwh on surface.of the=altar ‘This is the law of the grain offering: The sons of Aaron shall offer it on the altar before the Lord’ (Lev 6:7) (26) Fin Imperative šəmor ʾɛṯ ha=ʾiš haz=zɛ guard.imp.ms om the=man the=this ‘Guard this man’ (1Kgs 20:39) (27) Fin Imperative wə=ʾatta haqreḇ ʾel=ɛḵa ʾɛṯ ʾahăron ʾaḥ=iḵa and=you take.imp.ms to=2ms om Aaron brother=poss.2ms ‘Now you take Aaron your brother’ (Exod 28:1) 3.2 [+Indicative] Infinitival Clauses The form of the infinitive in the second type of construction is the Infinitive Construct. This form allows the attachment of pronominal clitics, something that is strictly disallowed in the Nom-inf construction, which has the Infinitive Absolute form. As we have seen, this difference is due to the fact that subject

8 Thus (24) is a general obligation, not restricted to any particular time and place, whereas (26) is restricted to the speech situation. The same contrast is found between the generic lō+Modal negation and the eventive ʾal+Jussive negation among Fin clauses.

- 978-90-04-44885-8

153

the infinitive in biblical hebrew

and object clitics attach to the relevant functional categories, which are present in the second type of construction but not in the Nom-inf construction. The first subtype, familiar from other languages, has a null pronominal anaphor subject (pro), typically controlled by another dp in the linguistic context. This is the pro-inf type. We will now see that it does not have temporal specification, i.e. no T functional category, yet it does have Asp/Mod specification. As it is not specified for T, the subject is not assigned case, and is hence pro. As it is specified for Asp/Mod, which is verbal inflection, it allows object clitics. The second subtype, Poss-inf, has an overt subject with possessive case,9 I will now argue that this construction is temporal and hence includes specification of the functional category T. As it has T specification, but not a finite one, it allows subject but not object clitics. It is distinguished from finite clauses, with a finite T (and hence both subject and object clitics). I assume that it is non-finite T which assigns possessive case to the subject, in parallel to the nonfinite -ing functional category which assigns accusative case to the subject of Acc-ing gerunds in English according to Reuland’s 1983 analysis.10 In the following examples of pro-inf and Poss-inf, notice the Infinitive Construct foms rəʾoṯ ‘see’ and šuḇ ‘return’ in (28)–(29) and (30)–(31), which differ from the corresponding Infinitive Absolute forms raʾo and šoḇ of the same verbs in (20)–(23) above. (28) pro-inf way=yišlaḥ šaʾul ʾɛṯ ham=malʾaḵim li=[rʾoṯ pro ʾɛṯ and=send.pret.3ms Saul om the=messengers to=[see.ic pro om dawiḏ] David] ‘Then Saul sent the messengers back to see David’ (1 Sam 19:15)

9

10

The possessive case is a marked case of the subject in other languages as well, such as Alaskan Yup’ik (Abney 1987: 28), Finnish (Kiparsky 2001), Ladakhi, Lak, Niue (Lander 2011: 590), Tagalog (Aldridge 2006, Collins 2017), Tzutujil Maya (Abney 1987: 31), and others. It has often been noticed that the bh Infinitive Construct subsumes properties of both infinitives and gerunds in other languages. pro-inf subsumes both the English infinitive and the pro-ing gerund. Poss-inf parallels the English Acc-ing gerund, despite the morphological difference between accusative and genitive. Poss-inf does not parallel the English Poss-ing, which is a nominal rather than a clausal construction (Pires 2001, 2006, 2007; Moulton 2004).

- 978-90-04-44885-8

154

doron

(29) Poss-inf wə=ḵi=[rʾoṯ šaʾul ʾɛṯ dawiḏ] … ʾamar ʾɛl ʾaḇner and=as=[see.ic Saul om David] say.perf.3ms to Abner ‘When Saul saw David …., he said to Abner, …’ (1 Sam 17:55) (30) pro-inf wa=yəšallaḥ ʾɛṯ hay=yona wə=lo yasəp̄ a and=send.pret.3ms om the=dove.f and=neg repeat.impf.3fs [šuḇ pro ʾel=aw ʿoḏ] [return.ic pro to=3ms again] ‘… and [he] sent out the dove, which did not return again to him again’ (Gen 8:12) (31) Poss-inf bə=[šuḇ yhwh ʾɛṯ šiḇaṯ ṣiyyon] hayinu when=[return.ic Yhwh om return.of Zion] be.perf.1cp kə=ḥolmim as=dream.ptc.mp ‘When the Lord brought us back to the city of Zion, we were like dreamers’ (Ps 126:1) The two constructions contrast sharply in distribution. The examples in (29) and (31) are temporal adverbials, and none of the examples in (28) and (30) are. This is not an accident, as it is the case in general that temporal prepositions only take Poss-inf complements, never pro-inf complements. This shows that Poss-inf clauses include T specification in their structure, whereas pro-inf clauses do not. Thus, only the former can serve as Specifier of the main clause T head (Cinque 1999). pro-inf clauses function as purpose clauses, as in (28) and (33), i.e., they are Asp/Mod phrases (am for short), which are Specifiers to the Asp/Mod head of the main clause:11,12

11

12

Purpose clauses are part of infinitival clauses which “are a group which displays a characteristic future-oriented, irrealis semantics” (Portner 1997: 183). Yet, as argued by Wurmbrand 2001, 2014, the seeming temporal relation of the infinitival clause to the main clause is not due to T but to Mod, which determines the inherent future orientation of purposes. Purpose clauses are distinct from rationale clauses (Jones 1985, Verstraete 2008), which can be expressed by the Poss-inf construction. The latter describes a result event, as in (i)

- 978-90-04-44885-8

the infinitive in biblical hebrew

155

(32) Poss-inf—Spec of T: temporal adverbial (cf. 6b) wə=ḵi=rəʾoṯ šaʾul ʾɛṯ dawiḏ ʾamar ʾɛl ʾaḇner and=as=see.ic Saul om David say.perf.3ms to Abner ‘When Saul saw David … he said to Abner’ (1 Sam 17:55)

below, not necessarily the outcome an agent’s intentions, unlike the intentional/modal characterization of purpose clauses: (i) Poss-inf: rationale clause wə=šaməru dɛrɛḵ yhwh ləmaʿan [haḇi yhwh ʿal and=keep.perf.mod.3mp way.of Yhwh in.order.to [bring.inf Yhwh on ʾaḇraham ʾeṯ ʾăšɛr dibbɛr ʿal=aw] Abraham acc rel speak.perf.3ms to=3ms] ‘they should keep the way of the Lord, …, in order that the Lord may bring to Abraham what He has spoken to him’ (Gen 18:19) One syntactic difference which distinguishes purpose and rationale clauses is that only the former allows an additional controlled empty category (glossed as ej), as in (ii): (ii) pro-inf: purpose clause ḥamal ha=ʿami ʿal meṭaḇ haṣ=ṣon wə=ha=baqarj ləmaʿan spare.perf.3ms the=peoplei on best.of the=cattle and=the=beefj in.order.to [zəḇoăḥ proi ej la=yhwh ʾɛ̆loh=ɛḵa] [sacrifice.inf proi ej to=Yhwh God-yours] ‘The people spared the best of the sheep and the oxen to sacrifice to the Lord your God’ (1Sam 15:15).

- 978-90-04-44885-8

156

doron

(33) pro-inf—Spec of Asp/Mod: purpose adverbial way=yerɛḏ yhwh li=rəʾoṯ ʾɛṯ ha=ʿir and=descend.pret.3ms Yhwh to=see.ic om the=city ‘The Lord came down to see the city’ (Gen 11:5)

Infinitival clauses also function as complements, and as such are selected by different types of verbs. Poss-inf clauses are propositional tp s, and are hence selected by propositional attitude verbs, such as know (Gen 19:35, Jer 15:15), remember (Jer 2:2, 18:20), consent (Gen 19:21), hear (1 Sam 14:27), see (Isa 52:8), illustrated in (34).

- 978-90-04-44885-8

the infinitive in biblical hebrew

157

(34) Poss-inf—Complement of propositional attitude verb zaḵarti l=aḵ … lɛḵt=eḵ ʾaḥăr=ay remember.perf.1cs to=2fs go.ic=poss.2fs behind=1cs ‘I remember … your coming after me’ (Jer. 2:2)

pro-inf clauses are Asp/ModP, and hence complements of aspectual verbs, e.g. begin (Judg 20:39), repeat (1Sam 15:35), stop (1Sam 23:13), finish (Lev 16:20), or modal verbs such as be able (Deut 7:22), want (1 Sam 19:2), intend (Exod 2:14), plan (Deut 19:19), refuse (Num. 20:21), give up (1 Sam 27:1), order (2 Sam 17:14), prevent (Num 22:16), illustrated in (35).13

13

As noted in Doron (2018), propositional attitude verbs receive a modal interpretation when they take pro-inf complements, e.g. know (1 Kgs 3:7), think (1Sam 18:25), hear (Gen 39:10), fear (Judg 7:10), remember (Ps 109:16) and others.

- 978-90-04-44885-8

158

doron

(35) pro-inf—Complement of Modal/Aspectual verb ha=lə=horḡ=eni ʾatta ʾomer Q=to=kill.ic=acc.1cs you intend.ptc.ms ‘You intend to kill me’ (Exod 2:14, cf. (2a))

Aspectual and modal verbs in the (35) structure are control verbs expressing root modality (ability, deontic). When the same verbs modify the aspectual and modal dimension of a state/event which is not determined by the actions or abilities of an agent, their modality is interpreted as circumstantial, they do not have an agent, and function as raising verbs (Hacquard 2011). The following examples describe the beginning (36), repetition (37), possibility (38) of an event/state, independently of an agent. The infinitival clause, which lacks T, undergoes restructuring with the main clause, and the subject of the infinitive is assigned nominative case by the main clause tam: li=[həyoṯ lə=śara ʾoraḥ k=an=našim] (36) ḥaḏal cease.perf.3ms to=[be.ic to-Sarah way as=the=women] ‘the way like woman had ceased to be for Sarah’ (Gen 18:11)

- 978-90-04-44885-8

the infinitive in biblical hebrew

159

(37) ki lo yuḵlu lə=[heraʾoṯ pro3mp] la=[ḇo pro3mp For neg can.impf.3mp to=[see.pass.ic pro3mp] to=[come.ic pro3mp ha=ʿir=a ] the=city=to ] ‘because they could not be seen coming to the city’ (2 Sam 17:17) (38) wat=tosɛp̄ ham=milḥama li=[həyoṯ ham-milḥama] and=recur.pret.3fs the=war.f to=[be.ic the-war.f] ‘And there was war again’ (Gen. 19:8) (39) Complement of a raising Modal/Aspectual verb, e.g., (38)

An additional point is that pro-inf, which has less structure, allows the amalgamation of the preposition l= ‘to’ into the syllabic structure of the verb, which results in the lack of spirantization of the middle root consonant, as in (40). Joüon (1923 §49f.) shows that this is different from the form of Poss-inf, i.e., the form of the verb in (41).14 (40) V+Asp/Mod li=npol li=šdoḏ li=ḇkoṯ li=špoṭ to=fall.ic to=rob.ic to=cry.ic to=judge.ic (Ps 118:13) (Jer 47:4) (Gen 43:30) (Exod 18:13) 14

I am indebted to Vince DeCaen for drawing my attention to this point.

- 978-90-04-44885-8

160

doron

(41) V+T ki=nəp̄ ol bi=ḡəḏol šəp̄ oṭ as=fall.ic when=grow.ic judge.ic (2Sam 3:34) (Exod 15:16) (Ruth 1:1) I summarize in (42) the morpho-syntactic characteristics of the different finite and infinitival clauses, where the relevant functional categories are ordered by the following hierarchy: T < Asp/Mod < Mood < Voice.

(42)

Phrasal Functional spine Force Verb form Subj. Subj. Obj. category case clitic clitic tpFin Fin Poss-inf tp pro-inf am-P Nom-inf MoodP

4

+TFin+am +Mood +T +am+Mood –T+am+Mood −T−am+Mood

+ − − +

Finite Inf. Constr Inf. Constr Inf. Abs.

Nom Poss − Nom

+ + − −

+ − + −

The Clausal Nature of the Infinitive Construction

The Hebrew grammatical tradition views the infinitive absolute as verbal and the infinitive construct (i.e., the bound form) as nominal. The European grammatical tradition views both infinitives as mixed nominal/verbal categories. But the approach above has analysed (i) the infinitive as V rather than N, not even a deverbal N, and (ii) the functional categories projected by V as clausal rather than nominal—similarly to what has been shown by Pires 2006 for the English pro-ing and Acc-ing gerunds, i.e., that they are clausal rather than nominal.15 There is good evidence for both points. First, the infinitive assigns accusative case to its direct object, as could be seen in all the examples above where the infinitive had a direct object.16 Moreover, object clitics attached to the infinitive are always accusative rather than

15 16

In English, Poss-ing gerunds are nominal. Modern Hebrew allows nominalized verbs to assign accusative case as well, which is a a marked option cross-linguistically. This phenomenon originates in Medieval Hebrew under Arabic influence (Blau 1990, Goshen-Gottshtein 1951/2006). Yet it is not found in Biblical Hebrew, where forms such as ʾahăḇa ‘love’, which were later recategorized as nouns, are still infinitives:

- 978-90-04-44885-8

161

the infinitive in biblical hebrew

genitive. In the case of nominal forms, such as the participle, one mostly finds genitive object clitics (fn. 3). Second, the infinitive has no nominal morphological inflection of gender, number, or definiteness.17 The infinitive is case marked in a few examples by the object marker ʾɛṯ, as in (43), but so are Fin cps as in (44): (43) wə=ṣeṯ=əḵa u=ḇoʾ=ăḵa yaḏaʿti and=exit.ic=poss.2ms and=enter.ic=poss.2ms know.perf.1cs wə=ʾeṯ hiṯraggɛz=ḵa ʾel=ay and=om rage.ic=poss.2ms to=1cs ‘But I know … your going out and your coming in, and your rage against me’ (2Kgs 19:27) (44) zəḵor ʾal tiškaḥ ʾeṯ ʾăšɛr hiqṣap̄ ta remember.imv.2ms neg forget.juss.2ms om rel provoke.perf.2ms ʾɛṯ yhwh ʾɛ̆loh=ɛḵa b=am=midbar om Yhwh God=poss.2ms in=the=desert ‘Remember! Do not forget how you provoked the Lord your God to wrath in the wilderness’ (Deut 9:7) Third, the infinitive is not modified by adjectives but by adverbs, such as the adverbs heṭeḇ ‘well’, ʿoḏ ‘more’, and maher ‘at once’ in (45)–(47): (45) wa=ʾɛkkoṯ ʾoṯ=o ṭaḥon heṭeḇ ʿaḏ ʾăšɛr daq lə=ʿap̄ ar and-crush.pret.1cs om=3ms grind.ia well until rel fine to=dust ‘and I crushed it and ground it very small, until it was as fine as dust’ (Deut 9:21) (46) ʾal tosɛp̄ dabber ʾel=ay ʿoḏ b=ad=daḇar haz=zɛ neg recur.juss.2ms speak.ic to=1cs more in=the=matter the=this ‘Speak no more to me of this matter’ (Deut 3:26)

17

(i) bə=ʾahăḇaṯ yhwh ʾɛṯ yiśraʾel lə=ʿolam in=love.inf Yhwh om Israel to=forever ‘Because the Lord has loved Israel forever’ (1 Kgs 10:9). There are few cases where the infinitive happens to have feminine morphology, such as love in the previous fn. There are even fewer cases where the infinitive is preceded by the article the.

- 978-90-04-44885-8

162

doron

(47) lo tuḵal kalloṯ=am maher neg be.able.impf.2ms destroy.ic=acc.3mp quickly ‘you will be unable to destroy them at once’ (Deut 7:22) Fourth, despite the genitive case marking of its subject, the infinitive in the Poss-inf construction is not a noun. It does not head a bound construction. Unlike the nominal bound construction where the bound state noun must be absolutely adjacent to its complement, the same is not true of the infinitive in the Poss-inf construction. Here, no adjacency is required. The subject of the infinitive is separated from the verb in many examples, something which never happens in a bound construction. The subject is separated from the infinitive verb hakkoṯ in (48) by the object marker ʾoṯo, and similarly in the other examples in (49)–(51): (48) lə=ḇilti [hakkoṯ ʿoṯ=o kol moṣʾ=o] to=neg [kill.ic om=3ms any find.ptc.ms=poss.3ms] ‘… lest anyone finding him should kill him’ (Gen 4:15) (49) wa=yəhi kə=[noăḥ ʿăl=ehɛm ha=ruăḥ] and=be.pret.3ms as=[rest.ic on=3mp the=spirit] way=yiṯnabbəʾu and=prophesy.pret.3mp ‘and it happened, when the Spirit rested upon them, that they prophesied’ (Num 11:25) (50) ha=[məšol b=aḵɛm šiḇʿim ʾiš] Q=[reign.ic at=2mp seventy man] ‘[Which is better for you?] Is it all seventy … ruling over you …?’ (Judg 9:2) (51) wə=haya la-[nus šamma kol roṣeăḥ] and=be.perf.mod.3ms to-[flee.ic there any murder.ptc.ms] ‘it shall be that any manslayer may flee there’ (Deut 19:3) We now turn to showing that embedded infinitival clauses have the distribution of embedded clauses rather than nominal projections. They are found as complements of prepositions, but only prepositions which take clausal arguments, including Fin cp s, for example the preposition kə- ‘as’ expressing similarity:

- 978-90-04-44885-8

the infinitive in biblical hebrew

163

(52) lo teṣe kə=ṣeṯ ha=ʿăḇaḏim neg exit.impf.3fs as=exit.ic the=slaves.m ‘she shall not go out as the male slaves go out’ (Exod 21:7) (53) u=mašaḥta ʾoṯ=am ka=ʾăšɛr mašaḥta ʾɛṯ and=annoint.perf.mod.2ms om=3mp as=rel annointed.2ms om ʾăḇi=hɛm father=poss.3mp ‘You shall anoint them, as you anointed their father’ (Exod 40:15) Prepositions like ʿim ‘with’, which only take dps complements and do not take Fin-cp complements, also do not take infinitival clauses. On the other hand, prepositions like yaʿan ‘since’, which do not take nominal complements in Classical bh but do take Fin-cp s, also take infinitival clauses: (54) yaʿan hiṯmakkɛr=əḵa la=ʿăśoṯ ha=raʿ bə=ʿene yhwh since sell.self.ic=poss.2ms to=do.ic the=evil in=eyes.of Yhwh hin=əni meḇi ʾel=ɛḵa raʿa behold=1cs bring.ptc.ms to-2ms calamity ‘Because your selling yourself to do evil in the sight of the Lord, behold, I will bring calamity on you’ (2Kgs 21:20–21) (55) yaʿan ki niḵnaʿ mip=pan=ay lo since that submitted.perf.3ms from=face=poss.1s neg ʾaḇi ha=raʿa bə=yam=aw bring.impf.1cs the-calamity in=days=poss.3ms ‘Because he humbled himself before me, I will not bring the calamity in his days.’ (1Kgs 21:29) The quantifier kol ‘all’, typically used with noun phrases, is found bound to infinitival clauses, but so it is with Fin cps: (56) li=šəmoăʿ ʾăl=ehɛm bə=ḵol qorəʾ=am ʾel=ɛḵa to=listen.ic to=3mp when=any call.ic=poss.3mp to=2ms ‘to listen to them whenever they call to you’ (1 Kgs 8:52)

- 978-90-04-44885-8

164

doron

(57) way=yošaʿ yhwh ʾɛṯ dawiḏ bə=ḵol ʾăšɛr halaḵ and=save.pret.3ms Yhwh om David where=any rel go.perf.3ms ‘So the Lord preserved David wherever he went’ (2 Sam 8:6) Other nouns as well, such as yom ‘day’, which are constructed to infinitival clauses, are also constructed to Fin cps: (58) ləmaan tizkor ʾɛṯ yom ṣet=ḵa in.order.to remember.impf.2ms om day.of exit.ic=poss.2ms me=ʾɛrɛṣ miṣrayim from=land.of Egypt ‘that you may remember the day you came out of the land of Egypt’ (Deut 16:3) (59) yom ʾăšɛr yəlaḏaṯ=ni ʾimm=i ʾal day rel bore.perf.3fs=acc.1cs mother=poss.1cs neg yəhi ḇaruḵ be.juss.3ms blessed ‘Let the day not be blessed in which my mother bore me!’ (Jer 20:14) Moreover, like Fin-cp s, infinitival clauses function as relative clauses. (60) has a Fin-cp relative clause, (61)—a pro-inf relative clause, and (62)—a Poss-inf relative clause. (60) Fin kol baśar ʾăšɛr yaqriḇu la=yhwh baśar all flesh [rel bring.impf.3mp to=Yhwh flesh] ‘all flesh which they bring to the Lord’ (Num 18:15) (61) pro-inf lɛḥɛm lɛḥɛm lɛ=ʾɛḵol bread to=[eat.ic pro bread] ‘bread to eat’ (Gen 28:20) (62) Poss-inf ha=ʿam mayim mayim li=štoṯ water to=[drink.ic the=people water] ‘water for the people to drink’ (Exod 17:1)

- 978-90-04-44885-8

the infinitive in biblical hebrew

165

Negation is found with infinitival clauses, and it can be shown that negation takes scope over the entire clause rather than just modifying the infinitival head. Only clausal scope can give the correct reading in (63). Sacrificing to the Lord is the purpose of sending off the people, not the purpose of not sending off the people. Therefore, negation attaches to the full clause letting the people go to sacrifice to the Lord rather than to the head letting go.18 (63) ʾal yosep̄ parʿo haṯel lə=ḇilti [šalaḥ pro ʾɛṯ neg repeat.juss.3ms Pharaoh deceive.ic to=neg [send.ic pro om ha=ʿam li=[zboăḥ pro la=yhwh]] the=people to=sacrifice pro to=Yhwh]] ‘But let Pharaoh not deal deceitfully anymore in not letting the people go to sacrifice to the Lord’ (Exod 8:25[29])

5

Conclusion

The paper shows that Biblical Hebrew infinitival constructions are clausal rather than nominal, and that the functional categories which determine the inflection of the finite verb also determine the feature specification of the infinitive. Moreover, the morphosyntax of the different infinitival clauses determines their distribution. Nom-inf clauses are root clauses with irrealis Mood, hence have the conversational force of imperatives. pro-inf and Poss-inf clauses are not specified for irrealis Mood, and thus have no conversational force. They therefore must be embedded clauses. The lack of T specification determines that the pro-inf clause cannot be interpreted as an independent proposition but is rather interpreted as part of the event denoted by the main clause, since it depends for its temporal anchoring on the temporal specification of the main clause. The Asp/Mod categoy of the pro-inf construction allows it to function as complement of aspectual and modal verbs, and as specifier to Mod/Asp heads, i.e. as purpose clauses. The Poss-inf clause, on the other hand, contains a T head, and hence denotes a separate proposition from the one denoted by the main clause. Accordingly, it functions as a complement of propositional attitude verbs or a temporal/ result specifier of the main-clause T. Moreover, the different categories T, Asp/Mod, and Mood in finite and infini-

18

In Modern Hebrew, the negative bilti has grammaticalized into a prefix which attaches to lexical items, in particular adjectives.

- 978-90-04-44885-8

166

doron

tival clauses have been shown to explain the various possibilities of subject and object cliticization in each type of clause.

References Abney, Steven. 1987. The English noun phrase in its sentential aspect. Ph.D. dissertation, mit. Aldridge, Edith. 2006. Absolutive case in Tagalog. Proceedings from the Annual Meeting of the Chicago Linguistic Society 42.2. 1–15. Chicago: Chicago Linguistic Society. Bauer, Hans and Pontus Leander. 1922. Historische Grammatik Der Hebräischen Sprache Des Alten Testaments. Halle: M. Niemeyer. Blau, Yehoshua. 1990. Hebrew and Arabic. Leshonenu Laʾam 40.5: 311–335. [in Hebrew] Blau, Yehoshua. 1979. The Phonology and Morphology of Biblical Hebrew. Second edition 2010. Jerusalem: The Academy of the Hebrew Language. Callaham, Scott N. 2014. Modality and the Biblical Hebrew Infinitive Absolute. Wiesbaden: Harrassowitz Verlag. Cardinaletti, Anna. 2008. On different types of clitic clusters. Pp. 41–82 in The BantuRomance Connection. A comparative investigation of verbal agreement, dp s and information structure, ed. Cécile De Cat and Katherine Demuth. Amsterdam: John Benjamins. 41–82. Cardinaletti, Anna and Ur Shlonsky. 2004. Clitic positions and restructuring in Italian. Linguistic Inquiry 35.4: 519–557. Cinque, Guglielmo. 1999. Adverbs and Functional Heads: A Cross-Linguistic Perspective. Oxford: Oxford University Press. Collins, James. 2017. Structure Sensitive Interpretation: A Case Study in Tagalog. Stanford PhD Dissertation. Cook, John A. 2006. The Finite Verbal Forms in Biblical Hebrew do Express Aspect. janes 30: 21–35. Cook, John A. 2012. Time and the Biblical Hebrew Verb: The Expression of Tense, Aspect, and Modality in Biblical Hebrew. Winona Lake IN: Eisenbrauns. Doron, Edit. 2016. The sources of Modern Hebrew syntax. Lecture presented at the Conference on the Emergence of Modern Hebrew. The Mandel Scholion Research Center. The Hebrew University. Doron, Edit. 2018. The infinitive construct as a verbal form. Paper presented at the Biblical Hebrew: Advances in Grammar and Lexicology conference, Jerusalem, January 28. [in Hebrew] Doron, Edit. 2019. The Biblical sources of Modern Hebrew syntax. Pp. 222–256 in Linguistic Contact, Continuity and Change in the Genesis of Modern Hebrew, ed. E. Doron,

- 978-90-04-44885-8

the infinitive in biblical hebrew

167

M. Rappaport Hovav, Y. Reshef, and M. Taube. Linguistik Actuell/Linguistics Today 256. Amsterdam: John Benjamins. Goshen-Gottstein, Moshe H. 2006. Syntax and Vocabulary of Medieval Hebrew: Under the Influence of Arabic. Jerusalem: Ben-Zvi Institute, edited and published by Shraga Assif and Uri Melammed on the basis of the unpublished 1951 PhD diss. [in Hebrew] Grimshaw, Jane. 1990. Argument structure. Cambridge, MA: mit Press. Hacquard, Valentine. 2011. Modality. pp. 1484–1515 in Semantics: An International Handbook of Natural Language Meaning, ed. C. Maienborn, K. von Heusinger, and P. Portner. hsk 33.2. Berlin: Mouton de Gruyter. Hatav, Galia. 1997. The Semantics of Aspect and Modality: Evidence from English and Biblical Hebrew. Amsterdam: John Benjamins. Hatav, Galia. 2008. The modal system of Biblical Hebrew. Pp. 163–191 in Theoretical Hebrew Linguistics, ed. G. Hatav. Jerusalem: Magnes Press. 163–191. [in Hebrew] Hatav, Galia. 2017. The Infinitive Absolute and Topicalization of Events in Biblical Hebrew. Pp. 207–229 in Advances in Biblical Hebrew Linguistics: Data, Method, and Analyses, ed. A. Moshavi and T. Notarius. Winona Lake, IN: Eisenbrauns. Johnson, K. 1988. Clausal Gerunds, the ecp, and Government. Linguistic Inquiry 19:4. 583–609. Jones, Charles. 1985. Agent, patient, and control into purpose clauses. Pp. 105–119 in Papers from the Parasession on Causatives and Agentivity at the Twenty-First Regional Meeting, Vol. 21.2, ed. W.H. Eilfort, P.D. Kroeber, and K.L. Peterson. Chicago, IL: Chicago Linguistic Society. Joosten, Jan. 2002. Do Finite Verbal Forms in Biblical Hebrew Express Aspect? janes 29: 49–70. Kiparsky, Paul. 2001. Structural case in Finnish. Lingua 111: 315–376. Lander, Yury. 2011. Varieties of genitive. Pp. 581–592 in The Oxford Handbook of Case, ed. A. Malchukov and A. Spencer. Oxford: Oxford University Press. Moulton, Keir. 2004. External arguments and gerunds. Toronto Working Papers in Linguistics 22: 121–136. Palmer, Frank R. 2001. Mood and Modality. Second Edition. Cambridge: Cambridge University Press. Pesetsky, David and Esther Torrego. 2001. T-to-C movement: causes and consequences. Pp. 355–426 in Ken Hale: A life in Language, ed. M. Kenstowicz. Cambridge, MA: mit Press. Pires, Acrisio. 2006. The Minimalist Syntax of Defective Domains: Gerunds and Infinitives. Amsterdam: John Benjamins. Portner, Paul. 1997. The semantics of mood, complementation, and conversational force. Natural Language Semantics 5: 167–212, 1997. Reuland, Eric J. 1983. Governing -ing. Linguistic Inquiry 14.1: 101–136. Verstraete, Jean-Christophe. 2008. The status of purpose, reason, and intended end-

- 978-90-04-44885-8

168

doron

point in the typology of complex sentences: implications for layered models of clause structure. Linguistics 46.4: 757–788. Wright, W. 1896. A Grammar of the Arabic Language. Third Edition. Cambridge: Cambridge University Press. Wurmbrand, Susi. 2001. Infinitives: Restructuring and Clause Structure. Berlin: Mouton de Gruyter. Wurmbrand, Susi. 2014. Tense and aspect in English infinitives. Linguistic Inquiry 45.3: 403–447.

- 978-90-04-44885-8

chapter 9

Light Verbs in Biblical Hebrew Todd Snider

1

Introduction*

There are a number of constructions in Biblical Hebrew in which two verbs, taken together, jointly describe a single event. In this paper, I present examples of these sorts of constructions and use these examples to illustrate the shared properties of these constructions, the ways they differ, and the ways they do not. My eventual goal here is to present a syntactic analysis of these constructions. Along the way, I consider four candidate analyses: (a) both verbs are main verbs, (b) both verbs jointly make up a serial verb construction, (c) the first verb is an auxiliary verb, and (d) the first verb is a light verb. Ultimately, I argue that only the light verb construction analysis accounts for the observed behavior we will see shortly. The constructions which are the focus of this investigation are not uncommon in the Hebrew Bible, and as such I take them to be a regular feature of the syntax of Biblical Hebrew. For the purposes of this examination, we will direct our attention to examples from narrative prose, as opposed to poetry, as poetic language (and in particular the poetry of the Hebrew Bible) often makes use of repetition in a way which increases the chances of verbs appearing together and obscures the relationship between such verbs. We will also be focusing mostly on examples from (pre-exilic) Standard Biblical Hebrew, so as to minimize the chances of this being a consequence of diachronic change or influence from Aramaic. Translations provided are from the King James Version (kjv) unless noted otherwise. Transcriptions and morphological glosses edited and significantly improved by Robert Holmstedt.

* My thanks to Bar Avineri, Molly Diesing, Edit Doron, Malka Rappaport-Hovav, Ruth Stern, members of the mabar Project, and audience members at bhlap2 for their helpful comments. Any errors are my own.

© Todd Snider, 2021 | doi:10.1163/9789004448858_010

- 978-90-04-44885-8

170 2

snider

The Data

To get a sense of the sort of construction on which we’ll be focusing here, it is instructive to look at a famous (if long) example: (1) way=yaškem ʾaḇraham bab=boqɛr and=awaken.pret.3ms Abraham in.the=morning way=yahăḇoš ʾɛṯ hămor=o way=yiqaḥ ʾɛṯ šǝne and=saddle.pret.3ms om donkey=his and=take.pret.3ms om two.of nǝʿar=aw ʾitt=o wǝ=ʾeṯ yiṣḥaq bǝn=o young.men=his with=him and=om Isaac son=his way=ǝḇaqaʿ ʿăṣe ʿola way=yaqom and=cleave.pret.3ms wood.of offering and=stand.pret.3ms way=yelɛḵ ʾɛl ham=maqom and=go.pret.3ms to the=place … ‘And Abraham rose up early in the morning, and saddled his ass, and took two of his young men with him, and Isaac his son, and clave the wood for the burnt offering, and rose up, and went unto the place of which God had told him.’ (Gen 22:3) Within this verse, the construction we are interested in is wayyaqom wayyelɛḵ, which the kjv translates as “and rose up and went”. But let us think carefully about how we should interpret the first verb wayaqom ‘he rose’. We already know that Abraham was awake (from the word wayyaškem ‘and he awoke’), so wayyaqom is not describing waking up/rising from sleep. And presumably Abraham was already standing up while chopping wood, mentioned in the immediately prior phrase. So either Abraham sat down between chopping wood and going—in a way entirely immaterial to the story being related—, or the meaning of the construction is something other than simply the combination of wayyaqom and wayyelɛḵ. The kjv provides the translation “rose up”, but in what sense is there a rising? One might be tempted here to offer a metaphorical or midrashic explanation, something about ‘rising to the challenge’ or ‘ascending in holiness’ (as Abraham went to follow God’s command), but I will instead argue that this is representative of a larger pattern, not a matter of particular contextual interpretation but rather an example of a regular grammatical pattern in Biblical Hebrew. And as we will see, this construction can be found in many other contexts. The question of the contribution of wayyaqom in this case presumes that this verb does have its own meaning to convey. For completeness, we can

- 978-90-04-44885-8

171

light verbs in biblical hebrew

demonstrate that qam ‘stand’ can indeed stand alone as a main verb, conveying meaning the way that verbs normally do: (2) way=yoʾḵǝlu way=yištu huʾ wǝ=ha=ʾănašim ʾăšɛr and=eat.pret.3mp and=drink.pret.3mp he and=the=men that ʿimm=o way=yalinu way=yaqumu ḇab=boqɛr with=him and=lodge.pret.3mp and=stand.pret.3mp in.the=morning way=yoʾmɛr šalḥu=ni la=ʾdon=i and=say.pret.3mp send.imp.mp=me to=master=my ‘And they did eat and drink, he and the men that were with him, and tarried all night; and they rose up in the morning, and he said, Send me away unto my master.’ (Gen 24:54) The same can be said of the other verb in the relevant construction in (1), halaḵ ‘go’, which can also stand alone as a main verb. So the example in (1) includes two verbs, each of which can stand alone, which together seem to describe one event. This sort of behavior is not restricted to the two relevant verbs we have seen in (1). That said, which verbs can take part in this construction are restricted, in a way to be discussed shortly. Before that, though, another pair of examples further illustrate the pattern: (3) way=yašuḇu way=yibku gam bǝne yisraʾel and=return.pret.3mp and=cry.pret.3mp also children-of Israel ‘The children of Israel wept again’ (Num 11:4) (4) wat=tǝmaher wat=torɛḏ kadd=ah and=hurry.pret.3fs and=lower.pret.3fs jug=her ‘She quickly lowered her jug’ (kjv: let her pitcher down) (Gen 24:18) The construction in (3) involves two verbs, wayyašuḇu ‘and they returned’ and wayyibku ‘and they cried’, but describes no physical returning, the way wayyašuḇu would if it stood alone as a main verb, as we can see by contrast with (5). (5) way=yašoḇ yosep̄ miṣrayma and=return.pret.3ms Joseph Egypt=ward ‘Joseph returned to Egypt’ (Gen 50:14) Instead, (3) conveys that the people cried again, describing a single event. Similarly, the verse in (4) involves a description of just one event, a jug-lowering

- 978-90-04-44885-8

172

snider

event, and not a hurrying event followed by a lowering event. The verb ‘hurry’, just like ‘return’, can function as a standalone main verb, as in (6). (6) way=ǝmaher aḇraham ha=ʾohɛ̆l=a and=hurry.pret.3ms Abraham the=tent=ward ‘Abraham hastened into the tent’ (Gen 18:6) We can note at this point that in all of these cases, it is the second verb of each construction (call it V2) which retains its full lexical content. (3) involves a crying event, and (4) involves a lowering event. In contrast, the first verb of each construction (call it V1) seems to be contributing some sort of information to our interpretation of the event described, but not the same meaning that it conveys when it acts as a standalone main verb. In addition, we can note that it is V2 which determines the valence of the combined verbal construction. If V2 is transitive, then the whole V1V2 construction will take an object, as in (4). If V2 is intransitive, then the V1V2 construction will also be intransitive, as in (3). It happens to be that most of the verbs which occupy the V1 position are themselves intransitive, as we have seen thus far, and it should not be surprising that an intransitive V1 would not add or require additional argument slots to be filled in these constructions. But even when the V1 is transitive, it does not affect the number of arguments associated with the entire V1V2 construction. As far as this comparison of the functions of V1 vs. V2 is concerned, the order of these two is strict. If the order of these two verbs are reversed, we end up with a two-events interpretation. This is always licit, because all of the verbs involved (both V1 and V2) can stand alone as main verbs. And this sort of reversal ends up getting a two-events interpretation because only some verbs can occupy the V1 slot. Consider, for example, the verse in (7), which has the verb wǝyašuḇu ‘they returned’—equivalent to the V1 verb from (3)—as the second in a pair of adjacent verbs. (7) yizkǝru wǝ=yašuḇu ʾɛl yhwh remember.impf.3mp and=return.impf.3mp to Yhwh ‘Remember and turn unto the Lord’ (Ps 22:27) In this position in (7), the verb wǝyašuḇu doesn’t convey the same meaning as in (11); the verse in (7) does not mean ‘Remember the Lord again’ (a one-event interpretation), but rather is a command both to remember and to turn (a twoevent interpretation). Similarly, the two adjacent verbs in (8) do not describe a command to go again, but rather two commands, to go and to return. (8) does not entail that the subject (Elijah) had been there before. - 978-90-04-44885-8

light verbs in biblical hebrew

173

(8) leḵ šuḇ lǝ=darkǝ=ḵa miḏbar=a dammaśɛq go.imp.2ms return.imp.2ms to=path=your wilderness.ward Damascus ‘Go, return on thy way to the wilderness of Damascus’ (1 Kgs 19:15) What about the subject of these verbal constructions, where does it appear? Because the text of the Pentateuch is primarily that of a narrative, it frequently describes one subject taking a number of actions in sequence. As such, and because Biblical Hebrew is a pro-drop language, the text frequently features chains of verb-object pairs in one sentence, with a single subject shared across each predicate (and not repeated after the first). This is demonstrated quite dramatically by the first example of this V1V2 construction we saw, as (1) can be schematized as VSVOVOVOV1V2O. When it comes to our V1V2 construction, this results in the subject of these V1V2 constructions appearing in one of three ways: (i) the subject can follow the entire construction, (ii) the subject can be interposed between the two verbs, or (iii) the subject can be dropped and not appear overtly. The lattermost case occurs whenever the V1V2 construction isn’t the first predication in a sentence, where the subject is understood as being shared with the prior predications. We saw this already in (1), with Abraham, and in (4), with Rebecca. We have also already seen an example of the subject following the entire construction, as in (3). Interposing the subject between V1 and V2 is also attested, as in (9): (9) way=yosɛp̄ ʾaḇraham way=yiqqaḥ ʾišša and=add.pret.3ms Abraham and=take.pret.3ms wife ‘Abraham took another wife’ (kjv: “again Abraham took a wife”) (Gen 25:1) Parallel to our other examples of this V1V2 construction, (9) does not describe two events (an adding event and a wife-taking event), but rather a single event. In fact, if anything is being ‘added’ here, it was through the wife-taking event that Abraham added to the size of his family. But the translation is quite telling here (unlike in (1)), making it clear that we are dealing with only one event, despite having two verbs. The dropping of the subject is determined by context, whether the V1V2 construction appears as the first or subsequent predication in a sentence. When the subject is overt, however, there is a choice as far as whether it follows the two verbs or appears between them. The variation between these forms might be influenced by narrative-driven choices about topic and focus, and they might also be subject to syntactic constraints on raising which are sensitive to other words in the sentence (e.g., gam ‘also’ as in (3)). I leave the question of the rhetorical differences between V1V2S and V1sv2 for future work. - 978-90-04-44885-8

174

snider

There is one more structural note to make about this construction before we try to characterize its meaning. So far, all of the examples of this V1V2 construction that we have seen have involved verbs introduced by waw. But we can also find examples of this construction that do not involve waw, including especially in imperatives. (10) qum hiṯhaleḵ baʾarɛṣ stand.imp.2ms walk.imp.2ms in.the=land ‘Arise, walk through the land’ (Gen 13:17) (11) šub šǝkab return.imp.2ms lie.imp.2ms ‘Lie down again’ (1Sam 3:5) Example (10) is not two commandments—one to rise and another to walk—, despite the kjv translation, but rather conveys a single instruction to bring about a single event. The same is true of (11), only now with the translation pointing toward the right interpretation: (11) describes not instructions to return and then to lie down, but rather an instruction to lie down again. The waw-less version of this V1V2 construction provides perhaps the best evidence for what one could call the de-lexicalization of V1, the use of what is normally a main verb but with a reduced/modified meaning; consider example (12). (12) qum na šǝḇa stand.imp.2ms please sit.imp.2ms ‘(Please) Sit (now)’ (Gen 27:19) Even if one were to successfully explain away example (10) as perhaps addressing a possibly-sitting Avram, or as calling for a metaphorical rising of his settled family, the same sorts of explanations do not work for (12). If we were to interpret both verbs as having their usual standalone meaning, (12) would be an instruction to perform an odd sequence of events, to stand and then sit (or to stand in order to sit?), in a way which isn’t usually taken to be the intended meaning of this verse. In fact, (12) is not a call to do some sort of aerobics; rather, there is simply no standing or arising involved in this instruction whatsoever. qum adds some meaning to this construction, but it is not the same meaning that it would convey if it were a standalone main verb. So what is the meaning contribution of V1 in these V1V2 constructions? The V1 in (11) adds an additive aspect to the interpretation of its V2. The V1s in (10)

- 978-90-04-44885-8

light verbs in biblical hebrew table 9.1

175

Verbs that can appear in first position in V1V2 constructions

Verb

Stand alone mean- Construction meaning ing

šaḇ hosip̄ hoʾil

to return to add to agree, be willing

miher hišqim hirba qam

to hurry to wake up early to increase to rise, get up

to do something again to do something again/more to do something willingly, to be content to do something to do something quickly to rush to do something to do something a lot to begin to do something, to do something immediately

and (12) add an inchoative aspect, a sense of immediacy to their V2s.1 Table 9.1 lists all of the verbs which appear in the V1 position of such constructions and their respective meaning contributions. One can see connections between these verbs’ stand alone meanings on the one hand to their contributions to this V1V2 construction on the other hand, but I leave the exact trajectory (or formal semantic modification) for each of these meanings to future work. These V1V2 constructions also connote a voluntary nature to the event that they describe: the event’s thematic agent is understood as having a volitional role in the event. We can take this to be part of the meaning contributed by V1, modified from its usual lexical meaning, in that V2 continues to carry the same meaning it would have were it not part of this construction. Interestingly, this volitional meaning is conveyed by all of the V1 verbs listed in Table 9.1, not just by hoʾil which does so as a result of its normal lexical meaning. Before we consider how best to analyze this construction, it is worth highlighting some other sorts of examples, which depending on one’s analysis might or might not constitute the same construction. First, one might find it notable that halaḵ ‘go’ is not listed as a V1 verb in Table 9.1. Depending on one’s analysis, halaḵ might in fact belong there: it can function as a stand alone main verb and can also appear adjacent to other verbs, and when it does it might

1 The particle naʾ in (12) also adds a cohortative meaning, though this is not specific to the V1V2 construction.

- 978-90-04-44885-8

176

snider

be interpreted as describing a single event. However, one might also take halaḵ to simply be retaining its normal lexical meaning, as physical movement can be argued to be involved in nearly every event or set of events described by halaḵ alongside another verb. Finally, one could analyze halaḵ as belonging to a similar class as the verbs in Table 9.1, but further along the historical trajectory which we will discuss later. If only to keep the set of verbs under consideration more cohesive—and thus the V1V2 construction which is of interest more uniform—we will set aside the question of the best analysis of halaḵ for the remainder of this paper. Second, there is another set of examples which might or might not be part of the same construction, depending on one’s analysis. All of the examples we have seen thus far have involved a pair of fully inflected verbs (including those which were imperatives). However, the verbs listed in Table 9.1 carry these same modified meanings when the V2 which follows them is infinitive, as illustrated in (13) and (14). (13) yašuḇ yhwh la=śuś ʿalɛ=ḵa lǝ=ṭob return.impf.3ms Yhwh to=rejoice.inf on=you for=good ‘God will again rejoice over you for good’ (Deut 30:9) (14) u=mǝmaher ha=ʾɛ̆lohim la=ʿăśot=o and=hurry.ptcp.ms def=God do.inf=it ‘God will soon do it’ (Gen 41:32) Example (13) involves no physical returning, and (14) involves no physical hurrying; both describe only one event each. For some theorists, these sorts of infinitive examples count as the same construction which is our focus here; for others, this disqualifies them. So for the moment, I merely flag their existence, but neither rule them in nor out; we will decide this issue on the basis of our eventual syntactic analysis. Finally, it is worth mentioning that the V1V2 construction which interests us is different from the construction wherein a single verb is repeated twice (in other words, where V1=V2), as in (15)–(17). (15) haqem taqim ʿimm=o lift.inf-abs lift.impf.2ms with=him ‘(if your brother’s donkey or ox falls down …) thou shalt surely help him to lift them up again’ (Deut 22:4)

- 978-90-04-44885-8

light verbs in biblical hebrew

177

(16) ki bǝ=yom ʾăḵol=ḵa mimmɛn=nu moṯ because on=day.of eat.inf=your from=it die.inf-abs tamuṯ die.impf.2ms ‘for in the day that thou eatest thereof thou shalt surely die’ (Gen 2:17) (17) we=ʾlohim paqoḏ yip̄ qoḏ ʾɛṯḵɛm and-God visit.inf-abs visit.impf.3ms om=2mp ‘and God will surely visit you’ (Gen 50:24) These sorts of constructions are fairly common in the Hebrew Bible, but they have a very different meaning contribution to the construction we have discussed thus far. They convey a sense of certainty or assuredness on the part of the speaker about the likelihood of the event described coming to pass. And quite different from our construction of interest, these V1=V2 constructions are available to nearly any verb (including but not limited to the verbs listed in Table 9.1). I take this to be an entirely different construction from our V1V2 construction. So ultimately what we aim to describe is a construction made up of two verbs which jointly describe a single event. Whatever analysis we end up with should help to explain the particular syntactic arrangement of the two verbs, the limited set of verbs which appear in the V1 position, the aspectual meaning contributed by each V1 and the additional volitional meaning which is contributed across the entire set of V1 verbs. We will explore several possible analyses in the following section.

3

Evaluating Possible Analyses

There are at least four analyses worth exploring when observing verbs working in concert as in our construction: i. They might be of different statuses, where V1 is a light verb ii. They might be of different statuses, where V1 is an auxiliary iii. They might both be main verbs, working together through normal composition iv. They might both be main verbs, operating within a Serial Verb Construction In this section, we will consider each of these analyses in turn. I will argue for the first—that what we are dealing with is a light verb construction—and explain the problems with each other potential account.

- 978-90-04-44885-8

178

snider

3.1 Light Verb Construction The term light verb was originally used to describe V+NP constructions (such as English take a swim, give a shout, etc.). Since then, it has somewhat expanded in use, and has also been used to describe: N+V constructions in Japanese (Grimshaw and Mester 1988) and Hindi (Mohanan 1994); and V+V constructions in Romance (Rosen 1990), Bengali (Ramchand 1991), Urdu (Butt 1995), and Yiddish (Diesing 1997, 2000, Taube 2008).2 Light verbs are also sometimes called vector verbs, delexical verbs, and a few other names. Considering the breadth of the literature on things which have been called light verbs, Butt 2003 presents a review of this literature and lays out the characteristic features of light verbs, especially as contrasting with auxiliary verbs, as the following: a. A light verb is always form-identical with a main verb in the language. b. Light verbs are identifiable as their own syntactic class. c. A light verb contributes ‘non-compositional’ meaning to a single monoclausal verbal complex. d. The meaning contributed often bears on the aktionsart of the complex predication, giving it a telic/complete/boundedness character. e. A light verb may also contribute meaning about benefaction, suddenness, forcefulness, or the like, to the verbal complex. We can review these features one by one to see how well these describe our Biblical Hebrew V1V2 construction. First, the verbs which can occupy the V1 slot can all also appear as standalone main verbs verbs in Biblical Hebrew. Second, do these V1 verbs constitute their own syntactic class? In order to answer this, it is helpful to get a sense of what sorts of markers demarcate light verbs as a syntactic class in other languages which have been argued to have them. For Mandarin and Urdu, light verbs display unique tonal patterns and selectional restrictions (Butt 2003), clearly separating them from other verbs. The same can’t be said for all languages, however. For Bengali, Yiddish, Wagiman, Bardi, and Jaminjung, light verbs don’t show overt morphological/phonological behavior, but they are drawn from small closed classes (Ramchand 1991, Diesing 1997, Butt 2003). This suggests that these are syntactically distinct classes, even if they aren’t identified with overt phonological or morphological markers. If this is reliable, then this seems to be sufficient for Biblical Hebrew: V1 verbs are not identified with any overt marking, but they do 2 See also Gamliel and Mar’i 2015 on Modern Hebrew and Arabic, though there they are classified as auxiliaries. The authors cite Taube and Butt, and do not argue against a light verb analysis.

- 978-90-04-44885-8

light verbs in biblical hebrew

179

seem to be drawn from a small closed class. At least relative to the languages reviewed in Butt 2003, Biblical Hebrew seems to be in good company in having a small closed class of such verbs even with no overt morphological marking. The third characteristic feature of Butt 2003 is that of ‘non-compositionality’, namely that the meaning contributed by the light verb in a light verb construction is different from the meaning it would contribute were it a standalone main verb.3 This, of course, is precisely the behavior which we observed above. Consider again for example the construction in (3). A normal composition of the meanings of ‘return’ and ‘cry’ wouldn’t give the meaning of ‘cry again’; one could return home and then cry without having ever cried before. The ‘again’ meaning which we get in (3) is the result of some process other than merely concatenating the two verbs. Our Biblical Hebrew construction, then, certainly involves this sort of ‘non-compositional’ meaning. Fourth, do our V1 verbs bear on the aktionsart of the verbal complex? The relevant process in Bengali described by this behavior is robust, and is how Bengali turns atelic predicates into telic ones (Ramchand 1991). For Yiddish, Diesing 1997 calls this process “perfectivization with an added twist”, the twist being a temporal truncation of the predicated event. For iterative events, this produces semelfactive interpretations, and for durative events this produces a telic interpretation with a shortened duration. And this perfectivization process also explains why Yiddish light verbs are incompatible with stative predicates.4 How does this Bengali and Yiddish behavior inform our analysis of this Biblical Hebrew construction? Well, telicity is not going to be the relevant factor for this construction, as Biblical Hebrew perfective aspect already encodes telicity on verbs, without the need for an additional light verb. But we can identify a certain temporal shortening in our V1V2 construction, in that it tends to convey a sense of ‘immediacy’. We can identify this sense of immediacy with a shortening the duration of what Butt 2003 calls the “causing event”: the inchoative “causing event” is shortened, which makes the potential start time of the main event earlier, thus giving the entire event an earlier ‘immediate’ feel. And we might connect this shortening of the causing event to the volitional agentdriven meaning contributed by the V1 verb: if the causing event is internallydirected, a matter of mental states rather than physical actions, then there is

3 This feature in no way implies that a light verb construction cannot be analyzed as conforming to a typical compositional Montague semantics. 4 Diesing’s (1997) generalization about stative verbs might be true for Biblical Hebrew as well, though it is difficult to be sure about this due to Biblical Hebrew’s flexible shifting of statives into inchoatives. If in fact statives are incompatible with our V1V2 construction, it might be due to the volitional contribution of this construction.

- 978-90-04-44885-8

180

snider

no reason to think that its runtime would be much longer than instantaneous. This would account for how a sense of volitionality might lead to a temporal sort of meaning contribution as well. If we consider this immediacy akin to the telic/complete/bounded character described by Butt 2003, then this is another reason to adopt a light verb construction analysis. Fifth and finally, Butt 2003 characterizes light verbs as often contributing an additional meaning beyond that of aspect. The examples provided in Butt 2003 deal with meanings which deal with benefaction, forcefulness, and—in an Urdu example which is most relevant to our Biblical Hebrew construction— volitionality. This feature, then, nicely describes the Biblical Hebrew V1V2 construction’s tendency to add a sense of volitionality to the interpretation of the V2 verb. Taken together, then, the Biblical Hebrew V1V2 construction which interests us here neatly fits Butt’s (2003) description of a light verb construction. The V1 verbs are form-identical with main verbs in Biblical Hebrew but form their own small closed class, and they contribute meaning beyond that of their standard lexical meaning, namely conveying senses of volitionality and timeboundedness. If we do adopt Butt’s (2003) characterization of light verb constructions, we might also choose to adopt her syntactic representation. Her model for the syntax of light verb constructions, based on prior work in Ramchand 2008, is as illustrated in in Figure 9.1. On this model, a single verbal complex is composed of (up to) three verbal projections: a cause (νP), a process (VP), and a result (RP). A given language might use all three of these verbal projections, or only a subset thereof (though presumably always including at least the main process VP projection). Because none of our V1 verbs contribute resultative aspect readings, we can posit that Biblical Hebrew only has a two-projection structure, and does not make use of the result RP layer. All of the V1s we have seen sit in the cause ν head, contributing to the inchoativity of the larger verbal complex. The main verb V2, on the other hand, sits in the process V head. And on this syntax, this will be the case whether V2 is inflected or infinitive; an infinitive is not a defective TP on this account, and so would end up including infinitive examples like (13) and (14). This would be an additional and broader consequence for how we analyze infinitives in Biblical Hebrew. I have argued here that, at least so far as Butt’s (2003) typological characterization is concerned, our Biblical Hebrew V1V2 construction fits the bill. But before we can be content with this analysis, we should consider other potential analyses; if the light verb construction analysis is correct, we should find reasons to reject alternative competing analyses. And that is precisely what we

- 978-90-04-44885-8

light verbs in biblical hebrew

181

figure 9.1 The structure of light verb constructions from butt 2003: 20

shall do over the next three subsections. Up next, we consider the possibility that these V1 verbs are not light verbs, but auxiliary verbs. 3.2 Auxiliary Verb + Main Verb Another potential analysis for our V1V2 construction is that they involve not a light verb in the V1 position, but rather an auxiliary verb (also called a helping verb). Auxiliary verbs are verbs which carry modal or aspectual information, like English do, may, has, can, and so forth. (18) Nanette may open the door. (19) Nanette has finished her tea. (20) Nanette does like art galleries. If indeed these V1 verbs were auxiliaries, it would explain how our verbal construction gets its additional aspectual meaning: the core function of auxiliaries is to carry tense/aspect/mood meaning, which are then ascribed to the auxiliary’s main verb. It would also explain why there are only a limited number of possible V1 verbs, as auxiliary verbs are a distinct class of verbs, a closed class,

- 978-90-04-44885-8

182

snider

as opposed to a special function that all verbs in a language can perform. And the reason that the particular verbs which can appear in the V1 position (those in Table 9.1) should have been recruited into this role—presumably grammaticalized from main verbs in the language—can be explained as a function of their normal standalone meanings. In these ways, an auxiliary verb analysis is tempting. However, this analysis runs into some difficulties, one of which is already raised by the above diachronic story. If indeed these V1 verbs were auxiliaries, we would have to say that they were in an ongoing process of grammaticalizing, that they were not yet completely grammaticalized as auxiliaries. For one thing, auxiliary verbs aren’t a subtype of main verbs, but rather their own distinct class, and auxiliary verbs typically cannot stand alone in a sentence without a main verb (and when they do appear alone, their counterpart main verb has been elided and is therefore interpreted as if present). The fact that our V1 verbs can appear alone suggests that these verbs, if auxiliaries, are still in the process of auxiliarizing. Relatedly, auxiliary verbs usually are not string-identical with main verbs in the language. Even though they are derived historically from main verbs, auxiliary verbs “quickly develop away from the main verb in form, function, and meaning” (Butt 2003). Depending on the duration of this “quickly”, it might be surprising for Standard Biblical Hebrew to have maintained this construction over the course of the Pentateuch’s creation, transmission, compilation, etc. Additionally, if we were to consider these V1 verbs to be auxiliaries, we would need some additional explanation for why all of these auxiliaries (across the class) contribute an additional volitional meaning. It would not be surprising for one auxiliary to convey such a meaning, but it would be very surprising for an entire class of auxiliaries to all convey such a meaning, above and beyond their individual meaning contributions. For these reasons, an auxiliary analysis is unsatisfying: it leaves open as many questions as it answers. While an auxiliary analysis explains the closed class nature of V1 verbs and why they each carry aspectual meanings, it does not account for why these V1 verbs are still string-identical with main verbs nor why the entire class shares an additional volitional meaning. This sort of auxiliary analysis is argued for in Chrzanowski 2011, which considers these V1 verbs to be a result of grammaticalization, towards status as auxiliary verbs. As a grammaticalization account, Chrzanowski 2011 cites Hopper and Traugott 1993 and in particular the cline in Figure 9.2. And though Hopper and Traugott 1993 includes vector verbs (= light verbs) as an optional stage in the grammaticalization path, Chrzanowski 2011 does not discuss or consider a light verb analysis for this construction.

- 978-90-04-44885-8

183

light verbs in biblical hebrew

full verb

> (vector verb)

> auxiliary

> clitic

> affix(es)

figure 9.2 Hopper and Traugott’s stages of grammaticalization (1993: 108)

main verb > auxiliary > clitic > affix(es) | light verb figure 9.3 A revised grammaticalization cline from butt 2003:16

Butt 2003 argues that the cline in Figure 9.2 is mistaken, for a few reasons. First, Butt 2003 asserts that non-aspectual contributions that light verbs make (“volitionality, benefaction, forcefulness, surprise, etc”; p. 13) should not be part of the path from main verb to auxiliary verb, as auxiliaries don’t have that function (nor do main verbs). And second, while auxiliaries “start out as a version of a main verb … but then quickly develop away from the main verb in form, function, and meaning”, light verbs remain string-identical and “do not lead to the development of functional categories”. For Butt 2003, then, light verbs do not represent a step on the way towards the behavior of auxiliary verbs. Instead, Butt 2003 proposes the alternative cline in Figure 9.3. On this alternative path, main verbs can develop into light verbs—retaining their stringidentical status, but weakening their lexical meaning in exchange for some additional aspectual meaning—or alternatively into auxiliary verbs—at which point they quickly diverge from their main verb origins both in meaning and in form. The two analyses we have considered thus far treated the two verbs as having importantly different statuses in the construction, where the V1 verb makes a different contribution from the V2 main verb. It is not always the case, however, that two adjacent verbs must be of different statuses; combining two verbs in a single verbal complex doesn’t entail that one must be reduced, deranked, or otherwise modified. The next two analyses to be considered take another tack on this question, treating the V1 and V2 verbs as having equal status in the construction. 3.3 Independent Main Verbs A third analysis to consider takes both the V1 and V2 to be main verbs, of the same status, combining together through normal compositional processes. This is at least somewhat attractive right off the bat, because we know that all of our V1 verbs can stand alone as main verbs in Biblical Hebrew, and so having a single unified function—and presumably lexical entry—for each of these verbs is certainly parsimonious. However the additional steps which would be

- 978-90-04-44885-8

184

snider

required to allow this analysis to capture the behavior of our V1V2 construction render this type of approach anything but parsimonious. For one thing, this type of analysis would require V1 verbs like qam ‘stand’ to either have a very unusual context-sensitive semantics (encoded in its lexical entry)—wherein it would have an aspectual meaning when followed directly by another verb or by a subject-verb sequence, and an eventive meaning in all other syntactic contexts—or, perhaps more likely, to be lexically ambiguous between an eventive reading and an aspectual reading. This in of itself isn’t necessarily a weakness for this analysis, as languages of the world do exhibit lexical ambiguity; there is no reason a priori to presume that Biblical Hebrew could not. And lexical ambiguity could handle the different valences of a V1 verb like hosip̄ ‘add’, as well: it would be transitive as an eventive main verb, and intransitive in its aspectual (but still main verb) use in our V1V2 construction. However, adopting an ambiguity analysis doesn’t itself explain why the class of V1 verbs is restricted to a small class, though one could argue that those are just the small number of cases where there happens to be an ambiguous aspectual form. Even more crucially, then, an ambiguity analysis gives us no way to explain why the whole class of such V1 verbs should share an additional volitional meaning. This behavior demands an explanation beyond that of coincidence, which would be the best that an ambiguity account could offer without additional tools or stipulations. Taking the V1 verbs to simply have aspectual main verb variant forms raises another question, though, that of comparison to the other aspectual main verbs of Biblical Hebrew, such as ḥalal ‘begin’. One difference between our V1 verbs and ‘other’ Biblical Hebrew aspectual main verbs is that of syntactic distribution: the latter appear with infinitive complements (as in (21)) and imperative complements (as in (22)), but not with (im)perfective complements as we saw in our V1V2 construction. ʾaḥaṯ lǝ=ḵull=am wǝ=zɛ ʾɛḥad wǝ=śap̄ a (21) hen ʿam they nation one and=language one to=them and=this la=ʿăśot haḥill=am begin.inf=their to=do.inf ‘the people is one, and they have all one language; and this they begin to do’ (Gen 11:6) (22) wǝ=ʾɛṯ ʾarṣ=o haḥel raš and-om land=his begin.imp.2ms possess.imp.2ms ‘and his land, begin to possess it’ (Deut 2:24)

- 978-90-04-44885-8

light verbs in biblical hebrew

185

If our V1 verbs are simply aspectual main verbs when in the V1V2 construction, then we have to do additional work to explain why they can combine with forms which other aspectual main verbs (outside of this construction) cannot. And to further illustrate the point, when ḥalal acts as a standalone main verb, it means ‘to profane/defile’ or ‘to pierce/wound’—even in the same binyan (e.g., Hiphil) (Köehler 2002)—as demonstrated by (23). (23) wǝ=lo ṯǝḥalel ʾɛṯ šem ʾɛ̆lohɛ=ḵa and=no profane.impf.2ms om name.of God=your ‘neither shalt thou profane the name of thy God’ (Lev 18:21) If anything, ḥalal presents an intuitively stronger call for an ambiguity account than do our V1 verbs. Where we can draw a fairly close connection between the meanings of qam, ‘to stand (physically)’ and ‘to stand (to do something)’, and where bridging contexts for those two meanings exist—cases where the two meanings coincide—as in (24), the same cannot be claimed as easily for the meanings of ḥalal. (24) wǝ=lot yošeḇ bǝ=šaʿar sǝḏom way=yarʾ lot and=Lot sit.ptcp.ms in=gate.of Sodom and=see.pret.3ms Lot way=yaqom liqraṯ=am and=stand.pret.3ms to=meet.inf=them ‘and Lot sat in the gate of Sodom: and Lot seeing them rose up to meet them’ (Gen 19:1) This suggests that while perhaps ambiguity is the right analysis for verbs like ḥalal, the case is certainly weaker for our V1 verbs like qam. An ambiguity account for our V1 verbs, then, raises new problems in that it fails to explain why these V1 aspectual main verbs behave differently from other aspectual main verbs in the language. It also leaves unexplained why our V1 verbs all convey an additional volitional meaning when in the V1V2 construction. 3.4 Serial Verb Construction Another potential analysis in which both the V1 and V2 verbs have the same status is one in which we analyze the V1V2 construction as a serial verb construction (SVC). These sorts of constructions, just like the construction which is our focus here, involve multiple verbs which jointly describe a single event. For example, consider the Nupe SVC in (25):

- 978-90-04-44885-8

186

snider

(25) Musa bé lá èbi. Musa came took knife ‘Musa came to take the knife.’ (Nupe, Tallerman 2014:(59)) In (25), the verb bé ‘came’ and lá ‘took’ work together to describe a single event (including the purpose thereof), and not both a coming event and a taking event. (Because of course, one’s coming to take doesn’t entail that any taking took place whatsoever.) One reason that serial verb constructions are taken to describe a single event, as opposed to a sequence of events, is that each verb in a SVC is either inflected identically, or else a single inflectional marker (on one verb) serves to mark the entire verbal construction. Crucially, if the verbs are marked differently, then only one is marked with tense/aspect/mood, and the rest are unmarked; SVCs do not allow verbs to be overtly marked with different tenses/aspects/moods. This, of course, accords with our V1V2 construction, which has both verbs marked identically (either as perfective or as imperative), or—if one’s analysis includes them—has overt marking on the V1 verb and leaves V2 an uninflected infinitive. Another reason to want to adopt a serial verb construction analysis relates to the relationship between this verbal complex and its subject. Many linguists describe SVCs as having a single shared subject (Collins 1993, Aikhenvald and Dixon 2006, though see Roberts 2009 for a dissenting view), and that is also true of our V1V2 constructions: we do not have any examples of our construction where V1 and V2 have different subjects. But another feature which is considered common to SVCs across languages is that they do not allow material to intervene between the verbs—and when they do, what is allowed to intercede are direct objects of transitive verbs. For example, the Yoruba and Vagala examples in (26) and (27), respectively, allow the object of the first verb to appear before the second verb in the construction. (26) ó mú ìwé wá he took book came ‘He brought the book.’ (Yoruba, Tallerman 2014: (61)) (27) ù kpá kíyzèé mòng ówl he take knife cut meat ‘He cut the meat with a knife.’ (Vagala, Tallerman 2014: (62)) But note that these are both still serial verb constructions, each describing only a single event. For example, (27) introduces the knife as an instrument of the

- 978-90-04-44885-8

light verbs in biblical hebrew

187

cutting event, not as a patient of a taking event. At first glance, this feature appears to be a point in favor of a SVC analysis for our V1V2 construction, as we for the most part do not see material intervening between V1 and V2. However, we have already seen subject intervene between the verbs, as in (9). And moreover, if any of the waws we have seen marking the second verb are wawconjunctive, or if waw-consecutive is properly analyzed as contributing conjunction in addition to its aspectual properties, then we have an overt meaningful intervener—in the form of conjunction—appearing between these ‘serial’ verbs, which runs contrary to what we expect for SVC s. Another difference between our construction and SVCs as they have been described in the literature arises in connection with the relative status of the verbs. In SVCs, each verb is of crucially equal status, both syntactically—in that neither can c-command the other—and semantically—in that neither contributes more than 1/# of the meaning of the entire verbal complex. And as we have seen, this is not the case for our V1V2 constructions, wherein V2 alone determines the construction’s valence and the type of event being described, while V1 is delexicalized and contributes only its aspectual meaning and a sense of volitionality. The two verbs in our V1V2 construction seem importantly unequal in status, which suggests that a SVC analysis might be ill-fitting. It is also worth noting that SVCs in the world’s languages can often contain more than two verbs which are serialized in the way we have seen thus far. For example, consider the Tariana SVC in (28), which has three serialized verbs. (28) phia-nihka phita pi-thaketa pi-eme you-rec.past.infer 2s.take 2s-cross+caus 2s-stand+caus ha-ne-na hyapa-na-nuku dem-distal-cl:vertical hill-cl:vertical-top.non.a/s ha-ne-riku-ma-se dem-distal-cl:loc-cl:pair-loc ‘Was it you who brought that mountain across (lit. take-cross-put.upright) (the river) to the other side?’ (asked the king) (Tariana, Aikhenvald and Dixon 2006: (6)) Meanwhile, we have no examples of V1V2V3 constructions in Biblical Hebrew which operate in parallel to the constructions we are investigating here. This suggests that, unless there is an additional syntactic constraint which prevents Biblical Hebrew SVCs from operating as flexibly as SVCs in other languages, our construction might not be best analyzed as a SVC. Finally, while a SVC analysis would explain the syntactic relationship between our verbs, it wouldn’t explain their individual aspectual contributions.

- 978-90-04-44885-8

188

snider

Note that SVCs tend to either build up complex meanings out of their parts (as in (26 and (28)) or deal with theta role assignment (as in (25) and (27)), as opposed to modifying aspectual information. A SVC analysis also would not explain why our Biblical Hebrew construction contributes a volitional meaning, nor why the V1 verbs are only drawn from a small restricted set. A SVC analysis, then, fails to account for all of the behavior of our Biblical Hebrew construction we had hoped to explain.

4

Summary

Biblical Hebrew makes use of a particular V1V2 construction wherein both verbs jointly are taken to describe a single event. In this construction, the first verb, V1, is somewhat delexicalized: instead of contributing its normal standalone main verb meaning, it contributes an aspectual meaning (which we can describe as duration shortening, or immediacy) as well as a non-aspectual meaning (namely volitionality). The verbs which can occupy this V1 position with this function are drawn from a small closed class, while any verb can occupy the V2 position. The V2 verb acts as the main verb in this construction, determining the valence of the overall verbal complex, as well as describing the type of event in question. After considering a number of possible analyses, I have argued here that these Biblical Hebrew V1V2 constructions are in fact best analyzed as light verb constructions, where V1 is the light verb. This type of analysis accounts for the observed behavior recounted above without needing to posit ambiguity, assume the existence of additional constraints on these verbs, or rely on additional explanations to account for the meanings contributed by the construction’s V1 verbs. This analysis fits into a larger picture of different verbal complex structures which have been attested across languages in general, as well as into the landscape of light verb constructions in particular. In addition to identifying the structure of this particular construction, it also provides insight into the syntax of the Biblical Hebrew verb system in general (and, for instance, how we treat infinitive verbs). Finally, this work bears on crosslinguistic diachronic work on grammaticalization paths, and how verb meanings can shift over time. This work may also change the way scholars of the Hebrew Bible read the text, because once one recognizes this particular V1V2 construction, one begins to see it throughout the text, contrary to received translations and despite even PP interveners…

- 978-90-04-44885-8

light verbs in biblical hebrew

189

(29) way=yaqom qayin ʾel hɛḇɛl ʾaḥ=iw and=rise.pret.3ms Cain to Abel brother=his way=yaharg=ehu and=kill.pret.3ms=him ‘Cain rose up against Abel his brother, and slew him’ (Gen 4:8)

References Aikhenvald, Alexandra Y. and Robert Malcolm Ward Dixon. 2006. Serial Verb Constructions: A cross-linguistic typology, vol. 2. Oxford: Oxford University Press. Butt, Miriam. 1995. The structure of complex predicates in Urdu. Stanford, CA: csli Publications. Butt, Miriam. 2003. The light verb jungle. Paper presented at the Workshop on MultiVerb Constructions, Trondheim, June 26–17. Chrzanowski, Jaroslaw. 2011. Verbal Hendiadys Revisited: Grammaticalization and Auxiliation in Biblical Hebrew Verbs. PhD thesis, The Catholic University of America, Washington, D.C. Collins, Christopher. 1993. Topics in Ewe syntax. PhD thesis, Massachusetts Institute of Technology, Boston, MA. Diesing, Molly. 1997. Light verbs and the syntax of aspect in Yiddish. Journal of Comparative Germanic Linguistics 1(2): 119–156. Diesing, Molly. 2000. Aspect in Yiddish: The semantics of an inflectional head. Natural Language Semantics 8(3): 231–253. Gamliel, Ophira and Abed al-Rahman Marʾi. 2015. Bleached verbs as aspectual auxiliaries in colloquial Modern Hebrew and Arabic dialects. Pp. 49–62 in Language Contact and the Development of Modern Hebrew, ed. Edit Doron. Studies in Semitic Languages and Linguistics 84. Leiden: Brill. Grimshaw, Jane and Armin Mester. 1988. Light verbs and θ-marking. Linguistic Inquiry 19(2): 205–232. Hopper, Paul J. and Elizabeth Traugott. 1993. Grammaticalization. Cambridge: Cambridge University Press. Koehler, Ludwig, Walter Baumgartner, and Johann Jakob Stamm, eds. 1994–2000. The Hebrew and Aramaic Lexicon of the Old Testament, trans. and ed. M.E.J. Richardson. 5 vols. Leiden: Brill. Mohanan, Tara. 1994. Argument structure in Hindi. Stanford, CA: csli Publications. Ramchand, Gillian. 1991. Complex predicate formation in Bangla. Pp. 443–458 in The Proceedings of the Ninth West Coast Conference on Formal Linguistics, ed. Aaron L. Halpern. Stanford, CA: csli Publications.

- 978-90-04-44885-8

190

snider

Ramchand, Gillian. 2008. Verb meaning and the lexicon: A first-phase syntax. Cambridge University Press Cambridge. Roberts, John R. 2009. How to find serial verbs in English: an rrg analysis of phase verb constructions. Paper presented at Role and Reference Grammar International Conference, Berkeley University, California. Rosen, Sara Thomas. 1990. Argument structure and complex predicates. PhD thesis, Brandeis University, Waltham, MA Tallerman, Maggie. 2014. Understanding Syntax. London: Routledge. Taube, Moshe. 2008. Verbal hendiadys in Yiddish. Unpublished manuscript.

- 978-90-04-44885-8

chapter 10

Argument Sharing Secondary Predicates in Biblical Hebrew Jacques E.J. Boulet

1

Introduction

This paper on secondary predicates in Biblical Hebrew (bh) represents a portion of a broader project to reappraise the Hebrew accusative using modern linguistic theory. Whereas the standard grammars tend to subdivide the accusative into two categories—for example, into objects and adverbials, or else into complements and adjuncts—I have argued elsewhere that we need at least three categories: arguments, modifiers, and secondary predicates (Boulet 2019). Recognizing secondary predication as a distinct syntactic category in bh will allow scholars to translate more accurately, to more confidently analyze the boundaries between clauses, and to recognize the equivalence of marked and unmarked variants of certain types of secondary predicates.1 Most linguistic work on secondary predicates is at least loosely compatible with Rothstein’s (2011, 1442) definition, which states that: Secondary predicates are one place non-verbal predicate expressions which occur under the scope of a main verb. Crucially, they share an argument with the main verb, the subject of the secondary predicate being either the subject or the direct object of the matrix verb. Typical English examples of secondary predication would be (1) and (2), which contain a depictive and resultative secondary predicate respectively, and where the subscript i marks coindexation with the shared argument. (1) Depictive: Esau ate the stewi hoti. (2) Resultative: Moses painted the linteli redi.

1 Some bh secondary predicates may be optionally marked by the particles bə- or lə-; see more on this below.

© Jacques E.J. Boulet, 2021 | doi:10.1163/9789004448858_011

- 978-90-04-44885-8

192

boulet

Why the adjectives hot and red are considered secondary predicates is made clear if we decompose the clauses into two parts (see Mikulskas 2017). Clause (1) can be broken down into (3): (3) Esau ate the stewi hoti. = Esau ate the stew. and (While Esau ate it) the stew (was) hot. Similarly, clause (2) can be broken down into (4): (4) Moses painted the linteli redi. = Moses painted the lintel. and (After Moses painted it) the lintel (was) red. At the most basic level, then, secondary predication is a case in which two predications are combined to form a single complex clause. In some cases, like those above, the equivalent statement could instead be made using two separate clauses, though this alternative is generally less efficient. Turning now to bh, the data that I have collected from the Pentateuch challenges three statements in Rothstein’s definition (Boulet 2019): first, that secondary predicates are “non-verbal” predicates, since bh participles can function as secondary predicates; second, that secondary predicates must “occur under the scope of a main verb” since in bh the primary predicate can be a copular predication, even a null copula; and third, that secondary predicates must “share an argument” with the main predicate, since there are both complement and adjoined varieties of secondary predication that share no argument. I therefore define secondary predication according to the definition in (5): (5) A secondary predicate is any non-finite predicate that occurs under the scope of another, primary predicate, whether by adjunction or as a complement. In this paper I focus on argument sharing secondary predicates. For bh I distinguish five types of argument sharing secondary predication: depictives, circumstantials, resultatives, explicit creation productives, and capacitives. The differences between the five types are captured using semantic representation. One may use both formal and informal systems of reprensentation. All argument sharing secondary predicates are adjoined. Those that are subjectoriented are adjoined in a higher position above the Verb Phrase (vp) and those

- 978-90-04-44885-8

argument sharing secondary predicates in biblical hebrew

193

that are object-oriented are adjoined in a lower position within the vp (see section 4). Before outlining the types of argument sharing secondary predication in bh (see section 3), I briefly summarize the state of affairs as regards the traditional understanding of secondary predicates in bh studies.

2

Previous Recognition of Secondary Predication in Biblical Hebrew

Hardly any work has dealt with secondary predicates in bh under the name ‘secondary predication.’ Perhaps the earliest mention available in English is found in the classic dictionary Brown-Driver-Briggs (1906, 89). There is a single mention there located under the entry for the particle beth, where they explain that beth may be used to introduce a primary predicate—that is, a copular clause—or a secondary predicate.2 Among their examples are (6) and (7) for primary and secondary predicates respectively, examples which are classic in the treatment of the so-called beth essentiae, which is better called beth of predication (Boulet 2018).3 (6) Beth Marking Copular Predicate wə=šem ha=ʾɛḥaḏ ʾɛ̆liʿɛzɛr ki ʾ ɛ̆lohe ʾaḇ=i and=name.cstr the=one Eliezer because God.cstr father=1cs bə=ʿɛzr=i pr=help=1cs ‘And the name of the (other) one (was) Eliezer, because “The God of my father (was) my help.”’ (Exod 18:4) (7) Beth Marking Depictive Secondary Predicate wa=ʾeraʾ ʾɛl ʾaḇraham ʾɛl yiṣḥaq wə=ʾɛl yaʾăqoḇ bə=ʾel and=1s.pst.appear to Abraham to Isaac and=to Jacob pr=El šadday Shaddai ‘I appeared to Abraham, to Isaac, and to Jacob as El Shaddai.’ (Exod 6:3)

2 I use the general word ‘particle’ in order not to pre-empt the analysis. Although bə- and lə- are very often prepositions meaning ‘in’ and ‘to,’ these particles sometimes have other functions. The function most relevant to this paper is that of ‘predicate marker’ which is glossed pr in the examples. 3 Abbreviations used in this paper for the linguistic glossing include (in alphabetical order): = ‘clitic boundary,’ adj ‘adjective,’ adv ‘adverb,’ caus ‘cause,’ cstr ‘construct,’ dom ‘differential

- 978-90-04-44885-8

194

boulet

In making these identifications, Brown-Driver-Briggs demonstrates strong intuitions, but they do little to define secondary predication. A more useful definition—though under a different name—is available from Joüon (1947, § 126a). In his description of the accusative in Biblical Hebrew, Joüon introduces the term accusatif prédicatif d’état (or ‘predicative accusative of state’)4 which he describes as the addition of a complementary phrase to a verbal proposition which expresses a state or quality of the subject or object. This definition is essentially the same as that given by contemporary linguists for depictive secondary predicates,5 and indeed I find that most of Joüon’s listed examples can be classified as depictives. Joüon gives the following bold-face examples of predicative accusative of state under adjectives (8), participles (9) and substantives (10) respectively: (8) Depictive Adjective:6 ʾăni məleʾ-a halaḵti wə=reqam hɛ̆šiḇa=ni 1cs full.adj-fs go.pfv.1s and=empty.adv caus.return.pfv.3ms=1sg yhwh Yahweh ‘I went (out) full, but Yahweh brought me back empty.’ (Ruth 1:21) (9) Depictive Participle way=yišmaʿ mošɛ ʾɛt ha=ʿam boḵɛ and=3msg.pst.hear Moses dom the=people.ms weep.ptcp.ms lə=mišpəḥoṯay=w ʾiš lə=p̄ ɛṯaḥ ʾohŏl=o by=clans=3ms man at=entrance tent=3ms ‘And Moses heard the people weeping throughout their clans, (each) man at the entrance of his tent.’ (Num 11:10) (10) Depictive Substantive gəḏuḏ-im wa=ʾăram yaṣəʾu and=Aram go.out.pfv.3cpl raider-mp ‘And Aram went out (as) raiders.’ (2Kgs 5:2) object marker,’ f ‘feminine,’ ia ‘infinitive absolute,’ irr ‘irrealis,’ juss ‘jussive,’ m ‘masculine,’ pfv ‘perfective,’ p ‘plural,’ pr ‘predication marker,’ pst ‘past,’ s ‘singular,’ ptcp ‘participle.’ 4 See also the English translation Joüon and Muraoka (2006, §126a). 5 For example, Rothstein (2011, 1442) writes that “depictives can be predicates of either subject or object of the main verb and typically express a property that their subject has while the event expressed by the main verb is going on.” 6 Here I follow dch, halot, and bdb in treating reqam as an adverb and the ending -am as adverbial. However, if reqam were analysed as an adjective, then it would also be a depictive.

- 978-90-04-44885-8

argument sharing secondary predicates in biblical hebrew

195

More recently secondary predicates have been tagged under the name ‘predicative adjunct’ in the etcbc database available for OakTree Software’s Accordance program (version 12.1.0).7 This database has 81 predicative adjuncts tagged for the books of Genesis through Deuteronomy.8 Although etcbc uses the single term ‘predicative adjunct,’ I have classified their tagged examples under seven categories: depictives (26×), capacitives (26×), circumstantials (12×), dynamic complementatives (10×), explicit creation productives (3×), static complementatives (1×), and direct objects (3×).9 It is clear that etcbc shares a similar view to mine regarding what counts as a secondary predicate, but the database is not yet systematic in its approach. First, etcbc could have tagged many more examples that are just like the ones they have tagged. The fact that so many have been overlooked suggests an ad hoc approach to clause analysis. Second, while many of the tagged examples could indeed be classified as adjuncts (i.e. the depictives, capacitives, circumstantials, and explicit creation productives), several cannot be (i.e. the dynamic and static complementatives). Besides my own work, I know of only one recent presentation abstract by Hatav (2017) that takes a modern linguistic approach to secondary predication. Hatav argues that the bh ‘infinitive absolute’ is a converb which may be used for depictive and resultative secondary predication. She presents (11) as an example of the infinitive absolute used as a resultative sp. hakke u=p̄ aṣoaʿ ha=ʾiš (11) way=yakke=hu and=3ms.pst.strike=3ms the=man strike.ia and=wound.ia ‘And the man struck him, wounding [him].’ (1 Kgs 20:37) It will not be possible to interact with Hatav’s work until more is available in writing. Her proposal is somewhat radical given the traditional interpretation

7 etcbc stands for the Eep Talstra Center for Bible and Computer, previously known as the Werkgroep Informatica Vrije Universiteit (wivu). 8 Note that an Accordance search for predicative adjuncts in those five books turns up 104 hits (as of Sept. 20, 2018). However, this larger number includes certain phrases as more than one hit, whereas I have counted each phrase once. 9 Depictives: Gen 7:1, 16; 15:2; 24:21; 25:25; 33:18; 35:29 38:11; Exod 13:18; 21:5; 22:3; Lev 6:9; 10:12; 16:10; 19:16; 20:20; Num 16:30, 33; 24:4, 16; 32:30, 32; Deut 3:18; 15:12, 13, 18. Capacitives: Gen 16:3; 17:8; 23:9; 26:34; 27:37; 30:4, 9; 34:8, 12; 41:45; Exod 6:8; 17:14; 22:12; Lev 3:14; 6:6; 7:34; 14:21; Num 18:24; 28:12 (x2), 13; 29:19, 25; 31:54; Deut 29:7; 31:21. Circumstantials: Exod 12:9; Lev 2:11; 3:1, 6; 5:6; 6:10; 19:19 (x2); 23:17; Num 2:16, 24; Deut 22:9. Dynamic Complementatives: Lev 13:3, 4, 10, 13, 20, 25; Num 5:21 (x2); Deut 26:19; 28:1. Explicit Creation Productives: Gen 1:27; 5:2; Exod 39:9. Static Complementatives: Gen 31:15. Direct Objects: Lev 2:14; Num 28:11; 31:3.

- 978-90-04-44885-8

196

boulet

of the infinitive absolute. In the meantime, I must prefer the approach of Cook and Holmstedt (2013) who use the term ‘adverbial infinitive’ for this non-finite verb form. This adverbial analysis is more consistent with the history of interpretation.

3

Argument Sharing Secondary Predicates

Although some scholars prefer to see all secondary predicates as subtypes of main categories like depictives and resultatives (e.g. Rothstein 2011), I argue that distinctions in either syntax or semantics need to be recognized (Boulet 2019). Even subtle distinctions in the semantic representation of two secondary predicates can lead to superior biblical interpretation. In this paper I present five distinct types of argument sharing secondary predicates: depictives, circumstantials, resultatives, explicit creation productives, and capacitives. Argument sharing secondary predicates share a common syntax, which distinguishes them from non-sharing adjunct and complement secondary predicates. However, the five types can be distinguished semantically using formal or informal means of representation. Here I use an informal system of English translation with semantic tags when possible. In order to distinguish capacitives I also use a formal Neo-Davidsonian representation. 3.1 Depictives Depictives specify the state of one of the primary predication’s arguments, either the subject or the object, at the time that the primary event unfolds. Thus, recalling example (1) we may say that in the sentence Esau ate the stewi hoti, the adjective hot specifies that the stew was hot at the time of the eating event. The stew may have been hot or cold at some time before the eating, or even after the eating. But at the time of eating, the stew was hot. Accordingly, the semantics of example (1) can be represented according to (12), which features two kinds of semantic tag. The first tag, ‘(while)’ in brackets, makes explicit the relationship between the primary and secondary predications, namely that the two ‘events’ represented by those predications occur concurrently.10 The second tag, ‘be’ in small caps, states that the secondary predication is a static predication (or state) that involves no change.

10

Some linguists prefer the term ‘eventuality.’ In either case, I use the terms broadly to include actions, changes, and states; see Cuervo (2003, 16) and Rothstein (2011, 1452).

- 978-90-04-44885-8

197

argument sharing secondary predicates in biblical hebrew

(12) Esau ate the stewi hoti. = Esau ate the stew (while) the stew be hot. In English, depictives tend to be realized by adjectives. Biblical Hebrew, on the other hand, allows noun phrases and participles in addition to adjectives. The following bold-face examples are object-oriented depictives from the adjective (13), noun (14), and participle (15) categories respectively. (13) way=yaḇeʾ yosep̄ ʾet dibbaṯ=am raʿa ʾɛl ʾăḇi=hɛm and=3ms.pst.bring Joseph dom report.fs=3mp bad.fs to father=3mp Lit: ‘And Joseph brought their report bad to their father.’ (Gen 37:2) (14) wə=laḇaš hak=kohen midd=o ḇad and=put.on.irr.3ms the=priest garment=3ms linen ‘And the priest shall put on his garment (being) linen.’ (Lev 6:3) (15) way-yaśśiḡu ʾoṯ=am ḥonim ʿal hay=yam kol and=3mp.pst.overtake dom=3mp camp.ptcp.mp on the=sea all sûs rɛḵɛḇ parʿo u=p̄ arašay=w horse.cstr chariot.cstr Pharaoh and=horsemen=3ms wə=ḥel=o ʿal pî haḥirot lip̄ ne baʿal ṣəp̄ on and=army=3ms on Pi Hahiroth before Baal Zaphon ‘And all of Pharaoh’s chariotry and his horsemen and his army overtook them camping by the sea near Pi Hahiroth, in front of Baal Zaphon.’ (Exod 14:9) Biblical Hebrew depictives may be unmarked, as in the above examples, or they may be marked by the particle bə- (i.e. the beth of predication). Two examples are given below: (16) with a noun phrase depictive and (17) with an adjective. (16) bə=šiḇʿim nɛp̄ɛš yarəḏu ʾăḇoṯe=ka miṣrayəm=a pr=seventy life go.down.pfv.3cp fathers=2ms Egypt=ward ‘Your ancestors went down to Egypt (as) seventy persons.’ (Deut 10:22) (17) wə=loʾ ḇiʿarti mimmɛn=nu bə=ṭameʾ and=not remove.pfv.1s from=it pr=unclean.ms Lit: ‘And I have not removed (anything) from it unclean.’ (Deut 26:14) Using the same informal representation as in (12), example (17) can be represented as in (18):

- 978-90-04-44885-8

198

boulet

(18) I have not removed [anything] from it (while) I be unclean. Note that the particle bə- appears to contribute nothing to the meaning; it serves only to mark the predicate. Just as in the unmarked cases above, the bethmarked depictive serves to specify a secondary predication which overlaps in time with the primary event. 3.2 Circumstantials Circumstantials superficially resemble depictives, but they can and should be distinguished semantically.11 Like depictives, circumstantials specify a state of affairs that overlaps temporally with the primary event. The difference between them is that circumstantials impose a condition on the primary event. For example, consider (19): (19) Esau eats stewi hoti (but not cold). Just as in (1) the eating of the stew by Esau occurs while the stew is hot. However, in (19) the temperature of the stew is a condition. Esau is happy enough to eat hot stew, but he refuses to eat stew if it is not hot. The informal semantic representation of (19) is as in (20): (20) Esau eats stew (if) stew be hot. The conditional nature of circumstantials means that they are compatible with verbs of particular tense, aspect, or modality. In Biblical Hebrew circumstantials occur with jussives, imperfects, and irreal perfects (also called wəqaṭal). Within my corpus, most of these occur within the legal prescriptions of the Pentateuch where God stipulates that rituals are to be done a certain way. Example (21) from Exodus 12 is typical: (21) ʾal toʾḵlu mimmɛn=nu naʾ u=ḇašel məḇuššal not 2mp.eat from=3ms raw and=boiled boil.pass.ptcp bam=mayim ki ʾim ṣəli ʾeš roʾš=o ʿal kəraʿay=w wə=ʿal in.the=water but if roasted fire head=3ms on legs=3ms and=on qirb=o innards=3ms ‘Don’t eat any of it raw or boiled in water, but only roasted in fire, (including) its head, as well as its legs and its innards.’ (Exod 12:9) 11

But see Rothstein (2011) who prefers to analyze circumstantials as a subtype of depictives.

- 978-90-04-44885-8

199

argument sharing secondary predicates in biblical hebrew

The above sentence is from the instructions for eating the Passover lamb. It includes three circumstantials: the adjectival phrases naʾ ‘raw’ and ḇašel məḇuššal bammayim ‘boiled in water,’ and the noun phrase ṣəli ʾeš ‘(thing) roasted in fire’.12 The first two set negative conditions on the consumption of the lamb and the last one sets a positive condition. The relevant portion of (21) can be represented as in (22): (22) Do not eat it (if) it be raw or boiled in water. and Only [eat it] (if) it be roasted in fire. 3.3 Resultatives Classic resultatives describe a state of one of the arguments of the primary predication that is directly caused by that primary event.13 Resultatives therefore have a temporal extension that is subsequent to the runtime of the primary event. They also represent a change predication, which is represented by the tag ‘become’ in small caps, instead of the static tag ‘be’ used with depictives and circumstantials. Example (2), which reads Moses painted the linteli redi, is a resultative. Moses was engaged in a painting action, and the result of Moses’ painting is that the lintel changed color. Accordingly, sentence (2) can be represented as in (23): (23) Moses painted the linteli redi. = Moses painted the lintel (causing) the lintel become red. As with depictives, English resultatives tend to be adjectives. So far all of my Biblical Hebrew resultative examples are nominal. Resultatives may be unmarked, as is the case of ṣippuy lammizbeaḥ ‘overlay for the altar’ in (24): (24) wa=yəraqqəʿu=m ṣippuy lam=mizbeaḥ and=3mp.pst.hammer=3mp overlay for.the=altar ‘And they hammered them (into) an overlay for the altar.’ (Num 17:4)

12

13

The final portion of (21), roʾšo ʿal kəraʿayw wəʿal qirbo ‘its head, as well as its legs and its innards,’ could potentially also be taken as a circumstantial, though I have taken it here to be in apposition to the attached 3ms pronoun on mimmɛnnu ‘from it.’ In Boulet (2019) I distinguish between transitive/unaccusative and unergative resultatives (cf. Bowers 1997). The resultatives treated in this presentation are transitive/unaccusative resultatives. Unergative resultatives do not share an argument. Structurally they are complements, though they are optional.

- 978-90-04-44885-8

200

boulet

Or, resultatives may be marked by lə-, like ləʿapar ‘to dust’ in (25): (25) wa=ʾekkoṯ ʾoṯ=o ṭaḥon heṭeḇ ʿaḏ ʾăšɛr daq and=1sg.pst.beat dom=3ms grind.ia do.well. ia until that fine lə=ʿap̄ar pr=dust ‘And I beat it—grinding (it) well until fine—into dust.’ (Deut 9:21) Example (24) may be represented semantically as in (26): (26) They hammered them (causing) them become an overlay for the altar. That is to say, Eleazar’s men took the the censers involved in Korah’s rebellion (see Numbers 16) and they beat them out with the result that they were changed into an overlay for the altar. 3.4 Explicit Creation Productives (ecps) Explicit creation productives (or ecps) are superficially like resultatives, and indeed, they have sometimes been treated as if they were resultatives. However, Rapaport (2018) has shown that they only seem to be resultative because the inherent semantics of the verb suggest a result. Specifically, ecps occur with verbs of explicit creation. English verbs of explicit creation are those like make, create, and build. Likewise, typical Hebrew verbs of explicit creation are bana, baraʾ, and ʿaśa. The objects of such verbs are produced, rather than simply acted upon.14 An example of an ecp in English is the adjective strong in (27): (27) The contractor built the housei strongi. Rapaport (2018) has taken ecps to be depictives rather than resultatives. However, I argue that they are distinct from both (Boulet 2019). Like resultatives, and unlike depictives, the runtime of an ecp coincides with the conclusion of the primary event, stretching forward from that point in time. If the event is building a house, then we can say that the product is only truly a house when the act of building is complete. With regard to (27), a half-built house could hardly be said to be a strong house.

14

Cf. the ‘effected-object accusative’ as discussed by Waltke and O’Connor (1990, §10.2.1f.) and Joüon and Muraoka (2006, 125p).

- 978-90-04-44885-8

argument sharing secondary predicates in biblical hebrew

201

Yet, like depictives, but unlike resultatives, an ecp is a static predicate (that is, be), not a change predicate (or, become). Example (27) does not involve a situation where a contractor worked on a weak house in order to change it into a strong house. Rather, the contractor started from scratch, and it was from the moment that building was complete and the product of building could be considered a house that the house was strong. The consequence of the above arguments is that ecps have a semantic representation distinct from that of both depictives and resultatives. The representation of (27) is (28): (28) The contractor built the house (such that) the house be strong. In Biblical Hebrew, explicit creation productives are mostly unmarked, but some few are marked with bə-.15 Example (29) has the unmarked ecp zaḵar unəqeḇa ‘male and female:’ (29) zaḵar u=nəqeḇa baraʾ ʾoṯ=am male and=female create.pfv.3ms dom=3mp ‘Male and female, He created them.’ (Gen 1:27b) A similar example from the very same verse has an ecp marked with the beth of predication (30): ʾoṯ=o ʾɛ̆lohim baraʾ (30) bə=ṣɛlɛm create.pfv.3ms dom=3ms pr=image.cstr God ‘Image of God, He created it [humanity].’ Both examples have the same syntax and the same basic semantics.16 The only significant difference is that in (29) the secondary predication is unmarked, while in (30) it is overtly marked by bə-. Beth contributes nothing to the meaning; it is merely a syntactic marker. In each case human beings are said to have been created such that, when their creation was complete, they were a certain way. Taken together, these clauses state that God created human beings to exist on the earth as ‘male and female’ and also as the ‘image of God.’ The semantic representation of (30) is as in (31): (31) God created [humanity] (such that) it be image of God. 15 16

All those marked with bə- identified so far occur in the early part of Genesis: Gen 1:26, 27 (2×); 5:1, 3; 9:6. In both (29) and (30) the ecp is fronted to a clause-initial focus position (see Holmstedt 2014).

- 978-90-04-44885-8

202

boulet

3.5 Capacitives Capacitives express the specific role or capacity in which an argument of the primary predication participates in that primary predication. An English example would be (32), where the word as introduces the secondary predicate (see Bowers 1993, 2001): (32) Samsoni served his people as judgei. The informal representation of (32) might be (33), though it leaves something to be desired, since it cannot in this way be sufficiently distinguished from a depictive: (33) Samson served his people (and) Samson be judge. At this point a formal Neo-Davidsonian event semantics is superior. The representation in (34) states that: there is a ‘serving’ event e1 such that the Agent of e1 is ‘Samson’ and the Patient of e1 is ‘his people,’17 and there is also a stative event e2 of ‘being judge’ such that the Holder of e2 is ‘Samson’ and such that e2 is the capacity in which the shared participant ‘Samson’ participates in e1.18 (34) ∃e1[serving(e1) & Agent(e1, Samson) & Patient(e1, his people) & ∃e2[be-judge(e2) & Holder(e2, Samson) & Capacity(e2, e1)]] Hebrew capacitives function the same way, though the majority are objectoriented. They may be unmarked, or else marked with lə- with the same meaning. An unmarked example would be ʾiššâ ‘wife’ in (35): (35) way=yiqqaḥ ʾišša ʾɛṯ yəhuḏiṯ bat bəʾeri ha=ḥitti and=3ms.pst.take wife dom Judith daughter.cstr Beeri the=Hittite wə=ʾɛt baśəmaṯ baṯ ʾelon ha=ḥitti and=dom Basemath daughter.cstr Elon the=Hittite ‘And he took (as) wife Judith, daughter of Beeri the Hittite, and (also) Basemath, daughter of Elon the Hittite.’ (Gen 26:34)19

17 18

19

Here Agent and Patient are meant in the proto-role sense of Dowty (1991). One may also add the statement ‘& T(e1) ⊆ T(e2),’ which means that the runtime of e1 is a subset (⊆) of the runtime of e2 (i.e. there is complete temporal overlap between the runtimes of e1 and e2). I consider the word ʾiššâ in (35) to be fronted to a focus position.

- 978-90-04-44885-8

203

argument sharing secondary predicates in biblical hebrew

We may compare the above to the marked ləʾiššâ in (36): (36) wa=ʾɛqqaḥ ʾoṯ=āh l=î lə=ʾiššâ and=1sg.pst.take dom=3fs for=1cs pr=wife ‘And I took her for myself as wife.’ (Gen 12:19) As stated above, argument sharing secondary predicates may be broken down into two clauses. This is the case in (37) which uses two clauses to express the same essential meaning as above: (37) way=yiqqaḥ ʾɛṯ riḇqa wat=təhi l=o lə=ʾiššâ and=3ms.pst.take dom Rebekah and=3fs.pst.be to=3ms pr=wife ‘And he took Rebekah. And she became a wife to him.’ (Gen 24:67) This last example supports the claim that the lə- introducing the noun ʾiššâ in (36) contributes little semantically. It is usual for Hebraists to interpret haya ‘to be’ plus lə- as ‘to become.’20

4

Syntactic Structure

The data reviewed in this paper has shown that argument sharing secondary predicates may be optionally marked by bə- or lə-, depending on the type, with little or no semantic contribution from the particle.21 I take this evidence to support the generalized theory of predication advanced by Bowers (1993, 2001) schematized in (38). According to Bowers, every predication has a Predication Phrase (PrP) structure with a predication head (Pr). The head serves to link a subject with its predicate, and this head may be overt or phonologically null. This same structure applies to primary copular and verbal predicates, as well as to secondary predicates.

20 21

See Wilson (2015) who argues that this construction indicates ‘change-of-state.’ In other work I show that bə- and lə- may also mark primary copular predicates and complementative secondary predicates (Boulet 2018, 2019).

- 978-90-04-44885-8

204

boulet

(38) Predication Phrase: Basic Structure

This analysis is supported by cross-linguistic data. According to Gardiner (1957), the Egyptian prepositions m ‘in’ and r ‘to’ are both used to mark predicates. It is remarkable that Egyptian employs prepositions for this purpose that are exact analogues of Biblical Hebrew bə- and lə-. Some scholars have already noted that the Arabic proclitic bi, a cognage of bə-, is used the same way (e.g. Manross 1954, Gordon 1981). Even outside the Afro-Asiatic family the preposition meaning ‘in’ is known to have been grammaticalized as a predicate marker in certain environments, in particular the Scottish Gaelic preposition ann (Adger and Ramchand 2003, 332) and the Irish preposition ina (Chung and McCloskey 1987, 179, n. 4). When translating the beth or lamed of predication into English, the semantic distinction between static (be) and change (become) predications may not come through. Frequently, if not always, the particles are best glossed using one of the standard strategies available to English for encoding secondary predicates. These include the unmarked option, as well as the marked options as, for, and to be (D’hoedt and Cuyckens 2017, 16–17). I have presented five distinct semantic types of argument sharing secondary predicate. All five types share two possible syntactic structures, depending on whether the secondary predicate is a predicate of the subject or the object of the primary predication. Both kinds involve adjunction of the secondary predication (PrP2) somewhere below the primary predication (PrP1). Subjectoriented predicates, like the PrP2 in (39), adjoin within the main predication PrP1, where it has access to the subject. Sharing of the subject argument is assumed to occur by movement from the specifier of PrP2 to the specifier of PrP1. Moved constituents are marked by angle-brackets.22

22

I assume that argument sharing happens according to the Movement Theory of Control (mtc) in one of its forms (Hornstein 1999; Bowers 2008).

- 978-90-04-44885-8

argument sharing secondary predicates in biblical hebrew

205

(39) Subject-Oriented Secondary Predicate:

Object-oriented secondary predicates are adjoined within vp, as in (40). The shared object is assumed to move from the specifier of PrP2 to the complement position of the verb. (40) Object-Oriented Secondary Predicate:

Given the two structures above, we may distinguish the (abstract) Case of secondary predicates based on orientation. Grammarians like Joüon (1947, § 126a) have previously taken examples like gədûdîm ‘raiders’ in (10) to be accusatives. However, I would rather expect subject-oriented secondary predicates to be nominative, just like the ‘predicate nominative’ of traditional grammar (see Waltke and O’Connor 1990, §8.3).

- 978-90-04-44885-8

206 5

boulet

Conclusion

In this study I have described five types of argument sharing secondary predication occurring in bh: depictives, circumstantials, resultatives, explicit creation productives, and capacitives. The significance of each of these types can be articulated by a distinct semantic representation, whether formal or informal. Argument sharing can be diagnosed through decomposition. The informal semantic representation used here is based on decomposition and simple semantic tags. However, informal representation has its limitations. To properly distinguish capacitives it was necessary to use formal Neo-Davidsonian representation. In general secondary predicates are never determined by the article, but are occasionally determined by a pronoun suffix or by a proper noun in a bound phrase. I have shown that secondary predicates may be optionally marked by be- or lə- with little or no difference in meaning compared to unmarked cases. Depictives and explicit creation productives may be optionally marked with be-, while resultatives and capacitives may be optionally marked with lə-. I have followed Bowers (1993, 2001) in employing a unified approach to predication. Within this approach the optional particles be- and lə- can be analyzed as overt realizations of the functional head Pr. Another advantage of this approach is that it allows straightforward mapping from syntax to semantics (see further Boulet 2019). Each Prx head in the tree is taken to introduce an event variable ex in a formal Neo-Davidsonain semantic representation. Event modification (e.g. temporal and locative modifiers) are assumed to adjoin within the corresponding PrP. Such adjoined modifiers have access to the event variable introduced by Prx and show up in the formal representation as predicates of ex.

References Adger, David, and Gillian Ramchand. 2003. Predication and Equation. Linguistic Inquiry 34 (3): 325–359. Boulet, Jacques E.J. 2018. The Biblical Hebrew Bēth Essentiae: Predicate marker. Paper presented at the H.H. Bingham Colloquium in New Testament Studies, Hamilton, June 15. Boulet, Jacques E.J. 2019. A Linguistic Reappraisal of the Biblical Hebrew Accusative. PhD thesis, University of Toronto, Toronto, Ontario. Bowers, John. 1993. The Syntax of Predication. Linguistic Inquiry 24 (3): 591–656. Bowers, John. 1997. A Binary Analysis of Resultatives. Pp. 43–58 in Texas Linguistic

- 978-90-04-44885-8

argument sharing secondary predicates in biblical hebrew

207

Forum 38: The Syntax and Semantics of Predication, edited by R.C. Blight, and M.J. Moosally. Austin, TX: Department of Linguistics, University of Austin Texas. Bowers, John. 2001. Predication. Pp. 299–333 in The Handbook of Contemporary Syntactic Theory, ed. Mark Baltin and Chris Collins. Malden, MA: Blackwell. Bowers, John. 2008. On Reducing Control to Movement. Syntax 11 (2): 125–143. Brown, Francis, S.R. Driver, and Charles A. Briggs. 1906. A Hebrew and English Lexicon of the Old Testament with an Appendix Containing the Biblical Aramaic. Oxford: The Clarendon Press. Chung, Sandra, and James McCloskey. 1987. Government, Barriers, and Small Clauses in Modern Irish. Linguistic Inquiry 18: 173–237. Cook, John A., and Robert D. Holmstedt. 2013. Beginning Biblical Hebrew: A Grammar and Illustrated Reader. Grand Rapids: Baker Academic. Cuervo, María Cristina. 2003. Datives at Large. PhD thesis, Massachusetts Institute of Technology, Cambridge, MA. D’hoedt, Frauke, and Hubert Cuyckens. 2017. The Development of the as-Secondary Predicate Construction. Language Sciences 59: 16–35. Dowty, David. 1991. Thematic Proto-Roles and Argument Selection. Language 67 (3): 547–619. Gardiner, Alan. 1957. Egyptian Grammar: Being an Introduction to the Study of Hieroglyphs. 3rd ed. Oxford: Griffith Institute, Ashmolean Museum. Gordon, Cyrus H. 1981. ‘In’ of Predication or Equivalence. Journal of Biblical Literature 100 (4): 612–613. Hatav, Galia. 2017. Secondary Predicates and Converbs in Biblical Hebrew. Paper presented at the Thursday Interdisciplinary Colloquium. Tel Aviv University, November 9. Holmstedt, Robert D. 2014. Critical at the Margins: Edge Constituents in Biblical Hebrew. Kleine Untersuchungen zur Sprache des Alten Testaments und seiner Umwelt 17: 109–156. Hornstein, Norbert. 1999. Movement and Control. Linguistic Inquiry 30 (1): 69–96. Joüon, Paul. 1947. Grammaire de l’ hébreu biblique. 2nd ed. Rome: Institut Biblique Pontifical. Joüon, Paul, and T. Muraoka. 2006. A Grammar of Biblical Hebrew. Rev. ed. Rome: Pontificio istituto biblico. Manross, Lawrence N. 1954. Bêth Essentiae. Journal of Biblical Literature 73 (4): 238–239. Mikulskas, Rolandas. 2017. Copular Constructions in Lithuanian. Amsterdam: John Benjamins. Rapoport, Tova. 2018. Secondary Predicates: Insights from Creation Verbs. Paper presented at the University of Calgary. Calgary, July 30. Rothstein, Susan. 2011. Secondary Predicates. Pp. 1442–1462 in An International Hand-

- 978-90-04-44885-8

208

boulet

book of Natural Language Meaning. Vol. 2, ed. Klaus von Heusinger, Claudia Maienborn and Paul Portner. Berlin: De Gruyter Mouton. Waltke, Bruce K., and M. O’Connor. 1990. An Introduction to Biblical Hebrew Syntax. Winona Lake: Eisenbrauns. Wilson, Daniel Joseph. 2015. Copular Predication in Biblical Hebrew. ma thesis, University of the Free State, Bloemfontain, South Africa.

- 978-90-04-44885-8

chapter 11

The Causative-Inchoative Alternation and the Semantics of Hiphil Kevin Grasso

1

Introduction

The causative-inchoative alternation is characterized by a single verb that can be expressed either as a causative or an inchoative, sometimes accompanied by a change in morphology.1 This is demonstrated in English by the following pair: (1) a. Causative—Jim broke the window. b. Inchoative—The window broke. Languages may or may not morphologically mark either variant, but the marking is not completely random, as shown by Haspelmath (1993). Following (Alexiadou, Anagnostopoulou and Schäfer 2006; Rappaport Hovav 2014b), I assume that both variants in the alternation represented in (1) should be analyzed as having a single lexical representation where the only argument required by the verb is the internal argument (i.e., the thing broken, the window). I argue below that both variants involve scalar change (Rappaport Hovav 2014a). Thus, the difference between the two variants is primarily syntactic: it pertains to whether or not the cause is syntactically present. The alternation above is, then, a Voice alternation (following (Alexiadou, Anagnostopoulou and Schäfer 2015)) that also requires two events. What we will see in Biblical Hebrew (bh) is that different kinds of Voice alternations are possible, as are the sources of bieventivity. In particular, the Voice alternation may be morphologically marked (as in the Qal-Niphal alternation) or unmarked (as in the Hiphil alternation),

1 I refer to the variant with an external argument as “causative” and the variant without an external argument as “inchoative”, though I will argue below that both variants are in fact causative semantically, even if there is no explicit causer in the syntax. Thus, “causative” in the syntactic sense simply means that there is an explicit “causer” in the syntax (e.g. Jim) responsible for the change of state, while “inchoative” means there is a change of state without mention of a specific causer. I define semantic causation below.

© Kevin Grasso, 2021 | doi:10.1163/9789004448858_012

- 978-90-04-44885-8

210

grasso

and bi-eventivity may be a product of the root or the binyan. I will also suggest, following (Alexiadou 2014), that some alternations lack Voice altogether, which will be relevant for one class of the alternation in Hiphil discussed below. This paper has two primary goals: 1. To provide an analysis of the zero-marked causative-inchoative alternation in Hiphil. 2. In doing 1, to shed some light on the syntax and semantics of the binyan system in Biblical Hebrew (bh), particularly the Hiphil binyan. In the rest of the introduction, I introduce the binyan system and the basics of argument alternations cross-linguistically and in bh. In section 2, I discuss the unmarked causative-inchoative alternation in bh and provide a syntactic analysis for it. I compare this analysis in 3 to the similar alternation found in Qal-Niphal, which I call the anticausative alternation following (Haspelmath 1993),2 and conclude this section by comparing the two alternations. I conclude with some implications for our understanding of the binyan system in general and provide some examples of exegetical issues related to the alternations discussed. 1.1 The Binyan System in Biblical Hebrew The Binyan system in bh is a difficult issue with many complexities. Although there are differences between the bh and Modern Hebrew (mh) systems, I present the system as outlined in (Doron 2003; Doron and Alexiadou 2012), omitting Piel, Pual, and Hitpael, since they are not part of the current discussion:

Voice

Simple

Causative

Active Passive Middle

a-a (Qal) ?3 n + i-a (Niphal)

h + i-i (Hiphil) h + u-a (Huphal)

2 I use the term “anticausative” as a strictly morphological term, not semantic. I reserve the term “inchoative” for the semantic interpretation of a change of state where the cause is not named and “causative” for a change of state where the causer is named. 3 Although there is evidence for a passive counterpart to the Paʾal form, it is fading out of use by the time of Biblical Hebrew and is no longer productive. The main form used to express the passive variant of Paʾal is the medio-passive Niphal.

- 978-90-04-44885-8

the causative-inchoative alternation

211

The idea that Niphal represents a middle and Hiphil a causative is by no means new (the observation goes back at least to Gesenius 1910).4 What becomes immediately obvious when looking at the system is that there are “gaps” in the marking: Qal has no passive counterpart, and Hiphil has no middle counterpart. Below, I describe how these “gaps” are handled by the system and the different ways in which they are accommodated. First, we must discuss a little more how the binyan system works, particularly how it interacts with roots. Work in Distributed Morphology has shown that roots should be separated from their categorizers, and this is particularly evident in a language like bh. A root like šbr is not in any of the categories noun, verb, or adjective, but it is “uncategorized.” A “categorizer” is normally called “little n/v/a”, depending on which category the word formed falls into. Examples of these in English include morphemes like -tion (e.g. construction; n categorizer), -ify (e.g. justify; v categorizer), and -able (e.g. buildable; a categorizer). To make a word in bh, the root combines with a “categorizer” (which in bh requires a particular vowel pattern, for example). Other work in Distributed Morphology has proposed that there may be different kinds of categorizers with different semantics. Within the verbal domain, different “flavors” of little v have been proposed which correspond to different semantic interpretations (Folli and Harley 2005). Although the details of my proposal cannot be spelled out, I suggest that Qal, Piel, and Hiphil are realizations of different “flavors” of little v. Thus, they both categorize the root and (potentially) provide their own semantics. Qal, I argue, is neutral—it only turns a root into a verb and does not contribute any semantic interpretation. Hiphil, on the other hand, contributes a result to the semantics of the root. Niphal and Huphal do not contribute any semantic interpretation, but they mark different Voices, medio-passive and passive respectively (this builds on the work of (Doron 2003) and (Arad 2005), but also departs from their analyses in some significant ways).5

4 The relevant sections are § 51 (Niphal) and § 53 (Hiphil). 5 Again, space does not permit an adequate treatment of the category of Voice, which has been discussed extensively in the linguistics literature. I only note here that the semantics of medio-passive voice is not limited to reflexive, though reflexive may be one of its interpretations. According to (Alexiadou and Doron 2012:4), non-active voice may receive a variety of different interpretations, such as inchoative (intransitive break), naturally reflexive verbs (shower (oneself)), naturally reciprocal verbs (meet), dispositional middles (this bread cuts easily), medio-passives (usually a combination of anticausative and passive), and passive (The window was broken). The interpretation of a middle form largely depends on whether it contrasts with another passive form as well as on the semantics of the root.

- 978-90-04-44885-8

212

grasso

1.2 Argument Alternations Cross-linguistically As noted above, the causative-inchoative alternation says nothing about the marking of the variants. In principle, there are at least three ways that the variants may be marked: the causative variant may be morphologically marked and the inchoative unmarked, the inchoative may be marked and the causative unmarked, or both may be unmarked (see (Haspelmath 1993:91–92) for the basic divisions). One additional possibility not ordinarily mentioned is to have both variants morphologically marked with causative morphology, which I call the Double Alternation below. These possibilities are shown in the following table (I adopt the terminology from (Haspelmath 1993)):

Inchoative Causative Causative Alternation Anticausative Alternation Non-directed Alternation Double Alternation

+ +

+ +

Importantly, languages may adopt more than one strategy to encode these alternations, so if a language has dedicated causative morphology (like Biblical Hebrew, for example), it need not be the case that causative morphology is the only way the causative variant is expressed (Italian, for example, uses at least the first three strategies (Alexiadou, Anagnostopoulou and Schäfer 2015)). This paper deals primarily with the Double Alternation where both the inchoative and causative variants are morphologically marked with a causative morpheme, i.e., the Hiphil binyan. Below, I show that all the alternations are possible before discussing the Double Alternation in detail. 1.3 Alternations in Biblical Hebrew Biblical Hebrew has a great deal of flexibility in marking causation, and it shows all of the variants described above, as shown in the following examples: 1.3.1

Causative Alternation

(2) və=ʾanu had=dayyaḡ-im və=ʾaḇəlu kol and=lament.irr-pfv the=fisherman-mp and=mourn.irr-prv all

- 978-90-04-44885-8

the causative-inchoative alternation

213

mašliḵe ḇa=yəʾor ḥakkah6 throwers-mp.of in.the=nile hook ‘The fishermen will mourn and lament, all who cast a hook in the Nile’ (Isa 19:8) (3) ko ʾamar ʾăḏonay yhwh bə=yom ridt=o šəʾol=a thus say.perv lord yhwh on=day.of descend.inf=3ms sheol=to hε=ʾɛ̆ḇalti caus=mourning.perv ‘Thus says the Lord God: On the day the cedar went down to Sheol I caused mourning’ (Ezek 31:15) 1.3.2

Anticausative Alternation

(4) və=ḵol kəli ḥereś ʾăšεr yippol me=hεm ʾεl toḵ=o kol and=all vessel.ms clay rel fall.irr from=3mp to inside=3ms all ʾăšεr bə=toḵ=o yiṭmaʾ və=ʾoṯo ṯišboru rel in=inside=3ms unclean.irr and=om=3ms break.irr ‘And any clay instrument which falls from them into it, anything in it will be unclean. You must break it.’ (Lev 11:33) (5) u=ḵəli ḥεrεś ʾăšεr təḇuššal b=o yiššaḇer and=vessel.ms clay rel cook.irr in=3ms break.irr.medp ‘And the clay instrument which you cooked with must be broken’ (Lev 6:21)

6 I have labelled the semantics of the verbal forms in bh according to the following scheme: qatal = Perfect/Perfective (perv), yiqtol = Irrealis (irr), weqatal = Irrealis-Perfective (irrpfv), wayyiqtol = Past-Perfective (past-pfv), and qotel = Progressive (prog). Justification for this system can be found in (Grasso 2016). perv is equivalent to the same operator found in (Condoravdi and Deo 2014). The most controversial of these forms is yiqtol, which is ordinarily labelled imperfective. See (Cook 2012) for the most recent survey of theories. The transcriptions only mark number and person on the nouns, and agreement with the verbs. I also mark Hiphil as caus for causative and Niphal as medp for Medio-passive. I only mark caus and medp on the relevant verbs. The relevant verbs in the examples are bolded.

- 978-90-04-44885-8

214 1.3.3

grasso

Non-directed Alternation

(6) və=niggəša yəḇimt-o ʾel=av lə=ʿεnε and=approach.irr-pfv brother.wife=3ms to=3ms to=eyes.of haz=zəqen-im və=ḥaləṣa naʿal=o me=ʿal raḡl=o the=elder-mp and=remove.irr-pfv sandal=3ms from=on foot=3ms ‘Then his brother’s wife shall go up to him in the presence of the elders and remove his sandal from his foot’ (Deut 25:9) (7) bə=ṣoʾn-am u=ḇi=ḇqar=am yeləḵu lə=ḇaqqeš ʾεṯ yhwh with=flocks-3mp and=with=cattle=3mp go.irr to=seek.inf om yhwh və=loʾ yimṣaʾu ḥalaṣ me=hεm and=not find.irr withdraw.perv from-3mp ‘With their flocks and herds they shall go to seek the Lord, but they will not find him; he has withdrawn from them.’ (Hos 5:6) 1.3.4

Double Alternation

(8) vay=yanniḥu ʾoṯ=o ʿaḏ hab=boqεr kaʾăšεr and=lay.aside.past-pfv om=3ms until the=morning as=rel ṣivva mošε və=loʾ hi-ḇʾiš command.perv Moses and=not caus=stink.perv ‘So they laid it aside till the morning, as Moses commanded, and it did not stink’ (Exod 16:24) (9) vay=yoʾməru ʾăle=hεm yerεʾ yhwh ʿăle=ḵεm və=yišpoṭ and=say.past-pfv to=3mp see.juss yhwh on=2mp and=judge.juss ʾăšεr hi-ḇʾaštεm ʾεṯ reḥe-nu ḇə=ʿene p̄ arʿoh rel caus-stink.perv om smell-1cp in=eyes.of Pharoah ‘and they said to them, “The Lord look on you and judge, because you have made us stink in the sight of Pharaoh” ’ (Exod 5:21) The first pair shows an unambiguous example of the causative alternation, since the verb ʾaḇal is causativized with the Hiphil binyan (and the English translation also requires the causative make do x construction). The verb in Hiphil requires an external causer that is not present in the Qal variant. The second pair is a classic case of the anticausative alternation. The verb šaḇar requires an external argument and specifies a result, while nišbar reduces the valency of the verb, making it require only the thing broken. The non-directed alternation is quite rare, and it seems to be only found with roots that have a

- 978-90-04-44885-8

the causative-inchoative alternation

215

locative argument.7 In the first example, ḥalaṣ has three arguments: a cause (someone removing), a theme (the thing removed), and a location. In contrast, the second example collapses the first two arguments into one, so that there is only a theme (the entity removed by itself) and a location. Finally, the last alternation will be the primary focus of this chapter. In the first example given, there is a state of smelling bad that the subject (theme) does not enter into, whereas in the second example, an external cause (Moses and Aaron) bring the people into a state of smelling bad to Pharaoh.

2

The Causative-Inchoative Alternation in Hiphil

2.1 Roots That Alternate As shown above in examples (8) and (9), Hiphil can participate in the causativeinchoative alternation without changing binyanim. However, this is not productive, i.e., most roots in Hiphil cannot participate in this alternation. Several studies have been conducted on Modern Hebrew to see which roots actually alternate (see (Kastner 2016) and the references there), but the types of roots that may alternate seems to be broader in bh. The alternation itself has already been noted in several grammars of bh (Waltke and O’Connor 1990; Joüon and Muraoka 2006), but it has not been treated extensively.8 The alternating roots can be seen in the following chart, which is, I think, exhaustive (the * marks verbs that allow for passive voice):9

7 Other examples include raʿah (Exod 34:3; Gen 30:31), paraṣ (Gen 28:14; Isa 5:5), šaraṣ (Gen 9:7; Exod 7:28), naṭa (Num 21:15; Exod 8:12), and šaḇ (Nah 2:3; Deut 30:3). All of the intransitive variants are listed first. This generalization also does not necessarily mean that all verbs in Paʾal that have a locative argument can alternate without changing binyanim. The list given is not exhaustive, and I am not sure why the alternation occurs with these kinds of verbs in the first place. 8 Waltke and O’Connor (1990) suggest that the Hiphil is semantically reflexive when it alternates. The reflexive analysis of the alternation was first proposed by (Chierchia 2004), and there has been extensive discussion on the analysis since (e.g. see the exchange in (Horvath and Siloni 2011; Koontz-Garboden and Beavers 2013)). Whether or not the reflexive analysis can apply to some verbs, it seems untenable as an analysis for all inchoative variants (see the discussion in [Lundquist et al. 2016]). 9 The following are examples for each of the roots; the verbs with one less argument are listed first: qrb (Exod 14:10; Num 3:6), nṭh* (Amos 2:8; 2 Sam 16:22), pwṣ (Exod 5:12; Gen 11:8), ngʿ (Gen 28:12; Isa 25:12), ʾwr (Ezek 43:2; Ps 18:29), ʿlm (Ps 10:1; Job 42:3), ṣmḥ (Deut 29:22; Isa 55:10), bʾš (Exod 16:24; Exod 5:21), prḥ (Job 14:9; Ezek 17:24), ṣwṣ (Ps 90:6; Num 17:23), ʿrk (Exod 20:12; 1Kgs 3:14), ybš (Joel 1:10; Josh 2:10), lbn (Isa 1:18; Dan 11:35), rḥb (Ps 25:17; Exod 34:24), ḥzq (Dan 11:32;

- 978-90-04-44885-8

216

grasso

Class of verbs

Roots

Verbs of Location Verbs of Emission and (dis-)Appearance Verbs from Degree Adjectives Verbs of Directed Movement Verbs of Exchange Verbs of Mental State/Emotion Verbs of Change of State (Misc.)

qrb, nṭh*, pwṣ, ngʿ ʾwr, ʿlm, ṣmḥ, bʾš, prḥ, ṣwṣ ʿrk, ybš, lbn, rḥb, ḥzq, rbh, šmn, gdl, ʿšr šwb*, nws mwr, ḥlp, yrš, nzr bwš, klm*, zyd, ḥrd, ṣrr, byn, ḥrš, ḥšh, ʿwr, śkl znh, ṣlḥ, ʿwz, tʿh, ʿwd*

As can be seen, there are a variety of different types of roots that can alternate, which makes a unified analysis difficult. I will divide the alternating roots into two broad classes based on whether they can passivize or not.10 Of course, in bh we do not have negative evidence, so it is not the case that all the roots above without an asterisk cannot passivize, yet given the many roots with no attested examples, it seems plausible that some cannot (particularly those that have been called internally caused change of state verbs; see Rappaport Hovav and Levin 1995). One class consists of verbs that only allow a causer argument and do not passivize like heṣiṣ ‘to blossom’ (which I will call blossom-verbs), and the other class consists of verbs that allow both a causer and agent argument and do passivize like hešiḇ ‘to return’ (which I will call return-verbs). Examples are found below in (10) and (11) with the inchoative examples listed first: (10) a. me=reḥa mayyim yap̄riḥa və=ʿaśa from=scent.of water flower.caus.irr and=produce.irr-pfv

10

2Sam 11:25), rbh (1 Chr 4:27; Deut 1:10), šmn (Neh 9:25; Isa 6:10), gdl (1Sam 20:41; Gen 19:19), ʿšr (Ps 49:17; Gen 14:23), šwb* (Jonah 1:13; Gen 29:3), nws (Judg 6:11; Exod 9:20), mwr (Ps 46:3; Jer 2:11), ḥlp (Job 29:20; Gen 31:41), yrš (Josh 17:18; Exod 34:24), nzr (Num 6:3; Lev 15:31), bwš (Jer 2:26; 2Sam 19:6), klm* (Jer 6:15; 1 Sam 25:7), zyd (Exod 21:14; Gen 25:29), ḥrd (Judg 8:12; 2Sam 17:2), ṣrr (Jer 48:41; Jer 10:18), byn (Mic 4:12; Isa 28:9), ḥrš (Exod 14:14; Job 11:3), ḥšh (2Kgs 2:3; Neh 8:11), ʿwr (Ps 57:8; Isa 13:17), śkl (Ps 2:10; Ps 32:8), znh (Hos 4:18; Lev 19:29), ṣlḥ (1Kgs 22:12; Gen 39:23), ʿwz (Isa 10:31; Exod 9:19), tʿh (Prov 10:17; 2Chr 33:9), ʿwd* (Mal 2:14; Isa 8:2). This distinction is inspired by (Alexiadou 2014) and her contrast between blossom-verbs and ferment-verbs, but the differences between the analyses are that her class of fermentverbs requires non-active Voice to alternate and what I call return-verbs do not.

- 978-90-04-44885-8

the causative-inchoative alternation

217

qaṣir kəmo naṭaʿ branch like young.plant ‘From the scent of water it [tree] will flower, and it will produce a branch like a young plant’ (Job 14:9) b. ho-ḇaš-ti ʿeṣ laḥ və=hi-praḥ-ti caus-dry up-1cs.perv tree moist and=caus-flower-1cs.irr-pfv ʿeṣ yaḇeš ʾăni yhwh tree dry I yhwh ‘I dry up the moist tree, and I make the dry tree flower. I am Adonai.’ (Ezek 17:24)11 (11) a. vay=yaḥtəru ha=ʾănaš-im lə=ha-šiḇ ʾεl and=row.past-pfv the=person-mp to=caus-return.inf to hay=yabbaša və=loʾ yaḵolu the=shore and=not able.perv ‘Nevertheless, the men rowed hard to return to dry land, but they could not, for the sea grew more and more tempestuous against them.’ (Jon 1:13)12 b. və=nεʿεsəp̄ u šamm=a ḵol ha=ʿăḏar-im və=galălu and=gather.irr-pfv there=to all the=flock-mp and=roll.irr-pfv ʿεṯ ha=ʾεḇεn meʿal pi hab=bəʾer və=hišqu ʾεṯ om the=rock from mouth.of the=well and=water.irr-pfv om haṣ=ṣoʾn və=he-šiḇu ʾεṯ ha=ʾεḇεn ʿal pi hab=bəʾer the=flock and=caus-return om the=rock on mouth.of the=well li=mqom=ah to=place=3fs ‘and all the flocks would gather there, and the shepherds would roll the stone from the mouth of the well and water the sheep, and return the stone to its place over the mouth of the well.’ (Gen 29:3)

11

12

I consider God here to be a kind of causer, more specifically a director. Isa 17:11 is the only other transitive example in the Bible, and it seems to be ironic and something that is impossible. Other examples outside the Bible include Sir 40:19 (with nεṭaʿ) and 1qh 16:6 (with ʿeṣ). This example could alternatively mean ‘return the boat’ with ‘the boat’ supplied by context. Other potential examples for this verb include Ps 85:4 and Ezek 18:32.

- 978-90-04-44885-8

218

grasso

Besides verbs that alternate like those in the list above, there are also some verbs that are never found with an external cause in the Bible, such as hiškim ‘to rise early’ in the example below. (12) vay=yaškimu ʾašdoḏim mim=moḥōraṯ and=rise early.past-pfv Ashdodites from=next.day ‘And when the people of Ashdod rose early the next day.’ (1 Sam 5:3) Thus, there is a three-way contrast between intransitive verbs in Hiphil. Verbs that undergo the Double Alternation may passivize, others cannot passivize, and some intransitive Hiphil verbs never allow for a causer or agentive subject to be added (which I will call rise-early-verbs). These three classes have different syntactic structures, as shown below in 2.2.2. 2.2 Analysis 2.2.1 Meaning of Cause/Result Before beginning the discussion on how the notion of causation interacts with roots in the syntax, we must first establish what is meant by causation. Many definitions have been offered, but I will follow Kratzer (2005) who states the following about causation: The intransitive vp in the sentence The sauce thickened, for example, would describe events of causing the sauce to be thick(er), where the prominent cause for that event can be (but doesn’t have to be) linked to properties inherent in the sauce itself. A commitment to a prominent external cause like an agent or a force of nature would be contributed by [active] voice. Chierchia’s and Levin & Rappaport’s proposals have the interesting consequence that the commonly posited become operator becomes superfluous in the decomposition of inchoatives, causatives, and anticausatives. Those three types of verbs are all causatives. They differ with respect to voice. There are two crucial points that Kratzer makes. First, she says that inchoatives, causatives, and anticausatives can all be subsumed under the notion of causation semantically. Her definition includes all these categories because Cause is defined as those events that entail a change/result in something (e.g. sauce becoming thicker). This definition of causation, then, is actually closer to the idea of Result, or scalar change, rather than Cause (at least, as the word cause is typically used in English, which normally presupposes a causer), since the change itself can be identified as the cause. I suggest that Hiphil is causative in

- 978-90-04-44885-8

the causative-inchoative alternation

219

this sense: Hiphil adds a result to the semantics of the root. I define Result as change along some scale (see (Rappaport Hovav 2014a) for discussion). Thus, I propose calling Hiphil the Result binyan rather than causative, which means that it may or may not include an external causer, and there is actually nothing in the semantics of the morpheme that would disallow inchoative interpretations.13 In fact, such interpretations are predicted for all roots that do not specify some property of the external argument (such as a verb like murder which requires intentionality) or require it for any other reason (such as a verb like destroy in English which doesn’t seem to require a property of an external argument, but nevertheless requires one to be present).14 Second, the key parameter in the alternations described above for Hiphil is not causation, but the type of voice. Thus, my analysis of the causativeinchoative alternation in Hiphil involves variations in how Voice combines with the verb, which implies that the head that introduces the external argument (Voice) is separate from the head that introduces the causing event (Pylkkänen 2008; Harley 2013). In addition, it suggests that Hiphil itself is not marked directly for Active Voice because not all verbs in Hiphil have an external argument. Although Active Voice may be the default, I assume this does not come from Hiphil itself, and other Voice options are available that do not introduce an external argument (this is similar to break in English, which may or may not realize an external argument, i.e., it has different Voice options that are not morphologically marked). A little more must be said about the Result semantics of Hiphil. In the standard grammars on bh, the Hiphil binyan is ordinarily characterized as causative, but little explanation is given to what causation actually means other than being the transitive version of verbs in Qal (see Gesenius § 53; Joüon and Muraoka 2006: §54; Waltke and O’Connor 1990: § 27). Most grammars then go 13

14

Of course, the way Kratzer defines causation also does not necessarily include the idea of an external causer, but this may be a confusing way to talk about causation. We normally think of causative events as requiring the cause to be named, so to avoid confusion, I am calling Hiphil result, which, of course, includes the inference that there was some cause. However, we often speak of results happening without any mention of the specific cause. In many dictionaries, the Hiphil binyan is glossed as something like ‘cause to …’ or ‘make x do …’. These glosses only work for those verbs in Hiphil that require an external argument because that particular causative construction in English does require an external argument. Hiphil actually overlaps with both the English ‘make do’ construction and the -en morpheme (as in verbs like lighten, fatten, shorten, lengthen, etc.) as well as -ify (as in verbs like justify, magnify, horrify, intensify, etc.). As will be shown below, these latter verbs in English (including the so-called degree achievements) represent ideas that are also expressed in Hiphil in Hebrew, and some alternate just like their English counterparts.

- 978-90-04-44885-8

220

grasso

on to note all the intransitive examples of Hiphil without a serious attempt to subsume these verbs under their notion of causation. This has led to some problematic characterizations of the Hiphil binyan, such as “to actively cause an event,” which Waltke and O’Connor (§27) suggest for the semantics of Hiphil following (Jenni 1968). It is simply not true that Hiphil always means to actively cause an event. For example, the verb heṣiṣ ‘to blossom’ does not mean there was some prior activity that produced the blossoming, nor is the result an event, since blossoming events result in the state of being in bloom. Simply put, this kind of definition cannot capture the many verbs like this found in Hiphil. There are two significant points to be made about how the semantics of Hiphil interacts with the root it combines with. First, the result component may be all kinds of different scalar changes (some of which are two point scales). Let us take several different kinds of verbs, such as hišliḵ ‘to throw’, hε̆εmin ‘to believe’, heṣiṣ ‘to blossom’, hilbin ‘to whiten’, and heriṣ ‘to make run’. All of these verbs involve a scalar change of some kind.15 hišliḵ ‘to throw’ involves a change in location of the thing thrown, hε̆εmin ‘to believe’ involves a change in attitude towards an object, heṣiṣ ‘to blossom’ involves a change in the state of the theme, hilbin ‘to whiten’ involves a change on the scale of color, and heriṣ ‘to make run’ involves a change in activity from not running to running. It should be noted that it is not necessarily the case that the changing part of the event is actually referred to every time Hiphil is used. Verbs involving a change in attitude, for example, denote the initial boundary of a state as well as the state itself,16 so either may be referred to in a particular context. So, a verb like heḇin ‘to understand’ may refer to the event of coming into the state of understanding, or it may refer to the state of understanding and presuppose that the change happened at some prior time to the state. Which part of the event is referred to depends on grammatical aspect and context.17

15

16 17

Another way to characterize Hiphil would be to say that it contributes a boundary happening in the sense of (Piñón 1997). I think such an analysis may actually better characterize Hiphil, but recharacterizing the binyan aspectually like this would take us too far afield. See (Marín and McNally 2011) for a discussion on similar verbs that denote both the initial boundary and the state. To see how this works, consider a verb like understand in English. If I say, John understood the question, it can have a stative or inchoative interpretation, that is, it can mean either John came into a state of understanding the question or the state of him understanding held at some prior time. In the simple present, John understands the question must be referring to a state that holds at reference time. Because the simple past in English is often interpreted perfectively, we can get the inchoative reading, but because the simple present ordinarily cannot be interpreted perfectively, we get the stative reading. In neither

- 978-90-04-44885-8

the causative-inchoative alternation

221

vHP vH

√P √ṣyṣ

theme

figure 11.1 blossom-verb Intransitive Structure

vHP causer

v H' vH

√P √ṣyṣ

theme

figure 11.2 blossom-verb Transitive Structure

Second, the kinds of verbs that do fit the description of “actively causing an event” are precisely those that never alternate in the Double Alternation. More accurately, verbs that name manners alternate only in the Causative Alternation and cannot be inchoative. The verb raṣ ‘to run’ denotes the manner of movement, so the Hiphil version cannot be inchoative and requires an external argument. This means that heriṣ ‘to make run’ cannot mean ‘to start running’. Why this is actually the case I leave open to further research. This brings us to the syntactic differences between the different verbs found in Hiphil. 2.2.2 Syntax of Alternating Verbs As discussed in the previous section, Result semantics does not entail the realization of a causer argument, as already pointed out by Pylkkänen (2008).18 In other words, the presence or absence of an agent or a causer in the syntax is not directly determined by the presence or absence of causative semantics (in the Kratzer sense, i.e., what I have called Result above). Hiphil always involves scalar change, but it does not always have a causer, contra (Doron 2003; Kastner 2019). Following (Kratzer 1996), I assume that Voice introduces the external argument of the verb, at least in the case of agents. According to (Alexiadou 2014), causers may be introduced in Spec,vP or in Spec,VoiceP, but verbs that disallow agents lack Voice and can only have Causers in Spec,vP. I also adopt this analysis. Those verbs that lack Voice do not have a passive, and they should

18

case has the semantics of understand changed in any way—all that has changed is what part of the event we are referring to. Pylkkänen uses “Cause” instead of Result, but she defined it just as Kratzer did above, which is equivalent to what I am calling Result.

- 978-90-04-44885-8

222

grasso

alternate only with causers. These are the blossom-verbs. Examples of the alternation were given in (10) above. Such verbs have syntactic representations like the ones found below, depending on whether the causer is present or absent (vH stands for Hiphil; I am ignoring everything above the vP that is irrelevant for the analysis). The presence or absence of the causer argument depends on a number of factors (see (Rappaport Hovav 2014b) for some relevant considerations). In addition to the alternating verbs in the blossom-class, there are also the nonalternating verbs in the rise-early-class. This type of unaccusative shows distinct syntactic differences with the blossom-class, most notably in our case the inability to combine with a causer. But why should this be the case? I suggest that these are a different class of unaccusative verbs, building on the analysis in Alexiadou and Schäfer 2011 (though my analysis differs from theirs in important ways—see their article for details). As opposed to the blossom-class where the root takes the theme as its argument and then combines with vH, roots in the rise-early-class do not take an argument directly, but the root combines with vH, and then the theme is introduced in Spec,vP. This has two consequences. First, because the theme is now occupying the place where the causer argument in blossom-verbs is introduced, these verbs cannot alternate. Second, this may account for why these verbs are also causer/agentive-like in some respects. A verb like hiškim ‘to rise early’ requires a sentient subject and requires at least some action, namely actually getting up. However, it seems that the subject is still not a full-blown agent. Something must still happen to the subject that is out of his or her control, namely actually waking up. Thus, I will call this theme an Active Theme. The structure for the rise-early-class of verbs is shown below in Figure 11.3. This leads us to our last class of verbs, the return-class, as shown in (11) above. There are only a handful of these verbs in bh. They differ from the two classes above by both alternating (unlike the rise-early-class) and allowing passivization (unlike both the rise-early-class and the blossom-class). Because they can passivize and allow agents, Voice must be present, and because they can participate in the Double Alternation, we need a place both for the agent/causer and the theme. This structure is found in Figure 11.4. While Figure 11.4 is the transitive structure for the return-class, I consider the intransitive structure to be equivalent to the rise-early-class shown in Figure 11.3. There are two important differences between the transitive structure below and that found for the blossom-class in Figure 11.2. First, the presence of Voice in Figure 11.4 allows for both agents and causers rather than just causers, as has been pointed out. Second is the position of the theme. I suggested above that the theme for the rise-early-class was more agentive than that of the blossom-

- 978-90-04-44885-8

223

the causative-inchoative alternation

vHP theme

v H' vH

√šḵm

figure 11.3 Non-alternating rise-early-class Structure

VoiceP agent/causer

Voice' Voice

v HP theme

v H' vH

√švḇ

figure 11.4 Transitive Structure for return-class

class, presumably because of its position in the structure (since it occupies the same position as the causer in the transitive variant of blossom-verbs). This leads to the prediction that the intransitive variant of the return-class should be semantically closer to rise-early-verbs than to blossom-verbs, and this seems to be the case. A verb like hešiḇ ‘to return’ in the intransitive reading normally involves some action on the part of the subject, just like the rise-early-class. Thus, the Double Alternation can come from one of two sources: the blossomclass where the theme combines directly with the root and allows a causer, or the return-class where the verb allows for Voice to introduce an external argument on top of what we have called an Active Theme.

3

The Anticausative Alternation in Qal-Niphal

The purpose of this section is not to provide a comprehensive analysis of the anticausative alternation in Qal-Niphal, but to simply point out some differences between this alternation and the Double alternation described above in Hiphil. I show that there are structural differences between the alternations that help to account for the differences in which roots participate in the respective alternations. 3.1 Some Roots That Alternate The causative-inchoative alternation in Qal-Niphal is different from the Double alternation described above in several respects. First, it is anticausative.

- 978-90-04-44885-8

224

grasso

In other words, the inchoative variant in the alternation is morphologically marked, i.e., by the nun-prefix. Second and following from this, there is no overt causative morphology. However, I argue below that there is a result/causative element in this alternation, but it is present in the root of the Qal variant. Haspelmath (1993) surveys a variety of languages to see how they encode the causative-inchoative alternation, and he notes that there are definite patterns with regard to which variant in the alternation is morphologically marked. He uses the results to suggest that there is a scale of “spontaneity” of causation that languages tend to mark morphologically: the less spontaneous the causing event is conceived to be, the more likely it will have an anticausative variant (a more precise definition of spontaneity is given below). The reverse is also the case—the more spontaneous the causing event is conceived to be, the more likely it will have a causative variant. Below, I show that the “less spontaneous” causatives in Biblical Hebrew conform to the pattern noticed by Haspelmath: causatives in Qal can generally be described as “less spontaneous”, and the anticausative variant is marked with Niphal. In Haspelmath’s survey, he lists ‘split’, ‘open’, ‘gather’, ‘break’ and ‘close’ as five of the most common verbs found in the anticausative alternation, and all of these participate in the Qal-Niphal alternation in Biblical Hebrew (the roots are the following: bqʿ, ptḥ, ʾsp, šbr, and sgr; an example for šbr is given in (4) and (5) above). This suggests that bh uses the Qal-Niphal alternation for these “less spontaneous” causing events. 3.2 Analysis 3.2.1 Result Roots Above, I mentioned that the result element of verbs in Qal comes from the root itself rather than being syntactically inserted. In the literature on the semantics of verbal roots, the manner/result complementarity hypothesis has played a prominent role, and it posits that some roots fall into the result category (see (Rappaport Hovav and Levin 2010) and many other works by the same authors). In essence, I assume that such roots involve some sort of change along a scale (Rappaport Hovav 2014a). These roots are also those that are considered by Haspelmath to be “less spontaneous.” The reasoning for calling these roots result, then, goes as follows. A root like šbr ‘√break’ names a not very spontaneous state. In other words, things do not generally break with no prompting or cause, i.e., spontaneously. If this is the case, states of being broken are usually results of some other event. This is the sense in which we can talk about some roots naming resultant states. Assuming that Qal contributes nothing to the semantics of the root other than turning it into a verb, verbs in Qal with an interpretation that involves a change along a scale (such as šaḇar ‘to break’) would have to receive their

- 978-90-04-44885-8

225

the causative-inchoative alternation

causative interpretation from the root itself. This also follows from the simple fact that many other states in Qal do not involve a result, so the result interpretation cannot be coming from the binyan (e.g. yaḏaʿ ‘to know’). The main difference between result verbs in Hiphil and result verbs in Qal, then, is the following: the former adds a result to a root, while the latter has a result element embedded in the root itself. Although the Qal binyan is not normally thought of as causative, there are two primary pieces of linguistic evidence that suggest that causative verbs (formed from these result roots) can indeed be found in Qal. First, following (Cinque 1999), Alexiadou et. al. (2015) show that causers can only be added in adjuncts to verbs that already have a causative semantics. If Niphal occupies the position of Voice in the syntax, it should not change the semantics of the verb, but it should only delete the external argument. This means that the corresponding anticausative verb in Niphal should still have the causative component that was in the Qal verb if present, since the causative component came from the root and Niphal doesn’t affect the semantics of the root. In other words, there should be some examples where a causer is added back into the syntax in the Niphal binyan if the causative semantics comes from the root, and this is exactly what we find, as shown in the example below: (13) va=hăqim-oṯi ʾεṯ bəriṯ=i ʾitə=ḵεm və=loʾ and=establish.irr-pfv-1cs om covenant=1cs with=2mp and=not yikkareṯ kol baśar ʿod mim=me ham=mabbul cut.off.irr.medp all flesh again from=water.of the=flood ‘And I will establish my covenant with you: never again will all flesh be cut off by the waters of the flood’ (Gen 9:11) In Gen 9:11, the causer, ‘the waters of the flood’, is introduced by the preposition min ‘from’ (I assume that causers are introduced by min ‘from’ and instruments by bə ‘in/by’ in bh). It is licensed by the result semantics present in the root krt ‘to cut off’. This shows that the root licenses causers, and assuming the above-mentioned authors to be correct, this can only be due to a verb having a causative semantics. The other piece of evidence that suggests that verbs in Qal may be causative is the fact that non-agent causers can be used as the external argument. This is shown in the below example: (14) ruaḥ haq=qaḏim šəḇar=eḵ bə=leḇ yamm-im wind the=east break.perv=2fs in=heart.of sea-mp ‘The east wind has broken you in the heart of the seas.’ (Ezek 27:26)

- 978-90-04-44885-8

226

grasso

VoiceP agent/causer

Voice' Voice

vQP vQ

√P √šbr

theme

figure 11.5 Qal Result Verb Structure

VoiceNP VoiceN

vQP vQ

√P √šbr

theme

figure 11.6 Niphal Anticausative Structure

The fact that a non-agentive, causative external argument can be added to a verb in Qal suggests that such verbs may indeed be causative.19 3.2.2 Syntax of Alternating Verbs In this section, I give a syntactic representation of the Qal-Niphal alternation and compare it to the structures given above for the Hiphil alternation. To begin with, the alternation between Qal and Niphal is a Voice alternation (Doron 2003), and the latter is, specifically, Medio-passive Voice. There are a variety of interpretations that a medio-passive can have (Doron and Alexiadou 2012), but they can often get an inchoative interpretation, which is the relevant interpretation for the causative-inchoative alternation, as shown in the structures above in Figure 11.5 and Figure 11.6. As shown in the structures, I assume that, at least in this alternation, the Niphal verb is derived from the verb in Qal. The only difference between the two structures is the presence of an external argument, which is introduced by the active Voice head in Qal and is deleted by the Middle Voice, Niphal. Both structures have the same semantics, namely causative (or resultative) semantics, as discussed in the previous section. 19

This also suggests that these verbs in Qal are bi-eventive, since causers require a bieventive event structure (Schäfer 2012). This means that the actual event structure of Qal and Hiphil may be the same: they can both be bi-eventive.

- 978-90-04-44885-8

227

the causative-inchoative alternation

4

The Anticausative Alternation vs. the Double Alternation

So what are the similarities and differences between the Double Alternation and the anticausative alternation described above? First, the most obvious difference between the two is Voice. blossom-verbs in the Double Alternation do not have Voice at all but they still alternate, so for this sub-class of the Double Alternation, the causative-inchoative alternation is not a Voice alternation. The return-class, on the other hand, does involve a Voice alternation, but it is the presence vs. absence of Voice that determines whether the external argument is present. In contrast to both, the anticausative alternation has a different type of Voice for Qal (active) and Niphal (medio-passive). Second, the alternations differ with regard to the semantics of the roots. Roots in the blossom-class do not name states that are results of events; the result interpretation of the state is contributed by the Hiphil binyan. While roots in the return-class do name states that involve scalar change, the theme is more actively involved in the event than in the blossom-class. Finally, roots in the anticausative alternation name states that are less spontaneous and require a prior event on their own, and the theme is not agent-like in any way. These contrasts are shown in the following table:

Blossom-class

Return-class

Anticausative alternation

Voice Roots

No Voice States

+/- Voice Result States (Active Theme)

Active/Medio-Passive Voice Result States (Non-Active Theme)

5

Conclusion and Implications

What conclusions have we come to in this paper? Here is a list: 1. Biblical Hebrew has many different ways of marking argument alternations. 2. Verbs in the Hiphil binyan may participate in the Double Alternation if they do not require an external argument or name a manner event, which means that verbs in Hiphil in general do not require an external causer, against some of the descriptions of Hiphil’s semantics found in the grammars. 3. Hiphil is better characterized as contributing a Result rather than a Cause (or a Kratzer-syle Cause). - 978-90-04-44885-8

228

grasso

4.

There is semantic overlap between verbs involving a Result in Qal and Hiphil, but they differ in regard to the source of the Result, either the root or the binyan. 5. Contrary to some studies, the Hiphil binyan does not need to cause an event. It may also cause states. There are several practical implications of the above discussion. First, there is a tendency for dictionaries to gloss verbs in Hiphil as “make/cause to x”, but we have seen that this English structure only corresponds to some of the verbs in Hiphil because the “cause to x” structure in English requires a causer. Many verbs in Hiphil are inchoative, suggesting Result is a better semantic description (or Cause understood in the Kratzer sense). There is thus no need to try to fit Hiphil into the “cause to x” template. Second, we have seen that Qal and Hiphil overlap at times, particularly when a verb in Hiphil is in the Double Alternation. And in fact, this can be seen in some Kethiv-Qere readings. For example, Judges 7:21 has the verb vynysv in Kethiv and vayyanusu in Qere. As mentioned above henis ‘to flee/make flee’ is an alternating verb (with a clear example in Judges 6:11), so this could be support that the original Kethiv reading is correct, or at least possible.20 Third and finally, the semantic similarity between Qal and Hiphil could have been a contributing factor in a semantic change from one form to another. For example, the root bin ‘to understand’ begins to be used more frequently in later texts with an explicit cause, meaning something like ‘to teach’. This could have originally resulted from the Qal stative verb being aspectually coerced into an inchoative, and as an inchoative, it would behave like a typical verb in Hiphil, i.e., as including a result. The external argument was then added to the inchoative Qal verb through the Hiphil binyan.

References Alexiadou, Artemis. 014. “The problem with internally caused change-of-state verbs.” Linguistics 52 (4): 879–909. Alexiadou, Artemis, Elena Anagnostopoulou, and Florian Schäfer. 2006. The Properties of Anticausatives Crosslinguistically. Pp. 187–212 in Phases of Interpretation, ed. Mara Frascarelli. Berlin: Mouton, 2006.

20

Other examples with Kethiv-Qere and this alternation are Jer 48:44 (nvs is the root again) and Ps 59:16 (nvʿ is the root). There are also several examples with the root švb (2Sam 15:8; Jer 33:26; 49:36; Joel 4:1; Ps 54:7; 73:10; Job 39:12; Prov 12:14).

- 978-90-04-44885-8

the causative-inchoative alternation

229

Alexiadou, Artemis, Elena Anagnostopoulou, and Florian Schäfer. 2015. External Arguments in Transitivity Alternations: A Layering Approach. Oxford Studies in Theoretical Linguistics 55. Oxford, UK; New York, NY: Oxford University Press. Alexiadou, Artemis, and Florian Schäfer. 2011. An unaccusativity diagnostic at the Syntax-Semantics interface: there-insertion, indefinites and restitutive again. Pp. 101–116 in Proceedings of Sinn & Bedeutung 15, ed. Ingo Reich, Eva Horch, and Dennis Pauly. Saarbrücken: Saarland University Press. Chierchia, Gennaro. 2004. A Semantics for Unaccusatives and Its Syntactic Consequences. Pp. 22–59 in The Unaccusativity Puzzle: Explorations of the Syntax-Lexicon Interface, ed. Artemis Alexiadou, Elena Anagnostopoulou, and Martin Everaert. Oxford, UK; New York, NY: Oxford University Press. Cinque, Guglielmo. 1999. Adverbs and Functional Heads: A Cross-Linguistic Perspective. Oxford Studies in Comparative Syntax. New York: Oxford University Press. Condoravdi, Cleo, and Ashwini Deo. 2015. Aspect shifts in Indo-Aryan and trajectories of semantic change. Pp. 261–292 in Language change at the syntax-semantics interface, ed. Chiara Gianollo, Agnes Jäger and Doris Penka. Berlin: De Gruyter Mouton. Cook, John A. 2012. Time and the Biblical Hebrew verb: the expression of tense, aspect, and modality in Biblical Hebrew. Linguistic Studies in Ancient West Semitic 7. Winona Lake, IN: Eisenbrauns. Doron, Edit. 2003. Agency and Voice: The Semantics of the Semitic Templates. Natural Language Semantics 11: 1–67. Doron, Edit, and Artemis Alexiadou. 2012. The Syntactic Construction of Two NonActive Voices: Passive and Middle. Journal of Linguistics 48 (1): 1–34. Folli, Raffaella, and Heidi Harley. 2005. Flavors of v. Pp. 95–120 in Aspectual inquiries. Dordrecht: Springer. Grasso, Kevin. 2016. A Semantic Analysis of the Biblical Hebrew Verbal System in Prophetic Literature. Unpublished ma Thesis, Graduate Institute of Applied Linguistics. Harley, Heidi. 2013. External Arguments and the Mirror Principle: On the Distinctness of Voice and V. Lingua 125: 34–57. Haspelmath, Martin. 1993. More on the Typology of Inchoative/Causative Verb Alternations. Pp. 87–120 in Causatives and Transitives, ed. Bernard Comrie and Maria Polinsky. Amsterdam: Benjamins. Horvath, Julia, and Tal Siloni. 2011. Anticausatives: Against Reflexivization. Lingua 121: 2176–2186. Joüon, Paul, and Takamitsu Muraoka. 2006. A Grammar of Biblical Hebrew. Rev. ed. Subsidia Biblica 27. Roma: Pontificio istituto biblico. Kastner, Itamar. 2016. The Zero-Derived Causative Alternation in Hebrew is Rare, but Systematic. Paper presented at The Word and the Morpheme, Humboldt Universität zu Berlin, Germany, September 22–24.

- 978-90-04-44885-8

230

grasso

Kastner, Itamar. 2019. Inchoatives in Causative Clothing: Change of State in Modern Hebrew heXYiZ. The Linguistic Review 36(3): 437–451. Kautzsch, Emil. 1910. Gesenius’ Hebrew Grammar. Trans. A.E. Cowley. 2nd English ed. Oxford: Clarendon Press. Koontz-Garboden, Andrew, and John Beavers. 2013. In Defense of the Reflexivization Analysis of Anticausativization. Lingua 131: 199–216. Kratzer, Angelika. 1996. Severing the external argument from its verb. Pp. 109–137 in Phrase structure and the lexicon, ed. Laurie Ann Zaring and Johan Rooryck. Dordrecht: Springer. Kratzer, Angelika. 2005. Building Resultatives. Pp. 177–212 in Event Arguments: Foundations and Applications, ed. Claudia Maienborn and Angelika Wöllstein. Berlin: Walter de Gruyter. Levin, Beth, and Malka Rappaport Hovav. 1995. Unaccusativity: At the Syntax-Lexical Semantics Interface. Linguistic Inquiry Monographs 26 Cambridge, MA: mit Press. Lundquist, Björn, Martin Corley, Mai Tungseth, Antonella Sorace, Gillian Ramchand. 2016. Anticausatives are semantically reflexive in Norwegian but not in English. Glossa: a journal of general linguistics 1(1): 47. 1–30 Marín, Rafael, and Louise McNally. 2011. Inchoativity, change of state, and telicity: Evidence from Spanish reflexive psychological verbs. Natural Language & Linguistic Theory 29 (2): 467–502. Piñón, Christopher. 1997. Achievements in an event semantics. Pp. in Semantics and Linguistic Theory 7: 276–293. Pylkkänen, Liina. Introducing Arguments. Cambridge, MA: mit Press, 2008. Rappaport Hovav, Malka. 2014a. Building Scalar Changes. Pp. 259–281 in The Syntax of Roots and the Roots of Syntax. In The Syntax of Roots and the Roots of Syntax, ed. Artemis Alexiadou, Hagit Borer, and Florian Schafer. Oxford Studies in Theoretical Linguistics 51. Oxford: Oxford University Press. Rappaport Hovav, Malka. 2014b. “Lexical Content and Context: The Causative Alternation in English Revisited.” Lingua 141: 8–29. Rappaport Hovav, Malka, and Beth Levin. 2010. Reflections on Manner/Result Complementarity. Pp. 21–38 in Syntax, Lexical Semantics, and Event Structure, ed. Edit Doron, Malka Rappaport Hovav, and Ivy Sichel. Oxford: Oxford University Press. Schäfer, Florian. 2012. Two Types of External Argument Licensing-the Case of Causers. Studia Linguistica 66 (2): 128–180. Waltke, Bruce K., and Michael O’Connor. 1990. An Introduction to Biblical Hebrew Syntax. Winona Lake, IN: Eisenbrauns.

- 978-90-04-44885-8

chapter 12

Hybrid Syntactic Constructions in Biblical Hebrew Tamar Zewi

1

Introduction

Biblical Hebrew grammars usually treat extensively standard syntactic constructions, which are used regularly and are the essence of Biblical Hebrew syntax. Examples are genitive phrases, nominal and verbal clauses, relative clauses, interrogative clauses, circumstantial clauses, various adverbial phrases and clauses, conditional sentences, and so forth.1 Here I would like to highlight Biblical Hebrew constructions which do not fully fit any of the conventional syntactic classifications and show some properties of more than one construction. Such constructions usually initially reflect syntactic fluctuation among standard constructions, and in this sense should be regarded as hybrid. Sometimes they may also be subject to reanalysis and can occasionally emerge as new constructions that deserve their own independent interpretation within the same language phase or in later phases. I focus on five hybrid constructions: (1) the genitive construction involving šɛl, (2) the inverted adjectival annexion, (3) the imperfect extraposition, (4) the object marker ʾɛṯ positioned before a subject, and (5) the third independent personal pronoun in nominal clauses. It will be shown that the common denominator of these constructions is their transition, at some point, from standard to non-standard syntactic constructions, which possess attributes of more than one syntactic pattern. Therefore, they may all be defined as hybrid, in certain cases in regard to their origin and emergence only, and in other cases all along.

1 Kautzsch 1910: 309–506; Joüon and Muraoka 2006: 325–614; Waltke and O’Connor 1990; Williams 2007.

© Tamar Zewi, 2021 | doi:10.1163/9789004448858_013

- 978-90-04-44885-8

232 2

zewi

Discussion of Hybrid Constructions in Biblical Hebrew

2.1 The Genitive Construction Involving šɛl A well-known example of the emergence and development of a secondary syntactic construction in post-Classical Hebrew is the periphrastic genitive consisting of šɛl. This construction pops up in Biblical Hebrew in seven examples only, six of which (examples 2–7) definitely come from Late Biblical Hebrew texts: (1) way=y=issaʿer lev mɛlɛḵ ʾăram ʿal and=trouble.ipfv.3ms heart.ms.cstr king. ms.cstr Aram on had=daḇar haz=zɛ way=yiqraʾ ʾɛl the=thing the=dem.ms and.then=call.ipfv.3ms to ʿăḇad=aw way=yoʾmɛr ʾăle=hɛm servant.mp=pron.3mp and.then=say.ipfv.3ms to=pron.3mp hă=loʾ taggiḏu l=i mi miš=šɛl=l=anu interr=neg say.ipfv.2mp to=pron.1cs who of=rel=to=pron.1cp ʿɛl mɛlɛḵ yiśraʿel to king.ms.cstr Israel ‘And the mind (literally, ‘heart’) of the king of Aram was greatly troubled because of this thing; and he called his servants and said to them, “Will you not show me who of us is for the king of Israel?” ’2 (2 Kgs 6:11) (2) way=yoʾmǝru ʾiš ʾɛl reʿ=ehu and.then=say.ipfv.3mp man.ms to friend.ms=pron.3ms lǝḵu wǝ=nappila ḡoraloṯ wǝ=neḏǝʿa come.imp.2mp and=cast.ipfv.1cp lot.mp and=know.ipfv.1cp bǝ=šɛl=lǝ=mi ha=raʿa haz=zoʾṯ la=nu on=rel=to=who the=evil.fs the=dem.fs to=pron.1cp way=yappilu goraloṯ way=yippol hag=goral and.then=cast.ipfv.3mp lot.mp and.then=fall.ipfv.3ms the=lot.ms ʿal yona upon Jonah ‘And they said to one another, ‘Come, let us cast lots, that we may know on whose account this evil has come upon us.’ So they cast lots, and the lot fell upon Jonah’ (Jon 1:7)

2 English translations are according to the rsv, with minor changes.

- 978-90-04-44885-8

hybrid syntactic constructions in biblical hebrew

233

(3) way=yoʾmɛr ʾăle=hɛm śaʾu=ni and.then=say.ipfv.3ms to=pron.3mp take.imp.2mp=pron.1cs wa=hăṭil=uni ʾɛl hay=yam and=throw.imp.2mp=pron.1cs to the=sea.ms wǝ=yištoq hay=yam me=ʿăl=eḵɛm ki and=quiet.down.ipfv.3ms the=sea.ms from=to=pron.3mp because yoḏeaʿ ʾani ki ḇǝ=šɛl=l=i know.ptcp.ms pron.1cs that because=rel=of=pron.1cs has=saʿar hag=gaḏol haz=zɛ ʿăl=eḵɛm the=tempest.ms the=great.adj.ms the=dem.ms to=pron.2mp ‘He said to them, “Take me up and throw me into the sea; then the sea will quiet down for you; for I know it is because of me that this great tempest has come upon you”’ (Jon 1:12) (4) ʾal tirʾu=ni šɛ=ʾăni šǝḥarḥorɛṯ neg gaze.ipfv.2mp=pron.1cs rel=pron.1cs swarthy.adj.fs šɛš=šɛ̆zap̄ aṯ=ni haš=šamɛš bǝne rel=scorch.pfv.3fs=pron.1cs the=sun.fs son.mp.cstr ʾimm=i niḥăru ḇ=i mother=pron.1cs angry.pfv.3mp with=pron.1cs śamu=ni noṭera ʾɛṯ hak=kǝramim made.pfv.3mp=pron.1cs keeper.ptcp.fs om the=vineyard.mp karm=i šɛ=l=li loʾ naṭarti vineyard.ms=pron.1cs rel=of=pron.1cs neg keep.pfv.1cs ‘Do not gaze at me because I am swarthy, because the sun has scorched me. My mother’s sons were angry with me, they made me keeper of the vineyards; but, my own vineyard I have not kept!’ (Song 1:6) (5) hinne miṭṭaṯ=o šɛl=li=šlomo šiššim gibborim behold litter.fs=pron.3ms rel=of=Solomon sixty.mp mighty.man.mp saḇiḇ l=ah mig=gibbore yiśraʾel around to=pron.3fs of=mighty.man.mp.cstr Israel ‘Behold, it is the litter of Solomon! Around it are sixty mighty men of the mighty men of Israel’ (Song 3:7) lǝ=p̄ an-ay šɛl=l=i (6) karm=i vineyard.ms=pron.1cs rel=to=pron.1cs for.pron.1cs šǝlomo u=maʾṯayim lǝ-ḵa ha=ʾɛlɛp̄ the=thousand.ms to=pron.2ms Solomon and=two.hundred.du

- 978-90-04-44885-8

234

zewi

lǝ=noṭrim ʾɛṯ piry=o to.keeper.ptcp.mp om fruit.ms=pron.3ms ‘My vineyard, my very own, is for myself; you, O Solomon, may have the thousand, and the keepers of the fruit two hundred’ (Song 8:12) (7) wǝ=raʾiṯi ʾɛṯ kol maʿăśe ha=ʾɛ̆lohim ki loʾ and=see.ipfv.1cs om all work.ms the=God because neg yuḵal ha=ʾaḏam li=mṣoʾ ʾɛṯ ham=maʿăśɛ ʾăšɛr can.ipfv.3ms the-man.1ms to=find.inf.cstr om the=work.ms rel naʿăśa ṯaḥaṯ haš=šɛmɛš bǝ=šɛ=l ʾăšɛr yaʿămol do.3ms under the=sun.ms because=rel=to rel toil.ipfv.3ms ha=ʾaḏam lǝ=ḇaqqeš wǝ=loʾ yimṣaʾ wǝ=ḡam the=man.ms to=seek.inf.cstr and=neg find.ipfv.3ms and=even ʾim yoʾmar hɛ=ḥaḵam la=ḏaʿaṯ thought claim.ipfv.3ms the=wise man.adj.ms to=know.inf.cstr loʾ yuḵal li=mṣoʾ neg can.ipfv.3ms find.inf.cstr ‘Then I saw all the work of God, that man cannot find out the work that is done under the sun. However much (literally, ‘because’) man may toil in seeking, he will not find it out; even though a wise man claims to know, he cannot find it out’ (Eccl 8:17) First, it is interesting that example 1, from 2Kgs. 6.11, the only one from a Classical Biblical Hebrew text, is found in direct speech of the king of Aram addressed to his servants. As indicated by Hurvitz, the form miššɛllanu “of us” may have been used in this case as a literary means, which intimates the Aramaic language spoken in that king’s court.3 This occurrence, albeit in a relatively early text, well supports the assumption that this construction emerged under the influence of Aramaic and due to intense contact between Hebrew and Aramaic in the first millennium bc, as it is inlaid in an Aramaic speech context. Moreover, in six of the examples (examples 1–6) the combination šɛ plus lǝ appears attached to the following words, that is: miššɛllanu “of us” (2 Kgs 6:11, example 1), bǝšɛllǝmi “on whose account” (Jon 1:7, example 2),4 ḇǝšɛlli “because of me” (Jon 1:12, example 3), šɛlli “my own” (Cant 1:6, example 4), šɛllišlomo

3 Hurvitz 1996: 87. 4 A similar example with an interrogative following ša instead of šɛ is šallama “for why” (Cant 1:7).

- 978-90-04-44885-8

hybrid syntactic constructions in biblical hebrew

235

“of Solomon” (Cant 3:7, example 5), and again šɛlli “my very own” (Cant 8:12, example 6). In these cases it can still be interpreted as reflecting the use of the late Biblical Hebrew relative particle šɛ, followed by the preposition lǝ, namely this is the construction of a common type of a concise relative clause which contains merely a prepositional phrase and indicates possession. Moreover, in examples 1, 3, 4, and 6 this combination introduces a pronominal suffix, and in example 2 the interrogative pronoun mi “who”. Only in one example, šɛllišlomo “of Solomon” (Cant 3:7, example 5), is it not followed by a pronominal suffix but by a noun—however, in a structure involving prolepsis by a cataphoric pronoun, miṭṭaṯo šɛllišlomo “the litter of Solomon”. In later Hebrew grammars this construction will be labeled ‘double annexion’. Furthermore, only in one example (example 7) is the combination šɛ plus lǝ separated and stands as an independent particle: bǝšɛl “even though” (Eccl 8:17) in the phrase bǝ=šɛ=l ʾăšɛr “even though”. This phrase actually contains both šɛ and ʾăšɛr, namely the Late Biblical Hebrew relative particle and the Classical Biblical Hebrew relative particle; this is still used in various Late Biblical Hebrew compositions, and is found in the late book of Ecclesiastes alongside šɛ, sometimes in the same verse. The redundant use in this verse of both šɛ + lǝ and ʾăšɛr points to a syntactic link still existent between the two relative particles in this period and to the as yet obscure nature of the emerging combination šɛ plus lǝ as an independent grammatical entity. All in all, these examples point to the formation of a new genitive construction in a certain Hebrew dialect contemporaneous with Late Biblical Hebrew, which became the standard in Mishnaic Hebrew. The syntax of this language phase was heavily influenced by Aramaic, which undoubtedly contributed to the emergence and distribution of this construction as an independent individual particle on the basis of a reanalysis of two existing syntactic function words, šɛ and lǝ, combined.5 Other parallel constructions usually found in Classical and Late Biblical Hebrew are compounds in which the first member stands in a construct state, like maʿăśe haʾɛ̆lohim “the work of God” (Eccl 8:17), phrases comprised of two nouns linked by the preposition lǝ, like mizmor lǝdavid “a psalm of David” (Ps 23:1), and concise relative clauses containing simply a prepositional phrase indicating possession, like yǝhi lǝḵa ʾăšɛr laḵ “keep what you have for yourself” (Gen 33:9). These remained in use in all phases of Hebrew, but the general reorganization of the linguistic system following the development and expansion of the constructions involving reanalysis of

5 Fassberg 2019: 54. More details on this particle in Post-Classical Hebrew see in Mor 2015: 75– 78, and more references there.

- 978-90-04-44885-8

236

zewi

an independent particle šɛl underwent a major revolution in Hebrew and was never the same again.6 While such constructions have been amply discussed in various Hebrew grammars, their explanation in terms of an ‘emergent construction’ involving ‘reanalysis’, possibly under the influence of Aramaic, deserves perhaps some more attention. 2.2 The Inverted Adjectival Annexion The second construction which I would like to discuss in this respect is the type commonly entitled ‘inverted adjectival annexion’. It is a genitive construction in which the first member in the construct state is a participle or an adjective and its role is twofold. On the one hand, the whole genitive construction forms an attributive relation with an implicit pronominal head, sometimes also explicitly expressed by a preceding external head. Accordingly, the initial member in the construct state agrees with this head in gender and number. On the other hand, the first member in the construct state also modifies the quality or state of the following second member in the annexion, although it does not agree with it in number and gender. This construction is prevalent in Biblical Hebrew.7 Examples in which the distinctive grammatical agreement between the first and second heads and the participle or adjective is evident are as follows. (8) wǝ=huʾ yaḡeaʿ u=rǝp̄e and=pron.3ms weary.ptcp.ms and=discouraged.ptcp.mp.cstr yaḏayim hand.fdu ‘While he is weary and discouraged’ Lit., ‘while he is weary and he is weak and his hands are weak’ (2 Sam 17:2) The first head in urǝp̄ e yaḏayim is masculine singular and second head is feminine plural.

6 See e.g. Joüon & Muraoka 2006: 444–448. For more on the dative l- in Hebrew Annexions see Goldenberg 2013b: 238–241. 7 This construction, entitled by Arab grammarians ‘impure annexion (‫ ’)إظافة غير حقيقية‬and by Goldenberg “adjectival annexion” is common in Arabic and Hebrew (Goldenberg 2013b:276– 277). On this construction in Biblical Hebrew see e.g. Joüon & Muraoka 2006:438–439, Shmidman 2013.

- 978-90-04-44885-8

hybrid syntactic constructions in biblical hebrew

237

(9) śar ḥămiššim u=nǝśuʾ p̄anim captain.ms.cstr fifty.mp and=ranked.ptcp.ms.cstr face.mp wǝ=yoʿeṣ wa=ḥăḵam ḥărašim and=counselor and=skillful.adj.ms.cstr spell.mp u=nǝḇon laḥaš and=wise.ptcp.ms.cstr charm.ms ‘The captain of fifty and the man of rank, the counselor and the skillful magician and the expert in charms’ Lit., ‘the captain of fifty and who is ranked and whose face8 is ranked and the counselor and who is skillful and whose spells are skillful and who is wise and whose charm is wise’ (Isa 3:3) This verse contains three examples. Twice, in the first and second phrases unǝśuʾ p̄ anim and waḥăḵam ḥărašim, first head is masculine singular (implicit in the participle) and second head masculine plural. In the third phrase unǝḇon laḥaš the first and second heads are both masculine singulars. (10) ʿam toʿe leḇaḇ hem people.ms err.ptcp.mp.cstr heart.ms pron.mp ‘They are a people who err in heart’ Lit., ‘they are a people who err and whose heart errs’ (Ps 95:10) The first head is masculine plural9 and the second head masculine singular in toʿe leḇaḇ. (11) zɛ hay=yam gaḏol u=rǝḥaḇ dem.ms the=sea.ms great.ptcp.ms and=wide.ptcp.ms.cstr yaḏayim hand.fdu ‘Yonder is the sea, great and wide’ Lit., ‘Yonder is the sea, great and wide and its hands are wide’ (Ps 104:25) In this example, first head is masculine singular and second head feminine plural. First head is masculine singular and second head, possibly, feminine singular in yǝšar darɛḵ in the following example. Note that contrary to the

8 The Hebrew word for “face” p̄ anim is feminine plural. 9 The noun ʿam “a people” is treated as both singular and plural in Biblical Hebrew. In this case it is obviously plural, as the agreement with the following participle dictates.

- 978-90-04-44885-8

238

zewi

rsv translations in the previous examples, the translation in this case is by a relative clause: (12) wǝ=ṯoʿăḇaṯ rašaʿ yǝšar and=abomination.fs.cstr wicked.adj.ms straight.adj.ms.cstr darɛḵ way.fs ‘But he whose way is straight is an abomination to the wicked’ Lit., ‘but he who is straight and whose way is straight is an abomination to the wicked’ (Prov 29:27) The following verse contains three examples. First heads of the first two examples, ṭoḇaṯ śɛḵɛl wip̄ aṯ toʾar, are both feminine singular, while their two second heads are masculine singular. First head is masculine singular in the third example, while second head is masculine plural in wǝraʿ maʿălalim. śɛḵɛl ṭoḇaṯ (13) wǝ=ha=ʾišša and=the=woman.fs good.adj.fs.cstr wit.ms qašɛ wi=p̄aṯ toʾar wǝ=ha=ʾiš and=beautiful.adj.fs.cstr look and=the=man.ms churlish.ptcp.ms wǝ=raʿ maʿălalim and=ill.ptcp.ms.cstr act.mp ‘The woman was of good understanding and beautiful, but the man was churlish and ill-behaved’ Lit., ‘the woman was good and her wit was good and she was beautiful and her look was beautiful and the man was churlish and ill and his acts were ill’ (1Sam 25:3) Obviously, when both heads are of the same gender and number, the disagreement in gender and number with the first noun rather than the second is less apparent. This is the case, or almost the case, in the following examples. (14) ʿam kɛḇɛḏ ʿawon people.ms laden.ptcp.ms.cstr iniquity.ms ‘A people laden with iniquity’ Lit., ‘a people laden with iniquity and whose iniquity is laden’ (Isa 1:4) In the example above, first and second heads are both masculine singular. In the following example, first and second heads are both plurals, although not in the same gender.

- 978-90-04-44885-8

hybrid syntactic constructions in biblical hebrew

239

(15) nǝharim yǝʾorim raḥăḇe yaḏayim rivers.mp streams.mp broad.ptcp.mp.cstr hand.fs.du ‘Broad rivers and streams’ Lit., ‘broad rivers and streams whose hands are broad’ (Isa 33:21) A related construction is the following. This example slightly deviates from the norms of the inverted annexion construction. The adjective is not in the construct but in the absolute state, and the noun in the genitive has an attached possessive pronoun. In this example, first head is masculine singular and second head feminine singular. (16) wǝ=hinne li=qraʾṯ=o ḥušay ha=ʾarki behold to=come.to.meet.inf.cstr=pron.3ms Hushai the=Archite qaruaʿ kuttant=o rent.ptcp.ms.abs coat.fs=pron.3ms ‘Behold, Hushai the Archite came to meet him with his coat rent’ Lit., ‘Hushai the Archite is rent and his coat (is rent)’ (2 Sam 15:32) While agreeing with Goldenberg’s indication that in this type of adjectival annexion “syntactic connexions cross the word boundary,”10 I would like to further define it in terms of a hybrid construction, in the sense that it reflects fusion of two originally separate constructions, in which two heads and their modifying participles or adjectives are syntactically, as well as morphologically, merged into one structure, while maintaining certain properties of each of the two original phrases in the new united construction. 2.3 The Imperfect Extraposition The next topic on which I would like to talk is the imperfect extraposition. The most inclusive contribution to the discussion of extrapositions in Biblical Hebrew, as well as Arabic, Aramaic, Akkadian, and Amharic, was made by Geoffrey Khan in his work “Studies in Semitic Syntax,” published in 1988.11 Khan took into consideration cases reflecting all properties of standard full extrapositions, but also embraced in his discussion imperfect types, which may have only partial properties of extrapositions or may possess redundant elements of extrapositions. Among these, I would like first to draw the attention to certain extrapositions which contain a redundant preposition, namely those in which

10 11

Goldenberg 2013b: 277. Khan 1988:67–104.

- 978-90-04-44885-8

240

zewi

the same preposition appears twice in the sentence, once before the extraposed member and again before the resumptive pronoun referring to this very member. Examples are these.12 2.3.1

Repetition of ʾɛṯ

(17) way=yoʾmɛr miḵayhu ḥay yhwh ki ʾɛṯ ʾăšɛr and.then=say.ipfv.3ms Micaiah live.pfv.3ms the Lord that om rel yoʾmar yhwh ʾel=ay ʾoṯ=o ʾăḏabber say.ipfv.3ms Lord to=pron.1cs om=pron.3ms speak.ipfv.1cs ‘But Micaiah said, “As the Lord lives, what the Lord says to me, that I will speak”’ (1Kgs. 22.14). 2.3.2

Repetition of lǝ

(18) wǝ=li=ṣlop̄ ḥaḏ bɛn ḥep̄ ɛr bɛn gilʿaḏ bɛn maḵir and=to=Zelophehad son.of Hepher son.of Gilead son.of Machir bɛn mǝnaššɛ loʾ hayu l=o banim ki ʾim son.of Menasseh neg be.pfv.3cp to=pron.3ms son.mp but only banoṯ daughters.fp ‘Now Zelophehad the son of Hepher, son of Gilead, son of Machir, son of Manasseh, had no sons, but only daughters’ (Josh 17.3). 2.3.3

Repetition of bǝ

(19) bǝ=šoʾa yippol b=ah in=ruin.fs fall.ipfv.3ms in=pron.3fs ‘Let them fall therein to ruin’ (Ps. 35.8). 2.3.4

Repetition of min (and bǝ to min)

(20) u=mip=pǝri ha=ʿeṣ ʾăšɛr bǝṯoḵ hag=gan and=of=fruit.ms.cstr the=tree rel in the=garden.ms ʾamar ʾɛ̆lohim loʾ ṯoʾḵǝlu mimmɛn=nu wǝ=loʾ say.pfv.3ms God neg eat.ipfv.2mp of=pron.3ms and-neg

12

For these and other examples see Khan 1988:75–76.

- 978-90-04-44885-8

hybrid syntactic constructions in biblical hebrew

241

ṯiggǝʿu b=o pɛn tǝmuṯun touch.ipfv.2mp in=pron.3ms lest die.ipfv.2mp ‘But God said, “You shall not eat of the fruit of the tree which is in the midst of the garden, neither shall you touch it, lest you die” ’ (Gen. 3.3). 2.3.5

Repetition of ʿal

(21) wǝ=ʿal dǝmuṯ hak=kisseʾ dǝmuṯ kǝ=marʾe and=on likeness.fs.cstr the=throne.ms likeness.fs as=form.ms.cstr ʾaḏam ʿal=aw mi=l=maʿla man.ms on=pron.3ms from=to=above ‘And seated above the likeness of a throne was a likeness as it were of a human form’ (Ezek 1:26) 2.3.6

Repetition of ʾɛl

(22) wǝ=ʾɛl mɛlɛḵ yǝhuḏa haš=šoleaḥ ʾɛṯ=ḵɛm and=to king.ms.cstr Judah the=send.ptcp.ms om=pron.2mp li=ḏroš ʾɛṯ yhwh ko ṯoʾmǝru ʾel=aw to=inquire.inf.cstr om Lord thus say.ipfv.2mp to=pron.3ms ‘But as to the king of Judah, who sent you to inquire of the Lord, thus shall you say to him’ (2Kgs 22:18) 2.3.7

Repetition of ʿim

(23) wǝ=ʿim ha=ʾămahoṯ ʾăšɛr ʾamart ʿimm=am and=by the=maid.fp rel say.pfv.2fs by=pron.3fp ʾikkaḇeḏa held.in.honor.ipfv.1cs ‘But by the maids of whom you have spoken, by them I shall be held in honor’ (2Sam. 6.22). Note that in six of the seven examples (17–22) the English translations present full extrapositions or simple sentences, since an exact translation reflecting the Hebrew structure would usually be considered ungrammatical. Only example 23 yields a double redundant translation of the Hebrew preposition ʿim by the English preposition “by.” As I have suggested elsewhere, these cases should be interpreted as hybrid in nature and should be considered imperfect and in transition, on the continuum from simple sentences not involving any type of extraposition at all to

- 978-90-04-44885-8

242

zewi

complete extrapositions.13 While they show all properties of a full extraposition by displaying an extraposed member and a resumptive pronoun referring to it, they still keep the status of the extraposed member as a direct or indirect object in the simple clause from which the extraposition has emerged. Second, I would like to mention another type of incomplete extraposition, which I also discussed in the past elsewhere.14 This type is attested not in main but in subordinate clauses in the role of objects. A full extraposition of this type is usually defined as prolepsis, namely a construction in which the extraposed sentence member not only precedes its predicate clause, as in a regular extraposition, but also comes before the subordinate particle introducing the subordinate clause. In such cases the extraposed sentence member gains a syntactic role of the first object of the verb in the main clause and the subordinate particle functions as a syntactic divider between the extraposed sentence member and the object clause, which is its predicated clause and a second object of the verb in the main clause. The verbs in this sentence type usually belong to the group known in Arabic grammars as ‫افعال‬ ‫“( القلوب‬verbs of perception”, literally, “verbs of the heart,” like raʾa “see”, yadaʿ and hikkir “know”).15 This pattern is demonstrated in the following examples: (24) way=yarʾ mošɛ ʾɛṯ ha=ʿam ki and.then=see.ipfv.3ms Moses om the=people.ms that p̄ aruaʿ huʾ broken loose.ptcp.ms pron.1ms ‘And Moses saw that the people had broken loose.’ Lit., ‘And Moses saw the people, that it had broken loose’ (Exod 32:25) (25) way=yoʾmɛr ḥušay ʾatta yaḏaʿta ʾɛṯ and.then=say.ipfv.3ms Hushai pron.2ms know.pfv.2ms om ʾaḇi=ḵa wǝ=ʾɛṯ ʾănaš=aw ki father.ms=pron.2ms and=om man.mp=pron.3ms that ḡibborim hemma u=mare nɛp̄ ɛš mighty man.mp pron.3mp and=bitter.ptcp.mp.cstr soul.fs hemma kǝ=ḏoḇ šakkul baś=śaḏɛ pron.mp like=bear.ms robbed.ptcp.ms in.the=field.ms

13 14 15

Zewi 1997:177–178, 1999a:246–248. Zewi 1996b:6–8. Zewi 1996b:6–8.

- 978-90-04-44885-8

hybrid syntactic constructions in biblical hebrew

243

‘Hushai said moreover, “You know that your father and his men are mighty men, and that they are enraged, like a bear robbed of her cubs in the field.”’ Lit., ‘Hushai said moreover, “You know your father, that he and his men are mighty men, and (that) they are bitter of soul, like a bear robbed of her cubs in the field”’ (2Sam 17:8) While this construction can still be described as a complete extraposition, albeit in a subordinate clause, there are two examples in Biblical Hebrew in which the subordinate particle repeats itself twice, both introducing and following the extraposed sentence member. These are the following: (26) way=yoʾmɛr dawiḏ lǝ=ʾɛḇyaṯar yaḏaʿti and.then=say.ipfv.3ms David to=Abiathar know.pfv.1cs bay=yom ha=huʾ ki šam doʾeḡ [Qr] ha=ʾăḏomi ki in.the-day.ms the=pron.3ms that their Doeg the=Edomite that haggeḏ yaggiḏ lǝ=šaʾul tell.inf.abs tell.ipfv.3ms to=Saul ‘And David said to Abiathar, “I knew on that day, that Doeg the Edomite, who was there, would surely tell Saul.”’ Lit., ‘And David said to Abiathar, “I knew on that day, that Doeg the Edomite, who was there, that he would surely tell Saul” ’ (1 Sam 22:22).16 (27) way=yoʾmǝru ʾel=aw ʿăḇaḏ=aw hinne and.then=say.ipfv.3mp to=pron.3ms servant.mp=pron.3ms behold naʾ šamaʿnu ki malḵe beṯ yiśraʾel ki now hear.pfv.1cp that king.mp.cstr house.ms.cstr Israel that malḵe ḥɛsɛḏ hem king.mp.cstr mercy.ms pron.3mp ‘And his servants said to him, “Behold now, we have heard that the kings of the house of Israel are merciful kings.”’ Lit., ‘And his servants said to him, “Behold now, we have heard that the kings of the house of Israel that they are merciful kings” ’ (1 Kgs. 20.31). These two occurrences should in fact be considered incomplete and hybrid extrapositions, in which the extraposed sentence part fluctuates between its 16

The rsv solves this syntactic conflict by turning the first part introduced by ki into a temporal clause: “And David said to Abiathar, ‘I knew on that day, when Doeg the Edomite was there, that he would surely tell Saul …’ ”

- 978-90-04-44885-8

244

zewi

status either as the subject of the object subordinate clause or as the first object of the main clause. 2.4 The Direct Object Particle ʾɛṯ Introducing a Subject I will now talk about the so-called ‘ʾɛṯ before subjects’ in Biblical Hebrew. One construction in which the particle ʾɛṯ introduces a subject, or at least a sentence member which can be defined as topic, is part of what has just been discussed. This is the type of incomplete extraposition in which both the extraposed sentence member, namely the subject or topic of the whole sentence involving extraposition, and the object of the following clause are introduced by the particle ʾɛṯ. This type was demonstrated in example 17 above. As suggested above, this example is similar in structure to other incomplete extrapositions in which prepositions other than ʾɛṯ repeat in the sentence twice, once before the extraposed sentence part and again before the resumptive pronoun referring to it. In this section, I would like to point out other Biblical Hebrew constructions in which the particle ʾɛṯ allegedly introduces a subject. Such examples have been interpreted by various scholars (e.g., Hoftijzer 1965) as introducing a subject, possibly for emphasis, or as reflecting certain qualities of an ergative construction, in which subject and object share a similar syntactic marking (this was partially done e.g. by Andersen 1971 and Khan 1984). Other scholars (e.g. Blau 1954, 1956, 1978, 1996a, 1996b and Zewi 1997) considered them as having only object marking, while sometimes involving another syntactic construction, process, or interference.17 Elsewhere, I treated the majority of these and similar examples as “hybrid sentences in transition” and pointed out that they also involve a change in the syntactic status of their grammatical object, which, being the logical subject of the sentence, may also acquire properties of a grammatical subject, while its marking as object still obtains.18 The first group of examples, which is generally uncommon in Biblical Hebrew, includes third person singular passive verbs as predicates. Such examples are interpreted as impersonal passives by Blau, who considers them equivalent to Arabic impersonal passives with indirect objects:19

17

18 19

On these constructions and their possible explanations see e.g. Blau 1954, 1956, 1978, 1996a,b Waltker 1955, Andersen 1971, Hoftijzer 1965, Muraoka 1985:146–158, Khan 1984, Zewi 1997, 1999a:249–252. For more references see especially Blau 1996a:137–138. As noted above, Blau (1996b:114–123) treats separately the group of invariable passives in which, he believes, the sentence member introduced by ʾɛṯ keeps its object status as a genuine property of the construction, which does not have a specific subject. Zewi 1997:173, 1999a:249. Blau (1996b:114–123).

- 978-90-04-44885-8

hybrid syntactic constructions in biblical hebrew

245

(28) way=yiwwaleḏ la=ḥănoḵ ʾɛṯ ʿiraḏ and.then=be.born.ipfv.3ms to-Enoch om Irad ‘To Enoch was born Irad’ (Gen 4:18)20 (29) way=yuggaḏ lǝ=riḇqa ʾɛṯ diḇr-e ʿeśaw and.then=be.told.ipfv.3ms to-Rebekah om word.mp.cstr Esau bǝn=ah hag=gaḏol son.ms=pron.3fs the=older.adj.ms ‘But the words of Esau her older son were told to Rebekah’ (Gen 27:42) (30) way=yiwwaleḏ lǝ=ʾahăron ʾɛṯ naḏaḇ wǝ=ʾɛṯ ʾăḇihuʾ and.then=be.born.ipfv.3ms to=Aaron om Nadab and=om Abihu ʾɛṯ ʾɛlʿazar wǝ=ʾɛṯ ʾiṯamar om Eleazar and=om Ithamar ‘And to Aaron were born Nadab, Abihu, Eleazar and Ithamar’ (Num 26:60) The second group, which is also rare in Biblical Hebrew, is comprised of examples with passive verbs which agree in number and gender with their subject introduced by ʾɛṯ. These examples were perhaps influenced by the construction of the examples above, but went further into acquiring subject properties in their agreement patterns: (31) ʾɛṯ ʾarbaʿaṯ ʾellɛ yullǝḏu lǝ=ha=rap̄ a om four.ms.cstr dem.cp be.descended.ipfv.3mp to=the=giant.mp bǝ=ḡaṯ in=Gath ‘These four were descended from the giants in Gath’ (2 Sam 21:22) (32) wǝ=loʾ ṯinnaṯen ʾɛṯ ha=ʿir haz=zoʾṯ and=neg be given.ipfv.3fs om the=city.fs the=dem.fs bǝ=yaḏ mɛlɛḵ ʾaššur in=hand.fs.cstr king.ms.cstr Assyria ‘And this city will not be given into the hand of the king of Assyria’ (2 Kgs 18:30)

20

In all rsv English translations the object obviously becomes a subject.

- 978-90-04-44885-8

246

zewi

(33) wǝ=ʾɛṯ ha=ʾaḥ lǝp̄ an=aw mǝḇoʿarɛṯ and=om the=brazier.fs before=pron.3ms burned.ptcp.fs ‘And there was a fire burning in the brazier before him’ (Jer 36:22) Other occasional examples, not involving passive verbs, which show ʾɛṯ before grammatical subjects, like the following two examples, should probably be explained as involving attraction to an adjacent transitive verb or as reflecting anacoluthon: (34) wa=yǝhi ha=ʾɛḥaḏ mappil haq=qora wǝ=ʾɛṯ and.then=aux.3ms the=one.ms fell.ptcp.ms the=log.fs and=om hab=barzɛl nap̄ al ʾɛl ham=mayim the=axe.head.ms fall.pfv.3ms to the=water.mp ‘But as one was felling a log, his axe head fell into the water’ (2 Kgs 6:5) (35) wǝ=ʾatta bǝ=raḥămɛ=ḵa ha=rabbim loʾ and=pron.2ms in=mercy.mp=pron.2ms the=great.adj.mp neg ʿăzaḇt=am bam=miḏbar ʾɛṯ ʿammuḏ forsake.pfv.2ms=pron.3mp in.the=wilderness.ms om pillar.ms.cstr hɛ=ʿanan loʾ sar me=ʿăle=hɛm bǝ=yomam the=cloud.ms neg depart.pfv.3ms from=on=pron.3mp in=day.ms lǝ=hanḥoṯ-am bǝ=had=dɛrɛḵ wǝ=ʾɛṯ ʿammuḏ to=lead.inf.cstr=pron.3mp in=the=way.fs and=om pillar.ms.cstr ha=ʾeš bǝ=layla lǝ=haʾir la=hɛm wǝ=ʾɛṯ the-fire.fs by=night.ms to=light.inf.cstr to=pron.3mp and=om had=dɛrɛḵ ʾăšɛr yelḵu ḇ=ah the=way.fs rel go.ipfv.3mp by=pron.3fs ‘You in your great mercies did not forsake them in the wilderness; the pillar of cloud which led them in the way did not depart from them by day, nor the pillar of fire by night which lighted for them the way by which they should go’ (Neh 9:19) The scarcity of examples of all these types in Biblical Hebrew supports their interpretation as hybrid constructions. 2.5 The Third Independent Personal Pronoun in Nominal Clauses The last topic on which I would like to talk on this occasion is one of the socalled ‘tripartite clauses’ in Biblical Hebrew and its possible development in Modern Hebrew. I have previously identified five patterns of nominal clauses in Biblical Hebrew, following Goldenberg’s classification of the Syriac sentence

- 978-90-04-44885-8

247

hybrid syntactic constructions in biblical hebrew

patterns: two short ones and three extended, which contain a third independent personal pronoun functioning, in our opinion, as a resumptive pronoun in an extrapositional pattern.21 (36) The following table summarizes all these types in Biblical Hebrew:22 Pattern

Biblical Hebrew

A Pred. – Subj.

ki naʿar ʾatta for youth.ms pron.2ms ‘For you are a youth’ (1 Sam 17:33)

A2 Subj. – Pred.

B Subj. – Pred. clause ↓ (Pred. – Subj. [3rd Pers. Pron.]) C Pred. clause – Subj. ↓ (Pred. – Subj. [3rd Pers. Pron.])

21 22

ki naʿar yiśraʾel when youth.ms Israel ‘When Israel was a youth’ (Hos 11:1) ʾattɛm ʿam qǝše ʿorɛp̄ pron.2mp people.ms stiff.ptcp.ms.cstr neck.ms ‘You are a stiff-necked people’ (Exod 33:5) ʾăšɛr ʾăḇanɛ=ha ḇarzɛl rel stone.fp=pron.fs iron.ms ‘Whose stones are iron’ (Deut 8:9) le=ʾlohim huʾ ki ham=mišpaṭ for the=judgment.ms to=God pron.3ms ‘For the judgment is God’s’ (Deut 1:17)

yhwh huʾ ṣaddiq righteous.adj.ms pron.3ms the Lord ‘The Lord is righteous’ (Lam 1:18)

Goldenberg 1983; Zewi 1994, 1999b, 2000, 2013, 2016:58–62. The Biblical Hebrew examples are cited from Zewi 2016:61, and see there a comparison with Syriac.

- 978-90-04-44885-8

248

zewi (cont.)

Pattern

Biblical Hebrew

D Subj. – Pred. clause ↓ (Pred. [3rd Pers. Pron.] – Subj.)

wǝ=yosep̄ huʾ haš=šalliṭ ʿal and=Joseph pron.3ms the=governor.ms over ha=ʾarɛṣ the=land.fs ‘Now Joseph was governor over the land’ (Gen 42:6)

Patterns A and A2 are simple nominal clauses and patterns B, C and D are extended types involving extraposition. I will not discuss all these types in detail here but will concentrate only on one, the fifth, which I named D, following Goldenberg’s classification of the Syriac sentence structure.23 I believe that this nominal sentence pattern should be analyzed in Biblical Hebrew as an extraposition, whose third personal pronoun is the predicate of the predicate clause, as well as a resumptive pronoun referring to the extraposed member. In this way, its role as predicate is similar to any other predicate in a nominal clause and its role as a resumptive pronoun is equivalent to the role of the third personal pronoun in patterns B and C. In no way in Biblical Hebrew in this language period does the third personal pronoun in pattern D carry the role of a copula, namely it does not represent the predicative relation (/ the nexus) of the sentence, which is represented by the mere positioning of the subject and predicate side by side.24 Accordingly, it is not hybrid and does not show any special development in Biblical Hebrew. In later phases of Hebrew another type of an extended nominal clause involving extraposition emerged, labeled here pattern E, in which the word order of the predicate clause allows also subject–predicate sequence instead of predicate–subject. Still the role of the third personal pronoun in this sentence pattern is that of a resumptive pronoun, which refers to the subject of the whole sentence, namely the extraposed sentence part. (37) Pattern E—scheme: Subject—predicate clause ↓ (subject—predicate) 23 24

Goldenberg 1983:106–107. Goldenberg 2005, Zewi 1996a.

- 978-90-04-44885-8

249

hybrid syntactic constructions in biblical hebrew

In the first four lines of the Modern Hebrew poem “What Does the Bird Care” by Hanoch Levin, the third personal pronouns are still dispensable, and good standard simple Modern Hebrew nominal sentences can be formed without them. Therefore they should not be considered fulfilling the role of a copula: (38) Pattern E—example (“What Does the Bird Care” by Hanoch Levin): ha=ʿeṣ huʾ gaḇoah, ha=ʿeṣ huʾ the=tree.ms pron.3ms tall.adj.ms the=tree.ms pron.3ms yaroq, green.adj.ms hay=yam huʾ maluaḥ, hay=yam huʾ the=sea.ms pron.3ms salty.adj.ms the=sea.ms pron.3ms ʿamoq, deep.adj.ms ʾim hay=yam huʾ ʿamoq, If the=sea.ms pron.3ms deep.adj.ms ma ʾiḵpat l=o l=a=ʿeṣ, what care.pfv.3ms to=pron.3ms to=the=tree.ms ma ʾiḵpat l=o lay=yam what care.pfv.3ms to=pron.3ms to.the=sea.ms šɛ=ha=ʿeṣ huʾ yaroq rel=the=tree.ms pron.3ms green.adj.ms ‘The tree is tall, the tree is green, The sea is salty, the sea is deep, If the sea is deep, what does the tree care, What does the sea care that the tree is green?’ All these nominal clause patterns continue to be used in Modern Hebrew. However, only in Modern Hebrew can we possibly speak of a new emergent construction, whose origins are in pattern E. In this construction, where the subject is usually generic, the third personal pronoun is indispensable, and the sentence cannot be formed without it. Two such Modern Hebrew examples are the following: (39) sep̄ ɛr / has=sep̄ ɛr huʾ ḥăḇer=o book.ms / the=book.ms pron.3ms friend.ms=pron.3ms haṭ=ṭoḇ bǝ=yoter šɛl ha=ʾadam the=best.ptcp.ms best.pp of the=man.ms ‘A book is a man’s best friend.’

- 978-90-04-44885-8

250

zewi

(40) yǝladim zɛ śimḥa / yǝladim hem śimḥa child.mp dem.ms joy.fs / child.mp pron.3mp joy.fs ‘Children are joy.’ In such cases, the third personal pronoun may often be replaced by a fossilized non-congruent masculine singular demonstrative pronoun zɛ, mainly in the spoken language register. Since it is impossible or almost impossible to omit either the third personal pronoun or demonstrative pronoun in these last two examples, these pronouns should be considered essential indispensable sentence members, which can be interpreted as fulfilling the role of a copula. This new development can perhaps also be observed in a preference in Modern Hebrew for the use of the third personal pronoun in translation of nominal clauses including a copula in languages in which it is an essential indispensable sentence part. This last pattern can possibly be interpreted in certain cases in Modern Hebrew as an emergent construction, which shows reanalysis of the third personal pronoun as a copula. The whole pattern containing the third personal pronoun between the subject and the predicate can thus be defined in Modern Hebrew as hybrid, in the sense that it sometimes has to be interpreted as an extraposition, whose third personal pronoun is a resumptive pronoun, and in certain limited conditions possibly as a nominal clause including a third personal pronoun in the role of a copula.

3

Final Words

In conclusion, the Biblical Hebrew constructions discussed here do not fit one simple syntactic definition and analysis. They all show syntactic innovation and fluctuation among more than one syntactic pattern in a certain phase of their development. One of them, the inverted adjectival annexion, stabilized as a regular widespread syntactic construction in Biblical Hebrew, as well as Classical Arabic, and continued to be used in all Hebrew periods. Two of them, the imperfect extraposition and ʾɛṯ introducing a grammatical subject, did not stabilize in Biblical Hebrew and have never evolved into standard Biblical or later Hebrew patterns. Two more of them, the genitive construction involving šɛl and to a much smaller extent the third independent personal pronoun in extended nominal clauses, show syntactic innovation in that they gradually underwent reanalysis in later phases of Hebrew, and acquired new syntactic status.

- 978-90-04-44885-8

hybrid syntactic constructions in biblical hebrew

251

References Andersen, F.I. 1971. Passive and Ergative in Hebrew. In: Goedicke, H. (Ed.). Near Eastern Studies in honor of William Foxwell Albright. Baltimore: The Johns Hopkins Press: 1–15. Blau, J. 1954. Zum angeblichen Gebrauch von ‫ את‬vor dem Nominativ. Vetus Testamentum 4: 7–19. Blau, J. 1956. Gibt es ein emphatisches ’ēṯ in Bibelhebraeisch? Vetus Testamentum 6: 211– 212. [Reprinted in Hebrew, ‫( כלום מציינת ֵאת במקרא גם שימוש נומי ָנטיווי‬pp. 137–148 in Studies in Hebrew Linguistics. Jerusalem: The Magnes Press, The Hebrew University, 1996a.)] Blau, J. 1978. On Invariable Passive Forms in Biblical Hebrew and Classical Arabic. Pp. 85–94 in Studies in Bible and the Ancient Near East Presented to Samuel E. Loewenstamm on His Seventieth Birthday, V. I., ed. Y. Avishur, and J. Blau. Jerusalem: E. Rubinstein’s Publishing House. (Hebrew). [Reprinted in Hebrew, ‫על הסביל הסתמי במקרא‬ (‫( )דיון משווה על רקע הסביל הסתמי בערבית הקלאסית‬pp. 114–123 in J. Blau, Studies in Hebrew Linguistics. Jerusalem: The Magnes Press, The Hebrew University, 1996b.)] Fassberg, S.E. 2019. An Introduction to the Syntax of Biblical Hebrew. Jerusalem: The Bialik Institute. (Hebrew). Goldenberg, G. 1983. On Syriac Sentence Structure. Pp. 97–140 in Arameans, Aramaic and the Aramaic Literary Tradition, ed. M. Sokoloff. Ramat Gan: Bar-Ilan University Press: 97–140. [Reprinted, pp. 525–568 in G. Goldenberg. 1998. Studies in Semitic Linguistics. Jerusalem: The Magnes Press, The Hebrew University.] Goldenberg, G. 2005. Pronouns, Copulas and a Syntactic Revolution in Neo-Semitic. Pp. 239–252 In Studi afroasiatici: Contributi presentati all’xi Incontro italiano di linguistica camito-Semitica (Bergamo, 5–7 giugno 2003), ed. A. Mengozzi. Milan: Franco Angeli. [Reprinted, pp. 97–108 in G. Goldenberg. 2013a. Further Studies in Semitic Linguistics. Münster: Ugarit-Verlag.] Goldenberg, G. 2013b. Semitic Languages: Features, Structures, Relations, Processes. Oxford: Oxford University Press. Hoftijzer, J. 1965. Remarks Concerning the Use of the Particle ’t in Classical Hebrew. Leiden: Brill. Hurvitz, A. 1996. Hebrew and Aramaic in the Biblical Period—the Problem of ‘Aramaisms’ in the Linguistic Research of the Hebrew Bible. Pp. 79–94 in Studies in Hebrew and Jewish Languages Presented to Shelomo Morag, ed. M. Bar-Asher. Jerusalem: The Bialik Institute. (Hebrew). Joüon, P. and T. Muraoka. 2006. A Grammar of Biblical Hebrew.2nd ed. Rome: Editrice Pontificio Istituto Biblico. Kautzsch, E. 1910. Gesenius’ Hebrew Grammar. 2nd ed. Trans. A.E. Cowley. Oxford: Clarendon Press.

- 978-90-04-44885-8

252

zewi

Khan, G. 1984. Object Markers and Agreement Pronouns in Semitic Languages. Bulletin of the School of Oriental and African Studies 47: 468–500. Khan, G. 1988. Studies in Semitic Syntax. Oxford: Oxford University Press. Mor, U. 2015. Judean Hebrew: The Language of the Hebrew Documents from Judea between the First and the Second Revolts. Jerusalem: the Academy of the Hebrew Language. (Hebrew). Shmidman, A. 2013. Construct, Inverted. Pp. 595–596 in Encyclopedia of Hebrew Language and Linguistics, vol. 1, ed. G. Khan. Leiden: Brill. Waltke, B.K. and O’Connor, M. 1990. An Introduction to Biblical Hebrew Syntax. Winona Lake, IN: Eisenbrauns. Waltker, N. 1955. Concerning the Function of ’ēth. Vetus Testamentum 5: 314–315. Williams, R.J. 2007. Williams’ Hebrew Syntax. Third ed. Rev. and exp., John C. Beckman. Toronto: University of Toronto Press. Zewi, T. 1994. The Nominal Sentence in Biblical Hebrew. pp. 145–167 in Semitic and Cushitic Studies, ed. Goldenberg, G. and Sh. Raz. Wiesbaden: Harrassowitz. Zewi, T. 1996a. The Definition of the Copula and the Role of 3rd Independent Personal Pronouns in Nominal Sentences of Semitic Languages. Folia Linguistica Historica 17: 41–55. Zewi, T. 1996b. Subordinate Nominal Sentences Involving Prolepsis in Biblical Hebrew. Journal of Semitic Studies 41: 1–20. Zewi, T. 1997. Subjects Preceded by the Preposition ̕et in Biblical Hebrew. Pp. 171–183 in Studien zur hebräischen Grammatik, ed. A. Wagner. Freiburg, Schweiz: Universitätsverlag and Göttingen: Vandenhoeck and Ruprecht. Zewi, T. 1999a. Incomplete Topicalization in Biblical Hebrew and Modern Hebrew and Several Arabic Bible Translations. Australian Journal of Linguistics 19: 241–253. Zewi, T. 1999b. Tripartite Nominal Clauses and Appositions in Biblical Hebrew. Ancient Near Eastern Studies 36: 36–47. Zewi, T. 2000. Is There a Tripartite Nominal Sentence in Biblical Hebrew? Journal of Northwest Semitic Languages 26: 51–63. Zewi, T. 2013. Nominal Clause. Pp. 830–839 in Encyclopedia of Hebrew Language and Linguistics, vol. 2., ed. G. Khan. Leiden: Brill. Zewi, T. 2016. On Several Linguistic Axioms and the Semitic Languages. Pp. 50–67 in Gideon Goldenberg—In Memoriam, ed. A. Shisha-Halevy. Jerusalem: The Israel Academy of Sciences and Humanities. (Hebrew).

- 978-90-04-44885-8

Index of Authors Abney, Steven 153, 166 Adger, David 204, 206 Aelbrecht, Lobke 84, 89, 101, 102 Aikhenvald, Alexandra Y. 186, 187, 189 Aldridge, Edith 153, 166 Alexiadou, Artemis 209, 210, 211, 212, 216, 221, 222, 225, 226, 228, 229, 230 Amir, Noam 15, 35, 38 Anagnostopoulou, Elena 209, 212, 228, 229 Andersen, Francis I. 244, 251 Anderson, Stephen R. 13, 34 Anstey, Matthew P. 11, 34 Arad, Maya 126, 128, 129, 140, 141, 142 Bajohr, Hannes 2, 7 Balcaen, M. Jean 24, 34, 42, 43, 57 Ball, Martin J. 15, 24, 34 Barnes, Jonathan 10, 13, 14, 15, 20, 21, 22, 33, 34 Barr, James 1, 3, 7, 8 Bauer, Hans 149, 166 Beavers, John 215, 230 Becker-Kristal, Roy 12, 20, 34 Becker, Michel 15, 34 Beckman, Jill N. 10, 14, 31, 34 Beckman, Mary E. 22, 34 Berkovits, Rochele 22, 27, 35 Blake, Frank 42, 57 Blau, Joshua (Yehoshua) 44, 57, 149, 160, 166, 244, 251 Bodine, Walter R. 3, 7 Borer, Hagit 126, 128, 129, 139, 140, 142 Borg, A. 63, 64, 82 Bosch, Anna 13, 15, 21, 24, 34, 35 Bošković, Željko 119, 123 Boulet, Jacques E.J. 191, 192, 193, 196, 199, 200, 203, 206 Bowers, John 199, 202, 203, 204, 206, 207 Brambatti Guzzo, Natália 15, 36 Breuer, Mordecai 40, 49, 50, 57 Butt, Miriam 178, 179, 180, 182, 183, 189 Bylinina, L. 81, 82 Byrd, Dani 22, 35 Callaham, Scott N. 132, 142, 150, 166 Cambier-Langeveld, Tina 22, 35

Cardinaletti, Anna 149, 166 Chierchia, Gennaro 215, 218, 229 Cho, Taehong 22, 35 Chomsky, Noam 33, 35 Chrzanowski, Jaroslaw 182, 189 Chung, Sandra 204, 207 Churchyard, Henry 9, 11, 35 Cinque, Guglielmo 33, 35, 154, 166, 225, 229 Clements, Nick 13 Cohen, Evan-Gary 15, 35 Cohn, Abigail 27, 35 Collins, Christopher 186, 189 Collins, James 153, 166 Condoravdi, Cleo 213, 229 Cook, John A. 144, 166, 196, 207, 213, 229 Corley, Martin 215, 230 Cowper, Elizabeth 103, 109, 110, 112, 117, 123, 125, 142 Crosswhite, Katherine M. 10, 13, 14, 20, 34, 35 Cuervo, María Cristina 196, 207 Cuyckens, Hubert 204, 207 D’hoedt, Frauke 204, 207 Dawson, Willa 18, 35 De Swart, Henriette 133, 142 DeCaen, Vincent 9, 23, 36 DeCaen, Vincent 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 103, 109, 110, 112, 114, 117, 123, 124, 125, 142 Deo, Ashwini 213, 229 Diesing, Molly 169, 178, 179, 189 Dixon, Robert M.W. 186, 187, 189 Dobrovie-Sorin, C. 76, 82 Doron, Edit 76, 82, 87, 89, 101, 119, 124, 125, 126, 128, 130, 135, 136, 139, 140, 142, 145, 157, 166, 210, 211, 221, 226, 229, 230 Dorvel, Benjamin 2, 7 Dowty, David 202, 207 Dresher, B. Elan 9, 12, 17, 24, 27, 36, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59

- 978-90-04-44885-8

254 Edwards, Jan 22, 34 Edzard, Lutz 44, 58 Even-Shoshan, Abraham 53, 54, 55, 58 Fassberg, Steven E. 235, 251 Flemming, Edward 12, 20, 34, 36 Folli, Raffaella 211, 229 Gamliel, Ophira 178, 189 Garcia, Guilherme 15, 36 Gardiner, Alan 204, 207 Geller, Stephen A. 84, 101 Gesenius, Wilhelm 12, 36, 125, 130, 131, 135, 138, 139, 142; see also Kautzsch, Emil Gleason, H.A. 3, 7 Goad, Heather 15, 36 Goerwitz, Richard L. 9, 36 Goldberg, Lotus M. 90, 101 Goldenberg, Gideon 132, 133, 142, 236, 239, 246, 247, 248, 251, 252 Gordon, Cyrus H. 204, 207 Gordon, Matthew 12, 15, 16, 17, 18, 19, 20, 22, 31, 33, 36 Goshen-Gottstein, Moshe H. 160, 167 Grasso, Kevin 213, 229 Gray, George Buchanan 84, 96, 101 Griffiths, James 93, 101 Grimshaw, Jane 167, 178, 189 Gussenhoven, Carlos 33, 36 Hacquard, Valentine 158, 167 Halle, Morris 33, 35, 42, 43, 58, 122, 124, 126, 140, 142 Hammond, Michael 22, 36 Hannahs, Stephen J. 15, 36 Harbour, Daniel 103, 104, 108, 109, 112, 116, 123, 124, 125, 129, 133, 141, 142 Harley, Heidi 126, 140, 141, 142, 211, 219, 229 Haspelmath, Martin 209, 210, 212, 224, 229 Hatav, Galia 103, 104, 108, 117, 124, 125, 130, 131, 132, 133, 134, 142, 143, 144, 150, 167, 195, 207 Hayes, Bruce 15, 19, 20, 21, 22, 35, 36, 42, 47, 58 Heller, D. 76, 82 Hendel, Ronald S. 3, 7 Hessling, Vincent 2, 7 Heuven, Vincent J. van 22, 35 Himmelreich, Roman 27, 33, 36

index of authors Hoftijzer, J. 244, 251 Holmstedt, Robert D. 49, 58, 95, 100, 101, 196, 201, 207 Hopper, Paul J. 182, 183, 189 Hornstein, Norbert 204, 207 Horvath, Julia 215, 229 Hurvitz, Avi 234, 251 Hyman, Larry M. 19, 20, 37 Idsardi, William J. 42, 43, 58 Ivlieva, N. 81, 82 Johnson, K. 167 Jones, Charles 154, 167 Joosten, Jan 144, 167 Joüon, Paul 39, 40, 41, 54, 58, 194, 200, 205, 207, 215, 219, 229, 231, 236, 251 Kastner, Itamar 215, 221, 230 Kautzsch, Emil 229, 231, 251; see also Gesenius, Wilhelm Kayser, Samuel 13 Khan, Geoffrey 11, 13, 23, 37, 39, 58, 59, 239, 240, 244, 252 Kim, Yoo-Ki 132, 143 Kiparsky, Paul 153, 167 Koerner, Konrad 1, 2, 7 Koontz-Garboden, Andrew 215, 230 Kratzer, Angelika 112, 124, 130, 143, 218, 219, 221, 227, 228, 230 Kroeze, Jan H. 143 Labelle, Marie 126, 143 Lander, Yury 153, 167 Leander, Pontus 149, 166 Legate, Julie Anne 112, 124 Levin, Beth 216, 218, 224, 230 Liberman, Mark 33, 37, 42, 58 Lindblom, Björn 13, 20, 33, 37 Lundquist, Björn 215, 230 Malone, Joseph 42, 58 Maman, Aharon 1, 8 Manross, Lawrence N. 204, 207 Marantz, Alec 122, 124, 126, 140, 141, 142, 143 Marín, Rafael 220, 230 Marʾi, Abed al-Rahman 178, 189 McCarthy, John J. 27, 35, 37 McCloskey, James 204, 207

- 978-90-04-44885-8

255

index of authors McNally, Louise 220, 230 McShane, Marjorie J. 85, 86, 89, 101 Merchant, Jason 84, 85, 90, 101 Mester, Armin 178, 189 Mikulskas, Rolandas 192, 207 Miller-Naudé, Cynthia L. 84, 85, 86, 87, 88, 89, 90, 91, 92, 97, 98, 100, 101, 102 Mohanan, Tara 178, 189 Moon, Seung-Jae 20, 33, 37 Mor, Uri 235, 252 Moshavi, Adina 63, 83, 133, 142, 143 Moulton, Keir 153, 167 Munn, Alan S. 119, 120, 124 Muraoka, Takamitsu 39, 40, 41, 54, 58, 194, 200, 207, 215, 219, 229, 231, 236, 244, 251 Murguía, Elixabete 90, 102 Najman, Hindy 3, 7 Naudé, Jackie A. 143 Nespor, Marina 22, 35, 47, 59 O’Connor, Michael P. 3, 7, 83, 84, 102, 103, 108, 118, 124, 125, 131, 134, 135, 137, 138, 143, 200, 205, 208, 215, 219, 220, 230, 231, 252 Ostendorf, Mari 22, 38 Padgett, Jaye 10, 13, 34, 37 Pak, Miok 135, 143 Palmer, Frank R. 152, 167 Perrier, Pascal 22, 38 Pesetsky, David 167 Piñón, Christopher 220, 230 Pires, Acrisio 153, 160, 167 Podobryaev, A. 81, 82 Pollock, Sheldon 2, 7 Portner, Paul 135, 143, 151, 154, 167 Price, Patti J. 22, 38 Prince, Alan 12, 27, 33, 37, 40, 41, 42, 58, 59 Pylkkänen, Liina 219, 221, 230 Qimron, Elisha 13, 27, 37 Ramchand, Gilian 178, 179, 180, 189, 190, 204, 206, 215, 230 Rapaport, Tova 200 Rappaport Hovav, Malka 16, 17, 27, 33, 38, 42, 59, 209, 216, 218, 219, 222, 224, 230 Rendsburg, Gary A. 44, 53, 59

Reuland, Eric J. 153, 167 Revell, Ernest J. 9, 23, 27, 38, 39, 57, 59 Roberts, John R. 186, 190 Roettger, Timo 15, 22, 36 Rosen, Sara Thomas 178, 190 Ross, John R. 102 Rothstein, Susan 63, 66, 70, 76, 79, 82, 83, 130, 143, 191, 192, 194, 196, 198, 207 Ryan, Kevin M. 15, 17, 19, 38 Sagi, Hagit 15, 38 Saltzman, Elliot 22, 35 Sampson, Geoffrey 3, 7 Sanders, C. 2, 8 Schäfer, Florian 209, 212, 222, 226, 228, 229, 230 Screnock, John 83 Selkirk, Elizabeth O. 47, 59 Sharvit, Yael 81, 83 Shattuck-Hufnagel, Stefanie 22, 27, 38 Shlonsky, Ur 149, 166 Shmidman, A. 236, 252 Silber-Varod, Vered 15, 35, 38 Siloni, Tal 215, 229 Smolensky, Paul 33, 37 Sorace, Antonella 215, 230 Stateva, P. 81, 83 Steiner, Richard C. 66, 83 Steriade, Donca 19, 35, 37 Strauss, Tobie 47, 59 Sudo, Y. 81, 82 Tabain, Marija 10, 13, 22, 34, 37, 38 Tallerman, Maggie 186, 190 Taube, Moshe 178, 190 Téné, David 1, 8 Torrego, Esther 167 Traugott, Elizabeth 182, 183, 189 Truckenbrodt, Hubert 33, 38 Tungseth, Mai 215, 230 Turk, Alice E. 22, 27, 38 Turner, James 2, 3, 8 van der Merwe, Christo H.J. 131, 132, 143 Vergnaud, Jean-Roger 42, 58 Verstraete, Jean-Christophe 154, 167 Vogel, Irene 47, 59 Vries, Mark de 93, 95, 96, 101, 102

- 978-90-04-44885-8

256 Waltke, Bruce K. 83, 103, 108, 118, 124, 125, 131, 134, 135, 137, 138, 143, 200, 205, 208, 215, 219, 220, 230, 231, 252 Waltker, N. 244, 252 Watson, Wilfred G.E. 84, 102 Weitz, Tabea 2, 7 Wickes, William 50, 59 Wightman, Colin W. 22, 38 Williams, Briony J. 15, 24, 34 Williams, Ronald J. 231, 252

index of authors Wilson, Daniel J. 203, 208 Wiltshire, Caroline 13, 15, 35 Wright, W. 151, 168 Wurmbrand, Susi 148, 168 Ylikaski, Jussi 137, 143 Zanuttini, Raffaella 135, 143 Zewi, Tamar 242, 244, 247, 248, 252 Zwiep, I. 1, 8

- 978-90-04-44885-8

Index of Scriptures Genesis 1.1 1.21 1.25 1.26 1.27 1.29 1.29–30 2.3 2.16 2.17 3.1 3.3 3.4 4.8 4.14 4.15 4.18 4.23 4.24 5.1 5.2 5.3 7.1 7.2–3 7.6 7.11 8.3 8.4 8.5 8.12 8.13 9.6 9.7 9.11 9.15 10.16 11.5 11.6 11.8 12.9 12.19 13.15 13.17 14.9 14.23

111 99 99 201 195, 201 99 92 64 103, 132 177 87 241 125 189 149 148, 162 245 99 99 201 195 201 195 99 195 68 132, 150 69 115, 131 154 69 201 215 225 145, 146 51 156 184 215 131 203 112 174 62 216

15.2 15.13 15.21 16.3 16.15 17.1 17.8 17.13 18.1 18.6 18.11 18.18 18.19 18.28 19.1 19.8 19.19 19.21 19.35 20.18 21.3 22.3 22.17 22.22 23.9 24.5 24.18 24.21 24.31 24.32 24.47 24.54 24.67 25.1 25.25 25.29 26.13 26.34 27.3 27.12 27.19 27.37 27.42 28.2 28.12 28.14

195 134 50 195 56 106 195 106, 134 134, 150 172 158 134 155 88 185 159 216 156 156 134 56 170 134 50 195 134 171 87, 195 99 99 99 171 203 173 195 216 131 195, 202 134 87, 99 174 195 245 164 215 215

- 978-90-04-44885-8

258 Genesis (cont.) 29.3 30.2 30.4 30.9 30.19 30.31 30.32 31.3 31.4 31.15 31.41 32.12 32.29 32.29 33.2 33.9 33.18 34.8 34.12 35.29 37.1 37.2 37.8 37.33 38.11 39.1 39.6 39.9 39.23 40.15 41.32 41.34 41.35 41.43 41.45 42.6 42.9 42.15 43.2 43.3 43.32 44.5 44.17 45.8 47.18 48.14 49.1 49.11

index of scriptures

63, 216, 217 61 195 195 61, 72 215 105, 109, 113 132 99 195 216 132 90 99 99 235 195 195 195 195 132 197 132 134 195 157 90, 99 90 216 134 176 69 53, 99 121, 122 195 248 145 90 134 159 99 132 112 90, 99 99 94, 99 99 99

49.13 49.17 50.14 50.24 50.35 Exodus 2.14 2.19 3.7 3.9 3.19 5.12 5.13 5.21 6.3 6.8 7.28 8.11 8.12 8.25 9.2 9.19 12.9 12.13 12.18 12.38 12.42 12.48 13.3 13.18 14.1 14.9 14.14 15.16 16.4 16.8 16.24 17.1 17.14 18.2–4 18.4 18.6 18.13 18.18 19.1 19.13 20.8 20.12

99 99 171 134, 177 134

146, 157, 158 132 150 99 99 215 99 214, 215 193 195 215 121 215 165 216 216 99, 195, 198, 199 146 65 99 147 106 108, 135 195 215 197 216 160 87 99 214, 215 164 195 93, 94 193 99 159 134 65 104 108, 135, 145 135, 215

- 978-90-04-44885-8

259

index of scriptures 21.5 21.7 21.14 21.23–24 22.3 22.12 22.23 23.18 24.4 25.2 25.12 25.17 26.1 26.3 26.35 28.1 28.19 28.21 29.37 30.19 31.17 32.6 32.25 33.5 34.3 34.24 36.1 36.13 36.16 39.9 40.15 40.36–37 Leviticus 2.11 2.14 3.1 3.6 3.14 5.6 6.1 6.3 6.6 6.7 6.9 6.21 7.34 10.12 11.33

195 163 216 99 195 195 99 53 99 99 99 108 99 99 99 152 72 99 78 72 63 106, 118 242 247 215 216 99 99 99 195 163 146

195 195 195 195 195 195 195 197 195 106, 135, 152 195 213 195 195 213

14.17 14.21 15.31 16.1 16.2 16.5 18.21 19.6 19.16 19.19 19.24 20.2 23.9 23.17 24.43 25.14 25.21 Numbers 1.52 2.16 2.24 3.6 5.21 6.3 6.19 6.23 6.25 6.26 7.24 7.66 7.72 7.78 11.1 11.4 11.15 11.19–20 11.23 11.25 11.32 12.14 13.3 14.33 15.24 15.35 16.3 16.13 16.17 16.33

53 195 216 195 157 99 185 64 195 195 62 195 99 195 147 122 64

99 195 195 215 195 216 99 134 112 112 72 68 68 69 194 171 107 99 87 162 107 104 132 75 99 106, 134 195 106 99 195

- 978-90-04-44885-8

260

index of scriptures

Numbers (cont.) 17.4 17.23 18.14 18.15 20.21 21.2 21.15 21.28 21.29 22.16 23.7 23.19 24.4 24.5 24.6 24.16 26.6 26.65 28.11 28.12 28.13 29.19 29.25 30.3 31.3 31.54 32.3 32.16 32.32 33.38 35.33

199 215 195 164 157 134 215 99 99 157 98, 99 91, 92 195 99 99 195 245 99 195 195 195 195 195 122 195 195 195 99 195 66 99

Deuteronomy 1.1 1.17 2.24 3.18 3.26 5.3 5.12 5.16 6.17 7.18 7.22 8.2 8.9 9.7 9.21

216 247 184 195 161 99 108, 135, 152 135 107 108, 115 157, 162 87 247 161 139, 161, 200

9.27 10.22 11.29 13.1 13.15 14.9 15.2 15.5 15.7–8 15.12 15.13 15.18 15.22 16.1 16.3 17.7 19.3 21.14 22.1 22.4 22.9 22.24 24.9 24.9 24.12–13 25.9 25.17 26.14 26.19 27.1 28.1 28.13 28.23 29.7 29.13–14 29.22 30.3 30.9 31.21 31.27 32.7 32.25 33.1 33.18 33.28 Joshua 1.13 2.1

145 197 99 99, 132 137 9 106 132 132 195 195 195 99 108, 135 164 99 162 132 132 132, 176 195 92, 93 108 135 132 214 135 197 195 108, 135 195 87 99 195 99 215 215 176 195 99 91 99 99 99 99

108, 135 215

- 978-90-04-44885-8

261

index of scriptures 6.3 6.13 6.16 7.7 9.2 10.16 14.4 14.7 17.3 17.18 20.7

134 105, 114 62 107, 115, 116 151 62 90 148 90, 240 216 99

Judges 1.8 2.22 4.14 4.24 6.11 6.37 7.1 7.19 7.21 8.12 9.2 9.8 10.8 14.9 20.35 20.39

48, 49 87 99 131 216, 228 98 157 106, 119 228 216 162 134 99 131 61 157

1Samuel 1.9 3.5 3.12 5.3 6.12 8.2 8.9 9.2 10.15–16 11.11 14.13 14.19 14.27 15.15 15.35 17.14 17.33 17.55

139 174 134, 151 218 131 56 132 64 133 118 99 131 156 155 157 82 247 154, 155

18.6 18.12 18.25 19.2 19.3 19.15 20.41 21.7 22.22 23.13 23.26 24.2 25.18 25.28 25.3 25.7 26.2 26.25 27.1 30.17 30.22 2 Samuel 1.6 1.11 3.16 3.34 5.1 5.19 6.11 6.22 8.2 8.6 11.7 11.25 15.3 15.8 15.32 16.22 17.2 17.8 17.14 17.17 18.29 19.6 21.2 21.22 24.12 24.13

99 157 157 157 89 153 216 90 243 157 99 134 61 134 238 216 99 132 157 99 90

134 89, 99 131 160 131 133 65 241 105, 113, 134 164 119 216 131 228 239 215 216, 236 242, 243 157 159 90 216 90 245 135, 137 149

- 978-90-04-44885-8

262

index of scriptures

2Samuel (cont.) 24.24 132 1Kings 3.7 3.14 6.1 6.38 7.12 8.52 10.9 14.25 15.1 15.28 16.18 16.24 17.1 19.15 20.31 20.37 20.39 21.29 22.12 22.14 22.18

157 215 80 65 64 163 161 63, 68 67 67 68 56 90 173 243 131, 195 152 163 216 240 90

2Kings 2.3 2.11 4.2 5.2 5.11 5.15 5.17 6.5 6.11 7.1 11.11 11.18 18.3 18.4 18.13 18.32 19.27 21.20–21 22.18

216 114 90 194 106 90 90 246 232, 234 90 134 107, 137 245 110, 111 70 114 161 163 241

Isaiah 1.3 1.4 1.11 1.18 1.21 1.26 1.27 3.3 3.16 5.5 6.1 7.11 8.2 10.31 13.17 17.11 19.8 19.22 25.12 28.9 32.17 33.21 35.6 40.12 43.18 45.13 52.8 52.15 55.1 59.1 59.9

96 238 55 215 89 84, 95 97 237 105, 108, 118 215 216 134 216 216 216 217 212, 213 131 215 216 137 239 86 54 96 91 156 96 215 87 91

Jeremiah 2.2 2.11 2.26 6.15 10.18 12.17 15.15 20.14 22.19 25.29 33.8 33.26 36.22 47.4 48.41

156, 157 216 216 216 216 131 156 164 134 129 54 228 246 159 216

- 978-90-04-44885-8

263

index of scriptures 48.44 49.12 49.36

228 129 228

Ezekiel 1.26 17.24 18.32 20.14 26.7 27.26 31.15 39.16 43.2 45.14

241 215, 217 217 147 78 225 213 56 215 55

Hosea 5.6 6.6 11.1

214 87 247

Joel 1.1 4.1

215 228

Amos 1.3 2.8 5.18 9.8

88 215 88 125

Jonah 1.7 1.12 1.13

232, 234 233, 234 216, 217

Micah 4.12 7.1

216 86

Nahum 2.3

215

Zechariah 9.5 12.13

91 116

Malachi 2.14

216

Psalms 2.1 2.4 10.1 15.1–5 18.29 18.33 18.42 18.49 19.11 20.8 22.27 23.1 24.3–4 25.17 32.8 33.1 35.8 46.3 47.5 49.8 49.17 51.9 54.7 57.8 59.16 60.2 73.1 76.5 85.4 87.7 90.6 95.1 97.6 100.2 104.25 109.17 118.13 119.96 126.1 132.8 137.1 139.6 148.6

216 86 215 89 215 149 91 149 87 85, 86, 98 172 235 89 215 216 54 240 216 39 125 216 55 228 216 228 39 228 87 217 54 215 237 88 88 237 157 159 54 154 86 145 87 55

Job 11.3 13.17 14.9

216 107 215, 216, 217

- 978-90-04-44885-8

264

index of scriptures

Job (cont.) 21.2 26.1 28.1 29.2 31.29 37.23 39.12 42.3

107 55 87 216 149 55 228 215

Prov 3.12 10.17 12.14 17.16 19.7 25.27 26.23 29.27 30.4

39 216 228 55 54 137, 138 55 238 56

Ruth 1.1 1.21

160 194

Song 1.1 1.6 1.7 3.7 8.12

78 233, 234 234 233, 235 233, 234, 235

Ecclesiastes 8.17

234, 235

Lamentations 1.5 1.18

55 247

Esther 1.12 1.17 1.19 4.16 9.15

51, 52 51 52 32 67

Daniel 11.32

215

Nehemiah 6.5 8.11 9.19

62 216 246

1 Chronicles 4.27 11.9 12.14 12.18 24.13–18 25.18–31 27.14–15

216 131 66 114 66 66 66

2 Chronicles 33.19 34.8

216 67

- 978-90-04-44885-8