Johnson 2004 - CDLI - UCLA

UNIVERSITY OF CALIFORNIA 

Los Angeles 

In the Eye of the Beholder: 

Quantificational, pragmatic and aspectual features 

of the *bi- verbal prefix in Sumerian 

A dissertation submitted in partial satisfaction of 

the requirements for the degree of Doctor of Philosophy 

in Near Eastern Languages and Cultures 

by 

Justin Cale Johnson 

2004

The dissertation of Justin Cale Johnson is approved. 

––––––––––––––––––––––––––––––––– 

Antonio Loprieno 

––––––––––––––––––––––––––––––––– 

Brent Vine 

––––––––––––––––––––––––––––––––– 

Robert K. Englund, Committee Chair 

University of California, Los Angeles 

2004 

ii

To Grace 

iii

TABLE OF CONTENTS 

Acknowledgements vii 

Vita xiv 

Publications and presentations xv 

Abstract xvi 

Abbreviations and translations 1 

Preface 4 

Chapter 1: The BNBV inal distributional class 22 

Section 1.1: Directive case and the oblique object hypothesis 26 

Section 1.2: The bare noun + *bi-√ verb [BNBV] diagnostic 33 

Section 1.3: BNBV inal as a low source applicative 46 

Section 1.4: BNBV inal verbs of perception 56 

Section 1.5: BNBV inal verbs of adversity 61 

Section 1.6: Other BNBV inal verbs 75 

Section 1.7: Possible motivations for the BNBV inal distributional class 80 

Section 1.8: Ergative possessors and the transfer of possession model 86 

Chapter 2: Decomposing lexical aspect: Telicity, the particle-verb 92 

construction and secondary predication 

Section 2.1: Three aspects of predication: Grammatical, lexical 97 

and secondary 

Section 2.2: D-stem/Gtn-stem reduplication in Sumerian 105 

iv

Section 2.3: BNBV inal reduplication: Depictive, resultative or tertium quid 120 

Section 2.4: The curious morphosyntax of progressive achievements 141 

Section 2.5: Introducing telicity, ßu, and inalienable possession under 158 

nominalization 

Section 2.6: BNBV inal as a particle-verb construction 176 

Section 2.7: The link between the BNBV construction and the rise of 188 

ergativity 

Chapter 3: Diagnosing presupposition in Sumerian 192 

Section 3.1: Presupposition and assertion, topic and focus 196 

Section 3.2: Focus and focal presupposition 206 

Section 3.3: Topicalized and identificational *bi-√ prefix verbs in *mini-√ 213 

Section 3.4: The Definiteness Effect 243 

Section 3.5: Head-internal relatives revisited 251 

Chapter 4: Focus-affected constituents in the *XP nam bi-√ construction 261 

Section 4.1: Focus in Sumerian 263 

Section 4.2: Focus-affected *bi-√ prefix constructions 268 

Section 4.3: Focus-affected compound verb constructions 286 

(excluding BNBV inal) 

Section 4.4: Focus-affected BNBV inal predicates 291 

Section 4.5: Seeing something that is no longer there: The demonstrative 299 

*-re and visual perception 

Section 4.6: Contrastive/identificational focus and the thetic/categorical 307 

opposition in Sumerian 

v

Chapter 5: Possession, animacy and the rise of ergativity in Sumerian 314 

Section 5.1: Inalienability and the historicity of the case-marking 315 

schemata 

Section 5.2: Possession and the development of ergativity 319 

Section 5.3: Conclusions and prospects 331 

References 333 

vi

ACKNOWLEDGEMENTS 

Each dance and its music belong to a time and place. It can be borrowed 

elsewhere, or later in time, but it will never be in its moment again. 

When these little cultural blooms are past, they become ethnic or nostalgic, 

but never quite fully present—manifesting the web of their original connections 

and meanings—again 

— Gary Snyder 

There are many people who have helped me in one way or another over the years. I 

cannot mention them all here, but I would like to thank a few people who made 

particularly lasting impressions on me at each stage of the journey. 

At UC Davis, where I did my undergraduate degree, two of my teachers were 

particularly influential: my undergraduate advisor, Naomi Janowitz, introduced me to the 

History of Religions, instilled in me the critical eye that is so often missing from work in 

the humanities, sent me to Cairo for a year to cure me of my parochialism and tried to get 

me to think twice about going to the Divinity School at the University of Chicago; Aram 

Yengoyan let me into the graduate courses that he taught when I was eighteen years old 

and had no right to be there, and my decision to go to graduate school was largely a 

product of the impression that those classes left on me. Kendall House, a graduate student 

vii

at UC Davis, talked Anthropological Theory with me whenever he had a spare moment 

and remained my theoretical interlocutor for years. Frances and Anniken taught me 

many, many things, and Jonah and Amira taught me everything else. 

The Divinity School at the University of Chicago was a strange and wonderful place 

and, through the good offices of Arthur Droge, Dennis Pardee and Michael Silverstein, I 

actually managed to learn a few things while I was there. Jonathan Z. Smith talked me 

out of being several things and into being a historian, all the while dispensing pearls of 

great price that I couldn’t have found anywhere else, and Pia found me a job at the 

Faculty Club and spent countless hours talking to me about History of Religions and 

Jonathan Z. Smith. 

When I came to UCLA, Antonio Loprieno welcomed me with open arms, gave me 

enough money to live on and waived every requirement that he could so that I would 

have more time to work on the languages that I wanted to know and came to UCLA to 

learn. Thanks to Giorgio Buccellati for introducing me to cuneiform and teaching me 

Akkadian and to Yona Sabar for teaching Aramaic and helping me understand the 

bureaucracy at UCLA. Bill Schniedewind spent hours and hours helping me to 

understand the Hebrew Bible and has continued, throughout my time at UCLA, to 

support my work in all kinds of ways. Likewise, whether dispensing sage advice or 

explaining the subtleties of topic and focus in Egyptian, Loprieno’s efforts on my behalf 

have been unceasing. Michael Fishbein and Michael Cooperson welcomed me into their 

Arabic classes, Elizabeth Carter taught me about seals and archaeology and Elam, and 

Paul Kroskrity, about language and culture and ideology. Asif Agha picked up my 

viii

education in anthropological linguistics, pragmatics and semiotics where Silverstein had 

left off and was always generous with his time and his optimism. 

In my third year at UCLA, I discovered Sumerian in the classes of Bob Englund and 

have spent the vast majority of my time here at UCLA ever since preoccupied by the 

languages, culture and history of Mesopotamia. Bob has done everything in his power to 

make me into a good Assyriologist. In the meantime, he has gone to a great deal of 

trouble to ensure that I was well fed, had a roof over my head and was able to ignore the 

daily rush and concentrate on learning to read cuneiform. Over the years, he has listened 

with some amusement I imagine to my theories, all the while insisting that I master, to 

the best of my ability, the philological and historical disciplines that constitute 

Assyriology as a field of study. For all of these things, I am eternally grateful. Thanks to 

Klaudia as well for always making me feel welcome when I was out visiting the boss. It 

is telling that Englund thought it important that I spent some time studying with at least 

one other leading Assyriologist so as to broaden my exposure to the different methods 

and the best practices of other leaders in the field. When I applied for and received a two- 

year fellowship to study with Professor Maekawa at Kyoto University, Englund 

supported it even through the territory was unfamiliar and my departure left him with no 

graduate students to help out with the preliminary stages of the Cuneiform Digital 

Library Initiative. 

I would like to take this opportunity to express my thanks to the Japanese Ministry of 

Education for providing me with two years of funding that allowed me to study Sumerian 

full-time in Kyoto without giving a thought to my financial situation; at the same time, I 

ix

would also like to express my heartfelt thanks to Professor Maekawa for acting as my 

sponsor and tutor throughout my time in Japan. On numerous occasions, he welcomed 

me into his home and would often answer my questions in excruciating detail for hours 

on end on an almost daily basis. My fellow students in Maekawa’s kenkyushitu, in 

particular Wakaha Mori, Naoko Ohgama and Harumi Horioka, helped me in so many 

ways and on so many occasions that I will not try to list them here—I will always be in 

their debt. Maekawa’s methods and those inculcated in my coursework at UCLA were 

quite different, but I feel that I have benefited tremendously from the best aspects of both 

Germanic scholasticism and detail-oriented philology and the Japanese emphasis on rapid 

and comprehensive reading of the entire corpora in conjunction with excruciatingly 

detailed lexical investigations of key terms. 

Professor Maekawa and Wakaha would regularly confront my overly optimistic, glib 

theories with their detailed and intricate knowledge of Sumerian grammar; time and again 

my theories were sent back to the drawing table, but in the process I came to appreciate 

the seriousness and diligence of the small band of Sumerologists in Japan. Thanks to Jun 

Ikeda, Tohru Maeda and all those who participated in the Sumerology Workshop 

meetings where I presented my first papers on Sumerian as well. 

When the rigors of Japanese, Sumerian or Sumerian translated into Japanese became 

too much for me, I would often find myself in downtown Kyoto having a drink with 

either Petra or Akhmar or both, or more likely, having dinner with a circle of friends 

including Shito, Haruko, Yoshiko, Hanako, Takada, Sachiko, Yamamoto and Imai. They 

x

adopted me, taught me Japanese, fed me strange food and otherwise kept sane during my 

stay in Japan, not to mention Christmas in Beijing. 

When I returned to UCLA in 2001, the department welcomed me back and Bob 

Englund had a position in the newly founded Cuneiform Digital Library Initiative waiting 

for me upon my return. Special thanks to Diane Abugheida who was invaluable in 

providing advice and calm reassurance during my doctoral exams and the preparation of 

this dissertation. Thanks to the Kinsey Administrative Group, the Institute for Social 

Science Research and the staff at the Young Research Library, particularly those who 

have dealt with my interlibrary loan requests. 

During my years at UCLA, I have been blessed with a wonderful group of friends: 

Lynn Dodd and her husband Ted have been giving me advice, inviting me over to their 

house for dinner, reading manuscripts, and helping me out in a hundred other ways for 

nearly ten years; my old friend Mel moved down to LA with me, shared an apartment 

with me for several years and has always, always, always been there when I needed her; 

Alessandra’s exuberant desire to know made me want to know too and then in the end, 

she also taught me that knowledge is not the most important thing in the world; Jess and 

John and Brian and Emily made sure that I had good sashimi at least once a month. Yoko 

and Adam have done so many things to help me and have stood by my side at all the 

crucial moments. My fellow Assyriology students, Jennifer Rashidi, Adam Johnson, John 

Lynch and Jeff Szuchman have always been wonderful and generous compatriots and the 

Hebrew Bible students in the department always made me feel welcome and one of them 

even married me. 

xi

Thanks to Madeleine Fitzgerald, the post-doc at CDLI, for all manner of help, advice 

and gossip over the years. I shared an office and almost every class with Jacob Dahl for 

years and years: we complained (sometimes about each other) and laughed about just 

about everything and sometimes even fought, but we always remained and still remain 

the best of friends. Key components of the hypothesis that I present in this dissertation 

were inspired by the content and/or readings in several classes taught by Dominique 

Sportiche and Hilda Koopman, but they should not be held responsible for the theoretical 

infelicities that are sure to be found herein. Over the years at UCLA, many people have 

sent me offprints, unpublished manuscripts and other sources of data and interpretation 

that I could not have acquired on my own: Alfonso Archi, J. Bauer, the late Jeremy 

Black, Jerrold Cooper, Bram Jagersma, Fumi Karahashi, Piotr Michalowski, Gonzalo 

Rubio, Gebhard Selz, Walter Sommerfeld, P. Steinkeller, H. Vanstiphout, Niek Veldhuis, 

Aage Westenholz and Claus Wilcke. 

I would like to thank the members of my committee, Bob Englund, my advisor, 

Antonio Loprieno and Brent Vine, for reading drafts and answering questions on short 

notice and under the constraint of unreasonable deadlines. Brent Vine in particular has 

been wonderfully generous with his time, even going so far as to read through the draft 

and send me several important corrections and references while on sabbatical in Kyoto. 

Piotr Michalowski read through nearly the entire dissertation and sent me pages of pages 

of comments and criticism that have only been able to incorporate in part. Fumi 

Karahashi read through much of the dissertation while on vacation in Europe and 

Gonzalo Rubio did likewise even as he was attempting to meet a deadline for his own 

xii

forthcoming book. Both provided me with important comments and suggestions. Lynn 

Dodd and Adam Smith read through and provided me with extensive comments on a 

chapter on the history of writing that did not make it into the final version. To all those 

who sacrificed time for their own work and/or time with their families in order to help me 

with my dissertation, I offer my humble thanks. 

My parents have supported me in every possible way throughout my education and, 

even when my interests turned to long dead civlizations and the most obscure of 

academic disciplines, their support for my work never wavered. I could never have 

finished (or even started) without them. Thanks as well to my brother Greg, Julie and the 

kids for all kinds of encouragement big and small. 

I met Grace about the same time that I started writing my dissertation and she has 

endured its difficulties and its distraction throughout our courtship, engagement and 

marriage with patience and good humor and love. Besides answering questions about 

Korean and Chinese at all hours of the day and night, she held me together when I was 

flying apart and restored me when there was nothing left. I was inspired by her in more 

ways than she could possibly know and I would never have finished without her. 

xiii

VITA 

July 13, 1971 Born, Chico, California 

1993 B. A., Religious Studies 

University of California, Davis 

Davis, California 

1995 M. A., History of Religions 

University of Chicago 

Chicago, Illinois 

1996 Departmental Scholarship 

Department of Near Eastern Languages and Cultures 

Los Angeles, California 

1997 Del Amo Scholarship 



1998-1999 Research Assistant 



1999-2001 Monbusho Fellowship 

Kyoto University 

Kyoto, Japan 

2000 M. A., Near Eastern Languages and Cultures 

University of California, Los Angeles 


2001-2004 Research Assistant 

Cuneiform Digital Library Initiative 


xiv

PUBLICATIONS AND PRESENTATIONS 

Johnson, Cale. December 4, 1999. Global case-marking in Sumerian: /mu-/, /ba-/, /i-/. 

Paper presented at the 6th West Asian Linguistics Meeting, Kyoto, Japan. 

––––––. December 23, 1999. Evidence of antipassivization in Sumerian. Paper presented 

at the 40th Sumerology Workshop, Tokyo, Japan. 

––––––. 2000. Evidence of antipassivization in Sumerian. The Bulletin of the 

International Institute for Linguistic Sciences Kyoto Sangyo University 21: 205-240. 

––––––. July 1, 2000. Medium as trope: On the poetics of “The debate between Bird and 

Fish”. Paper presented at the 7th West Asian Linguistics Meeting, Kyoto, Japan. 

––––––. August 31, 2000. The verbal infix /-ni-/ in the Sumerian of the Old Babylonian 

Period: Mediopassivization and pragmatic constraints on its interpretation. Paper 

presented at the 41st Sumerology Workshop, Kyoto, Japan. 

–––––– and Wakaha Mori. March 27, 2001. Detransitivization in Sumerian. Paper 

presented at the 42nd Sumerology Workshop, Tokyo, Japan. 

––––––. July 2, 2002. Complex graphemes in the proto-cuneiform corpus and the 

problem of phonological reconstruction. Paper presented at the 48e Rencontre 

Assyriologique Internationale, Leiden, The Netherlands. 

––––––. October 21, 2002. The invention of writing in Mesopotamia. Public lecture at 

Andrews University, Berrien Springs, Michigan. 

––––––. April 4, 2003. Sound/Image: The multimodal poetic structure of Bird and Fish 

102-117. Paper presented at the American Oriental Society, Nashville, Tennessee. 

––––––. 2004. Two Ur III Tablets from the Tulare County Library. Cuneiform Digital 

Library Bulletin 2004/2 (http://cdli.ucla.edu/pubs/cdlb/2004/cdlb2004_002.html). 

––––––. March 13, 2004. Decomposing lexical aspect in Sumerian: BNBV(inalienable) 

lexical class, particle-verbs and telicity. Paper presented at the American Oriental 

Society, San Diego, California. 

––––––. April 23, 2004. Diagnosing presupposition in Sumerian. Paper presented at the 

Sprache-Aegyptologie Roundtable, Basel, Switzerland. 

xv

ABSTRACT OF THE DISSERTATION 

In the Eye of the Beholder: 

Quantificational, pragmatic and aspectual features 

of the *bi-√ verbal prefix in Sumerian 

by 

Justin Cale Johnson 

Doctor of Philosophy in Near Eastern Languages and Cultures 

University of California, Los Angeles, 2004 

Professor Robert K. Englund, Chair 

Traditional models of Sumerian grammar suggest that the primary function of the *bi-√ 

verbal prefix is as a marker of locative agreement of one kind or another, but these 

models have, as a rule, described the *bi-√ prefix in purely morphological terms and 

failed to elucidate syntactic relations that hold between the *bi-√ prefix and other 

elements of the clauses in which it occurs. One of the consequences of the traditional 

avoidance of syntactic questions in the study of Sumerian is that almost nothing is known 

about several grammatical phenomena that appear to be coded through syntactic means 

such as definiteness, existential quantification, and the pragmatic opposition between 

xvi

topic and focus. This dissertation investigates the *bi-√ prefix as part of a larger effort to 

identify some of these basic morphosyntactic oppositions in the text-artifactual record of 

the Sumerian materials that originate from Old Babylonian Period (ca. 2000-1600 BCE) 

in Mesopotamia. I argue that the occurrence of a bare inalienable noun immediately to 

the left of a *bi-√ prefix verb forms a distinctive morphosyntactic construction which can 

be identified on the basis of both morphosyntactic and semantic critieria as a low source 

applicative construction. 

The identification of this subset of the occurrences of the *bi-√ prefix as a low 

source applicative construction has broad implications for other parts of Sumerian 

grammar, in particular, the development of ergativity in the language. The low source 

applicative construction is limited to a group of verbs that includes verbs of perception 

and adversity and can be juxtaposed to a type of causative construction in which a bare 

alienable noun occurs immediately to the left of the *bi-√ prefix. This opposition stems 

from the fact that, in the low source applicative construction, the ergative noun phrase is 

the possessor of the inalienable noun rather than the agent of the clause. In order to 

elucidate several possible correlations between the clausal possessive constructions 

underlying the use of the low source applicative and the development of ergativity, 

several other issues are also dealt with in considerable detail in the balance of the 

dissertation: the use of possessive pronominal suffixes to code definiteness and 

topicalization as well as the role of indefiniteness and the definiteness effect in verbs of 

perception, head-internal relative clauses and contrastive focus constructions. 

xvii

Abbreviations and Translations 

Assyriological abbreviations follow the list made available on the internet by CDLI at the 

following address: (http://cdli.ucla.edu/Tools/abbrev.html). All other non-Assyriological 

abbreviations are listed below: 

Abs Absolutive case 

Acc Accusative case 

Agr Agreement 

Alla Allative 

Anim Animate 

AP Adjective Phrase 

Appl Applicative 

Asp Aspect 

BNBV Bare Noun Bi-Verb construction 

BNBV inal 

Cause Causative 

Cop Copula 

Dat Dative case 

Decl Declarative mood 

Def Definite 

Dir Directive case 

DistAdv Distal Adverb 

Inalienable Bare Noun Bi-Verb construction 

DP Determiner Phrase 

Dur Durative grammatical aspect 

Erg Ergative case 

1

FocusP Focus Phrase 

ForceP (Illocutionary) Force Phrase 

Gen Genitive case 

GN Geographical Name 

HIRC Head-Internal Relative Clause 

Imprf Imperfective grammatical aspect 

Inan Inanimate 

Indef Indefinite 

Intrans Intransitive 

IO Indirect Object 

LoAp Low Applicative 

Loc Locative case 

LocTerm Locative-Terminative case 

Masc Masculine 

Neg Negative 

Nmz Nominalizer 

Nom Nominative case 

NP Noun Phrase 

Obl Oblique case 

OO Oblique Object 

Perf Perfective grammatical aspect 

PerfPrt Perfective Participle 

Pl(ur) Plural 

PN Personal Name 

Poss Possessive 

Pres Present tense 

ProAnim Animate Pronominal agreement marker 

ProInan Inanimate Pronominal agreement marker 

PP Prepositional/Postpositional Phrase 

2

Prt Participle 

Redup Reduplication 

Rel Relative marker 

Sg Singular 

Spec Specifier 

Subord Subordinate marker 

Top Topic 

TopicP Topic Phrase 

TP Tense Phrase 

Trans Transitive 

Vent Ventive 

VP Verb Phrase 

XP Phrase of unspecified type 

As a rule, the translations of Sumerian were initially taken from ETCSL or a published 

edition of the passage in question; it should probably noted that ETCSL translations are 

also, as a rule, taken from past editions of the text in more or less anonymous fashion. 

Once these translations entered the manuscript they began a process of transformation so 

that, as they stand, they are curious hybrids of the conventional wisdom of previous 

generations of Assyriologists and my own insights and misunderstandings. The 

translation of a passage from Streck 1999 on pages 129 and 130 is my own except for one 

phrase that Prof. Streck kindly offered me a translation of. In the examples, one asterisk 

indicates that the example is hypothetical or that the segment in question is a bound 

morpheme and would not ordinarily occur alone; two asterisks indicate the example in 

question is ungrammatical. 

3

Preface 

Before turning to the first chapter I would like to spend a few moments discussing the 

role of linguistic theory and the absence of tree diagrams and other kinds of linguistic 

formalism in this dissertation as well as providing a restatement of the hypothesis to be 

investigated. 

Theoretical preliminaries 

One of the most surprising aspects of the linguistic investigation of archaeologically 

recovered languages is the particular type of linguistic theory that is most often used in 

such investigations, primarily “broad” or “inclusive” typology of the Greenbergian 

variety and “psychological” and/or “extreme” functionalism (cf. Nichols 1984, 100-101, 

103) with few if any formal or distributional criteria whatsoever. These theories are not, 

for the most part, predictive, but, on the contrary, merely act to confirm the plausibility of 

a reconstruction that is based on either philological investigation or analogies based on 

the set of languages familiar to the analyst. The actual method of investigation often boils 

down to a rather selective reading of recent typological descriptions followed by an 

attempt to match a particular construction in the dead language with the linguistic type 

that most closely resembles it. The “identification” is rarely tested and distributional 

criteria that might link the construction in question to other morphosyntactic 

distributional patterns are usually omitted. I have attempted, wherever possible, to avoid 

4

such “typological/psychological” functionalism, but I have not always succeeded. But 

even after theories that do not make use of formal diagnostics are set aside, many 

competing theories remain to choose from. 

There are worlds of difference between the two main camps in syntactic research, 

formalists and functionalists, or so it seems at times, particularly in their respective 

treatment of pragmatic phenomena such as topic and focus, but recent developments in 

formalism such as Rizzi’s work on the fine structure of the left periphery (Rizzi 1997) 

have introduced large numbers of essentially “pragmatic” syntactic positions or landing 

sites within the left periphery and attempted to capture pragmatic phenomena in their 

syntactic trees. Recent work on the formation of verbal complexes (Koopman and 

Szabolcsi 2000, Julien 2002) now regularly moves the entire clause into the left periphery 

with various pieces of phrase structure moving from traditional syntactic positions such 

as verb phrase (VP) and tense phrase (TP) into the left periphery and ending up in places 

like (illocutionary) force phrase (ForceP), topic position (TopicP) and focus position 

(FocusP). In recent formalist work, one is left with the impression that pragmatic 

categories are an essential part of syntactic structure. But the view that pragmatic 

phenomena affect syntactic categorization is, in fact, not new at all: functionalists have 

long maintained just such a point of view and the inclusion of such “pragmatic” syntactic 

positions in recent formalist work amounts to an admission that the functionalists were, in 

certain respects, right all along. This paradigm shift in formalist work, which can be 

attributed, above all, to Kayne’s work on binary branching and antisymmetry (Kayne 

5

1994), has produced a degree of compatibility with certain schools of functionalism such 

as Role and Reference Grammar (Van Valin 1993). 

The fact that the text-artifactual record of any particular archaeologically recovered 

language is necessarily quite limited and of inconsistent quality would seem to require 

that highly predictive linguistic theories take the lead role. The two theories that have 

most influenced the hypotheses and argumentation in this dissertation are (i) Role and 

Reference Grammar and in particular work on case-marking and clause-linkage by 

Silverstein (1976; 1993a) and Agha (1993) and (ii) recent formalist work carried out on 

the basis of Kayne’s antisymmetry (1994) and, in particular, the generative literature on 

so-called “double object” constructions, culminating—at least for the time being—in 

Pylkkänen’s recent dissertation (2002) on low applicative constructions. My own 

theoretical orientation remains structural-functionalist and those functionalist theories 

that continue to make use of formal diagnostics for distributional classes are, in my view, 

the most trustworthy. But the vast majority of recent research on the several topics that 

this dissertation is about have been carried out almost entirely within generative 

linguistics, notably investigators working within Distributed Morphology (see Marantz 

1998 for orientation) and those investigating the right periphery—the region below, for 

example, 31 AspSingCompletive1 in Cinque’s cartography (Cinque 1999)—and its 

association with lexical aspect and quantization (Tenny 2000; Filip 2000). Although 

Pylkkänen’s work and theoretical works by a few others are discussed from time to time 

in this dissertation, the formal representations that have taken on such importance in 

syntactic investigation in the past few decades have been eschewed. 

6

There are several reasons for not using tree diagrams or other methods of 

representing detailed hierarchical phrase structure in this dissertation. The most important 

reason is that the kind of relatively subtle and contrastive semantic and pragmatic 

differences that would—under normal circumstances—provide the basis for choosing one 

tree diagram over another (types of syntactic binding, quantifier scope, etc.) are simply 

unavailable. Given the empirical constraints on an archaeologically recovered language, 

we can never be sure of the precise semantic or pragmatic features of any sentence—all 

the glosses and translations in this and all other investigations of Sumerian morphosyntax 

are only approximations. There is no way of producing additional examples, introducing 

measured variation in form and testing for grammaticality through elicitation with a 

native speaker. Thus, in a special sense, any tree diagram or other formal representation 

is deceptive in that it suggests that we know more about Sumerian syntax than we 

actually do. Although generic formal representations would, at least in certain places 

along the way, have made the argumentation more precise, they would also have 

remained entirely meaningless to the vast majority of those who investigate the Sumerian 

text-artifactual record. 

Instead of attempting to import a huge amount of linguistic theory and its associated 

terminology and formalism into the Assyriological discussion, I have chosen to follow a 

rather different method in presenting my findings. Since the most important component 

of any investigation of Sumerian morphosyntax is the identification of formal criteria that 

differentiate distributional classes in Sumerian (rather than the scholarly metalanguage of 

translations and glosses that are applied to the Sumerian forms), the description of each 

7

grammatical phenomenon begins with a purely formal criterion on the basis of which a 

set of examples can be identified without recourse to the meaning of the examples. Once 

the formal criterion and the examples that meet the criterion have been isolated, I attempt 

to characterize the set of examples on the basis of both traditional philological practice 

(including the use of analogous constructions in English and other well-known and 

typologically similar languages such as Korean and Japanese) as well as the most recent 

theoretical discussion of the relevant phenomenon in disciplinary linguistics. The 

theoretical discussions often include predictions about special restrictions on certain 

distributional classes and postulate associations with related phenomena and I have used 

such predicted behaviors to form hypotheses about related distributional classes in 

Sumerian. Much of the theoretical argumentation is, however, abbreviated or sometimes 

omitted entirely where I felt that the Sumerian materials provided sufficient justification 

for the meanings that I hypothesize. Ultimately, the only real justification for positing a 

morphosyntactic category (whether lexical classification, for example, or derivational 

morpheme) in Sumerian is the degree to which the hypothesized meaning of a particular 

linguistic form makes better and more detailed sense of the particular contexts in which it 

occurs; any linguistic theorizing or typological conformity merely provides a set of 

predictions which constrain the formation of hypotheses, but have no real effect on their 

validity in the Sumerian text-artifactual record. 

One of the more troubling aspects of recent work on Sumerian morphosyntax is that 

it has remained persistently morphological in orientation and avoided, wherever possible, 

syntactic argumentation or investigation. Although the best traditions of Americanist 

8

descriptivism and structural-functionalism in Europe have always treated morphology 

and syntax as, essentially, two varieties of marking within the same distributional system, 

the tradition in Sumerology grows out of the philological practices of the late nineteenth 

century and tends to value only morphological oppositions. At present, as noted above, 

certain varieties of functionalism such as Role and Reference Grammar [RRG] and 

certain varieties of formalism such as Distributed Morphology [DM] are actually in 

agreement to a great degree and differ primarily in terms of the geometric model that 

each adopts for the formal representation of syntactic phenomena: RRG adopts multiple, 

partially independent formalisms for distinct phenomena such as surface syntactic 

constituency, argument and focus structure, and case-marking whereas DM linearizes 

each of these phenomena into a single syntactic tree and invokes massive leftward 

movement to ensure that constituents travel through the nodes in the tree that code 

information such as focus structure and the like. The point is that the differences between 

these theories is relatively minor (at least from the perspective of the current state of 

Sumerian grammar), whereas they share the central assumption that there is no separate 

morphological subsystem in the grammar and that the entire morphosyntactic system is 

really just a syntactic system in which some distributional classes are coded by segmental 

morphology, while others are coded through configuration. The import of such an 

observation is that those interested in detailed morphosyntactic description of Sumerian 

would do well to attempt investigation of Sumerian on the basis of syntactic distribution 

as understood within either RRG or DM rather than the traditional reliance on inherited 

9

morphological oppositions that are based, above all, on the lexical and grammatical 

observations of Old Babylonian literati who were not native speakers of the language. 

10

The hypothesis 

Two unresolved problems with present-day reconstructions of Sumerian grammar have 

troubled me since I first came into contact with the language: (i) the correlation or lack 

thereof between topic/focus marking and (in)definiteness and (ii) questions of 

subordination and pragmatic backgrounding in a written tradition in which each line is 

typically a clause and the verb is regularly final, yielding few hints as to the syntactic 

relations that hold between clauses. With the important exception of Yoshikawa’s work 

(assembled in Yoshikawa 1993), few investigations of Sumerian have paid much 

attention to how topic and focus are coded or differentiated in Sumerian. 1 But 

comparative and typological studies have repeatedly demonstrated that the articulation of 

topic and focus in a language tell us almost as much about the syntax of the language as 

word-order and morphological patterns. At the same time, the articulation of topic and 

focus in a particular language often interacts in a highly rule-governed fashion with 

(in)definiteness. Take the two sentences in (1) and (2) for example (here and in the rest of 

the dissertation, topics appear in italics with indices such as “i” in (1) marking 

constituents that refer to the same entity; constituents in focus appear in capital letters). 

(1) As for Mary i, I don’t know what she i wants. 

(2) It’s THE CAR that needs to be fixed, (not the motor scooter). 

1 Rubio suggests that “the choice of [conjugation] prefix seems governed by focus” (Rubio forthcoming, 29) on the 

basis of Gragg’s suggestions to the same effect (Gragg 1973b, 93-94; Woods 2001, 394) and Vanstiphout’s discourse 

analytical study of the *i- prefix (Vanstiphout 1985; Woods 2001, 394-395). Gragg’s rather general statement could 

serve as a rather distant antecedent to my own views, translated into one of the contemporary theoretical programs, but 

the more specific articulation of Vanstiphout on the basis of ± person and ± locus is really a codification conventional 

Sumerological interpretations of the conjugation prefixes and should not serve as the basis for future investigations of 

focus in Sumerian. 

11

The topic-marked noun in (1), Mary, is resumed by the pronoun, she, later on in the 

sentence, whereas the focus construction in (2) makes use of a cleft-sentence in which a 

dummy pronoun followed by the copula in combination with a distinct intonation of the 

word car indicates that it is emphasized within the sentence, possibly in contrast to some 

other referent. Needless to say, topic and focus patterns often reconfigure the 

grammatical structures in a language quite dramatically. 

Once the articulation of topic and focus have been delineated within a language, it 

soon becomes apparent that clauses that include an explicit topicalization tend to occur in 

a rather different set of textual environments than clauses involving a focalization. 

Clearly, a clause that introduces a new topic will often be followed by a number of 

subsequent clauses in which the topic plays an important role, even if it is rather quickly 

replaced by another referent as topic. Clauses involving focalization are quite different in 

that they often occur at points of culmination or transition within, for example, an 

unfolding narrative. 2 The problem with attempting to use topic and focus to understand 

the tectonics of clause structure in Sumerian is that there are no hard and fast rules: each 

literary tradition adopts a limited number of what we might call rhetorical devices, which 

are built up out of both syntactic phenomena such as contrastive focus as well as textual 

conventions that have no direct syntactic correlation such as rules of elementary 

composition, viz. “the first sentence in a paragraph should be the ‘topic sentence’ and 

summarize the content of the entire paragraph.” Clearly, as soon as an entirely 

typographic phenomenon such as the paragraph is invoked, we are no longer dealing with 

2 The expression “points of culmination and transition” can be thought of as analogous to Vanstiphout’s foregrounding 

function (see Vanstiphout 1985, 12-13). 

12

matters of morphosyntax—we have entered, at that point, the realm of text-artifactual 

convention. Just as familiarity with present-day Indo-European languages (what Whorf 

famously called Standard Average European) often blinds us to the types of linguistic 

structure used in languages that bear a greater similarity to Sumerian than, say, English, 

likewise, our own conventions for the organization of clauses into paragraphs tends to 

outweigh the different ways in which other text-artifactual traditions organize textual 

material. Whether we think of the at times radically periodic textual structure of formal 

Japanese, in which what a speaker of English might call a topic sentence regularly occurs 

at the end of a paragraph if at all, or the sometimes curious set of textual devices such as 

Wiederaufnahme, i.e., “repetitive resumption,” and Seidel’s Law that organize the 

redaction of the Hebrew Bible (see Levinson 1997, 17-20, for an overview), the lesson is 

clear: the particular morphosyntactic phenomena and the way in which they are used in 

specific rhetorical devices is often highly constrained within a particular text-artifactual 

tradition. Even once the morphosyntactic phenomena are isolated and identified, the 

recognition of the rhetorical devices in which they occur and the overall textual pattern of 

such devices can only be identified through extensive contextual analysis of particular 

morphosyntactic forms. This dissertation is a case-study of one such morphosyntactic 

form in context: the *bi-√ verbal prefix. 

Having identified a number of reasonably coherent properties associated with a 

construction in which a bare inalienable noun immediately precedes a perfective verb 

bearing the *bi-√ prefix (hereafter Bare Noun Bi-√ Verb [BNBV inal]), the next step is to 

figure out what the *bi-√ prefix is and where it comes from. Given the array of 

13

descriptive characterizations applicable to the construction in question, this amounts to a 

relatively straightforward typological question. Is there some grammatical category or 

phenomenon that shows a similar set of properties, distributional and semantic? The 

answer came in the form of a recently completed MIT dissertation (Pylkkänen 2002). 

Pylkkänen, working within the Distributed Morphology branch of the generative tradition 

(see Marantz 1998 for orientation and bibliography), proposed a number of subdivisions 

within the group of phenomena known as applicative constructions, constructions in 

which an adpositional phrase of one kind or another loses its adpositional morphology 

and becomes directly involved in the syntactic processes of the verbal predicate. One 

particular type of applicative in the typology of applicative constructions proposed by 

Pylkkänen, the low source applicative, corresponds particularly well with the Sumerian 

phenomenon dealt with herein and I argue that the *bi-√ prefix in a BNBV inal construction 

should be classified as a low source applicative. The low source applicative hypothesis 

not only makes sense of the particular syntactic and semantic phenomena associated with 

inalienable nouns that immediately precede *bi-√ prefix verbs, but it also conforms to 

Zólyomi’s description of certain rather specific features of what is known as the directive 

case hypothesis (see Zólyomi 1999 for orientation). But the low source applicative 

hypothesis also predicts a number of previously unrecognized semantic phenomena that 

are associated with the subset of *bi-√ prefix compound verbs that include a bare 

inalienable noun, including the use of the ergative case to code an experiencer argument 

and an association of the locative-terminative or non-human dative argument with a 

perceived or adversely affected argument. Hence I argue that the “ergative” postposition, 

14

*-e, also marks (raised) inalienable possessors and that, where the nominal component of 

a BNBV compound verb is inalienable, the ergative case-marked noun is not the agent of 

the clause, but merely the possessor of the inalienable theme. Such a possessive 

construction forms an absolutive/oblique case-marking system in which the inalienable 

possessor and the bare inalienable noun form a single unit in the zero-marked, absolutive 

case, whereas the locative-terminative or locative argument functions as the oblique 

argument. This type of pattern is, in cross-linguistic terms, quite common with verbs of 

perception and experience. 

Bearing in mind the series of interdependent hypotheses proposed in the first chapter 

(the *bi-√ prefix in the BNBV construction as a low source applicative, the ergative 

postposition, *-e, as inalienable possessor rather than agent in BNBV inal constructions, 

and consequently the formation of absolutive/oblique predicates in which the “ergative” 

case-marked noun is an experiencer of an event rather than the cause), the second chapter 

investigates what effects, if any, the curious case-marking system of BNBV inal predicates 

has on lexical and grammatical aspect in Sumerian. Why, for example, does the 

introduction of an inalienable noun in a compound verbal construction regularly form a 

telic predicate? Telicity in more familiar languages such as English is usually coded 

through the use of particles indicating directions as in hammer out, tie up, shoot down, 

and so forth, but Sumerian seems to be using inalienable nouns to achieve the same 

function. Ultimately there are good reasons for suspecting that the use of particles to 

control telicity in English and other similar languages and the use of inalienable nouns in 

the position immediately before the verb in Sumerian are equivalent in a number of 

15

espects and differ only with respect to a single parameter: the head-initial sequence of 

languages like English as opposed to the strictly head-final order in Sumerian. Whether 

viewed in terms of dependency theory (Nichols 1986; Van Valin 2001, 86-107) or one of 

several XP-movement hypotheses currently under discussion (Koopman and Szabolcsi 

2000, Julien 2002), the most reasonable way of coding telicity in Sumerian would appear 

to be through the introduction of a noun immediately to the left of the finite verb that, due 

to its inalienability, is not highly referential. At the same time, the inherently relational 

nature of body-part nouns, taking the human body as its default compass, means that the 

directional quality of English particles such as “up,” “down,” “out” and “in,” can likewise 

be coded using inalienable nouns like gu 2 “neck,” giri 3 “foot,” a 2 “arm,” and ßa 3 

“stomach.” 

Whereas chapters 1 and 2 attempt to elucidate two perspectives on the BNBV inal 

construction (case and thematic role in chapter 1 and lexical aspect (telicity) in chapter 

2), chapters 3 and 4 identify two morphosyntactic environments in which *bi-√ prefix 

verbs such as BNBV inal predicates (but not limited to them exclusively) regularly occur. 

These two environments correspond to topicalizing and a particular kind of focalizing 

contexts of occurrence: the topicalized *bi-√ prefix construction is dealt with in chapter 

3, the focalized in chapter 4. Topics regularly presuppose that the referent of the topical 

phrase is known to both speaker and addressee, so topical phrases are often definite (even 

when indefinite, they tend to be highly specific). As I argue below, there are three 

primary ways of indicating that a nominal phrase is a topic in Sumerian: the two familiar 

methods are the use of the ergative postposition or an independent pronoun, but I would 

16

argue that there is a third mechanism, namely, the use of nouns modified by possessive 

pronouns in clause-initial position. In chapter 3, I argue that *bi-√ prefix verbs can also 

be topicalized through this third method. In particular, I suggest that verbs of the form 

*mini-√ derive from *bi-√ prefix verbs that have been modified by the third person 

animate possessive pronoun, *-ni. 3 This use of *bi-√ prefix verbs as the basis for a 

construction that, at least prototypically, is limited to nouns raises the possibility that *bi- 

√ prefix verbs are, in fact, a species of relative clause, in particular, the type of head- 

internal relative clauses [HIRC] frequently found in verb-final languages. I go on to argue 

that *bi-√ prefix verbs are indefinite HIRCs, whereas *mini-√ prefix verbs are the 

corresponding definite HIRCs, making use of the same system of possessive pronouns to 

code definiteness and topicality. 

To a certain degree, topic-marking in Sumerian is well understood, even if it has 

hardly ever been studied in detail. Clause-initial noun phrases in the ergative case and 

independent personal pronouns can be identified as topics due to the combination of 

clause-initial position and/or ergative case-marking, 4 but that is usually where the 

discussion ends. Although a somewhat more detailed discussion of topic-marking 

patterns appears in section 1.7 below, I would like to suggest that the missing piece in 

theories of Sumerian topic-marking is the possessive pronoun construction. Besides the 

3 The form that is typically given as third person animate possessive suffix is *-ani, but it seems fairly clear that, if the 

animate, *-ni, and inanimate, *-bi, possessive pronouns derive from the proximal, *-ne, and distal, *-be, 

demonstratives, the original form of the third person animate possessive pronoun must have been *-ni rather than *-ani. 

Although I do not develop the argument here, I believe that there is good reason to suspect that *-ni itself is an 

inalienable possessive pronoun and that the corresponding alienable possessive pronoun consists of a linking 

morpheme, *-a-, and the inalienable possessive pronoun, *-ni. 

4 The {e} that appears at the end of independent pronouns is identical to the ergative postposition in orthography and 

phonological form, but independent pronouns in Sumerian seem to exhibit nominal/accusative rection (Woods 2001, 

272). In terms of orthography, however, the defining criteria for topicalization can be stated as (i) clause-initial position 

and (ii), at least in orthographic terms, the “ergative” postposition. 

17

ergative case-marked lexical noun phrase such as lugal.e and the pronouns such as za.e, I 

propose that the third way of topic-marking in Sumerian is through the use of otherwise 

facultative possessive pronoun constructions such as lugal.œu 10 ‘my king’. Since 

Sumerian lacks determiners like those found in English and similar languages and over 

time the use of full demonstratives seems to have attenuated somewhat (cf. Woods 2001, 

25-184), I hypothesize that Sumerian might be using possessive constructions as a means 

of coding definiteness and that since topics are regularly definite, in its written 

incarnation, Sumerian might well have used a somewhat artificial convention such as 

possessive pronoun constructions to indicate that a particular noun phrase was topical. I 

have not attempted to defend this view in itself, but if the derivation of *mini-√ from *bi- 

√ can be taken as evidence, then it lends some support to my suggestion. 

If it is in fact the case that Sumerian codes definiteness through the addition of a 

possessive pronoun, does this imply that bare, unmodified nouns in Sumerian are 

indefinite? For the most part, bare nouns in Sumerian are simply unmarked for 

definiteness. But in one particular environment, bare nouns in Sumerian do seem to 

display what is known as a definiteness effect: a restriction in a particular 

morphosyntactic environment that allows only indefinite nouns to occur in that 

environment. Only when affected by focus as in the answer to a wh-question or in 

contrast with an alternative, do environments that display a definiteness effect allow a 

definite noun to occur. The classic example of this kind of alternation is the existential 

sentence in English with expletive there as in the following examples. 

18

(5) There is a rat in my soup. 

(6) *There is the rat in my soup. 

(7) A: What’s in your soup? 

% B: There’s the rat, the tomatoes and the creamed corn. 

Under ordinary discursive expectations (in which existential sentences are only used to 

introduce a new topic), the definite noun phrase in (6) is blocked, whereas the indefinite 

one in (5) is fine. The definite noun phrase is only permitted when it is focused as in 

answer to the question in (7A), and then it is limited to a distinct pair-list reading, in 

which the referent of the definite noun phrase is interpreted as one of a number of 

possible answers to the question (see sections 3.4 and 3.5 for further elaboration). 

Chapter 4 identifies and investigates one example of a focus-affected definiteness 

effect environment: the *XP nam bi-√ construction, a construction that I interpret as a 

negative contrastive focus construction. In the non-progressive/perfective aspect, the 

configuration of body-part noun followed by *bi-√ prefix has all the properties typically 

associated with an existentially-quantified, indefinite noun such as boundedness and 

homogeneity, but in the progressive/imperfective aspect, which is the usual form of the 

verb in *XP nam bi-√ environments, the predicate no longer refers to a bounded entity 

and can no longer be interpreted as equivalent to an indefinite noun. The unbounded 

version only seems to be permitted in an environment that also includes contrastive, 

indeed, negative contrastive focus of the form “it is not A, (but rather B) that ….” 

Therefore, in the midst of a number of broader arguments as to the nature of topic and 

19

focus, clause-linkage and rhetorical devices of one kind or another, much of the inner 

workings of the dissertation deal with a pair of relatively circumscribed issues: (i) 

definiteness in relation to topic and focus and (ii) the role of possessive constructions in 

the development of the animacy opposition and ergativity in Sumerian. 

Chapter 5 introduces several theories that others have proposed to explain the 

development of ergativity in Sumerian and the differentiation of the two postpositions 

that share the form *-e: namely, the locative-terminative postposition and the ergative 

postposition. I argue that several distinct stages led to the development of ergativity in 

Sumerian and that the correct analysis of the Sumerian phenomena is similar to both 

Garrett’s description of the development of ergativity in Hittite (1990) and Zólyomi’s 

description of the history of four-participant verbs in Sumerian (1999). 

At a somewhat more mundane level, the organization of chapters 1 through 5 can be 

thought of as a series of related distributional classes held together by the hypothesis 

outlined above. 

Chapter Distributional class Interpretation 

1 BNBV inal Absolutive/oblique case-marking 

2 *bi-√ Achievement lexical aspectual class 

3 *mini-√ Topic and definiteness 

4 *XP nam bi-√ Focus and indefiniteness 

5 *-e (LocTerm and Erg) Possession and the rise of ergativity 

20

Therefore, it should be apparent that, in terms of methodology, I follow a rather 

straightforward procedure: chapter 1 identifies a small and, hopefully, well-defined 

syntactic class (BNBV inal), chapter 2 looks at the contexts in which members of the class 

occur so as to identify some of the its semantic features, chapters 3 and 4 look at the 

occurrence of the BNBV inal class in two distinct morphosyntactic environments, *mini-√ 

and *XP nam bi-√ respectively, in the hope that some form of semantic compositionality 

can be identified in these environments, and chapter 5 formulates a diachronic hypothesis 

as to how ergativity arises in Sumerian, resulting in the differentiation of locative- 

terminative and ergative cases, both of which share the form *-e. 

21

1 

The BNBV inal distributional class 

Sumerian frequently makes use of a syntactic construction in which a noun phrase occurs 

immediately before the verb and seems to form a lexical or idiomatic unit with the verbal 

root. These noun/verb pairs are known as compound verbs and are often treated as 

distinct lexemes in lexicographical work such as the example in (1). 

(1) Enmerkar and the Lord of Aratta, l. 554 5 

en aratta ki .ke 4 gig.e igi bi 2.in.du 8 

en aratta-k-e gig-e igi [b] 5-[i] 10-[n] 11-√du 

lord GN-Gen-Erg wheat-LocTerm eye Inan-Dir-ProAnim-close.up 

The lord of Aratta saw the wheat 

5 Transcriptions of languages written in cuneiform and, in particular, transcriptions of Sumerian make use of a plethora 

of conventions, only a few of which will be outlined here: (a) graphemes are separated by periods if they belong to the 

same word or appear in superscript if they are purely semantic elements and lack phonological representation, (b) the 

numbers in subscript following some graphemes code correspondences between certain orthographies and phonological 

forms, but in themselves they do not represent any phonological diacritic. Elsewhere in this thesis, where necessary, the 

subdivisions in the verbal prefix signaled by the use of square brackets such as -[b] 5- are based on the segmentation and 

numbering in use at the Pennsylvania Sumerian Dictionary (Karahashi, personal communication), but for the sake of 

exposition the segmentation implicit in the work of Krecher, Jagersma, and Zólyomi (Zólyomi 1999) is used in this 

section. See the section on conventions and abbreviations in the front matter for the abbreviations in the glosses. 

22

In recent years the investigation of their syntax has become one of the most interesting 

areas of Sumerian grammatical research, primarily due to the postulation of a distinct 

case-marking pattern in which a directive case agreement morpheme [Dir] appears in the 

verbal prefix and is thought to agree with either a dative or a locative case-marking 

postposition following the corresponding noun phrase. The noun phrase that bears the 

dative or locative case in these constructions is often now described as the oblique object 

[OO] of the verb. 

In (1), under the OO hypothesis, igi is the syntactic object of the verb √du 8 and is in 

the zero-marked absolutive case, but since it is a compound verb, the logical object is 

coded in an oblique case such as the locative-terminative case, *-e [LocTerm], that 

follows gig “wheat” in the phrase gig-e. 

(2) Enki’s Journey to Nippur [1.1.4], l. 93 (Al-Fouadi 1969; Karahashi 2000: 29) 

d en.ki.ke4 gu 4 im.ma.ab.gaz.e 

enkik-e gud-Ø [i] 2-[b] 5-[a] 7-[b] 11-√gaz-e 

PN-Erg ox-Abs Vent-NH-Dat-ProNH-√kill-Dur 

Enki was slaughtering oxen 

In (2), however, as one might expect in an ergative language, the agent appears in the 

ergative case, d en.ki.ke 4, while the patient appears in the absolutive case, gu 4. The 

morphosyntactic pattern in (1) is thought to be limited to transitive verbs of low 

23

transitivity as opposed to the pattern in (2), which exhibits high transitivity (Zólyomi 

1999: 216-217 n. 2). 

In section 1.1, the OO hypothesis and its roots in work carried out, above all, by 

Krecher, is briefly examined and the rather heterogeneous and complex character of the 

“system” of verbal prefixes emphasized. In section 1.2, a proper subset of the many 

lexemes that are generally described as compound verbs is isolated on the basis of a 

simple diagnostic: the occurrence of a bare, unmodified nominal component of a 

compound verb that immediately precedes the verbal component with the verbal prefix 

[bi-√] (termed the Bare Noun + *Bi-√ Verb [BNBV] class) as in (1). The set of lexemes 

defined by the BNBV diagnostic, particularly if it is further restricted to inalienable 

nouns (hence the mnemonic [BNBV inal]), is interesting in that it corresponds to what are 

known as low source applicative constructions in other languages (Pylkkänen 2002) and 

their association with privative semantics (including verbs of perception which form the 

core of the BNBV inal class). Low source applicatives and their semantic peculiarities are 

described and compared to the contextual meaning of the subset of compound verbs 

defined as BNBV inal in section 1.3. The three following sections look at three types of 

BNBV inal verb in more detail: section 1.4 deals with verbs of perception, section 1.5 with 

what, for the time being, I term verbs of adversity and section 1.6 with the remaining 

examples. 

In section 1.7, a number of possible motivations for and/or correlations with the 

BNBV inal class are sketched out and two distinct phenomena are isolated as likely factors 

in the semantic effects displayed by the BNBV inal class: indefiniteness and transfer of 

24

possession. I propose that the *bi-√ prefix is the unmarked member of the rather 

elaborate set of dimensional infixes, that it is regularly used with an indefinite bare noun, 

and that, for the most part, we can use a transfer of possession model in which BNBV inal 

verbs code the transfer of the referent of an indefinite noun from a source to a goal (both 

indicated, for the most part, by an *-e postposition): the particular semantics of each 

subset within the BNBV inal group resulting in a somewhat different interpretation of 

source and goal. In section 1.8, I offer an alternative interpretation of the significance of 

the ergative postposition in BNBV inal constructions as inalienable possessor rather than 

ergator and draw a comparison to nominative/dative or absolutive/oblique case-marking 

subsystems that are used in a variety of languages to code the thematic relations in verbs 

of perception and experience. 

25

1.1 Directive case and the oblique object hypothesis 

Sumerian morphosyntax has traditionally been thought of in terms of agreement between 

nominal case-endings or postpositions and elements of the verbal prefix with relatively 

little consideration of any kind of external evidence or motivation for such agreement 

besides the phonological form of the agreeing elements themselves. The paradigms 

generated by this traditional practice have been subjected to increasingly sophisticated 

elucidation and testing in recent years, but the basic premise remains unchanged. In 

particular, models in which subcategorization of the verb replaces verbal agreement as in 

the head-marking, head-final type have not, as a rule, been investigated in any detail. 

Therefore I use the word agreement in the following account in a pretheoretical sense to 

mean simply regular co-occurrence without taking any position as to whether it is true 

agreement or subcategorization of the verb with a corresponding nominal. Regardless of 

theoretical presuppositions, an interesting contrast can be drawn between the two case- 

marking subsystems that involve agreement with dative and locative case-marked nouns: 

the indirect object [IO] subsystem as opposed to the oblique object [OO] subsystem. In 

the following account I attempt to follow the model of directive case-marking and OO 

proposed by Zólyomi (1999) without attempting to take into consideration differences 

between Zólyomi’s work and that of Jagersma, which is largely unpublished. The views 

adhered to by Zólyomi and Jagersma both reflect a line of thought that begins with work 

by Krecher (1985). I would also like to emphasize that the morphological segmentations 

in this section attempt to represent the hypothesis advanced in recent years by Zólyomi 

and Jagersma—they do not represent my own understanding of Sumerian morphological 

26

segmentation. Readers who find the directive case hypothesis unconvincing would do 

well to skip this section entirely—it is not a critical component of my own proposal. 

Like many languages, the primary opposition in the nominal system of Sumerian is 

between animate nouns that are capable of being an agent and those that are inanimate 

and only capable of filling an agentive thematic role in unusual circumstances; this 

corresponds to a basic division at or near ± human on a hierarchy of inherent lexical 

content or other measure of relative animacy (Silverstein 1976; 1993a; see Duranti 1997: 

181-192 for background; Lee 1997 for broader implications). The animacy opposition is 

realized in, for example, the dative case-marking postposition: the animate form is *-ra, 

the inanimate form is *-e (the dative and locative-terminative postpositions respectively; 

see Zólyomi 1999: 251-253 for justification). The ergative postposition, *-e, is identical 

in morphological form to the locative-terminative postposition, *-e, but since the ergative 

postposition usually only occurs with animate nouns and the two postpositions often 

differ in their placement within the clause (the ergative typically in initial position in the 

absence of focus, the locative-terminative not similarly constrained), we can be fairly 

sure of the opposition. The opposition is illustrated in the two following examples, where 

the salient postposition appears in bold: 

27

(3) Ean. 11 2:9-3:4 (Jagersma forthcoming, 7/22) 

d nin.gir2.su.ra gir 2.su ki mu.na.du 3 

ningirsuk-ra girsu [mu] 4-[n] 5/6-[a] 7-√du 

PN-Dat GN Vent-Anim-Dat-build 

He built Girsu for Ningirsu 

(4) Gudea Cyl. A 18: 8-9 (Zólyomi 1999: 252) 

e 2.e im.ma.œen 

e-e [i] 2-[b] 5/6-[a] 7-√œen 

house-Dat Vent-Inan-Dat-go 

He went to the temple 

Note in particular the agreement between the nominal postposition in each example and 

the corresponding segment of the verbal prefix: (3) NP-ra = [n] 5/6-[a] 7 and (4) NP-e = 

[b] 5/6-[a] 7, where [n] 5/6 and [b] 5/6 are animate and inanimate respectively. 6 The paradigm of 

(singular) dative elements in the verbal prefix are as follows: 7 

6 The animate and inanimate elements in the verbal prefix are often assigned to distinct slots, [ ]6 and [ ] 5 respectively in 

the verbal prefix in order to solve several incongruities having to do with the verb, but in the constructions involving 

the type of verb that concerns us here, there is no reason to maintain the opposition. There is a great deal of 

disagreement about the number, segmentation and ordering of the verbal slots and the numbering used here is purely 

conventional, hence the notation [ ] 5/6. 

7 The form of the second person dative alternates between *mura-√ and *mara-√ but the latter is simply a case of 

assimilation and should not be confused with the first person dative prefix or necessarily associated with the second 

person dative affix itself. 

28

*-’a- first person singular 

*-ra- second person singular 

*-na- third person animate singular 

*-ba- third person inanimate singular 

In opposition to the IO subsystem outlined above, another paradigm has been identified 

within the verbal prefix that is for the most part identical to the IO paradigm except for 

the vowel: the directive case agreement morphemes, *-bi-, *-ni-, *-ri-, which form part of 

the OO subsystem. These elements likewise occur in the verbal prefix, but they are not in 

complementary distribution with the dative agreement morphology and their pattern of 

agreement and interpretation remain controversial as illustrated in the following 

examples: 

(5) Nanna-Suen’s Journey to Nippur [1.5.1], l. 271 (Zólyomi 1999: 218, ex. 3) 

ud 5.de 3 niœ 2.ar 3.ra bi 2.ib 2.gu 7.en 

ud-e niœara [b] 5/6-[i] 10-[b] 11-√gu-[en] 13 

goat-Dat groats Inan-Dir-ProInan-eat-1/2Sg 

I will make the goats eat groats 

29

(6) Ean. 1 9:2-3 (Zólyomi 1999: 218) 

e 2.an.na.tum 2.ra lu 2 ti mu.ni.ra 

eanatum-ra lu ti [mu] 4-[n] 5/6-[i] 10-√ra 

PN-Dat man arrow Vent-Anim-Dir-hit 

Somebody struck Eannatum with an arrow, 

(7) Cyl. A 15: 19-21 (Zólyomi 1999: 242, ex. 61) 

kur œiß eren.na lu 2 nu.ku 4.ku 4.da gu 3.de 2.a en d nin.œir 2.su.ke 4 œiri 3 mu.na.ni.œar 

kur eren-ak lu nu-ku.ku-ed-a gudea-r 

mountain cedar-Gen man Neg-enter.Redup-ED-Loc PN-Dat 

en ningirsuk-e œiri [mu] 4-[n] 5/6-[a] 7-[ni] 10-√œar 

lord PN-Erg foot Vent-Anim-Dat-Dir-place 

Ningirsu has made Gudea make his way into the Cedar Mountain where 

nobody is to enter 

All three examples are causative constructions that exploit a compound verb, and since 

causativization of low transitivity verbs seems to be one of the basic features of alienable 

BNBV constructions (as already noted by Zólyomi [1999, 238-242]), any future work on 

the alienable BNBV class will have to investigate causativization in detail. The alienable 

class is not, however, dealt with in substantial fashion herein. 

But, at this point in the discussion, the most important thing to notice is the relative 

incoherence and analytical complexity of the verbal prefixes in (5), (6) and (7): examples 

(5) and (6) are roughly parallel, differing only in the animacy of the directive argument: 

30

[b] 5/6 is inanimate in (5), whereas [n] 5/6 is animate in (6). In (7), however, the directive 

case shows allophonic variation and appears in the form [ni] 10 which is formally 

equivalent to [n] 5/6-[i] 10 (Anim-Dir). Zólyomi analyzes the [ni] 10 in (7) as a locative 

morpheme that agrees with the locative postposition at the end of the phrase kur 

œeß eren.na lu2 nu.ku 4.ku 4.da rather than as the directive infix. But here Zólyomi seems to 

run into trouble, since one of the original motivations of the directive case hypothesis was 

the differentiation of the benefactive use of the animate dative postposition, *-ra, and its 

use to mark a causee. Zólyomi’s Table 5 (1999, 229) identifies the co-occurrence of the 

animate dative postposition, *-ra, with the “/n/ + /a/” verbal infix as the dative case (to be 

associated, primarily, with benefactive use), whereas the co-occurrence of the same 

postposition with the “/n/ + /i/” verbal infix is identified as the directive case. Since the 

interpretation of gu 3.de 2.a(r) as causee rather than benefactive (in accord with Zólyomi’s 

translation) would require associating gu 3.de 2.a(r) with [ni] 10 rather than [n] 5/6-[a] 7, under 

the directive case hypothesis, gu 3.de 2.a(r) is either dative, in which case, the clause is 

benefactive rather than causative and [ni] 10 can be interpreted as locative, or directive, in 

which case, (a) the clause is causative, but (b) [ni] 10 must be reinterpreted as the directive 

case infix rather than the locative infix, (c) the dative infix no longer has a argument in 

the clause to agree with, and (d) [ni] 10 has to be treated as an allophonic variant of [i] 10 so 

as to preserve Krecher’s Rule (see Zólyomi 1999, 221, table 1, slot II). 

The interpretation of [ni] 10 as an allophonic variant of the directive case infix, 

however, is clearly a loophole in the directive case hypothesis and Krecher’s Rule. The 

morphological segmentation of the verb in (7), at least in Zólyomi’s analysis, favors 

31

interpreting [ni] 10 as locative, whereas the contextual meaning of the clause clearly calls 

for a causative interpretation. None of the theories that have been applied to examples 

like (7) can be considered descriptively adequate, but the line of argument exemplified by 

the OO hypothesis, in particular, the fact that benefactive arguments seem to be 

differentiated from causee arguments (where the nominal phrases themselves are both 

followed by the animate dative postposition, *-ra) by the alternation between the dative 

infix (prototypically *-na-) and the directive infix (prototypically *-ni-) is presumably a 

valid characterization in empirical terms at some level of description. Any new proposals 

must make sense of these descriptive facts in the long run, but I have chosen to narrow 

the scope of this study to a proper subset of the verbs bearing the *bi-√ prefix, leaving the 

more elaborate alternatives involving animate causees and benefactives for future study. 

My point, however, is that the directive case hypothesis is overly complex in 

morphological terms and makes extensive use of homophonous segments such as /ni/ that 

are analyzed in distinctive ways (locative [ni] 10 as opposed to animate directive 

[n] 5/6-[i] 10) so as to resolve inconsistencies within the directive case hypothesis itself. In 

the following section, I propose an alternative solution to one subset of the 

morphosyntactic facts addressed by the directive case hypothesis. Where the directive 

case hypothesis postulates increasingly complex and empirically fragile morphological 

segmentations, my proposal juxtaposes lexical classes of several types to 

morphosyntactically defined components of the verbal prefix (such as “applicative” or 

“causative”) to yield meanings that are equivalent, or in some cases, even more precise 

than the values proposed in the directive case hypothesis. 

32

1.2 The bare noun + *bi-√ verb [BNBV] diagnostic 

The set of compound verbs in Sumerian is extensive, no doubt due in large part to the 

rather small set of verbal roots in Sumerian (approximately 200 verbal roots are identified 

by Thomsen ([1984] 2001, 295-323; Karahashi 2000, 16). But the compound verbs co- 

occur with a wide variety of verbal prefixes and cannot be easily characterized in 

semantic terms, so it is reasonable to assume that there are a number of distinct lexically- 

defined subsets and that compound verb is at most a relatively uninformative cover term. 

The only syntactic feature shared by the entire set is that atelic verbs of ongoing action 

(Activity verbs in the Vendlerian typology [Vendler 1967]) do not form compound verbs 

(this does not trivially follow from the fact that Activity verbs are generally intransitive, 

particularly given phenomena such as unaccusativity). 8 If, however, certain formal 

diagnostics are applied to the entire set of compound verbs, various subsets do seem to 

emerge: one of the clearest of these subsets is the one defined by the co-occurrence of a 

bare noun that is immediately followed by a verb that begins with the *bi-√ verbal prefix 

(the bare noun + *bi-√ verb [BNBV] diagnostic). In the subset of compound verbs 

defined by the BNBV diagnostic, certain verbs (œiß-√tuk “to hear,” igi-√du 8 “to see,” 

zu 2-√gub “to taste”) which indicate the achievement of a perception are included by the 

diagnostic as opposed to, say, an activity such as looking at or for something without 

perceiving it. In the following section (2.3), a particular syntactic description of the 

8 Unaccusative predicates take a single argument that is analogous, in semantic terms, to a patient or “direct object,” but 

that occurs in the same syntactic position as an agent or “subject.” Thus, in a nominative/accusative language like 

English, the syntactic form of an unaccusative is a simple intransitive sentence with subject preceding the verb, in 

semantic terms, however, the subject of an unaccusative is interpreted as if it were accusative and followed the verb. 

The generative tradition has long interpreted such forms by having the “subject” following the verb in underlying form. 

33

BNBV class will be proposed, but in this section the diagnostic test is specified and the 

set of lexemes defined by the diagnostic characterized. 

Before attempting to explain why the BNBV diagnostic is a useful distributional 

class and forms a relatively coherent, semantically-defined group, I would first of all like 

to contrast the syntactic behavior of the BNBV class with several other types of verbal 

construction in Sumerian. In each of the following two pairs, examples (8) and (9) as well 

as examples (10) and (11), the first example is a member of the BNBV class, while the 

second is not, even though the verbs mean very nearly the same thing. 

(8) Gilgamesh and the Netherworld [1.8.1.4], l. 302 (Karahashi 2000, 121, ex. 25) 9 

lu 2 izi.la 2 igi bi 2.du 8.am 3 igi nu.mu.ni.du 8 

lu izi-√la-r igi [b] 5/6-[i] 10-√du-am 

person fire-hang-Dat eye Inan-Dir-hold-Cop 

igi [nu] 1-[mu] 2-[n] 5/6-[i] 10-√du 

eye Neg-Vent-Anim-Dir-hold 

A: Did you see a man who had burned (to death)? 

B: I didn’t see such a man. 

9 Shaffer also records a slightly different verbal prefix in one variant: nu mu.de3.du 8.am 3. 

34

(9) Lament over Sumer and Ur [2.2.3], l. 23 (Karahashi 2000: 124) 

d en.lil2.le igi.ni ki kur 2.ra ba.ab.œar.ra.a.ba 

enlil-e igi-ni ki kur-a 

PN-Erg eye-3Poss place oppose-Loc 

[b] 5/6-[a] 7-[b] 11-√œar-a-bi-a 

Inan-Dat-Inan-place-Nmz-DistAdv-Loc 

After Enlil had looked at the enemy land ... 

(10) The Young Scribe and his Father, l. 117 (Sjöberg 1973, 112, 117-118, 128, 147; 

Karahashi 2000, 87) 

a.ra 2 na.me.ka ka.tar.ra.a.du 11.du 11 œiß li.bi 2.tuku.am 3 

ara name-k-a katara-du-du œiß li [b] 5/6-[i] 10-tuku-am 

time any-Gen-Loc praise-say-Redup “wood” Neg Inan-Dir-“have”-Cop 

I have never heard such praise. 

(11) Lugalbanda II [1.8.2.2], l. 160 (PSD A/3: 85) 

mußen.e œeßtu 2 mu.un.ßi.in.ak (var. mu.un.ßi.in.œa 2.ar) 

mußen-e œeßtu [mu] 2-[n] 5/6-[ßi] 9-[n] 11-√ak/œar 

bird-Erg ear Vent-Anim-Alla-Anim-do/place 

The bird listened to him 

35

Examples (8) and (9) obviously differ according the basic BNBV criteria, so in (9) the 

nominal component of the compound verb bears a possessive pronoun and does not 

immediately precede the verb. Both (8) and (9) are compound verbs in the general sense, 

however, and they take the perceived object in an oblique case (the apparent absence of 

oblique case on lu 2 izi.la 2 is only apparent, since both phonological and orthographic 

principles often result in the non-appearance of certain postpositions in the writing 

system, and it has been reconstructed in the gloss). With a handful of exceptions, the 

nominal component of any particular BNBV compound verb never bears a possessive 

pronoun or demonstrative clitic even in those cases where the verbal prefix is not the 

*bi-√ prefix. 

Example (11) differs from example (10) in that an allative [Alla] directional element 

[ßi] 9 occurs in the verbal prefix even though an allative postpositional phrase does not 

appear in the clause. Although older theories expected that dimensional infixes such as 

[ßi] 9 would “agree” with postpositional phrases formed from an etymologically related 

postposition, Gragg’s work has overturned this assumption (see section 1.3). I assume in 

the following that, in the majority of cases, the dimensional infixes can be thought of as 

the pronominal equivalent of the etymologically corresponding postpositional phrase. As 

a rule, the *bi-√ prefix excludes such directional/pronominal elements—this is in fact the 

intuition behind the most widely accepted reconstruction of the verbal prefix under the 

OO hypothesis: namely, that the *bi-√ prefix is made up of [b] 5/6-[i] 10, excluding the 

directional/pronominal elements (otherwise known as the dimensional infixes) in slots 

36

[ ] 7, [ ] 8 and [ ] 9. In other words, the occurrence of the *bi-√ prefix in the BNBV 

diagnostic necessarily excludes the other dimensional infixes which I would argue are 

associated with pronominalization of postpositional phrases and, as a result, the 

definiteness of the noun phrase that bears the postposition (see Gragg 1973a). All in all, 

the BNBV diagnostic systematically excludes from the BNBV set the most obvious cases 

of definite nouns that immediately precede the verb: namely, independent nominal 

phrases bearing possessive or demonstrative suffixes as well as the dimensional infixes. 

The reason for these restrictions on the definiteness of a phrase that appears immediately 

before the verb is that one of the primary goals of this investigation is to isolate 

constructions that include indefinite nouns and to track the morphosyntactic 

consequences of their indefiniteness. 

If one turns to the corpora being assembled by CDLI, several preliminary 

distributional questions should probably be addressed before looking at the semantic 

class defined by the diagnostic: (a) what kinds of constituent can occur immediately 

before *bi-√ verbs in general (what, in other words, is the BNBV class a subset of), (b) 

what kinds of constituents occur immediately before the nominal component of a BNBV 

compound verb in which the bare noun is alienable, and (c) what kinds of constituents 

occur immediately before the nominal component of a BNBV construction in which the 

bare nominal is inalienable. The following table presupposes a number of interpretative 

principles that need to be made clear: type refers to a distinct lexemic set of exemplars 

(which due to the extremely formulaic character of most early documents represents a 

single exemplar in certain respects, e.g., ßu.a bi 2.gi 4 with 142 tokens of the type, nearly 

37

all identical in syntax), XP 1 refers to the syntactic phrase that immediately precedes the 

*bi-√ prefix verb, and XP 2 refers to the phrase that immediately precedes XP 1, 

schematically: *XP 2 XP 1 bi-√. It should be emphasized that this is not an attempt to 

delineate the overall correlations between particular *bi-√ prefix verbs and any co- 

occurring syntactic phrase, but rather an attempt to locate any patterns involving locality 

with respect to the *bi-√ prefix itself. Where there is apparent alternation between the 

category of the preceding XP with respect to a particular verb and the alternation is 

between a postpositional phrase and a bare DP that ends in a vowel, all other things being 

equal, I have assumed that the bare DP is in fact a postpositional phrase of the 

corresponding type. 

Period XP 2 XP 1 Lexemes Types/tokens 

ED IIIb PP Loc/Erg DP Bare i 3-√ak, gin 2-√bar, ki-gar-√du 11, nu- 

banda 3-√gar, giß-√ra, na-√ri/ru 2, a- 

√ru, al-√du 3, si-√sa 2, giß/GIN 2.ÍE 3- 

√se 3, igi-√sag 5, za 3-√ßu 4, giß-√tag, 

38 

za 3-√us 2 

ED IIIb XP zero/DP/Erg/Loc DP Loc za 3-√ak, √du 8, √gar, ßu-√gi 4, √gu 7, 

√mu 7, √du 3 

Several important observations can be made about the overall distribution of *bi-√ 

prefixes in the different corpora: there is clear opposition in the ED IIIb materials 

14/143 

7/126

etween (i) a construction in which a bare noun phrase occurs immediately to the left of 

the verb and a locative or ergative postpositional phrase occurs immediately to the left of 

the bare nominal *PP Loc/Erg NP Bare bi-√, and (ii) a construction in which a locative 

postpositional phrase appears immediately to the left of the verb and the category of the 

constituent to the left of the locative postpositional phrase is entirely unconstrained *XP 

PP Loc bi-√. 10 Although the number of tokens of BNBV constructions in the ED IIIb corpus 

are roughly equivalent for the two distributional classes, I would like to focus on the 

distribution that shows greater formal complexity, the DP Loc/Erg DP Bare *bi-√ class, if for no 

other reason than the fact that the Old Babylonian materials that show the same 

distribution of preceding elements form a relatively coherent semantic class. Thus, 

anticipating my results somewhat, I would like to limit my attention to BNBV predicates 

of the form *DP Loc/Erg DP Bare *bi-√ in which the nominal component of the compound verb 

is an inalienable body-part noun. 

The *DP Loc/Erg DP Bare *bi-√ distributional class is very poorly represented in the Old 

Akkadian materials, if present at all, and the predominant feature of the OAkk exemplars 

of *bi-√ prefix verbs is that co-occurring NPs tend to lack any overt case morphology 

whatsoever. I take this as an underspecification perhaps under the influence of the 

emergent, non-native Sargonic scribal elites, but do not pursue the matter further here. 

The Ur III materials are also quite limited in that administrative and legal genres 

dominate the attested materials. There are a few examples of predicates that meet one of 

10 Note that the highly attenuated nature of verbal prefixes in the ED IIIb period may call into question some of these 

identifications, but the juxtaposition of the two syntactic patterns, [NP-e NP-Ø bi-√] vs. [XP NP-a bi-√], seems to 

represent a rather clear pattern. 

39

the two criteria mentioned above, namely an inalienable body-part noun that immediately 

precedes the *bi-√ prefix: igi bi 2.ib.sa 6 (ASJ 7,125, 27); ßu bi 2.bar.ra.ße 3 (BM 18078 = 

ASJ 9, 329, rev. i 2), ßu bi 2.ba (BM 23871 = ASJ 10, 88, 2; ITT 2/1: 924 = NG 213; 

MVN 2, 2), ßu bi 2.in.ba (SANTAG 6, 154); a 2 bi 2.dar (Nik 447), a 2 bi 2.in.dar (ITT 

3:6572 = NG 145; BM 106157 = SNAT 535; AO 6047 = NG 214); ßu bi 2.in.us 2 (Jean SA 

42). But in no case is the bare inalienable noun itself preceded by a noun followed by the 

locative-terminative postposition. If one assumes that ßu-√ba and ßu-√bar represent the 

same verbal root, there are 11 tokens of four types. Of these lexemes, most show the 

distributional pattern DP Loc/Erg DP Bare *bi-√ just as the ED IIIb materials do, but one, a 2- 

√dar, shows the same distributional pattern as the Old Babylonian examples that form 

one of the primary sets of data for this chapter: *DP LocTerm DP Bare bi-√. This last 

distributional pattern is the most robust and coherent subset within the class defined by 

the BNBV diagnostic in Old Babylonian materials from Nippur: those BNBV 

constructions in which the bare nominal is an (inalienable) part of the body and the 

nominal that precedes the bare nominal bears the locative-terminative postposition. This 

particular distributional category is, however, limited to a handful of questionable 

examples in the third millennium corpora such as igi-√sag 5 (ED IIIb) or igi-√sa 6(g) (Ur 

III): the only lexeme that fully conforms to the Old Babylonian pattern that is the 

centerpiece of this chapter is a 2-√dar, which is included in the appropriate section below 

in spite of its anachronism. One possible member of the BNBV class that seems to have 

been lexicalized is umbisag, “scribe,” which presumably derives from um bi 2.sag x, “one 

who strikes (= √sag 3) the DUB a/UM”—leaving open the question of the proper 

40

interpretation of UM, which like MA is a member of the set of graphemes in Sumerian 

without any clear lexical meaning and presumably derives from whatever language 

underlies the proto-cuneiform materials (see Englund 1998, 77, fn. 158, 79-81). In 

chapter 3, I argue that one function of the *bi-√ prefix is to form what are known as head- 

internal relative clauses and it appears that umbisag may represent just such a head- 

internal relative clause. 

If we now turn to the class defined by the BNBV diagnostic in the Old Babylonian 

materials from Nippur, in the materials studied by Karahashi (2000), which are limited to 

compound verbs in which the nominal component of the compound is a body-part, nearly 

a dozen and a half compound verbs belong to this narrowly defined subset of the BNBV 

class. 

Table of BNBV inal lexemes (excluding those involving ßu ‘hand’) 

Noun Verb Meaning Type 

a 2 √dar to seize illegally; to confiscate Adversity 

a 2 √œar to lose Adversity 

a 2 √su 3 to spread out; to sail away; to run off ? 

du 10 √œal 2 to fight Adversity 

dur 2 √œar to sit; to seat ? 

œiß √tuk to hear Perception 

gu 2 √du 3 to neglect; to despise Adversity 

gu 2 √e 3 to wear; to cover; to be clad in ? 

gu 2 √gur to bend downwards; to lodge; to spy Perception 

igi √du 8 to see Perception 

igi √kar 2 to examine Perception 

ka √keß 2 to put together; to tie up; to bind Adversity 

ßa 3 √dab 5 to feel hurt; to be angry; to be worried Adversity 

umbin √la 2 to scratch Adversity 

za 3 √tag to push away; to overthrow; to reject Adversity 

zu 2 √gub to taste Perception 

41

Note that two lexemes, saœ-√il 2 “to lift the head” and saœ-√se 3 “to entrust; to take care of” 

are not included although the bare nominal, saœ, “head,” is generally thought of as part of 

the body. This is due to the fact that saœ “head” presumably stands in opposition to saœ.ki 

“forehead” in that saœ is probably relational like an “sky; above” or ki “earth, below,” 

whereas saœ.ki is limited in meaning to a part of the body. There also seem to be some 

telling grammatical differences such the fact that compound verbs headed by saœ occurs 

in the *nam.mi-√ construction, whereas the other alienable nouns occur in the *nam bi-√ 

construction, an opposition that seems to depend on the inalienability of the head noun; I 

explore it detail in chapter 4. Note that in the most famous case of disembodied head in 

Sumerian literature, namely when Gilgamesh and Enkidu dump Huwawa’s head out in 

front of Enlil (Gilgamesh and Huwawa, version A, line 183), the possessive pronoun 

refers to the original owner, but the form is saœ.du.ni rather than saœ.œa.ni. This leads me 

to suspect that saœ itself does not directly refer to the head. 

kas 4-√kar “to sprint” is also excluded even though Karahashi included it in her study of 

compound verbs that involve a part of the body. Karahashi herself notes that “kas 4 

normally means ‘runner, courier’ but may have originally designated a part of the leg/foot 

which became obsolete during later times…” (Karahashi 2000, 129). 

The reason that few if any of these lexemes occur in texts earlier than the Old 

Babylonian period is, in my view, that the kinds of generic contexts in which verbs of 

perception, adversity and possession that constitute the vast majority of the members of 

the class simply did not exist in the written medium. In other words, the genres in which 

the particular semantics that I argue are associated with this distributional class are not 

42

attested in significant numbers prior to the Old Babylonian period. Only once these 

genres surface in text-artifactual form, does the distributional category that I investigate 

in this chapter appear in clear and robust fashion. But even if the lexemes that occur in 

the class defined by the BNBV diagnostic differ in some respects, the form of the 

syntactic constituents that seem to occur in BNBV constructions in the OB period can be 

derived in a fairly transparent way from the form seen in the ED IIIb period. Whereas the 

ED IIIb materials seem to suggest a syntactic structure such as 

[[… PP Loc/Erg [(NP Bare) bi]-√], the BNBV inal construction seems to suggest a syntactic 

ground plan such as the following: [… PP Loc/Erg [PP LocTerm NP Bare] bi]-√ with the inclusion 

of the an additional PP LocTerm argument of the verb (the oblique object in the terminology 

of Zólyomi [1999]). 

Even given this relatively sketchy set of distributional facts, a preliminary 

distributional template can be proposed: 

(12) [ XP … [[ Direct speech Ø Ø bi 2]-du 11]] 

(13) [ XP … [[PP Loc Ø Ø bi 2]-√]] 

(14) [ XP … [PP Loc/Erg Ø [NP Bare] bi 2]-√] 

(15) [ XP … [PP Loc/Erg [PP LocTerm [NP Bare]] bi 2]-√] 

Although the series of phrase diagrams running from (12) through (15) is a fiction in that 

they do not represent any single synchronic phase of the language, they are at least 

partially representative of the overall syntactic pattern that I am attempting to justify in 

43

this chapter. And, it is, after all, a syntactic pattern familiar from the center-embedding 

genitive construction in Sumerian. 

Example Syntactic form Genitive parallel 

13 [PP Loc bi 2]-√ [NP-Loc] 

14 [PP Loc/Erg NP Bare bi 2]-√ [NP NP-Gen] 

15 [PP Loc/Erg [PP LocTerm NP Bare (bi 2)] bi 2]-√ [NP [NP NP-Gen]-Gen] 

As I will argue in explicit terms later on, a number of factors indicate that the *bi-√ prefix 

is quite similar to the expletive there in English existential sentences such as “there is a 

dog in the parlor.” This obviously corresponds in certain respects to the traditional notion 

that the *bi-√ prefix “agrees” with a locative argument, but I would like to emphasize two 

fundamental distinctions before moving on to the next section. If the *bi-√ prefix is 

similar to the expletive there in English existential sentences, then one might expect it to 

regularly occur with locative arguments. But it needs to be remembered that there does 

not agree with the locative it occurs with, it simply occurs in certain syntactic 

configurations that also typically include locative arguments. In other words, it is a matter 

of syntax rather than a morphologically-driven agreement relation that holds between a 

morpheme on the verb and a locative phrase in the clause. The other difference between 

the traditional agreement-based interpretation and my suggestion that the bi-√ prefix is 

analogous to expletive “there” is that the use of expletive “there” in English has far 

reaching syntactic implications for case-marking, quantification and focus constructions. 

44

In the following section, I argue that the *bi- prefix is best understood from a syntactic 

point of view rather than the conventional, “morphological” interpretation of it as an 

agreement marker. Bearing these considerations in mind, let us turn to the first part of my 

redescription of the *bi-√ prefix. 

45

1.3 BNBV inal as a low source applicative 

Not only does the BNBV diagnostic, when its use is limited to compound verbs in which 

the bare noun is an inalienable part of the body (BNBV inal), identify a distinct syntactic 

class, but it also identifies a semantic field that includes verbs of perception as well as 

verbs that signify privation or loss with respect to an argument that represents the object 

of perception or the participant who is adversely affected. The natural counterpart of an 

argument that represents the source from which something is taken, an entity that 

receives or benefits from the privation of the source, is an experiencer argument. Recent 

work (Pylkkänen 2002) has identified a syntactic phenomenon known as a low source 

applicative 11 that exhibits the same set of semantic features as the BNBV inal class: this 

section offers an overview of low applicativity based on Pylkkänen’s dissertation that 

supports a redescription of the *bi-√ prefix in BNBV inal constructions as a low source 

applicative morpheme. 

In general terms, an applicative morpheme allows a noun phrase that would 

otherwise have appeared in an adpositional phrase to appear in a zero/unmarked case and 

to participate more directly in the syntactic behavior of a verbal predicate. This 

phenomenon exists, for example, in English in so-called double object constructions as 

exemplified in the following examples: 

11 Pylkkänen’s low source applicative hypothesis represents only the most recent and innovative phase of the 

investigation of so-called double object constructions. Previous milestones in the investigation of the double object are 

Larson 1988 and Pesetsky 1995, see Harley 2003 for the current state of double object construction analyses in 

generative terms, including a review of previous literature. 

46

(18) Sue bought a cake for Lou. 

(19) Sue bought Lou a cake. 

Although there is no distinct segmental morphology in English, nonetheless it is fairly 

clear that the dative case-marked argument in (18), for Lou, can drop its preposition and 

shift to a position immediately after the verb and before the patient as in (19). Building 

on previous work by, among others, Pesetsky (1995), Pylkkänen (2002) differentiates a 

number of different kinds of applicative on the basis of different syntactic criteria. The 

most basic opposition drawn by Pylkkänen is between high applicatives and low 

applicatives. The classic examples of applicatives from slightly older works are, as a rule, 

high applicatives such as the following: 

(20) Chichewa (Pylkkänen 2002, 18; Baker 1988, 354) 

Maruto a-na-umb-ir-a inpeni mtsuko 

PN Sp-Past-mold-Appl-Asp knife waterpot 

Maruto molded the waterpot with a knife 

Note that the instrumental adjunct, inpeni, “knife,” in (20) is a bare noun, whereas its 

English translation, with a knife, is a prepositional phrase. Comparison with other 

examples that lack the instrumental adjunct show the *-ir- morpheme indicates that the 

knife is an instrument instead of, say, the patient. Pylkkänen then differentiates between 

high applicatives such as (20) and two types of low applicative: low goal applicatives as 

47

in the English double object construction in (19) and low source applicatives as in the 

following example from Korean (Pylkkänen 2002: 21, example [21] in contrast with 

example [22]). 

(21) totwuk-i Mary-hanthey panci-lul humchi-ess-ta 

thief-Nom Mary-Dat ring-Acc steal-Past-Plain 

The thief stole a ring from Mary (and it was in her possession when he stole it) 

(22) totwuk-i Mary-ui panci-lul humchi-ess-ta 

thief-Nom Mary-Gen ring-Acc steal-Past-Plain 

The thief stole Mary’s ring (with no implication as to the presence of Mary at 

the theft) 

The basic criterion for differentiating a high applicative such as (20) from a low 

applicative such as (21) is whether or not there is necessarily a possessive relation 

between the direct object and the applied argument: high applicatives do not presuppose 

any kind of possessive relation between direct object and the oblique argument, whereas 

low applicatives require such a possessive relation. In (20), there is no reason to suspect 

the existence of any kind of possessive relation between the knife and the waterpot, but in 

(21) as well as the English example in (19), the oblique argument, Mary-Dat and Lou 

respectively, must either start out (21) or end up (19) as the possessor of the direct object. 

Although Pylkkänen does not make the connection explicit, readers familiar with either 

48

Zólyomi’s description of possessor raising in Sumerian (1999, 231-237) or the use of 

clausal possessive constructions for certain kinds of inalienable noun (see Nichols 1988 

for orientation) should recognize that the low source applicative is also a kind of 

possessor raising construction. In other words, the low source applicative takes a 

possessive construction involving a head noun, such as panci-lul in (22), and a possessor 

in the genitive case, such as Mary-ui in (22)—where only panci-lul, the direct object, has 

a syntactic relation to the main verb—and changes it into a raised possessor construction 

in which both the head noun, panci-lul in (21), and the raised possessor, Mary-hanthey 

in (21), are arguments of the verb. It is only because of the fact that Mary-ui is not an 

argument of the verb, whereas Mary-hanthey is, that we can speak of the construction in 

(21) as a low source applicative construction, since one defining feature of an applicative 

construction is that it takes a phrase that is not an argument of the verb and converts it 

into a phrase that is an argument of the verb. 

The parallel with the other types of applicative should, therefore, be fairly clear: the 

high applicative in (20) removes the instrumental case-marking from an instrument, 

inpeni, and indicates that it a core argument of the verb by adding a morpheme, *-ir-, to 

the verb; the low goal applicative in (19) changes a benefactive prepositional phrase, for 

Lou, into a bare noun, Lou, and incorporates the bare noun into the verb; likewise, the 

low source applicative in (21) changes a genitive possessive construction into a clausal 

possessive construction with a raised possessor and, in the process, makes the raised 

possessor an argument of the verb (for further discussion of possessor raising and its 

relation to the rise of ergative case-marking in Sumerian, see chapter 5). 

49

Various theoretical considerations that I will not go into here lead Pylkkänen (2002, 

21-23) to postulate that the low applicative is best understood as a higher order predicate 

that takes three arguments: the direct object, the oblique argument and the lexical verb. 

But more importantly from a descriptive point of view, Pylkkänen also offers a number 

of diagnostics for differentiating high and low applicatives, two of which I repeat here in 

full (Pylkkänen 2002, 23): 

DIAGNOSTIC 1: TRANSITIVITY RESTRICTIONS 

Only high applicative heads should be able to combine with unergatives [agentive, 

intransitive verbs, CJ]. Since a low applicative head denotes a relation between the 

direct and indirect object, it cannot appear in a structure that lacks a direct object. 

DIAGNOSTIC 2: VERB SEMANTICS 

Since low applicatives imply a transfer of possession, they make no sense with verbs 

that are completely static: for example, an event of holding a bag does not plausibly 

result in the bag ending up in someone’s possession. High applicatives, on the other 

hand, should have no problem combining with verbs such as hold: it is perfectly 

plausible that somebody would benefit from a bag-holding event. 

It should be fairly clear that the BNBV inal construction meets both of these criteria: given 

the close association between compound verbs and the OO constructions in an 

environment defined by the occurrence of a *bi-√ prefix verb, it is reasonable to assume 

50

that *bi-√ prefix verbs are generally transitive and involve both a direct object and an 

indirect object. Likewise, the fact that all of the members of the BNBV inal class involve a 

change of state indicates that Pylkkänen’s second diagnostic is satisfied as well. 

One the most well-known semantic effects of the constructions such as (12) and the 

corresponding form in Japanese (gapped adversity passives) is that the event described by 

the predicate affects the dative possessor in a negative way: this clearly results from the 

source argument implicit in low applicative constructions and is apparent in examples of 

the BNBV inal class (Pylkkänen 2002, 23; on Japanese gapped adversity passives and their 

similarity to (12), see Kubo 1992 and Pylkkänen 2002, 59-68). One the best examples of 

the role of privation in low source applicatives is the meaning that Landau has attributed 

to possessor dative constructions in Hebrew (1999) and Pylkkänen critical response to 

Landau’s initial proposal (Pylkkänen 2002, 43-49). 

Landau begins with a discussion of a raised possessor construction in Modern 

Hebrew in which the possessor is preceded by a dative preposition, while the possessed 

noun is unmarked except for case as in the following example (Borer and Grodzinsky, 

1986, ex. 12a; Landau 1999, 3, ex. 3). 

(23) ha-yalda kilkela le-Dan et ha-radio 

the-girl spoiled to-Dan Acc the-radio 

The girl broke Dan’s radio 

51

But Landau then draws a comparison with a construction in which the dative 

prepositional phrase is not the actual possessor, but rather a person who is negatively 

affected by the event described in the clause. 

(24) Gil ßavar le-Rina et ha-mißkafayim ßel Sigal 

PN broke to-Rina Acc the-glasses of PN 2 

Gil broke Sigal’s glasses on Rina, i.e. Rina was negatively affected by event 

Here in (24), the possessor of the glasses is indicated by ßel Sigal “of Sigal,” whereas the 

phrase that includes the dative preposition refers to Rina as the person negatively 

affected by the breaking of the glasses. The negative affect attributed to the referent of 

a noun in the dative case or a prepositional phrase should be familiar as one of the 

hallmarks of the low source applicative described above. The problem arises with a 

seemingly arbitrary restriction that would seem to contradict a low source applicative 

interpretation of verbs of direct perception. Landau notes that the possessor dative 

construction does not felicitously occur with verbs of perception such as ra’a “to see.” 

Landau offers the following example of an ungrammatical sentence in which the verb 

is ra’a “to see” in contrast with a similar clause in (26) where the verb means “to look 

at” (Landau 1999, 25-26, ex. 49a and 49b). 

(25) *Gil ra’a le-Rina et ha-bayit 

PN saw to-Rina Acc the-house 

Gil saw Rina’s house 

52

(26) Gil histakel le-Rina al ha-bayit 

PN looked.at to-Rina on the-house 

Gil looked at Rina’s house 

Various hypotheses, taking examples like (24) as exemplary, have argued that the crucial 

matter is affectedness: in (23) and (24), the dative possessor is affected by the event, 

whereas in (25), it is not, hence the ungrammaticality of (25). Landau himself argued 

otherwise, attributing the ungrammaticality to the fact that the perceiver in (25) is non- 

agentive, whereas in (26), it is agentive. Landau does comment, in a footnote, that there is 

a small class of “to see” verbs which are grammatical such as the following (Landau 

1999, 26, fn. 14; Pylkkänen 2002, 48). 

(27) Gil ra’a le-Rina et ha-pupik 

PN saw to-Rina Acc the-belly.button 

Gil saw Rina’s belly button 

Landau attributes the grammaticality of (27) to the fact that ha-pupik “the belly button” is 

a part of the body and, in Landau’s view, inalienable; this follows from a fairly common 

analysis among generativists, namely that inalienable possessors are generated at [Spec, 

NP], whereas alienable possessors are generated at [Spec, DP]; the import of such a 

structural description has long been recognized in functionalist circles at least since the 

53

work of Nichols who describes the diachronic origin of inalienability as follows (Nichols 

1988, 579): 

A single diachronic process appears to motivate all of the attested patterns involving 

‘[in]alienability’: tighter bonding of possessive affixes, fusion of possessive affixes 

to nouns, and earlier lexicalization of possession, take place with those nouns which 

are most often possessed—kin terms, body parts, and (in languages where they are 

lexicalized as nouns) inherently relational notions such as parts of wholes. 

Landau’s argument ultimately boils down to a purely structural argument about the 

location of inalienable as apposed to alienable nouns in certain positions within syntactic 

trees and he explains the grammaticality of (27) on the basis of the idiomatic character of 

inalienable nouns and the fact that inalienable possessors are more tightly bound to an 

inalienable noun than is the case with alienable nouns. But it turns out that Landau’s 

hypothesis is wrong: as noted by Pylkkänen (2002, 48), there are cases in which an 

alienable noun can occur as the perceived object with the verb ra’a “to see” and the 

criterion appears to have nothing to do with inalienability and everything to do with the 

privative semantics associated with low source applicatives; Pylkkänen describes the 

criterion as follows (Pylkkänen 2002, 48): 

Landau notes that the acceptability of the construction increases with intimate body 

parts and decreases with neutral ones. In other words, the semantic restrictions 

parallel the Finnish [low source applicative] ones precisely: the possessor dative is 

only acceptable with ‘see’ when the seeing-event makes something public that was 

private before. And, contrary to Landau’s generalization, the object that becomes 

54

public does not need to be inalienably possessed; it simply needs to be “secret”, in 

some way. 

Pylkkänen cites as evidence the following example in which the perceived object is 

decidedly alienable. 

(28) Gil ra’a le-Rina et ha-mismaxim 

PN saw to-Rina Acc the-documents 

Gil saw the documents that Rina had 

The requirement for non-public objects of direct perception follows directly from the fact 

that “source applicatives . . . always involve an implication of loss, but this loss does not 

need to be concrete” (Pylkkänen 2002, 48). It should be relatively clear that the model 

that underlies low source applicatives is a transfer of possession model: an object of 

possession starts out in the possession of one entity and moves into the possession of 

another. Although acts of direct perception are quite abstract, Pylkkänen seems to be 

arguing that the perception of an entity initially “belongs” to that entity and the act of 

perception transfers the perception from the perceived object to its perceiver as if it were 

the referent of a concrete noun. At this point, it may be useful to turn to the first 

collection of exemplars from Sumerian, since they can be divided two primary groups, 

verbs of perception and affect, as well as third group of exceptions (those involving ßu 

“hand” are dealt with in chapter 2). 

55

1.4 BNBV inal verbs of perception 

Bearing in mind the sometimes quite complicated distributional pattern of low source 

applicatives, especially those involving perception and (usually negative) affect, I turn 

initially to BNBV inal predicates that are verbs of direct perception. 

(29) Enmerkar and the Lord of Aratta [1.8.2.3], ll. 551-554 

(Karahashi 2000, 119; Vanstiphout 2003, 88-89) 

551. gig ni 2.bi mu 2.a guru 7 

[x (x)] ka 

The wheat grown by itself . . . granary . . ., 

552. en aratta ki .ra As for that which was brought to the lord of 

mu.na.ni.in.≠ku 4±.[ku 4] Aratta, 

553. kisal aratta ki .a.ka igi.ni.ße 3 

i.im.dub ? 

It was piled up in front of him in the court of 

56 

Aratta, 

554. en aratta ki .ke 4 gig.e The lord of Aratta saw the wheat, 

igi bi 2.in.du 8 

The example of a BNBV inal predicate in line 554 is the one that appears at the beginning 

of the chapter as a prototypical example of a BNBV inal predicate. For a hint as to the 

reasoning behind my translation of line 552 as a topicalized relative clause, see the 

discussion of *mini-√ in chapter 3.

(30) Lugalbanda II [1.8.2.2], ll. 207-208 (Wilcke 1969a; Karahashi 2000, 119) 

207. [mußen.e an].ta igi mi.ni.in.il 2 

As for the bird, who was looking from 

erin 2.e igi bi 2.in.du 8.ru above, she saw the troops, 

208. [lugal].≠ban 3±.da ki.ta As for Lugalbanda, who was looking from 

igi mi.ni.in.il 2 sa˙ar erin 2.e below, he saw the troops enveloped in 

du 8.a igi bi 2.in.du 8.ru dust, 

The lack of a direct object with the first of the two perception verbs in each line provides 

some evidence in favor interpreting igi-√du 8 as “to see” (achieved perception) as opposed 

to igi-√il 2 “to look at” (attempted perception). As in the previous example, some 

justification for the translation of igi mi.ni.in.il 2 as a relative clause is offered in chapter 4 

in an extended discussion of the *mini-√ prefix; my interpretation of igi mi.ni.in.il 2 as a 

relative clause also provides some justification for the progressive form of the verb in the 

English translation: since the relative clause in *mini-√ provide background information, 

they must be translated into English in the progressive aspect even through the Sumerian 

verbal form is ˙amt≥u and would, typically, be translated with a past or perfect verb. The 

fact that √du 8 is followed by the plural suffix /eß/ with the orthographic form du 8.ru 

indicates that the perceived object is erin 2.e “the troops” rather than a cloud of dust. 

Partial justification for interpreting du 8.ru as the verbal root followed by the plural suffix 

rather than, for example, the marû marker, *-e, is offered in chapter 2. For additional 

examples of igi-√du 8 in BNBV inal configurations, see Karahashi (2000, 118-122). 

57

(31) Eduba Regulations [PBS 1/2, 98 = PBS 12, 30; UET 6, 168; UET 6, *240; ISET 

1, 72 (Ni 4103); CBS 11786], the following based on PBS 12, 30, rev. 1-5 

(unpublished) Falkenstein 1947, 183-184; Karahashi 2000, 127) 

1. tur e 2.dub.ba.a ≠ßub±.ßub.≠ba± x ßu The little ones left at left . . ., 

[...] igi dub.ba na.du 11.am 3 gu 2 [...] 

2. ad.da e 2.dub.ba.a giri 17.zal da.bi The father has . . . his mouth at school, 

i 3.tuku x 

3. ßeß.gal.e u 3 ba.dab 5 dub.be 2 

igi bi 2.ib 2.kar 2.kar 2 

The older student, once he has taken them, 

58 

should inspect the tablets, 

4. ki gu sar SI nu.ub.kar 2 a ≠œa 2±.e Where he does not find . . ., he should 

an.na.ab.sa 2.e correct it . . . , 

5. tukum.bi tur e 2.dub.ba.a tur.re If a student at school . . . a student . . ., 

This composition is unpublished and difficult to make sense of. Since only line 3 makes 

much sense and the continuity of the passage remains unclear, it should be treated with 

caution. 

(32) The Young Scribe and his Father, ll. 114-117 

(Sjöberg 1973, 112, 117-118, 128, 147; Karahashi 2000, 87) 

114. œiß gid 2.da ad.da.œa 2.ße 3 geßtu 2 

ga.œar nu.ub.be 2.en my father,” 

You do not say, “I want to listen to the ... of

115. nam d en.lil 2.le lu 2.u 18.lu.a That the fate of humanity is determined by 

nam.bi tar.ra Enlil, 

116. dumu kiœ 2.gi 4.a ad.da.na.ka That a son should follow in the occupation 

i 3.ib 2.us 2 

59 

of his father, 

117. a.ra 2 na.me.ka ka.tar.ra.a Haven’t you heard these precious statements 

du 11.du 11 œiß li bi 2.tuku.am 3 

made innumerable times? 

This composition is better known as Der Vater und sein missratener Sohn after the 

edition of Sjöberg (1973). The meaning of the passage is still quite uncertain and I have 

taken the absence of an explicit agreement marker before the verbal root in combination 

with the copula following the verb as an indication that line 117 is a question (for a 

similar construction, see the series of igi bi 2.du 8.am 3 [“did you see . . . ?”] predicates at 

the end of Gilgamesh and the Netherworld). The final verb in line 115 also shows a great 

deal of formal variation: nam.bi tar.ra (K and EE), nam bi 2.ib.tar.re (Z), nam.bi 

in.tar.re (AA), nam.bi in.tar.ra.am 3 (FF), nam.bi tar.re (II), nam bi 2 [...] (JJ). It seems 

fairly clear that the variation revolves around the sequence /nam bi tar/ and one 

interpretation, nam bi 2.ib.tar.re (Z), belongs to a grammatical class dealt with in chapter 

4, namely the *XP nam bi-√ distributional class. 

(33) UM 29-15-367 I 5-6 (Civil 1964, 9, Karahashi 2000, 180) 

munus.e u 2.lal 3.am 3 zu 2 bi 2.in.gub ßa 3.gal ba.ni.in.œar 

It was the clover that the woman tasted, and it became her meal,

The special characteristic of BNBV inal predicates involving perception is that in many 

of the examples both the perceiver and the perceived object are followed by the *-e 

postposition. This is, in my view, the prototypical and most archaic form of the BNBV inal 

construction; the other groups of BNBV inal predicates, though hewing to the semantic 

characterization of BNBV inal predicates, diverge from the perception verb group in that 

the adversely affected participant usually appears in the locative case. 

60

1.5 BNBV inal verbs of adversity 

The earliest attestation of a BNBV inal predicate, a 2-√dar, dates to the Ur III period and is 

adversarial. The four exemplars of a 2-√dar are included at the beginning of this section, 

but it should be noted that a 2-√dar does not occur in BNBV inal in the Old Babylonian 

materials; the rest of the examples in this section are from the Old Babylonian period. 

(34) Nik 2, 447, ll. 11-13 (Karahashi 2000, 75) 

10. niœ 2.gur 11 ur.TAR.LUH.ka As for the property of UrTARLUH, 

11. ur.ab.zu u 3 dam ur.ab.zu.ka.ke 4 

12. a 2 bi 2.dar ba.an.de 6 

Ur’abzu and his wife 

(35) NG 145 (ITT 3, 6572), obv. 2 (Karahashi 2000, 75) 

1. ≠ße± [x x x] 

illegally seized it and took it away, 

2. ße lu 2.≠nin.œa 2.ka± na.≠gu± Nagu illegally appropriated Lunina’s grain, 

a 2 bi 2.in.≠dar± 

3. na.gu lu 2.sukkal.≠ke 4± ma.an.bala Nagu said, “one of the sukkal’s servants 

bi 2.in.du 11 

(36) SNAT 535 (BM 106157) 

7. 1(diß) ma.na 1(u) 5(diß) gin 2 

lugal.ußur 4.e 

8. ur. d ˙al.mu.DU.ra in.na.an.ßum 2 

61 

exchanged it with me,” 

Lugal’ushur gave 1 mana 15 gin 

to Ur-halmuDU,

9. 1/3(diß) ma.na 5(diß) gin 2 

in.da.an.tuku 

He still was owed 1/3 mana 5 gin 

10. dab 5.bi.ße 3 œen.na.ni When he came in order to seize it, 

11. e 2.œa 2 a 2 bi 2.in.dar He illegally seized my house, 

12. kißib 3 ßu.ga 2 ba.an.de 6 

(37) NG 214 (AO 6047), i 10 – ii 2 

i 

He took away the seal in my hand, 

10. ßu. d nin.mug.ga ugula (As for) Shuninmuga, the overseer, 

11. u 3 a.PU 3.ÍA.a.a ßar 2.ra.ab.du and Apuzraya, the sharabdu, 

12. a.ßa 3 dam a.ne.a.ti.ka 1(bur 3) “Lugalmelam illegally seized 

ii 

GAN 2.am 3 

1. lugal.me.lam 2.[e a 2] bi 2.in.dar the field of the wife of Ane’ati, measuring 

62 

1 bur,” 

2. bi 2.in.eß they said. 

These four examples are from legal texts from the Ur III period, but they do not exhibit 

the most common distributional pattern found in the Old Babylonian materials, namely 

the combination of ergative and locative-terminative (or, in the case of adversity 

predicates, locative) postpositional phrases. Turning now to examples from the Old 

Babylonian period, one of the clearest examples of a BNBV inal verb of adversity is 

a 2-√œar, “to overcome.”

(38) Enmerkar and Ensu˙girana [1.8.2.4], ll. 120-125 (Karahashi 2000, 77) 

120. a.na ga.an.na.ab.be 2 a.na “What shall I say to him? What shall I say 

ga.an.na.ab.be 2 

63 

to him? 

121. en unug ki ki 

.ga en kul.aba4 .ra What shall I say to the lord of Uruk, the lord 

a.na ga.an.na.ab.be 2 

of Kulaba? 

122. [gu 4].da.ni gu 4.œu 10.da ≠lirum ? ± As for his bull, it stood up to fight my bull, 

im.da.ab.zi 

123. gu 4 unug ki .ga.ke 4 a 2 bi 2.ib.œar The bull of Uruk overcame him, 

124. lu 2.ni lu 2.œu 10.da usu im.da.ab.ra 2 As for his man, he wrestled with my man, 

125. lu 2 unug ki .ga.ke 4 a 2 bi 2.ib.œar The man of Uruk overcame him,” 

(39) Lament over Sumer and Ur [2.2.3], l. 386-388 (Karahashi 2000, 77) 

386. na 4 gal.gal.e ni 2.bi.a pu.ud.pa.ad Large stones on themselves fell with 

im.mi.ni.ib.za great thuds. 

386a. u 4 ßu 2.uß iri ki .ta im.˙ul.e Daily the evil wind came back into the city, 

mu.un.da.an.gi 4.gi 4 

ki 

387. uri5 ne3.bi.ta nir.œal2 gab2.gaz.e Ur, confident in its own strength, stood 

ba.gub ready for the murderers, 

388. uœ 3.bi lu 2.erim 2.e a 2 bi 2.ib 2.œar As for its people, the enemy overcame them, 

œiß tukul.e la ba.su8.ge.eß they could not withstand their weapons,

(40) Inanna and Ebih [1.3.2], ll. 83-86 

83. kur.kur.ra muß ki.in.dar.ra.gin 7 

ßu ˙u.mu.da.dub 2.be 2.eß crevice, 

84. muß.saœ.kal kur.bi.ta e 11.da.gin 7 

May he destroy the lands as a snake in a 

May he make them slither around like a 

da.ga ˙u.mu.da.an.du sankal snake coming down from a 

64 

mountain. 

85. kur.ra a 2 he 2 bi 2.in.œar igi May he overcome the mountain, examine it, 

˙e 2.ni.in.bar gid 2.da.bi ˙a.ba.zu and know its length, 

86. ˙ar.ra.an ku 3 an.na.ka.ße 3 

May he go out on the holy campaign of An 

˙e 2.ni.e 3 buru 3.da.bi ˙a.ba.zu and know its depth, 

In the four examples in (38), (39) and (40), a noun phrase in the ergative case is 

positively affected by the event, whereas the negatively affected participant either does 

not appear explicitly or appears followed by the locative postposition. 

(41) Lament for Sumer and Ur [2.2.3], ll. 400-402 

400. ßa 3.bi.a niœ 2 ßa 3.œar.ra.ka On the inside, we are being finished off by 

i.im.til.le.de 3.en.de 3.en famine, 

401. bar.bi.a œiß tukul elam ki .ma.ka On the outside, won’t we be finished off by 

ga nam ba.[e.til].le.en.de 3.en Elamite weapons, 

ki 

402. uri5 .ma lu2.erim2.e a2 bi2.ib.œar Won’t we be finished off by the enemy that 

ga nam ba.til.e.de 3..de 3.en has overcome Ur?

(42) Ur Lament [2.2.2], ll. 396-398 

396. œalga niœ 2 aratta ki .ka ßu pe.el.la 2 

im.mi.ib.du 11.ga.re 

397. u 4 kalam.ma niœ 2 du 10 

im.mi.ib.ku 5.da.re the Land, 

Which defiled the weighty counsel, 

The storm which cut off all that is good from 

398. u 4 saœ œi 6.ga a 2 bi 2.ib.la 2.a.re The storm which pinioned the arms of the 

65 

black-headed people, 

This is only a brief passage from a long litany of descriptive characterizations or 

appellations that are applied to the raging storm in the Ur Lament. For an extensive 

discussion of this passage, see below in sections 4.4 and 4.5. 

(43) A praise poem of Enlil-bani (Enlil-bani A), ll. 92-100 (Kapp 1955, 76-87) 

92. ßagina.zu Your governors 

93. saœ ßul.a.lum la ba.ak suffer no injuries, 

94. erin 2.zu erin 2 kur 2.ra As for your troops, 

95. a 2 bi 2.ib 2.œar they overcome hostile troops, 

96. œiß tukul.zu As for your weapons, 

97. œiß tukul gaba.ri nu.tuku.am 3 

they have no rival weapons, 

98. tuß.a.zu.ne When you take your seat, 

99. kur.kur all the foreign lands 

100. gam.e.en bow down to you,

(44) Sumerian Proverbs, Collection [6.1.01], no. 25 

(Alster 1997, 11; Civil 1985, 78; sources A and B only, elsewhere e.gar) 

da.ga nam.ku 3.zu d lamma a 2 bi 2.ib.œar 

Organization and wisdom overcome good fortune 

(45) Lugalbanda in the Wilderness [1.8.2.1], ll. 42-46 (Karahashi 2000, 80; Vanstiphout 

2003, 106-107) 

42. zi.du ßa 3 kuß 2.u 3 d en.lil 2.la 2.ke 4 

When the righteous one who takes counsel 

66 

with Enlil (i.e. Enmerkar), 

ki 

43. kul.aba4 niœin2.na.bi took away the whole of Kulaba (on 

im.ma.an.kar.eß.ba expedition), 

? 

44. [x] ≠kur±.ra.ke4 u8 .gin7 They wended their way through the . . . of 

ba.an.gur.ru.uß (√gur-eß) the mountains like ewes, 

45. [x x] ≠zi.ga?± gaba They (-eß) ran over the high . . . of the 

˙ur.saœ.œa 2.ke 4 am.gin 7 

du 10 bi 2.in.œal 2.≠le± (√œal-eß) 

foothills like wild bulls, 

46. [x x] za 3.ße 3 i 3.kiœ 2.e ˙ar.ra.an Each (?) . . . was looking for a passage, 

i 3.zu.ne and they found a road, 

In (45), note that the plural suffix, *-eß, occurs in both lines 44 and 45, although the 

orthographic form is quite different in each case: ba.an.gur.ru.uß as opposed to du 10 

bi 2.in.œal 2.le. The alternative would be to analyze bi 2.in.œal 2.le as a verb with the marû

verbal suffix, *-e, but such an interpretation would make this verb nearly unique in the 

entire set of BNBV inal predicates, which do not, as a rule, occur in the marû. 

(46) Enlil Hymn A [4.05.1], l. 40 (Falkenstein 1959, 44; Karahashi 2000, 82) 

37. bara 2 nam.˙e 2 e 2.kur e 2 za.gin 3 

sa˙ar.ta ßa ba.ra.an.il 2 

He made the dais of abundance, the Ekur, 

67 

the shining temple, rise from the soil, 

38. ˙ur.saœ il 2.la.gin 7 ki sikil.la He has made it grow in the pure place like 

bi 2.in.mu 2 

39. nun.bi kur gal a.a d en.lil 2 

40. e 2.kur eß 3 ma˙.a bara 2.ba dur 2 

a mountain lifted up, 

As for its prince, the Great Mountain, father 

Enlil, 

He has taken his seat on the dais of the 

bi 2.in.œar Ekur, the lofty shrine, 

41. e 2.a me.bi diœir sag 2 nu.di As for the rituals of the temple, no deity 

scatters them, 

42. ßu.lu˙ sikil.bi ki.gin 7 nu.silig.ge As for its hand-washing rites, they are 

everlasting like the earth, 

(47) Collection of Temple Hymns [4.80.1], l. 466 (Sjöberg and Bergmann 1969, 136; 

Karahashi 2000, 82; cf. Michalowski 1993, 154) 

463. nun.zu d gir 3.ra lugal As for your prince, Gira, the 

meß 3.lam.ma lord of Meshlam, 

464. diœir ˙uß ki.a lugal u 4.ßu 2.[uß] The fierce god of the underworld, the 

lord of Udshush,

465. d nergal d meß 3.lam.ta.e 3.[a] Nergal, Meshlamta’eda, 

466. muß 3.za e 2 bi 2.in.gub He has erected a house in your precinct, 

bara 2.za dur 2 bi 2.in.œar and taken his seat upon your dais, 

467. 10 e 2 d nergal gu 2.du 8.a ki 

10 lines: the house of Nergal in Gudua, 

(48) The Marriage of Sud [1.2.2], ll. 72-76 (Civil 1983, 54, 59, 62; Karahashi 2000, 83) 

72. [lugal.zu.ur 2 ki.ur 3 ma˙].a.ni Inform your lord thus in his august Ki’ur, 

ur 5.gin 7 du 11.mu.na.ab 

73. [ d en.lil 2.ra itima ku 3 si.ga].≠na± and repeat this to Enlil in the privacy of his 

u 3.ga.na.de 3.ta˙ holy bedchamber, 

74. [...] ra a 2 mu.da.an.aœ 2.≠ta± After . . . had instructed, 

75. [...] ≠DU± d nuska dur 2 bi 2.in.œar and Nuska took his seat on it, 

Karahashi transliterates line 75 as follows: [œiß] gu.za ßu.niœin.na mu.un.na.gub d nuska 

dur 2 bi 2.in.œar, “She set out the chair of honor for him and seated Nuska on it,” but since 

neither ETCSL, nor Karahashi provide information on sources, it is difficult to decide 

between them—here I adopt ETCSL’s transliteration. 

(49) Schooldays 42 (Karahashi 2000, 95) 

41. um.mi.a.œu 10 ßu.zu As for my professor, he said, “your (scribal) 

nu.sag 5.sag 5.e.ße in.tud 2.de 3.en hand is not good,” and he caned me, 

42. nam.dub.sar.ra gu 2 bi 2.du 3 

You have neglected the scribal art, you have 

nam.dub.sar.[ra im.ßub bi 2.ak] let the scribal art drop, 

68

43. um.mi.a.œu 10 ka.ni As for my professor, he was not able . . . his 

nu mu.e.da.an [...] mouth, 

(50) Schooldays 70 (Karahashi 2000, 95) 

68. tug 2 mu.ni.mu 4 niœ 2 mu.na.an.ba He dressed him in a (new) garment, gave 

˙ar ßu.na bi 2.in.du 3 

69 

him a gift, and put a bracelet on his 

hand, 

69. um.mi.a ßa 3 ˙ul 2.la.ni.ta eme The professor with joyful heart gave him a 

mu.na.DU speech, 

70. lu 2 tur inim.œu 10.ße 3 gu 2 

Young man, because you did not neglect my 

li bi 2.du 3.a im.ßub li bi 2.in.ak word, did not forsake it, 

71. nam.dub.sar.ra saœ.bi us 2.sa Reach the pinnacle of the scribal art, 

za 3.bi.ße 3 til.la Finish it in its entirety, 

72. niœ 2.na.me la ba.gi 16.ba.na.za.am 3 

It is something that cannot be obstructed 

ßu.œu 10.ße 3 mu.e.ßum 2.ma that you have put into my hand, 

(51) Bird and Fish [5.3.5], ll. 13-15 (Karahashi 2000, 129) 

13. [a.a d ]≠en±.ki.ke 4 

ambar.ambar.ra ka bi 2.in.keß 2 

gi sumun gi ˙enbur 2 bi 2.in.mu 2 

Father Enki knit together the marshlands, 

making young and old reeds grow there, 

14. [x] x pu 2 uma˙ ma˙ ku 6 mußen As for the birds and fish made to live in 

mi.ni.lu.ug . . . pools and lagoons,

15. [x x] ki niœ 2 zi pa.an.pa.ne.ne gu 7 

naœ.bi.ße 3 ba.an.ßum 2 

He gave them . . . all kinds of living 

70 

creatures as their sustenance, 

(52) Inninshagura, manuscript D, ll. 104-108 (Sjöberg 1975, 188-189; Karahashi 2000, 

134) 

104. d a.nun.na giri 17.bi gam.e.eß giri 17 

As for the noses of the Anuna, they have 

ba.e.dur 2.ru.ne.eß prostrated themselves noses bent (to the 

105. ur gal.gal imin.bi ba.e.u 5 

an.na ba.e.e 3.de 3 

ground), 

You ride on seven great dogs, you come out 

from heaven, 

106. an gal.e muß 3.za ni 2 bi 2.te The great An feared your . . ., he was 

ki.tuß.zu im.mi.˙u.lu˙ frightened of your dwelling-place, 

107. ki.tuß an.gal.la.ke 4 dur 2 

He let you take your seat in the dwelling- 

≠ba ? .e ? ±.œar su nu mu.e.da.zi.zi place of the great An (and then) he did 

not fear your any more, 

(53) Gilgamesh and Huwawa, version A [1.8.1.5], ll. 183-187 (Edzard 1991, 229; 


183. tug2 a.≠œa 2±.la 2 bi 2.in.ßub saœ.du.ni They threw down the leather bag and made 

bi 2.in.e 3.de 3 

184. igi d ! 

en.lil2.la2.ße3 im.ma.ni.in.œar.re.eß Enlil, 

his head come out (of the bag), 

As for that which they had placed in front of

185. d en.lil 2.le saœ.du d hu.wa.wa As for Enlil, when he spotted Huwawa’s 

igi ba.ni.in.du 8.a head, 

186. inim d bil 4.ga.mes What Gilgamesh had done angered him, 

ßa 3 bi 2.in.≠dab 5± 

187. a.na.aß.am 3 ur 5.gin 7 

i 3.ak.en.ze 2.en 

“Why did you (pl.) act in this way?” 

(54) Heron and Turtle [5.9.2], l. 102-106 (Karahashi 2000, 174) 

102. gud 3 us 2.sa.a.œu 10 

ba.ni.in.su˙ 3.su˙ 3 

As for my nest, it was turned upside down, 

103. dumu.œu 10 a.e ba.ni.in.ra.ra As for my young, they were knocked into the 

104. mußen me.en lu 2 

igi ku 10.ku 10.ga.ke 4 

105. niœ.bun 2.na saœ.ki.œu 10 

umbin bi 2.in.la 2 

106. uß 2.œu 10 gaba.œa 2 

im.mi.in.ßu 2.ßu 2 

71 

water, 

It is me, the dark-eyed bird, 

The turtle scratched my forehead, 

As for my blood, it covered my chest, 

(55) Lament over Sumer and Ur [2.2.3], ll. 21-26 (Karahashi 2000, 176) 

22. u 4 an.ne 2 kur.kur.ra saœ.ki After An had frowned upon all the lands, 

ba.da.an.gid 2.da.ba

23. d en.lil 2.le igi.ni ki kur 2.ra After Enlil had countenanced an enemy 

ba.an.œar.ra.a.ba land, 

24. d nin.tu.re niœ 2.dim 2.dim 2.ma.ni After Nintu had scattered the creatures that 

za 3 bi 2.in.tag.ga.a.ba she had created, 

25. d en.ki.ke 4 i7 idigna i7 buranun.na After Enki had altered the course of the 

aß 2 bi 2.in.bala.a.ba Tigris and Euphrates, 

26. d utu ˙ar.ra.an kaskal.e nam After Utu had cast his curse on the 

ba.an.ku 5.da.a.ba highways, 

(56) Nippur Lament [2.2.4], ll. 89-93 (Karahashi 2000, 176) 

89. aœ 2.gig ur 5.ra.a. a.ße.er.ra Among dejection and lamentation, he has 

mu.un.du 3 

72 

made it a sacrilege, 

90. ta.ße eœer na.aœ 2.bi i.bi 2.ni.a After its fate, how long till his face would be 

la ba.an.barag 2 

streaked with teardrops? 

91. aœ 2 sikil du 3.a.gin 7 ur 5.re He rejected it as though it were a 

za 3 bi 2.in.tag blasphemy! 

92. sig 4.bi ˙ul 2.e a.na.aß ib 2.ta.an.e 3 

Why has joy left its brickwork? 

93. œi 6 an.bar 7.ba ßa 3.ba er 2 bi 2.in.si At midnight and midday, tears filled his 

heart,

(57) Sargon Legend, 3 N-T 296, ll. 23-27 (Cooper and Heimpel 1983, 75, 77; Karahashi 

2000, 181) 

24. [i 7] ≠ma˙ i 7 mud±.ße 3 œa 2.ra For me, she drowned you in a great [river], 

mu.un.≠gir 5±.re.de 3.en a river of blood, 

25. [x] ≠x± [d] ≠ur±.[ d za].ba 4.ba 4 

… Urzababa chewed his lips, became 

nundum zu 2 bi 2.in.≠gu 7± seriously afraid, 

26. [x x] x na ≠AN± [x] sukkal.a.ni He spoke to … his chancellor, 

≠gu 3± mu.na.de 2.e 

27. […] ≠nin 9.e±.œu 10 ku 3 d inanna.ke 4 

“… my royal sister, Inanna,” 

(58) SP, UM 55-21-278 = 3 N-T 179 (Karahashi 2000, 181) 

ka 5.a zu 2 bi 2.in.sud 2.sud 2 

A fox bites. 

The extensive set of examples in this section show a regular case-marking pattern 

(adversely affected argument in the locative, *-a, while the positively affected argument 

occurs in the ergative, *-e) that differs from the case-marking pattern in the previous 

section in one detail: whereas the perceived object in BNBV inal verbs of perception 

regularly occurs in the locative-terminative case, *-e, the negatively affected argument in 

BNBV inal verbs of adversity regularly occur in the locative, *-a. I suspect that in the older 

phases of the language, the negatively affected argument in BNBV inal verbs of adversity 

was also followed by the locative-terminative, but that, unlike verbs of perception, the 

*bi-√ prefix in verbs of adversity caused the case-marking on the negatively affected 

73

argument to be changed to locative in those cases in which an explicit postposition is 

written; in some of the cases, however, the negatively affected argument is modified by a 

possessive pronoun or is entirely bare. I would interpret this variation as a sign of the 

uncertainty many scribes faced in assigning a case to these examples. 

74

1.6 Other BNBV inal verbs 

The following examples involve verbs that are neither experiential, nor adversarial. Some 

of these predicates probably do not, properly speaking, belong to the BNBV inal class. 

(59) The Exploits of Ninurta [1.6.2], ll. 348-352 (Karahashi 2000, 78) 

348. d nin.urta dumu d en.lil 2.la 2.ke 4 

gal.bi i 3.œa 2.œa 2 

Ninurta, the son of Enlil, wanted to do 

75 

something big, 

349. gu.ru.um na 4 kur.ra mi.ni.in.ak As for the pile of stones that had been 

350. dungu diri.ga.gin 7 

accumulated in the mountains, 

Since it (pile of stones) stretched across 

a 2 bi 2.in.su 3.su 3.ud them (mountains) like floating clouds, 

351. bad 3 ma˙.gin 7 kalam.ma igi.ba Blocking the view of the land like a huge 

bi 2.in.tab.≠ba± wall, 

352. œiß bal ki.ßar 2.ra mu.ni.in.œar He installed a spigot (?) at the horizon, 

This passage appears several times in the course of the dissertation and the form of the 

verb in line 350 is cited in recent secondary literature in two forms: bi 2.in.su 3.su 3.ud and 

bi 2.in.su 3.su 3.e: ETCSL offers the former, Karahashi the latter (for further discussion of 

the form of the verb in the text-artifactual record, see example [49] in section 2.3 below).

(60) A praise poem of Shulgi (Shulgi B) [2.4.2.02], ll. 52-55 = 77-80 

52. niœ 2 ak.a.œa 2 ni 2.bi ga.buluœ 5 

As for what I myself have done, let me boast, 

53. kal.ga.œa 2 mu.bi a 2 bi 2.su 3.ud As for the fame of my power, it has spread 

76 

far and wide, 

54. œeßtu 2.œu 10 niœ 2 galam.ma As for my wisdom, it is full of subtlety, 

su 3.ga.am 3 

55. a.na.œu 10 niœ 2 ka.ge dib.ba nu As for my achievements, do they not surpass 

(61) Tree and Reed [5.3.4], ll. 2-6 (Karahashi 2000, 96) 

all qualifications? 

2. [ki] daœal.e ku 3 na4 za.gin 3.bi bar.bi Precious metals and gems were inlaid on the 

am 3.mi.ib.si outer wall of the vast place, 

3. na4 esi na4 nir 2 na4 gug sud.ra 2.aœ 2.œa 2 

ßu tag ba.ni.in.du 11 

4. [ki] u 2.ßim.e ˙i.li gu 2 bi 2.e 3 

Diorite, chalcedony, cornaline—all the 

shining ones were applied, 

The grassy ground is covered in luxury and 

nam.nun.ba mu.un.gub stands in its princeliness, 

5. ki ku 3 ki sikil.la an.ku 3.ra ni 2.ba The holy place, the pure place—for holy An 

mu.na.sig 7 

6. an diœir ma˙.e ki daœal.la du 10 

im.ma.ni.ib.nir sweet, 

it is by itself beautiful, 

An, the lofty god, made the good place 

The version (and much of the translation) reproduced above is based on manuscript A 

(= Ni 4463) in the unpublished edition of Vanstiphout. Manuscript LO (= AO 6715)

differs from manuscript A in a number of places: in line 2, bar.ba instead of bar.bi; in 

line 4, bi 2.ib 2.e 3 instead of bi 2.e 3 and nam.nun instead of nam.nun.ba; and in line 5, 

ni 2.bi instead of ni 2.ba. 

(62) Shulgi and Ninlil’s Barge (Shulgi R), manuscript A [2.4.2.18], ll. 90-92 

(Klein 1990a, 108-109; Karahashi 2000, 96) 

92. lugal me.lam 2 huß gu 2 bi 2.e 3 

The king is clad in a terrifying luminosity, 

Line 92 is a catchline meant to indicate the incipit of a following tablet; the first half of 

line 92 also occur in line 21 of the Ur III catalog at Yale [0.1.2]. In light of the fact that 

the line occurs without the benefit of context, there is probably not much this example 

can tell us. 

(63) Winter and Summer [5.3.3], ll. 195-199 (Karahashi 2000, 99) 

195. ßar.ra.ab.du 8 ni 2 bur 2.bur 2.ra ßa 3 

The bragging sharabdu who does not 

a.ßa 3.ga nu.zu know the extent of the field, 

196. e 2.me.eß ßeß.œu 10 a 2.kuß 2.a.œu 10.a Summer, my brother, after you have gone 

ni 2 buluœ 5 u 3 mu.ni.e 3 

77 

out boasting about my toil, 

197. za 3 mu.ka ße e 2.a u 3.mu.ni.in.ku 4 When at the turn of the year grain is brought 

guru 7 gu 2 bi 2.in.gur.gur into the houses in order to fill the 

198. diri u 3.me.de 6 tug2 bar.dul 5 

granaries, 

When you bring surplus, your bardul 

tug2 niœ2.lam 2.zu bal.a.kam and niglam garments are . . .,

199. uruda ˙a.zi.in 2(diß) ma.na lu 2 

u 3.mu.ra.an.ßum 2 edin.zu.ße 3 

ba.du.un 

When someone gives a two mana axe to you, 

78 

you go off to your steppe, 


288. en.te.en.e inim.ma gu 4 ma˙.gin 7 

saœ im.il 2.il 2.e to speak: 

289. a.a d en.lil 2 nam.ku 5.œal 2 

mu.e.ßum 2 a he 2.œal 2.la mu.e.de 6 

Like a great bull Winter raised his head 

"Father Enlil, you gave irrigation-control 

and brought abundant waters, 

290. a.gar 3.a.gar 3.ra za 3 ba.ni.tag.tag The plantations were set up, lit., adjoined, 

guru 7 gu 2 bi 2.gur.gur in order to fill the granaries, 

291. ße ab.sin 2.na gu 2 ba.ni.in.peß The grain grew thick in the furrows, 

292. d ezina 2.e ki.sikil sa 6.ga.gin 7 

And Ezina came forth in splendor like 

ni 2 pa e 3 bi 2.ak a beautiful maiden, 

ETCSL translates the line in question as “I made one meadow adjacent to another and 

heaped high the granaries,” whereas Karahashi translates: “I will remove (the floor 

water) from the cultivated fields and pile up silos” (Karahashi 2000, 99). a.gar 3 

presumably refers to what is known in German as der Flur, meaning a collection of 

farmlands; the use of a.gar 3.a.gar 3 here would seem, therefore, to refer to the entire 

expanse of collected farmland in a community. I have translated it as “plantations” for 

lack of a better term. This example is dealt with further in section 2.3, example (34).

Although the three semantic classes into which I have divided up the BNBV inal 

predicates show some variation in the form of the postposition following the perceived or 

negatively affected argument, the single most important criterion of the BNBV inal class is 

manifest throughout: as a rule, the constructions lack the causative semantics typically 

found in other *bi-√ prefix constructions and the ergative case-marked noun is an 

experiencer of the event rather than an agent responsible for bringing it about. 

79

1.7 Possible motivations for the BNBV inal distributional class 

Given the correlation between a formal test (bare nominal immediately before a *bi-√ 

prefix verb), a typological parallel (low source applicative constructions) and a relatively 

coherent semantics (verbs of perception and privation, which form a single coherent 

semantic class under the constraints of a low source applicative hypothesis), one might 

reasonably ask why these formal properties in particular are used to code this semantic 

category in Sumerian. I argue in the balance of this study that the two crucial factors 

determining the formal properties of the BNBV inal construction in Sumerian are 

indefiniteness (or its referential equivalent coded through other means) and transfer of 

possession from a source to a goal as indicated by the absence of resumptive 

pronominalization on the nominal component of a BNBV inal construction: any kind of 

possessive pronoun functioning resumptively would have to choose between resuming 

either the source argument or the goal argument, when the crucial factor in the BNBV inal 

construction is the transfer rather than the possession itself. 

Indefiniteness has not been investigated in Sumerian in any detail and Sumerian 

nouns are not explicitly identified as either definite or indefinite by distinctive 

morphology. Certain kinds of nouns and nominal constructions such as possessive 

constructions are often translated into European languages using definite nouns, but this 

may be misleading in that there is no necessary correlation between occurrence in a 

possessive construction and definiteness (Lyons 1999, 22-26). Nonetheless, if one 

assumes for the sake of argument that possessive pronouns in Sumerian are indicators of 

definiteness, certain phenomena fall into place: the use of nouns bearing possessive 

80

pronouns to form topic constructions can be explained rather straightforwardly if the 

addition of a possessive pronoun makes the noun definite, since discourse topics are 

regularly definite cross-linguistically. As emphasized by Woods, many of these 

supposedly possessive pronouns could be reinterpreted as demonstrative pronouns 

(Woods 2001, 87-94), but the same topic-marking function also occurs with first and 

second person possessive pronouns (which are, of course, deictic, but not demonstrative 

in the ordinary sense of the term). Compare, for example, (65), (66) and (67), setting 

aside for the moment questions of resumptivity in the anticipatory genitive in example 

(67): 

(65) Erßemma 88: 18-19 (Cohen 1981, 84; Woods 2001, 92) 

er 2 im.me [ Topic er 2.bi] gig.ga.kam ßir 3 im.me [ Topic ßir 3.bi] gig.ga.kam 

I weep. As for the tears, they are bitter. I sing. As for the song, it is bitter. 

(66) Gudea, Cyl. A 9:25-26 [ETCSL 2.1.7] (Zólyomi 1993, 133) 

[ Topic igi ˙uß.a.œu 10] kur.re nu.um.il 2 

As for my terrible look, no country can bear it. 

(67) Gudea, Cyl. A 5:4 [ETCSL 2.1.7] (Zólyomi 1993, 35) 

[ Topic e 2.a œiß.˙ur.bi] im.œa 2.œa 2 

As for the plan of the temple, he was setting it out. 

81

Note that I differ from Zólyomi in taking the entire anticipatory genitive in (67) as the 

topic—Zólyomi only includes e 2.a in the topic and sees the -bi of œiß.˙ur.bi as a 

resumptive pronoun. In examples (65), (66) and (67), the possessive pronoun demarcates 

a topic to the left from the rest of the sentence to the right: in (65), er 2 im.me constitutes 

a distinct clause and does not violate the generalization. It could be argued that in (65) 

and (67) the possessive pronoun is demonstrative in some fashion, but this is clearly not 

the case in example (66). Note that—in principle at least—every noun in the 

absolutive/nominative (zero-marked) case has available to it one of two possible 

“possessive” constructions: where the possessor is a lexical noun phrase (as opposed to a 

pronominal), the anticipatory genitive is regularly available for coding topicality through 

a possessive pronoun-final construction, whereas in the absence of a lexical noun phrase, 

the possessor can be coded more-or-less vacuously (since the possessive relation is 

presupposed and the addressee is presumably already aware of the possessive relation 

anyways) through the use of the ordinary possessive pronoun. 

Although I do not attempt to establish the truth of such a possessive pronoun theory 

of topic-marking any further herein, I think it does suggest the plausibility of associating 

possessive constructions in Sumerian with definiteness. This is also in conformity with 

typological expectations in that in languages in which determiners (a somewhat general 

category subsuming demonstratives) and possessives occupy the same structural position 

as in Sumerian, the occurrence of a noun within a possessive construction generally 

indicates that the noun is definite (similar languages include English and Irish, known as 

“determiner-genitive” languages in opposition to “adjectival-genitive” languages such as 

82

Italian and Greek, see Lyons 1999, 22-25 with references). Bare nouns in Sumerian, 

therefore, have at least the possibility of being indefinite. 

The other piece of, admittedly, indirect evidence in favor of linking the BNBV inal 

distributional class with indefiniteness has to do with the location of the so-called 

dimensional infixes (a class of directional morphemes that appear in the verbal prefix) 

within the Sumerian verb: in purely formal terms, the *bi-√ prefix excludes the entire set 

of dimensional infixes. In the system of numbered morphosyntactic slots proposed by 

Jagersma/Zólyomi, the dimensional infixes occupy slots [ ] 7 through [ ] 9, whereas the 

*bi-√ prefix, segmented into /b/ and /i/ fills slots [ ] 5/6 and [ ] 10. In the slightly different 

system in use at the Pennsylvania Sumerian Dictionary, the dimensional infixes fill slots 

[ ] 4, [ ] 6, and [ ] 8, while the *bi-√ prefix is split into slots [ ] 3 and [ ] 9. The point that is 

recognized by both systems is, as plainly stated by Gragg, “[t]he dimensional infixes, as a 

rule, do not occur with bí-, im.mi-, or al-” (Gragg 1973a, 15). But more importantly, 

Gragg’s seminal work on the dimensional infixes makes clear that there is no regular 

“agreement” between postpositions and dimensional infixes: 

In these literary texts there are approximately 700 occurrences of the postposition 

- ß è . But the infix - ß i - occurs only 142 times. … of these 142 occurrences of the 

infix, only 35 give the infix and the postposition together. While these statistics do 

not disprove the view just outlined [that the infixes arise through “agreement” with 

co-occurring postpositions], they do make it seem less likely; in fact, given these 

statistics one might well hypothesize that the infix and the postposition tend to be 

mutually exclusive rather than interdependent” (Gragg 1973a, 16) 

83

Gragg’s statistics, based on a corpus of Old Babylonian materials from Nippur is equally 

convincing (note that some of Gragg’s numbers are approximate). 

Postposition Infix Co-occurrences 

-ße 3 / -ßi- 700 142 35 

-ta / -ta- 380 100 29 

-da / -da- 170 430 26 

Gragg emphasizes that, with the exception of clauses that include a verb of motion, the 

postposition and its etymologically corresponding infix almost never occur in the same 

clause. 

Although further research into the distribution of postpositions and their 

etymologically related infixes is needed, I would guess that, excluding verbs of motion, 

the postposition regularly occurs with definite lexical noun phrases, the infix when the 

lexical noun is pronominalized and the *bi-√ prefix when the relevant adverbial argument 

is indefinite. This would make the *bi-√ prefix the unmarked term in the system of 

dimensional infixes, a system that is organized on the basis of oppositions of source (-ta) 

and goal (-ße 3), which coincides with the semantic field that I would associate with the 

*bi-√ prefix. The plausibility of *bi-√ as the unmarked dimensional infix follows quite 

naturally from its role as a low source applicative when it is immediately preceded by an 

inalienable noun. We can presumably juxtapose the importance of the source argument in 

the Sumerian low source applicative in BNBV inal across a number of distinct lexical 

classes: (i) verbs of perception include an indefinite noun in a raised possessor 

construction (with the experienced object as the raised possessor of an inalienable noun 

84

that refers to the medium through which the perception was achieved) as the source of the 

perception, (ii) verbs of adversity in which the negatively affected entity is the source in 

juxtaposition to the positively affected entity as the goal, and (iii) verba dicendi in which 

the object transferred from a possessor at the source to a possessor at the goal is simply 

the word that the speaker utters. In each case, the *bi-√ prefix indicates that there has 

been a transfer of possession from a source to a goal, where the object being transferred, 

at least in grammatical terms, is indefinite. 

85

1.8 Ergative possessors and the transfer of possession model 

The model underlying Pylkkänen’s low source applicative hypothesis is a transfer of 

possession model, but rather than attempting to describe the model in abstract terms, I 

would like to return to the Korean example (21) dealt with above, repeated below as (68). 

(68) totwuk-i Mary-hanthey panci-lul humchi-ess-ta 

thief-Nom Mary-Dat ring-Acc steal-Past-Plain 

The thief stole a ring from Mary (and it was in her possession when he stole it) 

In a transfer of possession model, the direct object of the verb starts out in a raised 

possessor construction, in which the source or the starting point of the transfer is the 

raised possessor, Mary-hanthey in (68), and the goal, recipient or endpoint of the transfer 

is the subject of the sentence, totwuk-i in (68). The possessive relation at the beginning of 

the event is coded through the use of a raised possessor construction, whereas the 

possessive relation at the end of the event is part of the lexical meaning, “to steal,” of the 

main verb: there are, therefore, in a rather special sense of the term, two predicates at 

work in any particular low source applicative construction, both of which encode a 

possessive relation of some kind. The Korean example, however, differs from the 

foregoing Sumerian examples in that the lexical meaning of the vast majority of the 

verbal roots used to form BNBV inal predicates in Sumerian encode, at most, existence in a 

location (œal 2, œar, gub) or attachment (du 3, keß 2, te, dab 5, la 2, tag), if even that (dar, gar 3, 

su 3, *tuk [derived from /œeßtug/ “ear,” not tuku “to possess”], du 3, œar, du 8, kar 2). Where 

86

then is the resultant possessive relation between the direct object and the recipient coded? 

The answer lies, I believe, in a little known phenomenon that occurs with the possessor of 

an inalienable noun in languages that make use of ergative case-marking. 

In West Greenlandic Inuit, the possessor of an inalienable noun that has been 

incorporated into the verbal complex regularly occurs in the ergative case, as in the 

following examples (Geenhoven 1998, 18, fn. 4; cf. Sadock 1991, 96-97; Rischel 1971, 

234). 

(69) kunngi-p panik-passna-qar-p-u-q 

king-Erg daughter-many-have-Indic-Intrans-3Sg 

There are many king’s daughters, i.e., princesses 

(70) hansi-p qimmi-p ame-qar-tip-p-a-a 

Hans-Erg dog-Erg skin-have-Cause-Indic-Trans-3Sg-3Sg 

Hans let him have a dog’s skin 

(71) sisimiut sissa-p naalaga-qar-p-u-q 

Sisimut shore-Erg chief-have-Indic-Intrans-3Sg 

Sisimut had a chief of the shore 

In these examples, it is clear that the ergative case-marked noun in (69) and (71) and the 

ergative nominal that is closest to the verb in (70) are not, in any sense, agents in the 

87

sentences in which they occur, but rather that the particular means of coding an 

inalienable possessor in West Greenlandic Inuit is through the use of the ergative case. 

The importance of the analogy is not simply that the resultant possessive relation in 

the transfer of possession in BNBV inal predicates is coded by the ergative case-marking 

on the ultimate possessor, it also makes sense of the fact that the ergative case-marked 

nominal in BNBV inal constructions is an experiencer, whereas the ergative case-marked 

nominal in other BNBV constructions that do not involve an inalienable noun are causers, 

prototypical agents. The primary function, therefore, of the ergative case-marking in 

BNBV inal constructions is to indicate the possessive relationship between the inalienable 

noun and its possessor rather than to mark the nominal argument which brings about the 

event in question: there are consequently two raised possessors in BNBV inal constructions 

and the bare noun is transferred from the possession of one to the possession of the other 

without the intervention of an agent of any kind. 

Source Goal 

Locative-terminative Ergative inalienable possessor 

Note, however, that even though both possessors occur in raised possessor constructions, 

the resultant possessive relationship is coded by an element of the nominal syntax, 

whereas the initial possessive relationship is coded through the use of a clause-level 

raised possessor construction such as “there is a DP possessed at (the location of) DP possessor” or 

“DP possessed belongs to DP possessor.” At the same time, both of the raised possessor 

88

constructions that occur in the BNBV inal construction are prototypical environments for 

indefinite nouns that exhibit what is known as the definiteness effect (a syntactic 

phenomenon that restricts the noun in certain syntactic positions to an indefinite noun, 

see chapter 3 for details). 

Within a wide variety of different types of case-marking systems, one often finds a 

distinctive case-marking subsystem that uses the unmarked or default case (nominative or 

absolutive) in opposition to a dative or other oblique case so as to code verbs of 

perception and/or volition. The main characteristic of such predicates is the presence of 

two arguments, typically described as “theme” and “experiencer” and the association of 

these two arguments with a nominative/dative or absolutive/oblique case-marking 

system. Givón, for example, speaks of “verbs of cognition, sensation or volition, whose 

object registers—logically—no discernable impact or change. In fact, it is the 

experiencer-subject which registers some internal/cognitive change. Common verbs in 

this class are ‘see’, ‘hear’, ‘know’, ‘understand’, ‘think of’, ‘want’, ‘feel’, and most 

commonly they are semantically states rather than actions” (Givón 1984, 100). Givón 

notes that many such verbs of cognition, sensation and volition occur in English in a 

construction that makes use of an expletive ‘it’ and a dative experiencer such as the 

following examples (Givón 1984, 101, ex. 39). 

(72) a. It seems to me that … (cf. I see/think that …) 

b. It appears to me that … (cf. I find it apparent that …) 

c. It is necessary that … (cf. I insist that …) 

89

In the examples in (72), a dummy pronoun appears in the subject position, while the 

experiencer appears in a dative prepositional phrase immediately before that which has 

been perceived, i.e., the theme. Predicates of this type also bear a strong resemblance to 

existential verb constructions such as, for example, “there is a rat in the soup” (see below, 

section 3.4, for further explication). In both types of construction, we find three crucial 

components: (i) a dummy pronoun or expletive in the subject position before the verb, (ii) 

a theme or perceived object/situation that follows the verb in some cases but cannot be 

characterized as a logical object or semantic patient, and (iii) an oblique argument that 

refers to either the location in which the object/situation occurs (as in the existential 

sentence) or the person to whom the object/situation is apparent as in (72). Given the use 

of the ergative postposition to refer to an inalienable possessor as outlined above, one 

way of thinking about the case-marking relations that hold between the different nominal 

phrases in a BNBV inal clause is that the locative-terminative source argument represents 

an oblique case and that the combination of the bare inalienable noun and its raised 

inalienable possessor act as a single nominative/absolutive case-marked argument. 

In my view there are two important corollaries that follow from the case-marking 

relations that hold between the zero-marked nominative/absolutive experiencer + 

inalienable noun and the oblique case-marked perceived object: (i) the non-agentive 

character of the perceiver in combination with the stative and/or resultative semantics of 

the predicate and (ii) the fact that surface form of BNBV inal predicates resembles the 

surface form or an ordinary transitive clause with ergative agent, *-e, and zero-marked 

absolutive patient (where the perceived object in the locative-terminative case is 

90

analogous to the ergative agent of a transitive clause and the bare inalienable noun + 

inalienable possessor is analogous to the zero-marked absolutive patient), but that the 

thematic roles of the participants are theme and experiencer rather than agent and patient. 

The first of these two corollaries is dealt with in the following chapter, where the non- 

agentivity of the experiencer and the stative/resultative nature of the BNBV inal class are 

dealt with at length; the second corollary is taken up in detail at the end of the dissertation 

(see chapter 5 for details). 

91

2 

Decomposing lexical aspect: 

Telicity, the particle-verb construction and secondary predication 

Having outlined a rather involved set of case-marking relations in the BNBV inal 

construction at the end of chapter 1—the transfer of possession from a raised locative- 

terminative source to a goal in the ergative case—this chapter seeks to clarify several 

nagging issues that have to do with lexical aspect and whether or not aspectual properties 

interact with this curious case-marking regime in any palpable way. I argue that, in fact, 

the BNBV inal construction provides a useful diagnostic class for resolving a number of 

outstanding problems in previous explanations of the lexical aspectual system in 

Sumerian, notably Yoshikawa’s proposal that the introduction of ßu “hand” into an 

otherwise atelic predicate (forming, in many cases, a BNBV inal construction) converts it 

into a telic predicate. But prior to turning to Yoshikawa’s proposal, several preliminary 

steps are necessary: in section 2.1, a number of basic theoretical tools such as the 

opposition between lexical and grammatical aspect, the Vendlerian classification of 

lexical aspectual categories and the opposition between depictive and resultative 

secondary predication are introduced; in section 2.2, an analogy is drawn between 

92

depictive and secondary predicates in English and what Yoshikawa termed Gtn-stem and 

D-stem reduplications in Sumerian (based on co-occurring Akkadian verbs in bilingual 

texts); and in section 2.3, the classification of BNBV inal predicates in terms of secondary 

predication is investigated with largely negative results. 

The fact that BNBV inal predicates do not fit into Yoshikawa’s Gtn-stem/D-stem 

opposition implies, for a number of reasons, that the BNBV inal predicates are 

achievements in the typology of lexical aspectual classes formulated by Vendler (1967) 

and subsequently extended in a number of ways by Dowty (1979). The classification of 

BNBV inal predicates as achievements (which can be characterized as a telic event in which 

the change between the initial state and the final state is instantaneous) fits quite nicely 

with the particular semantics of the members of the class, particularly when the class is 

defined in somewhat narrower terms using the additional criteria identified at the end of 

section 2.3. But a formal diagnostic of some kind is still needed if only so as to prevent 

circular argumentation since there is no independent source for semantic evaluations 

other than the attestations of the particular verbs in context, that is to say, there is—in a 

strict sense—no independent source for semantic evaluations whatsoever. There are only 

two substantial tests for diagnosing an achievement predicate: achievements do not 

normally appear in the progressive aspect (Rothstein 2004, 36-58), and when they do 

appear in the progressive aspect, it is because they have been converted into a derived 

accomplishment (Rothstein 2004, 123-147). Although “progressive aspect” (the term 

used by Rothstein) is not necessarily equivalent to the “imperfective (marû) aspect” 

familiar to most Assyriologists, in the following I use the term “progressive/imperfective 

93

aspect” as a rough equivalent to “progressive aspect” and “non-progressive/perfective 

aspect” as its opposite. Derived accomplishments, in turn, come in at least two varieties 

in English, progressive/imperfective achievements (“Dafna is finding her shoes” 

[Rothstein 2004, 36, ex. 2b]) and light verb constructions (“John gave the floor a clean 

sweep” derived from the resultative “John swept the floor clean”) and derived 

accomplishments in Sumerian seem to exhibit some properties of both these 

constructions. In section 2.4, the type of derived accomplishment constructions that are 

criterial for the identification of achievement predicates are introduced on the basis of 

English data and the non-agentive character of achievements is elaborated, whereas in 

section 2.5, the distributional facts associated with derived accomplishments in Sumerian 

are presented in some detail and lead, in turn, to a redescription of the function of ßu in 

certain BNBV inal constructions. 

The penultimate section of the chapter differs from the earlier ones in that it 

correlates three of the phenomena associated with BNBV inal constructions earlier in the 

dissertation (transfer of possession, inalienability and telicity) with a particular, 

typologically common syntactic pattern known as a particle-verb construction. Particle- 

verb constructions, which are extremely common in English, are one of the regular means 

of controlling telicity in SVO and VSO languages: the isolation of an adposition in the 

vicinity of the verb that lacks a nominal complement is used to transform atelic predicates 

into telic ones as in the frequent use of directional particles in English (“pour” as opposed 

to “pour out,” “fill” as opposed to “fill in” and so forth). Whereas the raised ergative 

possessor of an inalienable noun in a BNBV inal construction is not presumably involved in 

94

the particle-verb features of the construction (since the ergative case is a “core” or 

“structural” case analogous to the nominative or accusative cases and the combination of 

inalienable noun and its possessor seems to form a single unit in terms of reference and 

thematic role), the raised possessor in the locative-terminative case, however, seems to 

form an existential construction in conjunction with the low source application 

morpheme *bi-√ and, in this section, I propose that the combination of the raised 

locative-terminative possessor, the inherently relational bare noun and the low source 

applicative morpheme act as a unit in SOV languages like Sumerian to introduce telicity 

to otherwise atelic predicates in the same way that stranded adpositions do in SVO or 

VSO languages. On the basis of Szabolcsi’s proposal that weak quantifiers— 

such as the bare noun in BNBV inal constructions—can act as a kind of catalyst for the 

formation of verbal complexes in verb-final languages, I apply the model of complex 

verb formation described by Koopman and Szabolcsi (2000) to the formation of verbal 

complexes in Sumerian, yielding a relatively coherent description of a particle-verb 

construction in an SOV language. The interesting corollary of this proposal is that the 

particular semantics of the set of indefinite body-part nouns that form the nominal 

component of BNBV inal constructions, particularly in terms of their inherently relational 

and orientational features, correspond quite closely to the relational and orientational 

features of the directional adpositions typically found in particle-verb constructions in 

other languages. 

In the last section, I make a first attempt at incorporating the telicity of the BNBV 

construction—in particular, the similarity between particle-verb constructions and 

95

esultative secondary predicates—into Trask’s model for the development of Type B 

ergativity. It looks like word-order (at least with respect to the order of lexical verb and 

verbal prefix) acts as a fairly good indicator of the scope of the existential quantification 

introduced by the *bi-√ verbal prefix. If so, I argue that the shift from ROOT-*bi to 

*bi-ROOT order, which corresponds to the shift from underlying, “imperative” word- 

order to the word-order found in finite verbs, 12 corresponds to a narrowing of the scope of 

the *bi-√ prefix and that this change is a direct result of the grammaticalization of *-e as 

an ergative postposition. In other words, the change in word/morpheme-order resulted 

from the development of ergativity and can be explained as a way of indicating a 

corresponding change in the hierarchical relation (and hence scope relations) between 

verbal prefix and lexical verb. 

12 Michalowski describes the process that forms the imperative construction in Sumerian as “[c]opying the root to the 

front of the verbal form” (Michalowski 2004, 43). But even if Michalowski is correct, the diachronic structure of the 

language need not mirror its synchronic structure. 

96

2.1 Three aspects of predication: Grammatical, lexical and secondary 

As a way of introducing the often subtle interactions between grammatical aspect such as 

the progressive in English (be + -ing as in “Beth is reading a book”) and lexical aspect 

such as, for example, the difference between “to bite” and “to gnaw” in English (the 

opposition is sometimes also labeled using the German term Aktionsart, lit., “type of 

action”), I would like to turn initially to a few examples from English, where the two 

phenomena and their interaction have been studied in considerable detail. To simplify 

greatly, grammatical aspect draws a basic opposition between a perfective event, which is 

represented as a single point in time with no duration, and an imperfective event, which is 

represented as ongoing, extending in time. This opposition need not correspond to our 

own tense-tainted collective imagination: an activity such as “to run” (without a 

durational object or goal) can appear in either the progressive/imperfective or the non- 

progressive/perfective aspect, representing—or so the story goes—two different 

perspectives on the event, internal and external respectively. 

(1) Past perfective Past imperfective 

a. Sam ran. Sam was running. 

b. Sam drove the car. Sam was driving the car. 

c. Sam died. Sam was dying. 

d. Sam built the house. Sam was building the house. 

97

In (1), factoring out tense as much as possible (which is necessary in English since 

present tense perfectives and imperfectives interact in complex ways with the origo of the 

deixis [“now”]), 13 the same predicate can appear in either the unmarked perfective or the 

imperfective. 

What is known as lexical aspect has to do with types of predicates (where 

“predicate” is used as a cover term to refer to both bare verbs as well as verb + direct 

object combinations) classified on the basis of a number of seemingly unrelated 

questions: 

(a) Does the event that the predicate refers to have a natural endpoint? 

(b) Does the predicate refer to an event that requires a certain span of time in which 

it takes places? 

(c) Can the event referred to by the predicate be broken down into distinct stages 

that differ in the degree to which the event is completed in each stage? 

(d) Does the event referred to by the predicate take place because of the activity of 

the subject of the sentence? 

Based on a lengthy tradition of study that, at least in its modern phase, begins with 

Vendler (1967) and passes through a crucial transformation in Dowty’s redescription 

(1979), Agha (1993, 127-153) has proposed that these differences between predicates can 

13 An “origo of deixis” is a point with respect to which deictics of one kind or another are evaluated: given two acts of 

deictic reference oriented with respect to the same origo of deixis, the choice of which is a “near” deictic, “this,” and 

which is a “far” deictic, “that,” is determined by spatiotemporal proximity to the origo of deixis. 

98

e delimited on the basis of four features corresponding to the four questions above: (a) ± 

punctate, (b) ± durative, (c) ± segmental change of state, and (d) ± agentive. If the 

examples in (2) are recast in tabular format and classified on the basis of these features, a 

rough idea of the lexical aspectual classes commonly in use can be elucidated in short 

order. 

Predicate ± punctate ± durative ± seg. change state ± agentive 

(1a) run – + – + 

(1b) drive a car – + – + 

(1c) die + – – – 

(1d) build a house + + + + 

The set of features attributed to these four examples can be described in a plethora of 

ways, but certain classifications and their associated terminology are relatively standard 

and will be used throughout the discussion. Vendler defined four categories based on the 

first three features above: “states” lack all three features ([– punctate, – durative, 

– segmentable change of state]), “activities” such as (1a) and (1b) are ongoing events 

with no inherent endpoint ([– punctate, + durative, – segmentable change of state]), 

“achievements” are events which have both a duration and a telos but are not cumulative 

([+ punctate, + durative, – segmentable change of state]), whereas “accomplishments” 

can be characterized as having all three features ([+ punctate, + durative, + segmentable 

change of state]). The fourth feature, agentivity, is often incorrectly characterized 

99

separately on the basis of whether or not the single nominal argument of an intransitive 

predicate is analogous to the agent of a transitive predicate as in (1a) or patient as in (1c): 

agentive [+ agentive] predicates are often termed “unergative” on the basis of parallels 

with the distinctive case-marking that is used to mark agents in ergative languages, 

whereas patientive [– agentive] predicates are known as “unaccusatives” since their 

single nominal argument is analogous to the distinctive case-marking used to indicate the 

patient of a transitive predicate in accusative languages. In terms of lexical aspect, 

unaccusatives are often quite similar to telic transitive predicates, while unergatives often 

pattern along the lines of transitive predicates that lack telicity. 

Rothstein, whose most recent work (2004) seems to represent a high point in recent 

studies of lexical aspect describes these aspectual oppositions (excluding the ± agentive 

feature) in terms of just two properties in the following table (after Rothstein 2004, 192, 

ex. 1). 

[± stages] 

[± telic] 

= occurs in progressive = naturally heads telic VP 

States – – 

Activities + – 

Achievements – + 

Accomplishments + + 

On the one hand, therefore, activities and accomplishments have internal duration and 

extend in time, whereas states and achievements do not, while on the other hand, 

achievements and accomplishments have a natural endpoint, whereas states and activities 

do not. 

100

In the past few years, investigations of lexical aspect, particularly in formal 

semantics, have also uncovered a third domain of aspectual categorization that, as it turns 

out, is particularly relevant to investigations of Sumerian: the primary examples of this 

phenomenon in the linguistics literature—dominated, as it is, by English 

morphosyntax—are resultative and depictive secondary predication in English. Older 

investigations of secondary predication can be associated quite clearly with what are 

known as small clauses such as “John leave” in “Mary saw [ SC John leave]” (Stowell 

1981; 1983; cf. Stowell 1995 and references therein), but the importance of secondary 

predication as a lexical aspectual phenomenon has only been clearly identified and dealt 

with in the past few years, so I would like to take a moment to introduce it before moving 

on to the specifically Sumerian questions that occupy the rest of the chapter. If we apply 

one of the standard tests for telicity (co-occurrence with bound [“in three hours”] and 

unbound [“for three hours”] temporal adverbials) to a series of increasingly complicated 

English predicates, an interesting pattern emerges (although the following example has 

become the standard one used in the secondary literature, its proximate source is 

Rothstein 2004). 

(2) a. He hammered it for three hours / **in three hours. (“atelic”) 

b. He hammered it out for three hours / in three hours. (“telic”) 

c. He hammered it flat **for three hours / in three hours. (“resultative”) 

101

In the three examples in (2), contrasting temporal adverbials are used to differentiate the 

lexical aspectual properties of the complex predicates “to hammer,” “to hammer out” and 

“to hammer flat.” In (2a), the bare verb can only occur with the unbounded temporal 

adverbial “for three hours,” whereas the bounded temporal adverbial is ungrammatical: 

this indicates that the bare verb “to hammer” is atelic. In (2b), the introduction of a 

particle in a particle-verb construction allows the complex verb “to hammer out” to 

appear with either of the two temporal adverbials, indicating that there is an available 

telic endpoint in the semantics of the verb, but that it need not have been reached. The 

example in (2c), however, is inherently telic or better “resultative” and excludes the 

unbounded temporal adverbial. Note therefore that, even in English, so-called lexical 

aspect is in fact a grammatical phenomenon: particle-verbs as in (2b) show one type of 

telicity, while resultative secondary predicates (see below for definition) as in (2c) show 

another more restricted type of telicity. 

There are in fact a number of different types of what has come to be called 

secondary predication, some of which are illustrated in (3) below (the following 

examples correspond to Rothstein 2004, 60, examples 2a, 2b, and 3). 

(3) a. John i drove the car drunk i. 

b. Mary drank the coffee i hot i. 

c. John painted the house i red i. 

102

In each of these examples, the secondary predicate is in bold and occurs in sentence final 

position in English—the subscripted indices indicate which words are coreferential. 

Examples (3a) and (3b) are “depictive” secondary predicates in that they qualify a 

particular nominal argument throughout the entire duration of the event denoted by the 

main predicate: the implication is that John is drunk throughout the period of time in 

which he drove and that the coffee, likewise, is hot throughout the time it took to drink it. 

The “resultative” secondary predicate in (3c), unlike the depictive ones in (3a) and (3b), 

is only true upon completion of the event denoted by the main predicate, so the house is 

red in its entirety only at the completion of John’s act of painting and, in somewhat 

informal terms, the degree to which the direct object has the property attributed to it by 

the adjective that acts as the secondary predicate corresponds to the degree to which the 

event referred to by the main predicate is complete (see Rothstein 2004 for detailed 

criticism of the view that the direct object can serve as a true index of the relative 

completion of the event, cf. Krifka 1989). The literature on secondary predication is still 

somewhat inchoate and difficult to follow, but Rothstein notes several important points: 

(a) whereas resultative secondary predicates are limited in application to adjectival 

modifiers of direct objects, depictive secondary predicates can modify either subject or 

object of a transitive verb, (b) where the primary verb is intransitive, it can be 

transitivized so as to make available a direct object for secondary predication either 

through the use of a “fake-reflexive” (John laughed himself i sick i [Rothstein 2004, 60, ex. 

4c]) or a “non-thematic object” (John sang the baby i asleep i [Rothstein 2004, 60, ex. 4d]), 

and (c) the use of adjectives as the basis for secondary predication means that the 

103

nominal modified by the secondary predicate is necessarily a non-agentive theme 

argument with respect to the secondary predicate. 14 

In the Sumerological tradition, telicity is nothing new: much of Yoshikawa’s work 

centers on telicity (Yoshikawa 1988 [1993, 72-94]). It is clear from his collected works 

that his use of the two terms, “telicity” and “momentariness,” represents his attempt to 

make sense of, on the one hand, the fact that certain verbs have natural end points 

(“telicity”) such as “to build a house,” while others such as “to run” do not, and, on the 

other hand, the fact that certain verbs lack any inherent duration (“momentariness”) such 

as “to snap” or “to pop.” The difficulty that one faces in attempting to assimilate 

Yoshikawa’s aspectual metalanguage of telicity and momentariness to ongoing work in 

the study of aspect is substantial, but I have attempted in the following to take the lexical 

classes of ˙amt≥u reduplication identified by Yoshikawa and redescribe them in terms of 

secondary predication and the aspectual terminology in present-day use. I should 

emphasize, however, that Yoshikawa’s typology of ˙amt≥u reduplication relies above all 

on semantic criteria that are dependent on the reliability of Old Babylonian scribal 

traditions maintained by non-native speakers. 

14 This follows from much of the discussion about internal as opposed to external arguments in the generative tradition: 

since telicity is typically associated with the VP, it makes some sense that it is only nominal arguments that occur 

within the VP that can be modified by a secondary predicate. This would also explain why unaccusatives tolerate 

resultative secondary predicates, whereas unergatives do not. 

104

2.2 D-stem/Gtn-stem reduplication in Sumerian 

In the wake of Yoshikawa’s revolutionary description of grammatical aspect in Sumerian 

(see Yoshikawa 1968a, 1968b, 1974, 1989, 1993; written in dialogue with, at least to 

some degree, Edzard’s several iterations of a competing theory [1967, 1971, 1972], cf. 

Krecher 1995), the debate devolved to a class of reduplicated verbal roots that came to be 

called “˙amt¬u reduplications” due to the fact that they did not seem to belong to 

Yoshikawa’s marû reduplication class; Edzard termed these examples of reduplication 

freie Reduplikation, identifiable in formal terms by, prototypically, full CVC 

reduplication in opposition to the reduced CV reduplication of Yoshikawa’s marû 

reduplication class and the use of the same pattern of pronominal agreement as non- 

reduplicated ˙amt≥u roots. Much of Yoshikawa’s subsequent work, therefore, focused on 

questions of lexical aspect (primarily Yoshikawa 1979 [1993, 287-308] and 1981 [1993, 

309-322], cf. 1988 [1993, 72-94]). Yoshikawa argues that two distinct classes of freie 

Reduplikation forms can be distinguished (besides marû reduplications, which would 

constitute a third non-freie Reduplikation class [Yoshikawa (1979, 119 [1993, 308])]) on 

the basis of the form of the corresponding Akkadian verb in bilingual contexts: punctate 

verbs (including, prototypically, what are often now termed semelfactives in the 

literature, cf. Rothstein 2004, 28-29) such as “to hit,” “to break,” and “to snap,” which 

generally take Akkadian translations in the D-stem, can be juxtaposed to atelic verbs of, 

for the most part, repetitive motion such as “to crawl,” which are usually translated by 

Akkadian verbs in the Gtn-stem. Yoshikawa further notes that the opposition between 

these two classes corresponds to the transitivity of their respective predicates: those 

105

translated by Akkadian D-stem verbs are generally transitive, whereas those translated by 

Akkadian Gtn-stem verb are generally intransitive (Yoshikawa 1979, 119 [1993, 308]). 

The D-stem reduplications (subsuming the several semantically-defined types 

proposed by Yoshikawa) are best thought of as transitive and resultative, referring to the 

resulting state of the direct object: they correspond in a number of ways to the English 

resultative noted above (3c) such as “he hammered it flat” in that the two major 

subgroups within the D-stem reduplication set happen to correspond to the two 

components of an English resultative secondary predicate. In a resultative such as “he 

hammered it flat,” two components are necessary: a direct object, “it” (often resulting in 

reflexive or so-called fake objects with intransitive verbs), and an adjective, “flat,” that 

modifies the direct object and refers to the end state of the referent of the direct object 

after all telic points have been reached. In those cases in which the original meaning of 

the verbal root is transitive: the “it” of “hammered it flat” is already present, but the 

resulting state of “(something) flat” is introduced through the use of the reduplication. 

(4) An-ta-gal, E, iv 10'-11' (Cavigneaux et al. 1985 [MSL 17], 212; Yoshikawa 

1993, 291) 

10'. sa.a ma-ßa 2-du to full (cloth) 

11'. sa.sa muß-ßu-du to full (cloth) 

This lexically-attested use of √sa is nowhere to be found in third millennium and Old 

Babylonian sources, but its semantics do correspond to a resultative predicate in that the 

106

eduplicated Sumerian verbal form corresponding to the D-stem does not indicate the 

plurality of the direct object, but rather that the direct object is beaten to a point of 

completion: fulled cloth being cloth that has been beaten until it reaches a point at which 

it is fully cleansed and thickened. Thus the reduplication in this case contributes the 

adjectival values of “to be clean” and “to be thick.” 

(5) Nabnîtum J 331-342 (Finkel 1982 [MSL 16], 165; Yoshikawa 1993, 293; 

Molina and Such-Gutiérrez 2004, 4) 

331. ze 2 ba-qa-mu to pluck 

332. ze 2.ze 2 

… 

(ba-qa-mu) 

342. ze 2.ze 2 bu-uq-qu-mu to pluck clean 

Here likewise, the reduplication of the verbal root does not seem to indicate plurality of 

the direct object, but rather a resultative which includes the adjectival meaning “to be 

clean,” yielding something like “(a sheep) plucked clean” as its resultative meaning. But 

as should be apparent from the distance between lines 331 and 342, the simple opposition 

envisioned by Yoshikawa is not so simple: while it is true that basic and reduplicated 

stems follow directly in lines 331 and 332, the blank space in the translation column of 

line 332 presumably indicates that it is to be translated with a G-stem Akkadian verb just 

like the non-reduplicated form in line 331; the D-stem reduplication occurs ten lines later 

and is not preceded by a non-reduplicated form. This seems to indicate that reduplicated 

107

forms are, in some cases at least, radically ambiguous: the G-stem form in line 332 would 

seem to correspond to the common use of verbal reduplication to code the plurality of an 

absolutive argument, whereas the later reduplication in line 342 corresponds to the 

D-stem reduplication identified by Yoshikawa. 

The other type of D-stem reduplication is complementary in that, rather than taking a 

verb with a direct object and introducing an adjectival component, what Yoshikawa calls 

“denominative reduplication” takes an adjectival root and introduces a direct object, 

resulting in a kind of causative, or to be more precise, resultative construction. 

Yoshikawa notes that “[m]ost of the verbs assignable to this group seem to derive 

themselves from adjectives, in particular, those denoting colour terms, luminosity, [and] 

cleanliness” (Yoshikawa 1993, 305). 

(6) Groups III, E (Yoshikawa 1993, 305) 

82. tam.ma eb-bu-um bright, shiny 

83. tam.tam.ma ub-bu-bu-um to make bright, shiny 

84. dag 2.ga el-lum pure 

85. dag 2.dag 2.ga ul-lu-lum to purify 

108

Whereas D-stem reduplication introduced an adjectival component to the transitive verbs 

above, with respect to adjectival roots, which are necessarily intransitive, the D-stem 

reduplication introduces a direct object through causativization. 15 

(7) OBGT XI (MSL IV; Yoshikawa 1993, 305) 

11. [gun 3].≠gun 3±.na.ab bu-≠ri-im± make colored! 

12. [sig 7].sig 7.ga.ab bu-ri-≠iq± make green! 

Likewise in example (7), the reduplication of the verbal root yields a resultative form and 

in the case of adjectival roots results in a causative semantics. Note, then, that in purely 

formal terms, the examples of originally transitive predicates in (4) and (5) form a 

resultative secondary predicate through simple reduplication of the root, whereas the 

adjectival examples in (6) and (7) start out as a verbal root followed by the 

nominalizing/resultative suffix, *-a, and form the resultative secondary predicate by 

verbal reduplication that preserves the nominalizing/resultative suffix in final position. 

The fact that the resultative D-stem construction of both the transitive and the 

adjectival roots yields a semantic complex that includes both a direct object and an 

adjectival predicate that modifies the direct object should probably be taken as evidence 

that it is a resultative secondary predication (see Rothstein 2004, 59-90 and references 

therein). Yoshikawa drew a fundamental contrast between the D-stem ˙amt¬u 

reduplications dealt with above and the Gtn-stem ˙amt¬u reduplications that I turn to now 

15 Whether the causativization of the predicate follows from reduplication alone or is related in one way or another to 

the *-b- morpheme remains an open question: this is the generalization that best fits the grammatical texts (Kowaki 

1982), but it probably needs to be investigated more thoroughly in non-lexical environments. 

109

(Yoshikawa 1993, 308). Whereas D-stem reduplications share a number of important 

properties with resultative secondary predicates, Gtn-stem reduplications bear a number 

of similarities to depictive secondary predicates in English. Yoshikawa refers to a number 

of examples from Late Old Babylonian lexical lists such as the following. 

(8) Izi, E (Yoshikawa 1993, 297) 

329. u 2 ki tag pa-ßa-lum to crawl 

330. u 2 ki tag.tag pi-ta-ßu-lum to crawl continually 

(9) Izi, F (Yoshikawa 1993, 297) 

119. gu 2 gid 2 ßa-da-hu to walk 

120. gu 2 gid 2.gid 2 ßi-≠ta±-[du-du] to walk continually 

(10) Izi, G (Yoshikawa 1993, 298) 

249. gu 4.ud ßa!-˙a-≠t¬u± to jump 

250. gu 4.ud.gu 4.ud ßi-ta˙-˙u-≠t¬u± to jump continually 

In each case, the verb is intransitive and agentive and the reduplication of the Sumerian 

verbal root is translated by an Akkadian verb in the Gtn-stem, indicating repetitive action. 

The meaning of such repetitive Gtn-stem reduplications corresponds to the depictive 

secondary predicates described above in that the period of time referred to by the Gtn- 

stem reduplication is the entire period of time in the relevant domain, likewise depictive 

110

secondary predicates like (3a) and (3b) above (repeated below) indicate that the content 

of the secondary predicate is true throughout the period of time within which the main 

predicate takes place. 

(11) John i drove the car drunk i. 

(12) Mary drank the coffee i hot i. 

In (12), for example, the depictive secondary predicate “hot” is true throughout the entire 

period of time during which Mary was drinking the coffee, not only at the end of the 

event as was the case with the resultative secondary predicates. Likewise, in example 

(11), John was presumably drunk throughout the period in which he drove the car, 

although the implication is somewhat less clear than in (12) due to the fact that “to drive 

a car” is atelic: secondary predicates that modify the subject of transitive atelic predicates 

are fine, but secondary predicates that modify the subject of a transitive, telic predicate 

are somewhat problematic, presumably because the occurrence of a direct object 

interferes with a secondary predicate that is meant to modify the subject of the clause. 

(13) Mary i drank the coffee % drunk i/ % sober. 

In (13), “drunk” and “sober” seem somewhat odd, but not entirely ungrammatical; they 

do, however, seem to imply that the drinking of coffee might have resulted in “sober” 

111

coffee—only the inanimacy of coffee blocks the resultative interpretation and forces a 

construal with Mary as depictively modified. 

The import of these parallels between D-stem/Gtn-stem reduplications and secondary 

predication are difficult to clarify, but two points are worthy of mention: (a) secondary 

predicates tend to be associated with the nominal that corresponds to a natural end point 

or telos for the verb and (b) the D-stem and Gtn-stem reduplications both refer to the non- 

dynamic period of time after any and all telic points have been reached: where the verb is 

telic, such a non-dynamic period of time corresponds to the resultant state (analogous to 

resultative secondary predication), but when the verb is atelic, the non-dynamic period of 

time refers to the entire event (analogous to depictive secondary predication). As 

Rothstein reiterates: “… we see that when an AP [adjective phrase] is used as a 

secondary predicate, then the property it expresses must hold of the denotation of its 

subject for the whole time that the matrix [main] event is going on (for depictives) or for 

the whole time that the culmination of the matrix [main] event is going on (for 

resultatives)” (Rothstein 2004, 63). In other words, a secondary predicate in English 

picks out the nominal that serves as a natural end point for the verb and acts as an 

adjectival modifier, qualifying the nominal accordingly. 

What is the formal mechanism in Sumerian that allows reduplication of a verbal root 

to correspond in semantic terms to the introduction of a distinct, secondary adjectival 

form in the English examples? It just so happens that several types of adjective in 

Sumerian seem to be formed through the reduplication of a verbal root. The group of 

112

adjectival roots that form Yoshikawa’s “denominative reduplication” subset is one such 

group (Yoshikawa 1993, 305), repeated in (14). 

(14) a. bar 6.bar 6 /babbar/ “to be white” 

b. sig 7.sig 7 /sissig/ “to be green” 

c. dag 2.dag 2 /daddag/ “to be clean” 

d. gun 3.gun 3 /guggun/ “to be speckled” 

Note as well that the regular form of bar 6.bar 6 is babbar, that of dag 2.dag 2, dadag and 

that these two forms always occur in reduplicated form. 

Mandarin uses reduplication to code two distinct, but in my view, related 

morphosyntactic phenomena that may illuminate a distinct class of adjectives in 

Sumerian that is also formed through reduplication: what Li and Thompson term “the 

delimitative aspect” and certain forms of diminutive adjectives. Li and Thompson 

characterize the meaning of the delimitative aspect as “doing the action ‘a little bit,’ or 

for a short period of time” (Li and Thompson 1981, 232). 

(15) Mandarin Chinese (Li and Thompson 1981, 233, ex. 195) 

ni3 shi4 (yi) shi4 kan4 

you try one try see 

Try it a little and see! 

113

(16) wo3 wen4 (yi) wen4 zai4 jue2-ding 

I ask one ask then decide 

I’ll decide after I inquire a little. 

Although Li and Thompson do not make the connection, the form in English that seems 

to correspond to the delimitative aspect construction in Mandarin is a light verb 

construction in which a higher ditransitive predicate is introduced into the construction 

and the original main verb is changed into an indefinite noun—examples (15) and (16) 

are repeated below with English light verb constructions in the glosses. 

(17) ni3 shi4 (yi) shi4 kan4 

you try one try see 

Give it a try and see! 

(18) wo3 wen4 (yi) wen4 zai4 jue2-ding 

I ask one ask then decide 

I’ll decide after I make some inquires, lit., an inquiry. 

While the light verb construction in (18) involves a verb of creation, “make,” typically 

associated with verbs of speaking, the light verb construction in the gloss to (17) makes 

use of a low goal applicative construction. Light verbs that resemble a low source 

applicative such as “take a look at” also occur in English and bear a number of important 

114

similarities to the members of the BNBV inal class dealt with above, but the two should not 

be equated due to the highly restricted semantics of BNBV inal predicates. Pylkkänen states 

at one point that “light verb constructions are not syntactically low applicatives” (2002, 

38), based on the fact that, in her view, low applied arguments (Mary in “Lou gave Mary 

a cake”) cannot be modified by depictive secondary predicates, but I find her 

argumentation at this point forced and I suspect that the restriction is due to the 

incorporation of the applied argument into the predicate (i.e., in the middle of verb and 

direct object) in English. Although it is, as Pylkkänen notes, “an open research question,” 

I see no reason to exclude light verb constructions from a low applicative analysis (on the 

connection between light verb constructions, secondary predication and adjectival roots, 

see section 2.5 below). 

Turning back to reduplicated adjectives in Mandarin, here the idea of “doing 

something a little bit” is carried over into the adjectival part of the lexicon where it 

exhibits the well-known grammatical category of diminutive. “The diminutive form . . . 

expresses the speaker’s feeling that the referent is small, or subtle; that it endears the 

speaker—in short, a speaker evaluation of undersize, restricted and affectively positive,” 

whereas “the augmentative form . . . expresses the speaker’s feeling that the referent of 

some lexical item is large for what it is, or to excess, if an activity; that it is repulsive to 

the speaker—in short, a speaker evaluative of oversize, overmuch, and affectively 

negative” (Silverstein 1981, 8). In most languages, diminutives and augmentatives are 

formed through sometimes quite massive phonological deformation; in Mandarin 

Chinese, however, the diminutive is indicated by reduplication of the adjectival root. 

115

(19) ta1 de lian3 hen3 bai2 

she Poss face very white 

Her face is pale. 

(20) ta2 de lian3 bai2-bai2 de 

she Poss face white-Redup Nmz 

He face is beautifully white. 

In (19), the single adjectival root is neutral, but it presumably takes on a negative 

connotation in opposition to the reduplicated forms in (20). The reduplicated form in (20) 

does not indicate plurality or intensive meanings as we might expect on the basis of 

Sumerian or Akkadian comparisons, but rather that the referent of the adjective is, in an 

affective way, viewed as small, delicate or pretty. The contrast is highlighted when the 

adjective is question is something like “fat” as in (21) and (22). 

(21) na4 ge (hen3) pang4 de xian1-sheng 

that person very fat Nmz adult.male 

That fat man … 

(22) na4 ge pang4-pang4 de hai2-zi 

that person fat-Redup Nmz child 

That chubby (and cute) child … 

116

In terms of stereotypical usage, (21) would typically be used in reference to an adult and 

would tend to be a negative statement, whereas (22) would normally be used for a chubby 

baby or, perhaps, as a term of endearment for a girlfriend who has gained a little weight. 

Although it is unclear whether or not “reduplication as diminutive” is productive in 

the Sumerian preserved in the text-artifactual record, it does appear in two highly 

lexicalized adjectival expressions. 

(23) Diminutive-related CVC reduplication 

di 4.di 4.la “small” /diddil-a/ cf. tur “small,” gal “big,” ma˙ “huge” 

ku 7.ku 7.da “sweet” /kukkud-a/ cf. dug 3 “sweet,” sis “tart,” mun 4 “bitter” 

In (23), the contrast is two-fold: on the one hand, reduplicated diminutives contrast with 

ordinary adjectives that have the same basic meaning but lack the diminutive connotation 

(tur and dug 3 respectively), on the other hand, diminutives contrast with adjectives 

having the polar opposite meaning such as gal and ma˙, or sis and mun 4. 16 It should be 

emphasized that adjectives that signify “positive” qualities such as sa 6 “fine, good,” zid 

“just,” kalag “strong” would not normally be found in the diminutive since they lack the 

“cuteness” factor that is central to diminutive semantics. Note as well the close formal 

similarity between the adjectival roots in resultative secondary forms such as (6) and (7) 

and the diminutives in (23): in all four cases, the adjectival root is reduplicated and the 

16 Note that the two terms for “big” (gal and ma˙) and the two for “bitter” (sis and mun4) may well represent 

corresponding augmentative contrasts coded through phonological deformation, but without a clearer understanding of 

Sumerian phonology and a detailed investigation of the contexts in which the different forms occur, we cannot be sure 

of such an interpretation. 

117

eduplicated stem is followed by the nominalizing/resultative suffix, *-a. There are many 

open research questions pertaining to the history of the Sumerian adjective that I will not 

attempt to address here, but it seems fairly clear that reduplication in the examples in this 

section does not indicate plurality of individuals as is typically assumed for ˙amt≥u 

reduplication. 17 

The types of verbal/adjectival reduplication dealt with in this section, therefore, stand 

in opposition to the use of verbal reduplication to indicate plurality, certain types of 

pluractionality, or (in the case of “reduced” reduplication) progressive/imperfective 

aspect. The simplest explanation for the opposition between the reduplicated forms dealt 

with in this section as opposed to the more familiar plural/pluractional forms is that the 

forms dealt with in this section are limited to secondary predicational, adverbial or 

backgrounding functions, whereas the plural and pluractional uses of reduplication can be 

associated with primary predication, verbal or foregrounding discursive functions. In 

terms of secondary predication in Sumerian (if that is in fact the proper explanation for 

the reduplicated forms dealt with in this section), we should bear in mind that other 

studies of reduplication have shown that the particular grammatical phenomena indicated 

by reduplication are often determined by the lexical aspectual class of the reduplicated 

verbal root in conjunction with the particular morphology of the reduplication or the 

morphosyntactic environment in which the reduplicated form occurs (see Edzard 1971; 

1972; 1976; Yoshikawa 1993, passim; Krecher 1995 for the morphological forms 

17 The use of reduplicated attributive adjectives such as gal.gal in diœir gal.gal.e.ne “the greatest gods” is probably 

related to diminuative/delimitative reduplication in that superlatives regularly pick out a limited number of referents 

(see Thomsen 2001 [1984], 65; for a brief overview of verbal reduplication in Mandarin and its similarities to the 

attenuative prefix /po-/ in Russian, see J. Kim 2003). 

118

involved in Sumerian verbal reduplication; cf. Newman 1990, 53-130; 2000, 508-521 for 

an interesting set of materials for comparison). In a study of verbal reduplication in 

Niuean, Haji-Abdolhosseini, Massam and Oda have demonstrated that considerations of 

the morphological form of reduplication as well as the lexical aspectual class of the 

reduplicated verb regularly predict the meaning or grammatical function of reduplication 

in each case (Haji-Abdolhosseini et al. 2002, 488-489); similarly, the contrast in English 

between “he painted the red house” and “he painted the house red” demonstrates the 

difference a morphosyntactic environment makes: the adjective in the first of the two 

examples is an attributive adjective in a run-of-the-mill noun phrase, but the adjective in 

the second sentence is a resultative secondary predicate. 

In the following section, I look at the relatively small number of examples in which 

the verbal root of a BNBV inal predicate is reduplicated. I argue that reduplication of the 

verbal root in these cases is an indication of the fact that they are secondary and/or 

adverbial rather than an indication of plurality. This in turn raises the issue of which 

lexical aspectual class reduplicated BNBV inal roots belong to. I think that there are good 

reasons for assigning BNBV inal predicates in which the root is reduplicated—at least to 

the degree that they are secondary predicates—to a third lexical aspectual alongside the 

accomplishments that form the resultative (D-stem) class and the activities that form the 

depictive (Gtn-stem) class. Although the particular secondary or adverbial nuance 

remains unclear, I suggest that one possibility is that it forms an adverbial clause of cause 

or result and corresponds to a particular use of the iptaras form in Akkadian. 

119

2.3 BNBV inal reduplication: Depictive, resultative or tertium quid? 

In light of the opposition between D-stem, resultative reduplications and Gtn-stem, 

depictive reduplications, the aspectual properties of BNBV inal predicates do not seem to 

conform to the expectations of either of these two classes. This section presents the 

relatively few examples of BNBV inal predicates in which the verbal root is reduplicated 

and formulates one possible solution to their meaning and morphosyntactic function. The 

conventional interpretation of these forms is that they indicate plurality of a noun in the 

nominative/absolutive case. Although the interpretation of them as plural does explain 

some of the following examples, others do not seem to involve plurality. Since there is no 

independent evidence of plurality in these cases (other than the reduplication itself), I 

tentatively propose an alternative analysis of “reduplicated” BNBVinal predicates as a 

form of secondary or adverbial predication. If BNBV inal “reduplication” is secondary or 

adverbial, the simplest hypothesis would be that it belongs to either the resultative (D- 

stem) reduplication class or the depictive (Gtn-stem) reduplication class, but neither form 

of reduplication seems to correspond to the following examples of BNBV inal 

“reduplication.” 

In Yoshikawa’s paper “Verbal Reduplication in Sumerian” (Yoshikawa 1979 [1993, 

287-308]), in which he breaks down the various ˙amt¬u reduplications noted by Edzard 

(1971; 1972) into groups on the basis of the form of parallel, Akkadian verbs in 

bilinguals (largely lexical lists), the lone example of a BNBV inal predicate in Yoshikawa’s 

typology stands out as an exception and remains unclassified: œiß-√tuk (Yoshikawa 1993, 

120

290). Yoshikawa presents the rather sketchy material without further comment as 

follows: 

(24) Kagal, tablet E, IV (Yoshikawa 1993, 290) 18 

79. œiß tuk ße-mu-ú to hear 

80. œiß TUK.TUK ße-me-e a-ú to hear twice 

The English translation of the Akkadian gloss in line 80 is Yoshikawa’s—I have been 

unable to make sense of either the Akkadian gloss or Yoshikawa’s translation into 

English. Since Yoshikawa’s interpretation and its implications remain unclear to me, I 

have treated the form TUK.TUK, which could be read as either tuk.tuk or du 12.du 12, as a 

simple case of full verbal reduplication, namely œiß tuk.tuk. 

The verbal form œiß-√tuk.tuk occurs two times in the Old Babylonian materials 

made available through the ETCSL project: in the Sargon Legend, segment B, line 16 and 

Shulgi C, segment B, line 50, neither of which elucidate its meaning in any great detail. 

(25) The Sargon Legend [2.1.4], segment B, line ll. 14-18 

14. ku 3 d inanna.ke 4 ma.mu 2 i 7 mud.ße 3 

mu.un.gir 5.gir 5 

121 

In the dream, holy Inanna drowned him 

(= Ur-Zababa) in a river of blood, 

15. m ßar.ru.um.ki.in dum.dam.ma.na The sleeping Sargon gnawed the ground, 

zu 2 ki.≠ße 3± ba.da.ab.ra.≠a˙± groaning, 

18 No edition in MSL XIII; Civil notes that, “these texts will be published with corrections and additions in HAR-ra 

VII” (Civil 1971, 252).

16. dum.dum.bi lugal d ur. d za.ba 4.ba 4 

œiß tuk.tuk.da.ni heard them, 

17. lugal.ra ki ku 3.ga.ni.ße 3 

122 

As for these cries, because king Ur-Zababa 

He had them bring him into the 

im.ma.da.an.≠sun 5±.ne.eß king’s holy presence, 

18. m ßar.ru.um.ki.in ki He had them bring Sargon into the presence 

d ur. d za.ba4.ba 4.ße 3 

im.ma.da.an.sun 5.ne 

of Ur-Zababa, 

Whatever is happening in line 14 and whatever its connection to the overall sequence of 

the passage, the cries (dum.dum.bi) of Sargon clearly provide the reason that the king has 

Sargon brought before him, hence the introduction of “because” in the translation of line 

16. 

(26) Shulgi C, segment B, line 50 

48. x gi 4 x.ta ˙a.ma.ta.˙u.˙u.lu˙.eß 2 

… may they be terrified, 

49. erin 2.ne 2 ni 2 ˙a.ba.ab.te and may his troops be frightened, 

50. [x] x x x x … because they heard …, 

œiß ? ˙a.ma.ab.tuk.tuk 

51. x lugal kalam us 2.sa se 3.se 3 

in.ga.me.na.ta on track, 

Since I am a king who puts the homeland

The example in line 50 is quite badly broken and adds nothing of substance (and my 

partial translation of the line is pure supposition on the basis the preceding example), but 

it should be noted that, like the previous example, the temporal relation between the lines 

is unclear and deserves further investigation. 

In fact, the number of examples of ˙amt¬u reduplication of BNBV inal predicates can be 

counted on precisely one hand (five exemplars) and some members of the group are quite 

anomalous: a 2-√su 3 “to sail; to spread” and the two occurrences of gu 2-√gud “to pile up” 

do not seem to fit into the semantic field of the other BNBV inal predicates presumably 

because the nominal component of these examples is alienable rather than inalienable 

despite the existence of an homophonous inalienable noun in each case. Only two 

examples show reduplication of predicates that seem, at least upon initial inspection, to 

be BNBV inal predicates: zu 2-√sud 2 “to bite, to gnaw” and igi-√kar 2 “to examine.” 

(27) The Exploits of Ninurta [1.6.2], ll. 348-352 (Karahashi 2000, 78) 



123 



349. gu.ru.um na 4 kur.ra mi.ni.in.ak As for the pile of stones that had been 


accumulated in the mountains, 

Because they were spread all over like 

a 2 bi 2.in.su 3.su 3.ud drifting clouds, 

351. bad 3 ma˙.gin 7 kalam.ma igi.ba Blocking the view of the land as if they 

bi 2.in.tab.≠ba± were a huge wall,

352. œiß bal ki.ßar 2.ra mu.ni.in.œar He installed a spigot (?) on the horizon, 


196. e 2.me.eß ßeß.œu 10 a 2.kuß 2.a.œu 10.a Summer, my brother, after you have gone 

ni 2 buluœ 5 u 3 mu.ni.e 3 

197. za 3.mu.ka ße e 2.a u 3.mu.ni.in.ku 4 

124 

out boasting about my toil, 

Because, at new year, having brought grain 

guru 7 gu 2 bi 2.in.gur.gur into the house, the silos are full, 

198. diri u 3.me.de 6 tug2 bar.dul 5 

When you have brought in the surplus, your 

tug2 niœ2.lam 2.zu bal.a.kam bardul and niglam garments are ... 

199. uruda ˙a.zi.in 2(diß) ma.na lu 2 u 3 

When someone has given you a two-mana 

mu.ra.an.ßum 2 edin.zu.ße 3 ba.du.un axe, you go off to your steppe, 

Although the meaning of line 198 is somewhat unclear, the surplus presumably results 

from the abundance of food in the granary. In the following example (29), Winter makes 

an appeal to Enlil, the judge in Winter’s dispute with Summer. 


289. a.a d en.lil 2 nam.ku 5.œal 2 

mu.e.ßum 2 a he 2.œal 2.la mu.e.de 6 

"Father Enlil, you gave irrigation-control 

and brought abundant waters, 

290. a.gar 3.a.gar 3.ra za 3 ba.ni.tag.tag The plantations were set up in order to fill 

guru 7 gu 2 bi 2.gur.gur the granaries, 

291. ße ab.sin 2.na gu 2 ba.ni.in.peß The grain grew thick in the furrows,

292. d ezina 2.e ki.sikil sa 6.ga.gin 7 

125 

And Ezina came forth in splendour like 

ni 2 pa e 3 bi 2.ak a beautiful maiden, 

This example also occurs above in chapter 1, where I provide some justification for my 

interpretation of a.gar 3.a.gar 3.ra as “plantations.” Here, in the midst of a pair of ba.ni-√ 

verbs, the reduplicated BNBV inal construction seems to form a purpose clause rather than 

a causal one. The only difference, however, between a causal relation and a purpose 

relation is between past and future tense, so the causal predicate in 290 simply seems to 

be occurring in a future context rather than a past one. 

(30) Sumerian Proverbs, UM 55-21-278 = 3 N-T 179 [6.2.1] 

(Alster 1997, 305; Karahashi 2000, 181) 

ka 5.a zu 2 bi 2.in.sud 2.sud 2 

Because a fox bites, 

saœ nam.mi.bul.bul you should pay attention to it (to avoid being 

bitten) 

In general terms, the inherently decontextualized nature of proverbs makes any definitive 

interpretation of them quite difficult, but the basic lexical meaning of both verbs is quite 

obscure as well: von Soden cites a few texts in which BUL is associated with naßû and 

later in the same entry saœ bu.bu is juxtaposed to the Akkadian nu-uß qaq-qa-di (V R 16, 

47 cd; ZA 43, 243, 268; AHw 762); CAD identifies re¢ß naßû as an idiom meaning “to

check on quality or quantity of fields, materials, staples, animals,” and I have 

interpreted the second line of the proverb as an extended sense of this meaning. 

(31) Eduba Regulations, HT C3 (Karahashi 2000, 127) 

ßeß.gal.e u 3 ba.dab 5 dub.be 2 

igi bi 2.ib 2.kar 2.kar 2 

126 

The big brother, after taking the tablet in 

order to examine it, 

Although there is little if any direct evidence for the interpretation of reduplicated 

BNBV inal predicates as cause or result clauses (primarily due to the fact that there are very 

few cases and the contextual meaning is unclear in almost every case). If in future 

additional evidence can be educed in favor of such an analysis, it would result in a rather 

elegant system of adjectival/adverbial reduplication: atelic predicates would form the 

equivalent of depictive secondary predicates (Gtn-stem reduplication), accomplishments 

would form the equivalent of resultative secondary predicates (D-stem reduplication), 

and achievements would form cause or result clauses with respect to the main predicate, 

another form of secondary predication in a certain sense (for some orientation on the 

topic of causal and purpose adverbial clauses, see Larson 2004, 10-12; Sawada and 

Larson 2004). 

Why this might be the case can be clarified to some degree by looking at the 

behavior of achievement predicates in the unmarked or “present” tense in English: due to 

the fact that achievements lack duration of any kind, ambiguous forms in English are 

regularly interpreted so as to refer to either past or future events.

(32) a. Becky won the poker game in half an hour. 

b. 

% Becky wins the poker game in half a hour. 

c. Becky will win the poker game in half an hour. 

(33) a. Susan noticed the suitcase in 5 minutes. 

b. 

% Susan notices the suitcase in 5 minutes. 

c. Susan will notice the suitcase in 5 minutes. 

In both (32) and (33), each of the three English tenses are exemplified, but the present 

tense form in each example, (32b) and (33b) respectively, is not particularly felicitous in 

pragmatic terms, yielding either a curious historical present or a kind of apodictic 

promise about future events. These two—otherwise quite anomalous—readings 

correspond quite precisely, in my view, to the type of cause and purpose adverbial 

meanings associated with reduplicated BNBV inal forms mentioned earlier in this section. 

As secondary/adverbial predicates go, this makes a certain amount of sense: (a) activity 

predicates which are inherently durational but lack an endpoint correspond to depictive 

secondary predicates that hold throughout the duration of the main predicate, (b) 

accomplishments, which have both duration and an endpoint, correspond to resultative 

secondary predicates in that the durative part of the accomplishment limits the temporal 

span of the secondary predicate to the same span as the main predicate, while the 

endpoint further limits the duration of the secondary predicate to the final state of the 

main predicate, but (c) achievement predicates lack the cumulative duration of the 

127

accomplishment, so as secondary predicates they can refer to events that occur outside 

the temporal constraints of the main predicate, but still relate to the final state of the main 

predicate. 

(34) a. Joan washed the car drunk. 

b. Joan washed the car clean. 

c. (The car was really dirty.) Joan washed the car therefore 

= “Joan washed the car, because it was dirty. 

The first two examples in (34) are examples of depictive (34a) and resultative (34b) 

secondary predicates, whereas in (34c), the sentence-final form “therefore” corresponds 

in certain aspects to the interpretation of achievement secondary predicates that I propose 

above in that “therefore” occupies the same position that depictive and resultative 

secondary predicates occupy (although it can of course occur elsewhere), but refers to a 

prior sentence in the discourse and identifies it as the cause of the clause in which 

“therefore” occurs. 

I would expect that the Akkadian parallel to these kinds of secondary/adverbial 

achievement predicates would be forms in either the Gt-stem or G-stem perfects, which 

often occur in the apodoses of legal texts, for example, and thanks to Streck (2003, 98), 

we have a good list of cases in which a Gt-stem in Akkadian parallels a reduplicated 

verbal form in Sumerian, so I turn initially to Gt-stems and afterward to G-stem perfects. 

Streck identifies sixteen cases of Akkadian Gt-stems that correspond to reduplicated 

128

forms in Sumerian (the forms in parentheses are the corresponding reduplicated Sumerian 

forms cited in either the standard dictionaries or Streck 2003; the number in brackets that 

follows each lexeme refers to the number assigned to the lexeme in Streck 2003): itkupu 

(a.gi 6 mu.un.du 7.du 7) [5], mit˙us¬u [8], mitguru (ße.ße.ga) [20], atkulu (cf. participle su.bi 

in.gu 7.gu 7!) [26], ritkubu (ßed 3.ßed 3) [37], ßitrus¬u (ßu sa 2.sa 2) [40 (mistakenly listed as 

itru≈u on page 98)], ti≈butu (sa ˙ir.˙ir ir ) [41], atwû (du.ud.du // du 12.du 12, di.di, 

bala.bala) [43], mitluku (ad ba.ni.ib.gi 4.gi 4, cf. Gtn-stem mu.un.da.ab.sig 3.sig 3) [44], 

rit˙u≈u (giri 3 te.te) [62], ˙itbus¬u (si.su˙.bi ma.az.ma.az) [114], ˙itmut¬u (ul 4 sa 2.sa 2) 

[116], ßitpû (perhaps ßu du 7.du 7 although not attested, cf. Gtn-stem e.da.an.gub.gub.ba) 

[129], it˙usu (ir 2.ra im.mi.in.ße 8.ße 8 // ibakki it-˙u-sa) [134], etpußu (perhaps ka du 3.du 3, 

but again not attested) [142], and k/˙itru≈u (kid x.kid x [NUN.NUN], kid 6.kid 6 

[KAD 5.KAD 5]) [175]. 

Streck characterizes the Gt-stem verbs that parallel Sumerian reduplications, in 

particular, as “intensive” without much further in the way of interpretation (Streck 2003, 

98-99, cf. 110), but many of semantic fields identified by Streck in relation to Gt-stems in 

Akkadian correspond in a very general way to the semantic field of the BNBV inal class. 

But it should be noted that (a) no members of the BNBV inal class appear in parallel with 

Gt-stem verbs in Akkadian, (b) the Gt-stem reduplications identified by Streck never 

occur with the *bi-√ prefix, and (c) the semantics of the examples cited by Streck is by 

and large depictive or resultative. It is possible, of course, that adverbial clauses of cause 

or purpose are hiding in Streck’s “intensive” category, but if so I have not been able to 

identify sufficient context for the Gt-stem reduplications to characterize them as such. 

My suspicion is that reduplicated BNBV inal predicates (and perhaps reduplicated verbs 

129

earing the *bi-√ prefix generally) do correspond to certain discursive functions of G- 

stem perfect verbs in Akkadian. Streck identifies three primary functions of iptaras, the 

Akkadian perfect: (a) “it designates anteriority in the future in main clauses, subordinate 

clauses and conditional clauses,” (b) it is used “to express the temporal progress in main 

clauses, subordinate clauses and conditional clauses,” and (c) to refer to “the isolated past 

in main clauses of direct speech” (a usage that Streck limits the letters of Hammurapi 

[Streck 1998, 190; cf. 1999, 103]). Streck draws a parallel between the first of these uses 

and the Sumerian prospective, illustrating the usage (also found in Codex Ur-Namma 

§§13f and Codex Lipit-Ißtar §§11, 14, 20b, 30, 32a) with Codex Lipit-Ißtar §14. 

(35) Codex Lipit-Ißtar §14 

tukum.bi ir 11 lu 2.ke 4 lugal.a.ni.ir nam.ir 11.da.ni ba.an.da.gur lugal.a.ni.ir nam.ir 11.da.ni 

a.ra 2 2(diß).am 3 un.ge.en ir 11.bi al.bur 2.e 

If a man’s slave has contested his slave status against his master, and it has been 

proven that his master has been compensated for his slavery two-fold, that slave 

shall be free. 

which Streck compares to §196 from Codex Hammurapi (Streck 1998, 190): 

130

(36) Codex Hammurapi §196 

ßumma aw®lum ®n ma¢r aw®lim u˙tappid ®nßu u˙appadu¢ 

If a man (will have blinded =) has blinded the eye of another man, they shall blind 

his eye. 

The third type mentioned by Streck only occurs in the letters of Hammurapi and need not 

concern us here (see Streck 1999, 113-115, where a number of examples are presented). 

The second type of Akkadian perfect, however, what Streck describes as “zeitlicher 

Progreß der Vorzeitigkeit” or “temporal progress,” 19 bears a strong resemblance to my 

proposal that the reduplication of a Sumerian achievement verb in *bi-√ yields an 

adverbial clause of cause or result. In characterizing a corpus of “temporal progress” 

examples drawn from the letters of Hammurapi, Streck writes as follows: 

Temporal progress of anteriority (= zeitlicher Progreß der Vorzeitigkeit) is 

characterized by the sequence iprus + discourse particle + iptaras, whereby the 

number of iprus-elements as well as the number of iptaras-elements varies. The 

situations represented by iprus and iptaras are separated from one another 

temporally: in temporal terms, iptaras follows iprus and is thereby posterior to it. 

The progress is linked to neither a specific tense, nor a clause-level syntactic 

construction type: present and future tense examples are attested. In terms of clause- 

level syntactic construction type, only occurrences in main clauses are to be found in 

the following corpus, but attestations from outside the corpus show that the iptaras 

19 This may be equivalent to the traditional notion of consecutio temporum but I have not attempted to ascertain 

whether Streck equates the terms in his writings. 

131

denoting progress also occurs in subordinate clauses and in conditional sentences 

that begin with ßumma. The particle -ma functions as the discourse particle marking 

temporal progress in 20 cases, u in one case and inanna in three cases. We translate 

-ma and u with “(and) then,” “and (after that),” and with a contrastive nuance as 

“(since) however,” “then, however,” “but later.” We translate inanna with “now” 

(Streck 1999, 108). 

As examples of Streck’s temporal progress, I offer here the following two examples 

(equivalent to Streck 1999, 109, ex. 53 and 55): 

(37) AbB 4, 94: 12-14 

eqlam ... a˙˙® ab®ja imkusu¢-ma ≠ÍE±-am itbalu¢ 

My uncles have gathered up (the taxes on) the field, and then taken the grain for 

themselves. 

(38) AbB 4, 16: 8-11 

Í eqel b®t ab®ja ®kamanni-ma ana re¢dîm ittadin 

Í. has taken a field that belongs to the household of my father away from me, and 

then given it to a soldier. 

The glosses in (37) and (38) are rough translations of Streck’s own, but if I had my 

preference, I would rather translate many of Streck’s examples of “temporal progress” as 

result clauses as in (39) and (40), which repeat (37) and (38). 

132

(39) AbB 4, 94: 12-14 

eqlam ... a˙˙® ab®ja imkusu¢-ma ≠ÍE±-am itbalu¢ 

My uncles have gathered up (the taxes on) the field, so that they can take it for 

themselves. 

(40) AbB 4, 16: 8-11 

Í eqel b®t ab®ja ®kamanni-ma ana re¢dîm ittadin 

Í. has taken a field that belongs to the household of my father away from me, so that 

he can give it to a soldier. 

Akkadian iptaras forms, therefore, seem to function along the lines of injunctives 

elsewhere in Akkadian (as well as other relatively old Semitic languages). Injunction 

sequences expressing result or purpose, likewise, make use of the particle -ma, yielding 

sequences such as the following (after Huehnergard 2000, 147). 

(41) kaspam ßua¢ti piqdam-ma ekallam lu¢pul-ma b®t® la¢ i≈abbatu¢ 

Provide me with the aforementioned silver, so that I can pay the palace, so that they 

do not seize my estate. 

As Huehnergard describes the phenomenon: 

In a sequence of two or more clauses in which the first verb is an injunctive form, 

i.e., an Imperative, a Precative, a Prohibitive, or a Vetitive, and the following verb or 

133

verbs are also injunctive forms, and the verbs are connected either by -ma or, less 

often, asyndetically . . ., the second and following clauses are often to be translated 

as purpose clauses (i.e., ‘so that’, ‘in order that’, ‘that’). (Huehnergard 2000, 147) 

But I would differentiate purpose clauses, which are regularly indicated in Akkadian by 

the preposition ana + infinitive, from causal and result clauses. The traditional 

description of linked non-indicative clauses, as rehearsed by Huehnergard above, does 

use terms such as “purpose clause,” but it would be far better to characterize the relation 

as a causal one: “A causes B” or “B results from A” without any imposition of 

intentionality (or the deontic modality associated with it). Clearly the causal relation 

between a series of events (“A, so B, so C”) is independent of the deontic modality coded 

by what are traditionally called moods such as “precative” or “imperative.” I would argue 

that the indicative form of a causal sequence in Akkadian (the unmarked term within the 

system of deontic modality) is represented by iptaras, the Akkadian “perfect,” in 

sequential forms. 20 

Bearing this correlation in mind, I would like to propose that the Akkadian form that 

should correspond to reduplicated achievement predicates in Sumerian is the iptaras form 

in its cause or result functions. Such a correspondence would yield a rather elegant 

system of secondary or adverbial reduplication in Sumerian: (a) reduplicated activity 

predicates (Gtn-stem reduplications) would be similar to depictive secondary predicates, 

(b) reduplicated achievement predicates (iptaras reduplications) would correspond to 

20 The form corresponding to iptaras + ma in other older Semitic languages is clearly the use of wa- in both Eblaite and 

Northwest Semitic; whether /ma/ and /wa/ are to be equated remains unclear to me, but it is certainly a possibility that 

deserves further investigation. Particularly interesting in this regard is the apparent variation in Eblaite texts between 

the use of WA and U 3 as clause-initial connectors. 

134

either causal or result clauses depending on the aspectual contexts in which they occur, 

and (c) reduplicated accomplishments (D-stem reduplication) would be analogous to 

resultative secondary predicates. Since both activities and accomplishments are thought 

to include a [+ stages] feature (in Rothstein’s decomposition, see section 2.1 above), 

which allows them to freely occur in the progressive/imperfective aspect, when used as a 

secondary or adverbial predicate (perhaps better termed a backgrounding function), 

activities and accomplishments provide a durational extension within which the event 

referred to by the main predicate can take place. But achievements lack the [+ stages] 

feature and provide no such durational extension within which the main predication can 

take place. The resultative semantics, however, which originate from the telic point in the 

secondary predicate, remain in force even through the secondary predicate fails to 

provide a durational extension for the main predicate, yielding a resultative that, in 

temporal terms, either precedes the main predicate or follows it: preceding iptaras- 

reduplication resultatives could then be translated with “since” or “because” whereas 

following iptaras-reduplication resultatives would be translated as result or even, 

somewhat loosely, as purpose clauses such as “so that” or “in order to.” 

Other than the relatively few examples cited above from monolingual sources, I have 

not attempted to test the iptaras-reduplication hypothesis further. Bilingual examples are 

non-existent in The Return of Ninurta to Nippur (Cooper 1978), but there are a couple 

achievement reduplications bearing the *bi-√ prefix and paralleled by Akkadian iptaras 

verbs to be found in The Exploits of Ninurta. 

135

(42) The Exploits of Ninurta [1.6.2], l. 184 (van Dijk 1983, vol. 2, 78) 

A 1 

d a.nun.na.ke4.ne e 2.gar 8.e ba.ab.˙a.˙a.za 

B 2 [...] nun.na.ke 4.ne e 2.≠gar 8± [...] 

K 4 

d a.nun.na.ke4 e 2.gar 8.e ≠ba± [...] 

w [...] ra bi 2.in.˙a.˙a.za 

x 

n 2 

[... i-ga-ra]-ti i-ta˙-zu 

d a.nun.na.ke4.e.ne e 2.gar 8.ra bi 2.ib.˙a.˙a.za 

d a-nun-na-ki i-ga-ra-a-ti i-ta˙-za 

d a.[nun].na.ke4.e.ne e 2 [...] 

d a-nun-[na]-≠ki± i-ga-ra-a-≠ti± [...] 

In (42), although the Old Babylonian sources that preserve the relevant part of the line 

(A 1 and K 4) both have a *ba-√ prefix verb, the two first millennium sources (w and x) use 

the *bi-√ prefix and translate the verb with an iptaras form. The reason for the first 

millennium innovation is apparent from the context of line 184. 

(43) The Exploits of Ninurta [1.6.2], ll. 182-186 

182. an ib 2.dub 2 ba.tuß ßa 3 

136 

An was overwhelmed, crouched, pounded 

ßu ba.ßi.ib.ri his hands against his stomach, 

183. d en.lil 2 ib 2.ur 4.ur 4 da.ße 3 

im.ßi.ib.ri 

Enlil groaned and threw himself in a corner, 

184. d a.nun.na.ke 4.ne e 2.œar 8.e The Anuna flattened themselves against 

ba.ab.˙a.˙a.za walls,

185. tu mußen .gin 7 e 2 ni 2.te.ta gu 3 

im.ma.ni.ib 2.be 2 

137 

The house was full of fearful sighing as of 

pigeons, 

186. kur gal d en.lil 2.le d nin.lil 2.ra The great mountain Enlil was crying to 

gu 3 mu.na.de 2.e Ninlil, 

Whereas the Old Babylonian form of line 184 simply portrays it as one among a number 

of events in lines 182-185 that were not sequential, the first millennium redactor posits a 

causal relation between Enlil’s groaning in line 183 and the disappearance of the Anuna 

in line 184. Because line 184 has been reinterpreted as a result clause, its prefix seems to 

have been “corrected” to the *bi-√ prefix. If we were to privilege the first millennium 

version, an appropriate translation might be something like “Enlil groaned and hid 

himself, and because of this, the Anuna flattened themselves against the walls.” 

(44) The Exploits of Ninurta (1.6.2), line 350 (van Dijk 1983, vol. 2, 101) 

H dungu diri.ga [...] a 2 bi 2.in [...] 

F 1 

dungu diri.ga.gin 7 a 2 bi 2.≠in.su 3±.su 3.ud 

H 1 dungu diri.ga.gin 7 a 2 bi 2.in.su 3.su 3.ud ! (E) 

I 1 

J 1 

! 

dungu diri.ga.gin7 a2 bi2.≠in± [...] 

dungu diri.ga.≠gin 7± [...] 

K 1 ≠dungu± diri.ga.gin 7 a 2 ≠bi 2±.in.su 3.≠su 3± [...] 

L 1 

b 1 

dungu [...] 

[dungu diri].ga.gin 7 ni 2.bi mu.un.su 3 

≠ki±-ma ≠er-pe±-e-ti muq-qal-pi-ti ina ram-ni-ßu 2 i-ßad-di-˙u

The example in (44) was presented in abbreviated form above. As noted earlier, the 

sources that preserve the full form of the verb (F 1 and H 1) are evenly split in a sense: F 1 

preserves the form a 2 bi 2.in.su 3.su 3.ud (the variant used by ETCSL), while H 1 preserves 

the form a 2 bi 2.in.su 3.su 3.e (the variant used in Karahashi 2000, 78). Although the 

Akkadian translation of this line translates a different Sumerian predicate, ni 2.bi 

mu.un.su 3, and is not, therefore, of probative value in analyzing the *bi-√ prefix verb in 

question, the previous example can be taken as a model in favor of reduplication and the 

form in H 1 emended to a 2 bi 2.in.su 3.su 3.ud ! (E). 




138 



349. gu.ru.um na 4 kur.ra mi.ni.in.ak As for the pile of stones that he accumulated 


in the mountains, 

Because they were spread all over like 

a 2 bi 2.in.su 3.su 3.ud drifting clouds, 

351. bad 3 ma˙.gin 7 kalam.ma They blocked the view of the land like a 

igi.ba bi 2.in.tab.≠ba± huge wall, 

352. œiß bala ki.ßar 2.ra mu.ni.in.œar He installed a spigot (?) on the horizon, 

The reduplicated *bi-√ prefix verb in line 350 takes gu.ru.um na 4 “a pile of stones” as its 

subject and describes the result of Ninurta’s activities, perhaps best translated using a

esult clause: “He made a pile of stones in the mountains, so that/with the result that it 

stretched across the mountains like floating clouds, like a huge wall . . . .” 

Overall, therefore, the best candidate for the tertium quid mentioned at the beginning 

of the section is an interpretation of reduplicated BNBV inal predicates as adverbial clauses 

of cause or result. The evidence is quite slim, but what little evidence does exist seems to 

point in that direction. Other possibilities certainly exist, however: I suspect that future 

work on quantification and/or the mass/count noun opposition in Sumerian may 

ultimately yield a more conclusive result. Of the four lexical aspectual types identified at 

the beginning of the chapter, three seem to determine the interpretation of the 

adjectival/adverbial constructions found in ˙amt≥u reduplication: activities (Gtn-stem 

depictive secondary predicates), achievements (iptaras-stem causal secondary predicates) 

and accomplishments (D-stem resultative secondary predicates). In the previous section, I 

argued that certain subsets of adjectival roots were formed through verbal reduplication 

in earlier historical phases of the language and, furthermore, that in the period 

represented by the text-artifactual record or just before, the formation of adjectival roots 

through reduplication yielded a number of distinct semantic classes, but primarily a 

diminutive semantics as also seen in Mandarin Chinese. Whether the 

diminutive/delimitative lexical aspectual class as evidenced in Mandarin exists in 

Sumerian, remains an open question. But it seems fairly clear that ˙amt≥u reduplication 

can be associated with adjectival/adverbial formation at all levels of the grammar; the 

particular grammatical effects of such adjectival/adverbial formations remain, however, 

to be worked out in future studies. 

139

The last area of lexical aspectual concern is the ± agentive feature. Recently, 

Rothstein has argued that achievement predicates in the progressive/imperfective aspect 

exhibit a number of special characteristics, notably the absence of an agentive thematic 

role. The following section investigates whether or not this is the case in Sumerian. 

140

2.4 The curious morphosyntax of progressive achievements 

Although lexical and grammatical aspect show various kinds of interaction, one of the 

more dramatic is actually negative in form: achievement predicates such as the members 

of the BNBV inal class do not usually occur in the progressive/imperfective aspect. As 

already noted above, what is generally termed the “progressive” aspect in disciplinary 

semantics is presumably equivalent to the “imperfective” (marû) aspect in Sumerian and 

in the following sections, as above, “progressive/imperfective” should be considered a 

cover term for “progressive” and “imperfective.” 21 Rothstein identifies the basic 

incompatibility of achievements and progressive/imperfective aspect as follows: 

Since achievement verbs denote eventuality types which are near instantaneous and 

which are over as soon as they have begun, they ought to be incompatible with the 

progressive mode which asserts that an event is ‘in progress,’ and which we expect 

to be restricted to activities and accomplishments. (Rothstein 2004, 36) 

But the curious thing about achievement predicates is that they do at times occur in the 

imperfective (Rothstein 2004, 23-24). This section presents the BNBV inal class as 

prototypical exemplars of the achievement lexical aspectual class, argues that they do not 

regularly occur in the progressive/imperfective (marû) aspect, and ends with a few 

suggestions as to further possible evidence having to do with the case-marking and non- 

agentivity of their subjects. 

21 Strictly speaking, the marked category in the Sumerian grammatical aspectual system is the imperfective (marû) and 

in accord with convention should be described with a positive descriptor such as “progressive” or “continuative” (I 

prefer the latter). Since markedness is privative and not equipollent (Agha 1993, 132-133), the unmarked term in the 

system (˙amt≥u) would then be described as non-progressive or non-continuative. But I worry that changing established 

terminology on the basis of non-Assyriological convention (which is not consistently followed even in disciplinary 

linguistics) would only multiply terms and confuse matters, so I have retained the traditional terminology in part. 

141

As an example of the restriction on progressive/imperfective aspect, I turn initially to 

the following two examples (after Rothstein 2004, 36, ex. 1a and 1b). 

(46) a. Jane reached the summit of the mountain. 

b. Jane is reaching the summit of the mountain. 

(47) a. Mary spotted her friend at the party. 

b. Mary is spotting her friend at the party. 

In each pair of examples, the first sentence, (46a) and (47a) respectively, is a perfective 

achievement predicate in the past tense, whereas the second sentence in each pair, (46b) 

and (47b) respectively, is a progressive/imperfective achievement in the present tense. 

The sentences in (46b) and (47b) are much less “natural” and force the rather curious 

impression that one is watching a film of Jane or Mary with a running narration 

(Rothstein 2004, 56-57). Rothstein notes that “[t]hese are what have been called ‘slow- 

motion’ or ‘film-strip’ readings, first noticed by Sandro Zucchi, where a normally 

instantaneous event is perceived as being ‘stretched’ over time” (Rothstein 2004, 37, 

presumably referring to Zucchi 1999). Much of the peculiar semantics of these kinds of 

examples follows from the rather unusual relation that exists between spectator and 

narrated events (Lee 1997, 344). While so-called explicit primary performatives also (a) 

resist occurring in the progressive/imperfective aspect, (b) yield a spectator effect, and (c) 

regularly occur with the *bi-√ prefix due to its use with direct speech, although verba 

142

dicendi do not appear to belong to the BNBV inal class. To my knowledge, this kind of 

slow-motion interpretation of progressive/imperfective achievements (the only 

interpretation available for prototypical achievements like “notice” and “spot”) does not 

exist in the Sumerian text-artifactual record. 

On the basis of both the rather indirect and weak evidence of causal and result 

clauses in the previous section and the general semantic cast of the members of the 

BNBV inal class, I would hypothesize that the remaining members of the BNBV inal class 

are prototypical achievements and that they should not occur in the 

progressive/imperfective (marû) aspect except under rather unusual circumstances. 

Somewhat less prototypical achievement such as “to die” and “to find,” regularly occur in 

the progressive/imperfective without the “film commentary” interpretation as in the 

following examples (after Rothstein 2004, 36, ex. 2b and 2d). 

(48) a. The cat is dying. 

b. Dafna is finding her shoes. 

But non-prototypical achievements such as (48a) and (48b) do not occur in the BNBV inal 

class, although some of them do occur with the *bi-√ prefix. On the whole, BNBV inal 

predicates do not in fact seem to occur in the progressive/imperfective aspect at all: (a) 

the one BNBV inal predicate with a suppletive marû root (zu 2-√gub) never suppletes, (b) all 

BNBV inal reduplications can be interpreted as ˙amt¬u reduplications, and (c) I demonstrate 

in the following section that the relatively few BNBV inal roots which are occasionally 

143

followed by *-e (which could be an indication of progressive/imperfective aspect) are all 

to be interpreted as examples of the ergative or locative-terminative postposition 

following an indefinite relative construction. Although this is essentially negative 

evidence, the absence of progressive/imperfective achievements (or at least their 

restriction to highly exceptional contexts) is probably the best evidence available for the 

classification of BNBV inal predicates as achievements. 

Given the frequent use of metaphors that draw on imagery from the natural world 

(see Black 1998 for background, cf. Heimpel 1968) both in Sumerian literature in general 

as well as in the epic tale Gilgamesh and Huwawa in particular, the assertion in line 103 

from Gilgamesh and Huwawa, version A, that Huwawa’s brow devours reed-beds 

(according to the ETCSL translation) is surely a little bit odd. In the following, I break 

with my own convention and repeat the ETCSL translation verbatim. 

(49) Gilgamesh and Huwawa, version A [1.8.1.5], ll. 100-104 (Edzard 1991, 200-201 

[equivalent to lines 99-103 in Edzard’s edition]) 

100. ur.saœ ka.ka.ni ka His pugnacious mouth is a dragon’s maw, 

ußumgal.la.kam 

101. igi.ni igi piriœ.œa 2.kam His face is a lion’s grimace, 

102. ŒIÍ.GABA.a.ni a.œi 6 du 7.du 7.dam His chest is like a raging flood, 

144

103. saœ.ki.ni œiß.gi bi 2.gu 7.a lu 2 

145 

No one dare approach his brow, which 

nu.mu.da.te.œe 26.e.dam devours the reed-beds, 

(var. nu.mu.un.da.kar.re.de 3) 22 

104. lugal.œu 10 za.e kur.ße 3 u 5.a œa 2.e Travel on, my master, up into the 

iri ki .ße 3 ga.u 5 

mountains!—but I shall travel back to 

the city, 

The fact that œiß.gi “reed-bed” is in the absolutive case has led previous translators to 

interpret it as the patient of the verb √gu 7 “to eat.” Why, however, the brow of a forest 

guardian would be expected to devour reed-beds and why such an image would strike 

fear into the heart of a listener (as it was clearly intended to do) is beyond me: I would 

rather propose that the particular syntactic configuration of a bare noun followed by the 

*bi-√ prefix, where no pronominal agreement morpheme such as *-n- or *-b- occurs 

before the verbal root, be seen as an indefinite relative construction. In this case, the 

construction is particularly clear in that the head of the indefinite relative is set apart from 

the topic of the line through the use of the possessive pronoun topic-marking construction 

and the indefinite relative offers an elaboration as to “why no man dares to approach it.” 

At the discursive level, this particular configuration of topic-marked phase followed by a 

parenthetical comment in apposition to the topical noun, forms a tripartite rhetorical 

structure as in (50) that will reappear in a slightly altered guise in chapter 4. 

22 The variant probably arises from the fact that te.œe26, the marû form of the root √te(œ), is written TE.ŒA 2 and if it 

was misunderstood as a nominalized verbal root such as te.œa 2, it might, at some point in its transmission, have been 

written te.a. TE.A could have then been misunderstood as a diri writing for the verbal root √kar followed, as in the 

original, by the *-ed suffix, yielding kar.re.de 3. Since version B, in example (56) below, has a non-finite verbal form, 

te.œe 26.dam, the misinterpretation could easily have arisen on the basis of the several possible interpretations of TE 

ŒA 2 DAM.

(50) [ Topic saœ.ki.ni [ Apposition œiß.gi bi 2.gu 7.a]] [ Comment lu 2 nu mu.da.te.œe 26.e.dam] 

As for his brow, a devouring reed-bed, no ones dares to approach it, 

But how does my suggestion that œiß.gi bi 2.gu 7.a forms an indefinite relative clause 

speak to the broader question of whether or not BNBV inal predicates occur in the 

progressive/imperfective aspect? First of all, although homophonous with the relative 

marker, the *-a suffix that follows bi 2.gu 7 (the predicate inside the indefinite relative 

construction) is probably not a relative marker, but rather the locative postposition that so 

often can be found following an inanimate noun. This is demonstrated—or at least 

suggested—by the fact that the same line in version B of the same composition shows the 

locative-terminative postposition rather than the locative. 

(51) Gilgamesh and Huwawa, version B [1.8.1.5.1], l. 93 

saœ.ki.ni œiß.gi bi 2.gu 7.e lu 2 nu te.œe 26.dam 

As for his brow, a devouring reed-bed, no one approaches it, 

The alternation between locative and locative-terminative postposition is indicative of an 

ambiguity inherent in the metaphor introduced by the indefinite relative: since the 

conservatism of Old Babylonian redactors would never have allowed the removal of 

mu.da if the trajectory were from mu.da.√te.œe 26.e.dam to te.œe 26.dam, what we must 

take as the original form of the line in (51) preserves the locative-terminative postposition 

following bi 2.gu 7. We might expect a locative-terminative in such a construction with a 

146

discrete, bounded entity such as the brow of Huwawa, whereas the locative in version A 

would be more appropriate to a relatively unbounded entity such as a bed of reeds: 

ultimately the shift in verbal form from te.œe 26.dam to mu.da.te.œe 26.e.dam may well 

have resulted from the shift in nominal postposition from locative-terminative to locative. 

Since the change in nominal postposition results from the ambiguity inherent in the 

metaphor, namely “his brow is a reed-bed,” both saœ.ki.ni and œiß.gi presumably refer to 

the same entity. More importantly, the alternation between locative and locative- 

terminative shows that the /-a/ in version A is not a relative marker (it would be 

surprising if indefinite relative clauses made use of an explicit relative marker, cf. Old 

Babylonian indefinite relatives in warad i˙liqu “an escaped slave” (Huehnergard 2000, 

188, cf. Kienast 2001, 447) and the non-specific “indefinite” Middle Egyptian relative in 

gmj.n=j h≥f3w pw jw=f m jj.t “I found that it was a serpent which was coming” (Sh.S 61- 

62; Loprieno 1995, 158). Since the marker, whether locative or locative-terminative, is 

simply a nominal postposition, the form bi 2.gu 7.e cannot be a progressive/imperfective 

(marû) verbal form. 

Further examples of the indefinite relative are somewhat more difficult to identify 

but they are regularly found in a rhetorical device in which an indefinite relative is 

followed by a definite noun having the same referent as in the following two examples. 

(52) A tigi to Enki for Ur-Ninurta (Ur-Ninurta B) [2.5.6.2], ll. 25-28 

25. en ma˙ an ki.a za 3 bi 2.dib mu.zu August lord, you excel in heaven and earth, 

pa bi 2.e 3 

147 

your name shines forth,

26. d en.ki me a.na œal 2.la mu.u 8.ur 4.ur 4 Enki, you gathered up all the divine powers, 

abzu.ße 3 mu.u 8.œar lit., whatever (of them) existed, and 

148 

stored them in the abzu, 

27. ki.tuß ku 3 ßa 3.ge bi 2.pa 3.da.zu Your holy dwelling, chosen by heart, 

abzu eß 3 ma˙-a x x the abzu, the august shrine …, 

28. me.bi me.a diri œiß.˙ur.bi a.re.eß As for the plans of these Me, which exceed 

gub.ba me.en all others, you have made them 

praiseworthy, 

Although the interpretation of *-a as either a genitive or locative remains somewhat 

unclear in this example, its translation is fairly clear: “of/with regard to a holy place 

favored by you.” In the following example, the role of the nominal postposition is 

somewhat clearer. 

(53) A balbale to Inanna and Dumuzi [2.08.a] 

30. lugal ßa 3.ge bi 2.pa 3.da.zu For your king, chosen by heart, 

31. d dumu.zi dumu d ≠en.lil 2±.ra Dumuzi, the son of Enlil, 

32. e 2.tur 3.e i 3 ga ˙e 2.en.da.ab.be 2 

May the cattle-pen produce ghee and 

cheese, 

Here the familiar pattern of an indefinite descriptor followed by a personal name is 

repeated. The English translation is somewhat misleading and a more literal if somewhat 

stilted translation might be as follows: “a king, your favorite one, (namely) Dumuzi, the

son of Enlil, for him, …,” where the gradually increasing referential specificity and the 

application of the benefactive postposition to the most referentially specific phrase 

d dumu.zi dumu d en.lil2 is much more felicitous in pragmatic terms than the translation 

offered in (53). The indefinite relative also serves as a basis for the relative/ 

circumstantial construction (for further details, see below, section 3.3). 

If we broaden, for a moment, the area of investigation to include achievements in 

general (not limited to BNBV inal predicates), other somewhat less prototypical 

achievement predicates do allow the formation of the progressive/imperfective and I turn 

to these examples at this point so as to illustrate some directions for future research. The 

classic example of a progressive/imperfective achievement is its use with an intransitive 

verb of motion to provide background information for some other event as in the 

following example. 


(Cohen 1973, 71, 98, 120, 175-177; Vanstiphout 2003) 

163. inim gal d inanna gi zu 2.lum.ma Where will he carry such an important 

DU me.a mu.na.ab.tum 3 

164. ˙ur.saœ zubi.ße 3 bi 2.in.e 11.de 3 

149 

sharply-intoned message from Inanna? 

As he was bringing it up the Zubi 

Mountains, 

165. ˙ur.saœ zubi.ta As he was bringing it down from the Zubi 

im.ma.da.ra.ab.e 11.de 3 

166. susin ki .e kur an.ßa 4.an ki .a.ße 3 

Mountains, 

From Susa to the land of Anshan,

167. peß 2 tur.≠gin 7± kiri 3 

ßu mu.na.ab.œal 2 

150 

Everyone prostrated themselves before 

it/him like tiny mice, 

This is the last of three examples involving the verb √e 11(d) “to go up/down,” here in a 

causative *bi-√ verb construction with inim gal as causee: lines 164 and 165 provide 

background information for the main assertion that “from Susa to Anshan, everyone 

prostrated themselves (before it/him) ….” The three repetitions in Enmerkar and the Lord 

of Aratta differ in certain ways, but crucially not in the form of the verb bi 2.in.e 11.de 3. 

(55) Enmerkar and the Lord of Arattta [1.8.2.3], ll. 73 // 108 // 164 

73. ˙ur.saœ zubi.ße 3 ˙e 2 bi 2.in.e 11.de 3 3¸ precative 

108.˙ur.saœ zubi.ka ˙e 2 bi 2.in.e 11.de 3.en 2¸ precative 

164.˙ur.saœ zubi.ße 3 Ø bi 2.in.e 11.de 3 3¸ indicative 

(56) Enmerkar and the Lord of Aratta [1.8.2.3], ll. 74 // 109 // 165 

74. ˙ur.saœ zubi.ta ˙e 2 im.ma.da.ra.e 11.de 3 3¸ precative 

109.˙ur.saœ zubi.ta ˙e 2 ≠im±.ma.da.ra.an.e 11.de 3.en 2¸ precative 

165.˙ur.saœ zubi.ta Ø im.ma.da.ra.ab.e 11.de 3 3¸ indicative 

There are two types of variation in the three occurrences of each of these two lines: (a) 

the relatively minor alternation between zubi.ße 3 and zubi.ka in line 108 and (b) a 

fascinating opposition between the pronominal agreement marker immediately before the 

verbal root, which remains consistently *-n- in the progressive/imperfective

achievements in lines 73, 108, and 164, but changes in each occurrence of the 

progressive/imperfective activity predicates in lines 74, 109, and 165. Since all six lines 

provide background information in each of the three passages, it makes good sense that 

they all remain consistently in the progressive/imperfective (marû) aspect in each case. 

With the exception of the second person singular copula in line 108, “let it be you who 

…,” the precative/future forms of bi 2.in.e 11.de 3 differ from the indicative form only in the 

addition of the modal he 2. This is in stark contrast to the quite systematic variation in the 

pronominal agreement marker (*-Ø-, *-n-, *-b-) in each of the three examples of the 

progressive/imperfective activity: ˙e 2.em.ma.da.ra.e 11.de 3, ˙e 2.em.ma.da.ra.an.e 11.de 3, 

and im.ma.da.ra.ab.e 11.de 3 in lines 74, 109, and 165 respectively. 

However we choose to interpret the pronominal agreement morphology that 

immediately precedes the verbal root in these examples (for the standard theory, see 

Michalowski 1980a; cf. Woods 2001, 263-285), one conclusion is unavoidable: whereas 

the agreement in the activity predicates in lines 74, 109, and 165 is clearly affected by 

person (second in 109, third in 74 and 165) and modality (precative in 74 and 109, 

indicative in 165), the agreement morphology of the progressive/imperfective 

achievements remains completely unaffected by these changes in person and modality. 

This indifference to modality and person results, in my view, from the fact that the 

progressive/imperfective achievements in lines 73, 108 and 164 are non-agentive and, 

since they are non-agentive, the expected alternation between ergative animate *-n- in the 

perfective and accusative inanimate *-b- in the progressive/imperfective (at least in the 

151

context available in these three passages) is not available. 23 Since the standard models of 

agreement do not work particularly well when it comes to *bi-√ prefix predicates, I will 

not hazard a guess (at least at this point) as to what phenomenon underlies the pervasive 

use of *-n- in these constructions. Nonetheless, the invariance of *-n- in the achievements 

does conforms quite well to Rothstein’s characterization of achievements as non-agentive 

(Rothstein 2004, 138, ex. 38), since one might expect an agentive verbal form to vary in 

pronominal agreement according to grammatical aspect, as is regularly postulated for 

Sumerian agreement patterns in general. 

(57) a. Mary reached the summit of the mountain unconscious and on a stretcher. 

b. The supplies reached the summit of the mountain by helicopter. 

As these two examples makes clear, the theme of an achievement need not show any 

agentivity whatsoever. This fits well with the fact that BNBV inal predicates are 

achievements in that the theme of BNBV inal predicates is the nominal component of the 

compound verb, while the ergative case-marked nominal is its inalienable possessor, 

rather than its ergative agent. As I argued at the end of chapter 1, the ergative case- 

marking in BNBV inal constructions indicates that the ergative case-marked nominal is the 

inalienable possessor of the body-part noun that forms the nominal component of the 

23 There is also a slim possibility that the constructions in question are intransitive and do not take inim gal d inanna as 

an argument, even though the typically causative semantics of *bi-√ prefix verbs would generally suggest the transitive 

interpretation. If, for whatever reason, the construction is intransitive, the opposition between “an ergative, animate 

*-n- in the perfective and accusative, inanimate *-b- in the imperfective” would no longer be valid as stated, but the 

contrast between invariant *-n- in the achievements as opposed to the variation in the corresponding activity predicates 

supports a transitive, indeed causative, interpretation of the constructions in question. 

152

BNBV inal predicate: it is, in other words, non-agentive. This yields a zero-marked, 

absolutive/nominative experiencer such as lugal.e igi “the king’s eye” in opposition to an 

oblique, typically locative-terminative case-marked experienced object such as gig.e 

“wheat” at the end of the event in conformity the expectation that the theme of an 

achievement such lugal.e igi “the king’s eye/perception” is non-agentive. 

The last point I would like to bring up in this section has to do with the relative 

agentivity and affectedness of the theme or subject in a progressive achievement. 

Achievements have a curious way of taking as a direct object a noun the referent of 

which is not affected by the verb in any way at all: in “Mary reached the summit” (cf. 

Rothstein 2004, 138), the summit is not affected by Mary’s action at all, likewise in “Paul 

spotted the train,” nothing has happened to the train and it is Paul who has been affected 

by the event. Rothstein starts out, in a rather involved discussion of the issue with the 

following sentence (Rothstein 2004, 138, ex. 36). 

(58) Mary is reaching the summit. 

Rothstein points out that since progressive achievements bear a number of similarities to 

accomplishments in that the progressive/imperfective aspect adds an activity predicate 

meaning “she is going” to an otherwise momentary achievement “she reached the 

summit.” Because of this, (58) can be thought of as either an activity or an achievement: 

as an activity, Mary would be the single, non-agentive argument of an intransitive 

predicate, but the agentivity or affectedness of Mary in an achievement predicate is much 

153

less clear. Achievements are quite regularly transitive in a certain sense, but since the 

direct object is not affected by the predicate, it is hard to imagine that in an ergative 

language like Sumerian, the logical subject of an achievement would appear in the 

ergative case or the direct object in an absolutive case. 

If we turn from BNBV inal predicates to BNBV predicates in which the nominal 

component is alienable, or even *bi-√ prefix verbs that are not preceded by a bare noun, a 

few examples seem to provide some justification for interpreting *bi-√ prefix verbs as 

achievements more generally. The primary difference between BNBV inal predicates and 

other lexical classes is that—in terms of case-marking—the absolutive noun (in 

conjunction with its raised possessor in the ergative case) is an experiencer rather than a 

patient; where the absolutive noun is not inalienable, and, consequently, the ergative 

noun cannot be a possessor, we should expect that the ergative noun is an agent of some 

kind and the absolutive noun is a patient. Given such a combination of achievement 

semantics and an absolutive case-marked patient, there is the possibility that the logical 

object of an achievement predicate might undergo dethematization (demotion of an 

absolutive case-marked nominal to an adpositional case), resulting in a semantic “patient” 

appearing in an oblique case; there appear to be a couple of examples in which such a 

dethematization seems to have taken place. 

(59) Nippur Lament [2.2.4] (Tinney 1996, 110-111, 163) 

193. si.ga.bi mu.un.du 3.un.na.gin 7 

154 

Just as he silenced you, when he made 

˙ul 2 u 4 mi.ni.in.ku 4.ra joy enter again,

194. za.pa.aœ 2 kurun 2 lal 3 diri.ße 3 

bala.bala.e na.aœ 2.ße 3 

bi 2.ib.tar.re 

155 

The sound of choice beer and syrup poured 

to overflow he decreed as your fate, 

195. ib 2.si u 4 en 3.ße 3 er 2 gul.la e.ne “Enough!, lit. it is full!, It is time to stifle the 

e.ra.an.du 11.ga lament,” he said to you himself, 

(60) Nippur Lament [2.2.4] (Tinney 1996, 118-119) 

279. œarza su˙ 3.a bir.a.bi si bi 2.in.sa 2 

As for its disordered and scattered rites, 

he put them in order, 

280. ku 3.ku 3.ga ßub.ba ßu pel 2.la 2.a.ke 4 He restored the most sacred things, 

ki.bi.ße 3 in.œar.ra neglected and defiled, 

281. ßuku u 4 ßub.ba eßa zi 3 dub.dub.bu He ordains the offering of daily rations and 

na.aœ 2.ße 3 bi 2.ib.tar.re the grinding of fine meal and flour, 

282. œiß banßur ninda lu ninda ßar 2.re.da He decided to make bread plentiful at 

ßa 3.ga.ni im.de 6.a table, to make loaves numerous! 

These two cases are largely parallel and in both of them the direct object in a compound 

verb construction, nam-√tar “to decide someone’s fate; to decree” is followed by the 

allative postposition, even though there is no motion toward nam “fate.” Two other 

examples (from a single passage) seem to show the same syntactic pattern: the 

occurrence of the direct object of compound verb in the allative case instead of its usual 

absolutive form in conjunction with a progressive/imperfective verb with the *bi-√ 

prefix.

(61) A hymn to Haya for Rim-Sin (Rim-Sin B) [2.6.9.2] 

39. en œi 6.par 3.e mu.ni.ib.huœ.en You appoint the high priest for the gipar and 

saœ.se 3 mu.ni.ib.œar.re.en install him there as its caretaker, 

40. sipa kalam.ma.ra œidru ma˙ You make the shepherd of the Land hold the 

u 4 sud.ra 2 ßu.ni.ße 3 

bi 2.ib.dab 5.e.en 

156 

august sceptre until distant days, 

41. d ˙a.ia 3 diœir kalam.ma a.ra.zu Haia, you are the god of the land who 

uœ 3 ßar 2.ra œiß tuku me.en listens to the prayers of all people, 

42. za.e diœir gal.gal.e.ne You make him (= the king) hold the 

uœ 3 daœal.la ßu.ni.ße 3 

bi 2.ib.dab 5.e.en 

widespread people for the great gods, 

I should note, as this point, that dethematization does not seem to be limited to *bi-√ 

prefix verbs, the following example, likewise, exhibits dethematization of the nominal 

component of a compound verb but with mu.e.da-√ as prefix rather than bi-√. 

(62) Dialogue between Enkihegal and Enkitalu 194 (Dialogue 2; unpublished, cf. 


eß 2.gar 3 ˙e 2 bi 2.ßid niœin ˙e 2 bi 2.gi.ne a 2.ße 3 mu.e.da.aœ 2 

You ordered me to have the task computed and the grand totals corrected,

All the texts in which dethematization (-ße 3 immediately preceding *bi-√ prefix verbs) 

occurs date to the Isin-Larsa period: The Nippur Lament examples dating to perhaps a 

century before the Rim-Sin B examples (for other exceptional uses of the terminative 

postposition in the Isin-Larsa period, see Brisch 2003, 138-139; Thomsen 2001 [1984], 

103). If they can be corroborated by earlier examples, they may provide evidence of the 

demotion of the underlying theme (dethematization) so as to allow for underlying subject 

to appear in the absolutive/nominative case in progressive/imperfective achievements. 

The appearance of the “agent” in the nominative case in the progressive/imperfective 

(marû) aspect is to be expected regardless of lexical aspectual class, so the only real 

evidence for dethematization is the demotion of the patient from absolutive to allative 

case with a progressive/imperfective achievement verb. 

157

2.5 Introducing telicity, ßu, and inalienable possession under nominalization 

Among the many morphosyntactic proposals that originate from Yoshikawa’s work, one 

of the most surprising is that otherwise atelic predicates can, under certain circumstances, 

be made telic through the introduction of ßu “hand” in a compound verb construction of 

some kind (Yoshikawa 1993, 82-84). The clearest source of derived telicity, in my view 

however, is the use of the *bi-√ prefix. This leads to a rather interesting aporia in that, as 

argued above, the BNBV inal predicates are the prototypical examples of achievements and 

they include (by definition) a *bi-√ prefix verb as well as an inalienable noun such as ßu: 

the obvious question is which of these two elements is responsible for the telicity of these 

predicates. The plausibility of Yoshikawa’s proposal has been obscured up to now by my 

artificial exclusion of BNBV inal predicates in which ßu functions as the nominal 

component from my initial survey in chapter 1, but these examples will be presented at 

the conclusion of this section. 

At the beginning of the chapter, I spoke of D-stem reduplication in Sumerian as 

analogous to resultative secondary predication, but at the time I was hesitant to classify 

D-stem reduplications as resultative secondary predicates in the strict sense. One of the 

reasons for my hesitation was that the D-stem reduplications differ from ordinary 

resultatives in a number of little ways, but the main reason for my hesitation was that 

there is an extremely small group of constructions in Sumerian which resemble English- 

style secondary predicates even more and that may also provide some insight into which 

of the two elements, *bi-√ or ßu, codes telicity: the two examples of this phenomenon 

that I am aware of follow in (63) and (64). 

158

(63) Enki and Ninmah [1.1.2], ll. 53-55 

53. d nin.ma˙.e d en.ki.ra gu 3 

mu.na.de 2.e 

159 

Ninmah calls to Enki: 

54. me.dim 2 na.aœ 2.lu 2.ulu 3.ta sa 6.ge “Man’s body can be either good or 

˙ul ma.al.la.a.kam bad, 

55. ki ßa 3 gi 4.a.œu 10 na.aœ 2.tar Whether I make a fate good or bad 

bi 2.ib.se 3.ge bi 2.ib.˙ul.e depends on my will,” 

(64) Ninurta and the Turtle, segment B [1.6.3], ll. 49-51 

49. œa 2.ra saœ.œiß ? .ra.œu 10.uß igi.zu You who have decided to kill me, 

mu.e.œar.ra.≠am 3± 

50. x bil 2.ga gal.gal di œa 2.e . . . who makes big claims: “I put things in 

bi 2.≠ib 2±.œa 2.œa 2 œa 2.e bi 2.ib 2.zi.de 3 

place, I make things right,” 

51. za.e œa 2.ra a.gin 7 igi.zu From you who set your sights on me 

mu.œar.ra like this, 

At first glance these two examples do not seem to resemble the resultative secondary 

predicates that were under discussion earlier in the chapter (“Julie painted the house 

red”), but this is largely due to the fact that Sumerian is a verb-final language. In other 

verb-final languages, it is the resultative secondary predicate (typically based on an 

adjectival root) that occurs in final position in the sentence and bears the inflectional 

morphology typically associated with a “main verb.” Thus the sequence of lexical verb

followed by resultative seems to be invariant, which is quite surprising given the inverse 

ordering of constituents in SOV languages under Kaynean antisymmetry. 

(65) Japanese -te shimau construction (Kaiser et al. 2001, 518) 

nantoka shinai to go¢suto taun-ni nat-te shima-u 

something do-Neg Comp ghost town-Loc become-Inf end.up-Indic 

Unless we do something, it’ll end up becoming a ghost town. 

(66) Korean poryo-ta construction 

ceolsoo-ga ne ca-rul pa-ra boryot-ta 

PN-Sub 1Poss car-Acc sold-Inf end.up-Indic 

Chelsu ended up selling my car (negative meaning) 

The only way that I know of to identify the particular morphological element that 

introduces telicity would be to identify an adjectival root that has undergone two 

derivational phases: the first involving the shift from state (adjective) to derived 

achievement and the second from achievement to derived accomplishment. If one 

assumes, as I will for the sake of the argument, that each of these two derivations can be 

tied to some kind of morphological element or syntactic configuration, the derived 

achievement and the derived accomplishment should share at least one morphosyntactic 

feature that is not found in the simple adjective, while the accomplishment should bear at 

least one morphosyntactic feature that neither the state (adjective), nor the achievement 

160

exemplify. Rather than speaking in generalities, I would like to introduce an example 

from English which may serve as an informal model of the phenomenon I have in mind. 

English is capable, for example, of converting a predicate of uncertain telicity into a 

resultative through the introduction of an adjective as a secondary predicate that modifies 

the direct object and describes the resultant state in some fashion (“he swept the floor” > 

“he swept the floor clean”). A derived achievement such as “he swept the floor clean” 

can then be transformed into an accomplishment through the formation of a light verb 

construction: in one kind of English light verb construction, the main verb is changed into 

an indefinite noun that is modified by an attributive adjective that corresponds to the 

adjective used to form the resultative secondary predicate; the indefinite noun derived 

from the underlying main verb becomes the direct object of a higher, ditransitive verb 

which is introduced as the new main verb, while the original or underlying direct object 

becomes either the recipient or the source in a low applicative construction (“he swept the 

floor clean” > “he gave the floor a clean sweep”). Note that in the following parallel the 

lexical meaning of the English and Sumerian forms is completely unrelated (clean ≠ ˙ul, 

˙ul = “bad, rotten, false”)—they are only parallel in morphosyntactic form, if that, see 

below for further discussion. 

(67) a. The floor is clean ˙ul 

b. He swept the floor clean ßu bi 2.˙ul 

c. He gave the floor a clean sweep ßu ˙ul bi 2.du 11 

161

In (67), I present the two-part derivation (“state > achievement” followed by 

“achievement > accomplishment”) as exemplified by three English sentences alongside 

what Yoshikawa’s theory of ßu as the telicizing component might predict: ßu ˙ul and ßu 

˙ul bi 2.du 11 both include ßu as the telicizing component, while the derived 

accomplishment in (67c) also includes a plausible light verb, namely bi 2.du 11. 

The problem is that no such sequence seems to exist in the text-artifactual record. 

The contrast that is represented exists between achievements such as (63) and (64) above 

and nominalized structures such as the following examples (example [63] repeated below 

as [68a]). 

(68) a. Enki and Ninmah [1.1.2], l. 55 

ki ßa 3 gi 4.a.œu 10 na.aœ 2.tar Whether I make a fate good or bad 

bi 2.ib.se 3.ge bi 2.ib.˙ul.e depends on my will, 

b. Nippur Lament [2.2.4], l. 263 (Attinger 1993, 707) 

u 4 ki.en.gi ki.uri lu 2.erim 2.e The enemy, having done Sumer and Akkad 

ßu ˙ul bi 2.in.du 11.ga a terrible deed, 

An additional example of the ßu ˙ul bi 2.in.du 11.ga pattern occurs in line 299 of the 

Nippur Lament [2.2.4]. Since √˙ul is not a BNBV inal predicate, it shows causative 

semantics as expected: “to make (a fate) bad.” Whereas (68a) is equivalent in some 

respects to a resultative like “he swept the floor clean,” the ßu ˙ul bi 2.in.du 11.ga 

162

construction is (68b) bears a number of similarities to a light verb construction such as 

“he gave the floor a clean sweep”: the adjectival root in (68a), √˙ul, is nominalized as an 

indefinite noun, ßu ˙ul, that serves as the patient of a newly introduced ditransitive verbal 

root, namely √du 11. The environment in which √˙ul occurs in (68b), *ßu . . . bi-√, 

indicates that the nominal phrase formed by √˙ul in combination with the inalienable 

noun ßu is indefinite, but I suspect that the use of ßu also allows for the inclusion of an 

agentive adjunct such as lu 2.erim 2.e in a clause that would, otherwise, exclude agentive 

arguments of any kind. 

(69) a. Ninurta and the Turtle, segment B [1.6.3], l. 50 

x bil 2.ga gal.gal di œa 2.e bi 2.≠ib 2±.œa 2.œa 2 œa 2.e bi 2.ib 2.zi.de 3 

. . . who makes big claims: I put things in place, I make things right, 

b. Bird and Fish [5.3.5], l. 88 

ma˙ gu 7.gu 7 saœ ßu zi bi 2.ib 2.du 11.ga kisal.e ße 10 si.si 

Gluttonous, malformed, filling the courtyard with droppings, 

In (69a), there are two expressions in parallel: œa 2.e bi 2.ib 2.œa 2.œa 2 “I myself am one who 

puts (things) in place,” and œa 2.e bi 2.ib 2.zi.de 3 “I myself am one who makes (things) 

right.” Although I translate the two expressions as equational sentences, the *bi-√ prefix 

verbs are actually independent expressions in apposition to the first person singular 

pronoun (see below, chapter 3, for a discussion of the appositional character of head- 

163

internal relative clauses such as these). Since, again, the *bi-√ prefix predicates are not 

BNBV inal, their meaning is causative: “put (things) in place = cause (things) to be in 

place,” and “make (things) right/straight” respectively. In the light verb construction in 

(69b), however, the meaning of the expression is somewhat more difficult. The ETCSL 

translation of the expression saœ ßu ZI bi 2.ib 2.du 11.ga is simply “malformed.” But this 

translation follows almost entirely from the fact that the series of epithets are all negative 

characterizations of Bird’s appearance and the word saœ, “head,” appears in this 

particular epithet. The ETCSL transliteration offers some insight into the particular way 

that ETCSL editors imagined the head in question to be malformed: ETCSL transliterates 

ZI as zig 3, meaning “to rise up” and forming a syntagm saœ √zi(g) that has generally 

positive connotations in Sumerian literature if not elsewhere (cf. the story of the 

cupbearer and the baker in Genesis 40). The parallel that I have drawn in (69), however, 

would necessarily interpret ZI as the verbal root √zi(d), “to be straight; to be just,” and I 

would interpret the epithet to mean that Bird’s head (including its beak) is straightened 

and angular as opposed to the gracefully curving head of Fish. My interpretation would 

seem to be in line with Fish’s generally negative attitude toward the angular body-parts of 

Bird, if the ETCSL translation of line 86, ka ur 2 gu 2 guru 5 giri 3 su.ul.su.ul ka ˙a.la eme 

sal.sal as “[c]hopped-off beak and legs, deformed feet, cleft mouth, thin tongue!” has any 

merit. 

Although the number of cases in which the same adjectival root appears in both the 

bi 2.in.˙ul 2.le construction and the ßu ADJECTIVE bi 2.in/ib.du 11.ga construction is 

164

limited to two pairs of exemplars, 24 the details so precisely correspond to the two-step 

derivation discussed above (state [adjective] > achievement > accomplishment) that some 

consideration of the forms must be allowed. Again, using the three example sentences in 

English as scaffolding, the derivational correspondences can be understood as follows: 

(70) a. The floor is clean. ˙ul 

b. He swept the floor clean. bi 2.in.˙ul 

c. He gave the floor a clean sweep. ßu ˙ul bi 2.in.du 11.ga 

The presence of the *bi-√ prefix in (70b) and (70c), the achievement and the 

accomplishment respectively, would seem to indicate that it is the *bi-√ prefix that codes 

telicity, while the addition of ßu under nominalization seems to derive the 

accomplishment (70c) from the achievement in (70b). In formal terms, there are a 

number of intriguing similarities between the English and the Sumerian constructions in 

(70c) such as nominalization (*-a), the demotion of the original verbal root to a nominal 

argument position (˙ul) and its replacement by a newly introduced ditransitive verb 

(du 11), but one of the primary differences, which I would like to associate with the 

introduction of ßu in (70c), is that unlike (70b) the English sentence in (70c) is agentive. 

The contrast can be seen more clearly if the subject of the English sentences in (70b) and 

(70c) is changed to an inanimate noun. Thus a (resultative) achievement such as “the 

wind swept the floor clean” is fine with a non-agentive subject like “wind,” but “the wind 

24 Although not brought into discussion here, the larger set of examples of the form ßu su˙3.a ba.ab.du 11 (Lament over 

Sumer and Ur 67) presumably offer a minimal contrast with complexes of the form ßu ˙ul bi 2.in.du 11.ga. 

165

gave the floor a clean sweep” only works if the wind is personified a great deal. This 

seems, therefore, to conform to Rothstein’s expectation that the subject of achievements 

is non-agentive, while the subject off accomplishments is necessarily agentive. 

Thinking back to my earlier discussion of non-agentivity in achievements and the 

role of ergative possessor and inalienable body-part noun as a single constituent in the 

absolutive case, the reason for the use of the ßu may be elucidated somewhat: since ßu is 

the most agentive of the possible body-part nouns in a BNBV inal construction and the 

derived accomplishment in (70c) must use some means of reintroducing an agent into a 

previously non-agentive achievement predicate, ßu is the natural choice. The crucial 

factor, however, in differentiating such agent-introducing ßu constructions from 

ordinary BNBV inal constructions in which the nominal component happens to be ßu 

seems to be nominalization. The BNBV inal predicates can appear without any form of 

nominalization, whereas pseudo-agentive ßu in derived accomplishments only seems to 

occur under some form of nominalization. But, given the paucity of sources and the 

difficulty of differentiating them within such a relatively unstructured corpus, it has 

been difficult to identify this opposition in the text-artifactual records. Beyond the 

examples of the ßu ˙ul bi 2.in.du 11.ga construction above, I have been able to locate 

only a few other examples of ßu, where it introduces an agent rather than provides 

some specific semantic content to the predicate. 

(71) Ur Lament [2.2.2], l. 258 (Karahashi 2000, 149) 

ki 

256. egir3.re eß3 uri2 ˙ul.a.na Over her devastated shrine Urim, the 

gig.ga.bi im.me princess bitterly declares: 

166

257. an.ne 2 uru 2.œu 10 na.aœ 2 

ha.ba.da.an.ku 5 uru 2.œu 10 

˙u.mu.da.gul 

258. d mu.ul.lil 2.le e 2.œu 10 ßu ˙e 2 

167 

“An has indeed cursed my city, my city has 

been destroyed before me, 

Mullil has indeed transformed my house, 

bi 2.in.bala œiß al.e ˙a.ba.ra it has been smitten by pickaxes. 

259. sig.ta di.œa 2 izi ˙a.ba.ni.in.ßub On my ones coming from the south 

a uru 2.œu 10 ˙u.mu.da.gul he hurled fire. Alas, my city has been 

destroyed before me, 

(72) Lament over Sumer and Ur [2.2.3] (Michalowski 1989, 113; Karahashi 2000, 149) 

23. d en.lil 2.le igi.ni ki kur 2.ra After Enlil had looked favourably on an 

ba.an.œar.ra.a.ba enemy land, 

24. d nin.tu.re niœ 2.dim 2.dim 2.ma.ni za 3 

After Nintu had scattered the creatures 

bi 2.in.tag.ga.a.ba that she had created, 

25. d en.ki.ke 4 i7 idigna i7 buranun.na After Enki had altered the course of the 

ßu bi 2.in.bala.a.ba Tigris and Euphrates, 

26. d utu ˙ar.ra.an kaskal.e nam After Utu had cast his curse on the roads 

ba.an.ku 5.da.a.ba and highways, 

(73) Gilgamesh, Enkidu and the Netherworld [1.8.1.4], ll. 247-251 (Karahashi 2000, 167) 

247. tukum.bi a 2 aœ 2.œa 2 kur.ra (Enkidu:) If I tell you the order of the nether 

mu.ra.ab.[be 2.en] world,

248. [za].e tuß.a er 2 œa 2.e ga.tuß You, sit down and weep! (Gilgamesh:) I 

ga.≠er 2± shall sit down and weep! 

249. [x] ßu bi 2.in.tag.≠ga± ßa 3.zu …, which your heart rejoiced to touch, 

ba.e.˙ul 2 

250. […] ≠ßi ? ±.du.un bi 2.in.du 11 

168 

is …, 

251. [… tug 2 sumun].≠a±.gin 7 u˙ Worms infest it like an old garment (?), 

bi 2.in.≠tag± 

Karahashi reconstructs the line as su.mu ßu bi 2.in.tag.ga ßa 3.zu ba.e.hul 2, “When you 

touched my flesh, you felt happy,” but the line is broken and passage is difficult. 

Nonetheless, if we can assume that the meaning of the verb ßu bi 2.in.tag.ga is “to touch,” 

then it is clear that the ßu does not contribute to the lexical meaning of the verb since the 

verbal root √tag alone also means the same thing. But since ßu would serve as a natural 

semantic component of a verb meaning “to touch,” the precise lexical meaning of the 

verbal root √tag is probably in need of further investigation. 

One example is exceptional in that it is clear that the ßu does not play any particular 

role in the semantics of the predicate, but the predicate appears with ßu even though it 

contributes no lexical meaning to the predicate.

(74) Bird and Fish [5.3.5], ll. 7-11 (Karahashi 2000, 164) 

7. i7 ≠idigna± i7 buranun.na za 3.ga He laid out side by side the Tigris and the 

ba.an.la 2 a kur.kur.ra Euphrates, and caused them to bring 

mi.ni.in.tum 3.uß water from the mountains, 

8. i 7 tur.≠tur±.ra ßu lu˙ ba.an.ak He scoured out the smaller streams, and 

sur 3.sur 3 mi.ni.œar.œar positioned the other watercourses, 

9. [a.a d ]≠en±.ki.ke 4 tur 3 amaß Father Enki made the sheepfolds and cattle- 

ßu bi 2.in.daœal sipa unu 3 

169 

pens teem, and provided them with 

bi 2.in.tuku shepherds and herdsmen, 

10. ≠iri± ki a 2.dam ki.a He founded cities and settlements 

bi 2.in.œar.œar saœ gi 6 

throughout the earth, and made the 

mi.ni.in.lu black-headed multiply, 

11. lugal.e nam.sipa.bi.ße 3 

mu.un.tuku nam.nun.bi.ße 3 

mu.un.il 2 

He provided them with a king as shepherd, 

elevating him to sovereignty over them, 

Karahashi transliterates the verb in question as ßu bi 2.in.niœin, whereas ETCSL has ßu 

bi 2.in.daœal. The copy in PBS 10, 2, 21 is quite difficult to read and collation may yield 

NIŒIN, but the copy in UET 6/1, 38 is clearly AMA, hence daœal. I believe that this is an 

example of the use of ßu to introduce an agent into an achievement predicate; the reason 

for the absence of a nominalizing/resultative postposition is that the verbal root, √daœal, 

is adjectival and it is a good examples of a derived achievement as discussed above.

Since it is an achievement predicate, we might expect that there would be no ergative 

case-marked agent, but the introduction of ßu allows the agent to appear in the clause as 

the inalienable possessor of ßu in the ergative case. Thus the contrast between pseudo- 

agentive ßu compound verbs and true BNBV inal compound verbs in ßu on the basis of 

whether or not the main verb bears the nominalizing/resultative suffix, *-a, only seems to 

apply to non-adjectival verbs. Of course, it is not particularly surprising that fientive 

verbs bearing the nominalizing/resultative suffix and adjectival roots are more or less 

interchangeable. 

(75) Lament over Sumer and Ur [2.2.3], ll. 407-410 (Karahashi 2000, 153) 

407. lu 2.kar.ra.bi dub 3 nu.um.zil.e As for its refugees, they were unable to flee, 

bad 3 za 3.ga bi 2.in.dab 5.be 2.eß they were trapped inside the walls, 

408. uruda alan uzug x(AN.ZAG).ge The statues that were in the treasury were 

si.a.bi gu 2.guru 5 ba.an.ne.eß cut down, 

409. d nin.i 3.gara 2 aœrig ma˙.e The great stewardess Ninigara abandoned 

erim 3.ma ßu bi 2.in.dag the storehouse, 

410. œiß gu.za.bi igi.bi.ta ba.e.ßu 2 

170 

As for its throne, it was cast down before 

sa˙ar.ra ba.da.an.tuß them, (and) she threw herself down into 

the dust, 

Other examples of the ßu bi 2.in.dag construction include Enmerkar and the Lord of 

Aratta [1.8.2.3], lines 558-562, Eridu Lament [2.2.6], line 6/7, and Nippur Lament 

[2.2.4], line 88 (Karahashi 2000, 153-154).

(76) Lament over Sumer and Ur [2.2.3], l. 445 (Michalowski 1989, 64-65; Karahashi 

2000, 155) 

443. œiß na 2 gi 4.rin.na nu.um.gub u 2 

za.gin 3 nu.mu.un.bara 3 

171 

The divine bed was not set up, it was not 

spread with lapis-colored grass, 

444. alan AN.ZAG.ge 4 si.a.ba The statues that were in the treasury were 

gu 2.gur 5 ba.an.ne.eß cut down, 

445. engiz ensi kißib 3 œal 2.bi eß.da The temple cook, dream interpreter, and 

ßu li bi 2.in.du 7.uß seal-keeper didn’t prepare the rite, 

446. gu 2 ki.ße 3 œal 2.la.bi They stood in submission and were carried 

ba.e.su 8.su 8.ge.eß kur 2.re off by the foreigners, 

ba.ab.la˙ 5.e.eß 

447. uz.ga ku 3 ßu.lu˙ dadag.ga The holy uzga-priests of the sacred 

ßa 3.gada.la 2.bi.e.ne lustrations, the linen-clad priests, 

(77) Lament over Eridu [2.2.6], section B, ll. 7-11 (Green 1975, 338, 346 [ll. 5/3-5/7 in 

Green’s edition]; Karahashi 2000, 157) 

7. nin.bi ab 2.zi ar˙uß.a ku 3 

As for its lady, the faithful cow, the 

d dam.gal.nun.na compassionate one, the pure one, 

Damgalnuna, 

8. gaba.ni ≠i 3.˙ar.re igi±.ni She claws at her breast; she claws at her 

i 3.˙ar.re gu 3 mir.ra im.me eyes; she utters a frenzied cry, 

9. ßu 2(diß).a.na ≠œir 2± ba.da.ra She held dagger and sword in her two 

ßu bi 2.in.du 8 teß 2.bi i 3.gu 7.e hands; they consume each other,

10. siki.ni u2 numun 2.bur.gin 7 i 3.ze 2.e She tears out her hair like rushes, uttering a 

i.lu gig œa 2.œa 2 

172 

bitter lament, 

11. uru 2 munus.zu mu.≠til±.[la “My city, your woman does not reside (in 

a].≠la±.zu nu.gi 4.a.œu 10 me.a you); your lilting song does not 

er 2.zu ba.gig resound. Where is a lament uttered 

bitterly for you?” 

(78) Kesh Temple Hymn 109 (Karahashi 2000, 157, Gragg in Sjöberg and Bergmann 

1969, 174) 

107. e 2.e lugal.bur.ra am 3.mi.gub The lugalburra-priest stepped up to the 

108. en du 10 ßa 3 eß 2.la 2 am 3.mi.in.la 2 

109. a.tu.e ßibir ßu bi 2.in.du 8 

110. tu.e a.ur 4.a am 3.mi.tum 2 

temple, 

The good en-priest . . . held the lead-rope 

suspended, 

The atu-priest held the staff, 

The . . . brought the gathered waters, 

111. lal 3.e ki ku 3.ga am 3.mi.in.tuß The . . . took his seat in the holy place, 

(79) Inanna’s Descent 290-292 (Karahashi 2000, 157) 

290. d inanna kur.ta ba.e 11.de 3 

So Inanna left the netherworld, 

291. lu 2 igi.na sukkal nu.me.a The one in front of her, though not a sukkal, 

œiß gidru ßu bi2.in.du 8 

held a scepter, 

292. bar.ra.na ra.gaba nu.me.a The one behind her, though not an escort, 

œiß tukul ur2.ra bi 2.in.la 2 

carried a mace at his hip,


(Vanstiphout 2003, 84-85; Karahashi 2000, 165) 

501. kiœ 2.gi 4.a ka.ni dugud As for the messenger’s mouth, it was heavy, 

ßu nu mu.un.da.an.gi 4.gi 4 

173 

so he could not repeat it (the message), 

502. bar kiœ 2.gi 4.a ka.ni dugud Since the messenger’s mouth was heavy and 

ßu nu mu.un.da.an.gi 4.gi 4.da.ka he couldn’t repeat it (the message), 

ki 

503. en kul.aba4 .a.ke4 im.e The lord of Kulab patted some clay and put 

ßu bi 2.in.ra inim dub.gin 7 

the words on the tablet instead (-gin 7), 

≠bi 2.in±.gub (of saying them to the messenger) 

504. u 4.bi.ta inim im.ma gub.bu Before that time, no one had ever put words 

nu.ub.ta.œal 2.la on clay, 

Lines 501 and 502 actually illustrate quite nicely the contrast between two forms of cause 

or result clause: in line 501, the heaviness of the messenger’s mouth results in his 

inability to repeat the message (“his mouth was heavy, so . . .”) where the reduplication 

of √gi 4 presumably indicates that it is an adverbial/subordinate clause, but in line 502, the 

entire content of line 501 is taken as a cause (rather than a result) of the creation of 

writing (“Since . . ., the lord of Kulab . . . .”). The causal subordination is marked with the 

familiar bar . . . /ak/ construction.

(81) Enmerkar and Ensu˙girana [1.8.2.4], ll. 75-79 (Karahashi 2000, 165) 

77. [im].gin 7 ßu bi 2.in.ra im.gin 7 

174 

He patted it into (the form of) a tablet and 

igi i.ni.in.bar looked at it like a tablet, 

78. [e.ne] d inanna.da e 2.za.gin 3 

aratta.ka ˙u.mu.da.an.ti Aratta, 

He may dwell with Inanna in the Ezagin of 

79. œa 2.e [x] ≠an±.ta ki.a gub.ba.ni But I dwell with her . . . as her earthly 

˙u.mu.da.an.ti.en companion, 

(82) The Marriage of Sud [1.2.2] ll. 83-87 (Karahashi 2000, 171) 

83. [ßu] ≠ki±.ta.ni sukkal The minister opened his left hand and gave 

mu.un.ßi.in.bar gi 16.sa her the jewellery, 

mu.na.an.ßum 2 

84. [...] x.ka.ni niœ 2.nam œar.œar . . . everything . . . and set it before her, 

igi.[ni].ße 3 mu.na.an.gub 

85. [...] umbin niœ 2.ba ßu bi 2.in.ti She took the gift with fingertips, 

86. [...] ≠nibru± ki .ße 3 œiri 3 ba.an.dab 5 

He . . . directed his steps to Nibru, 

87. [...] x d en.lil 2.ra [ki] mu.un.su.ub . . . kissed the ground before Enlil, 

Although the examples are relatively few and difficult to differentiate, it does appear that 

those examples in which ßu serves an entirely syntactic purpose (so as to reintroduce an 

agent into an achievement predicate that would normally exclude an agentive argument) 

can be distinguished by the fact that the verbs that they appear in conjunction with are

either bear the nominalizing/resultative suffix, *-a, or—in one case at least—are based on 

an adjectival root. In the remaining cases, ßu appears before a fientive (non-adjectival) 

root that does not bear the nominalizing/resultative suffix and, at the same time, seems to 

contribute to the lexical meaning of the compound verb. Only the latter should probably 

be thought of as true members of the BNBV inal class, while the former adopt a BNBV inal 

construction for syntactic reasons and should probably be spoken of as derived BNBV inal 

constructions. 

175

2.6 BNBV inal as a particle-verb construction 

Up to now, I have concentrated on the semantic properties of the BNBV inal construction, 

particularly in terms of lexical aspect and certain interactions between lexical and 

grammatical aspect. Here in the penultimate section of this chapter, I describe the formal 

or distributional features that code the meanings suggested above. What morphosyntactic 

means, in other words, are used in Sumerian to code telicity and the phenomena related to 

it? I argue in the following that the BNBV inal construction is a particular kind of particle- 

verb construction (familiar from a variety of present-day European languages) and that 

the *bi-√ prefix is analogous to the particle in a particle verb construction and contributes 

telicity to the meaning of predicates in which it occurs. It has long been recognized that 

in English, for example, particles—prepositions that lack a nominal complement—are 

often used to control the telicity of verbs. 

(83) He hammered it Ø for three hours / *in three hours. 

(84) He hammered it out for three hours / in three hours. 

In these examples, as is standard practice, a temporal adverb can be used to differentiate a 

telic event from an atelic one: in (83), an atelic verb can only occur with an unbounded 

temporal adverb, for three hours, whereas the telic verb in (84), which is followed by the 

particle out, can also occur with a bounded temporal adverb, in three hours. Since 

Sumerian is strictly verb-final, the possibility of a particle-based system for controlling 

the telicity of a verb seems out of reach. But recent work on the formation of verbal 

176

complexes (Kayne 1994; Koopman and Szabolcsi 2000; Julien 2002) has argued that 

verbal complexes are regularly derived through movement of phrases that appear to the 

right of the verb in languages like English into positions on the left of the verb in 

languages like Sumerian. This section argues that just such a phenomenon is at work in 

Sumerian. 

In an SVO language like English, the syntax of particle-verb constructions is fairly 

clear: although the exact position of the adposition that serves as the particle varies, the 

crucial feature is that the adposition never takes a complement, which results in a “bare” 

adposition (hereafter referred to as a “particle”). 

(85) a. Amanda poured out the bottle of wine 

b. Amanda poured the bottle of wine out 

Although one might imagine that the preposition “out” in (85a) takes the noun phrase 

“the bottle of wine” as its complement to form a prepositional phrase “out (of) the bottle 

of wine,” the variant form in (85b) shows that this is, in fact, not the case (leaving aside 

possible preposition complement relations at underlying levels of representation). The 

variant in (85b) has an ordinary noun phrase, “the bottle of wine,” as the direct object of 

the main verb, “poured,” and a particle at the end of the sentence, “out,” without a noun 

phrase complement of any kind. The fact that the two sentences in (85) have more-or-less 

the same meaning indicates that—in all likelihood—the preposition in both sentences is 

to be interpreted as a particle that has more to do with controlling the telicity of the verb 

177

than indicating the location or adverbial case of the bottle of wine or its contents. The 

telicity of the construction probably arises from the fact that the particle forms what is 

known as a small clause along the same lines as the resultative secondary predicates 

discussed above. 

(86) a. Mary drank [ small clause the bottle [ adjective dry]] 

b. Mary poured [ small clause the bottle [ particle out] 

The similarity between these two constructions is not surprising, particularly in light of 

the fact that both the adjective “dry” and the particle “out” are intransitive states and both 

refer to the defining property of the endpoint of the event that the predicate refers to. 

It should probably be noted as well that, in English at least, resultative secondary 

predicates (“Mary drank the bottle dry”) and particle-verb constructions show complex 

interactions and co-occurrence restrictions with low applicative constructions. Den 

Dikken, for example, offers the following paradigm (1995, 173). 

(87) a. **They sent [ Indirect object Bob [ Direct object a package [ Particle off]]]. 

b. **They fixed [ Indirect object John [ Direct object a date [ Particle up]]]. 

c. **They handed [ Indirect object the children [ Direct object candies [ Particle out]]]. 

178

(88) a. I gave [ Indirect object John [ Direct object his books [ Particle back]]]. 

b. Hey, toss [ Indirect object me [ Direct object that wrench [ Particle up]]], will you? 

c. Would you saw [ Indirect object me [ Direct object a piece [ Particle off?]]] 

Den Dikken (1995, 173-174) elaborates the matter in some detail, but the reason that the 

examples in (87) fail is presumably due to the details of the fine structure of phrasal 

movement in English and need not concern us here. My point in raising examples such as 

(88c) is that they also display a rather consist set of word-order variations in English. 

(89) a. Hey, toss [ Direct object that wrench [ Particle up [ Indirect object to me]]], will you? 

b. Hey, toss [ Particle up [ Direct object that wrench [ Indirect object to me]]], will you? 

c. Hey, toss [ Indirect object me [ Direct object that wrench [ Particle up]]], will you? 

d. Hey, toss [ Indirect object me [ Particle up [ Direct object that wrench]]], will you? 

e. **Hey, toss [ Particle up [ Indirect object me [ Direct object that wrench]]], will you? 

What the series of examples in (89) demonstrate is that given an underlying word-order 

such as [Direct object [Particle [Indirect object]]] in (89a), the verb can “incorporate” 

either the particle as in (89b) or the indirect object as in (89c), but both can only be 

incorporated if they are in a particular order: (89d) is good, whereas (89e) is bad. 

The parallels with the BNBV inal construction are manifold, but somewhat difficult to 

describe due to differences in morphosyntactic structure. 

179

(90) Hey, toss [ Indirect object me [ Direct object that wrench [ Particle up]]], will you? 

(91) gig.e igi bi 2.in.du 8 

[ Indirect object gig-e [ Direct object igi [ Particle bi [ Verb n-du]]]] 

Just as in the BNBV inal construction, the transfer of possession model is paramount in the 

English example, due obviously to the necessary presence of a recipient or benefactee, 

which is incorporated into the verb. The particle (“up”), likewise, codes two distinct 

functions: the relational or orientational value (“up, out, in, back”) that holds between the 

source (the addressee in [90]) and the goal (“me” in [90]) and the telicity of the verb 

through its modification of the direct object. The differences between (90) and (91) are 

also interesting, however, particularly in light of the fact that Sumerian and English differ 

in terms of ergative vs. accusative case-marking systems and basic word-order. Since 

Sumerian uses an inalienable body-part noun to code the relational/orientational value, 

the body-part noun remains in the absolutive case and uses two distinct forms of raised 

possessor construction (locative-terminative and ergative as inalienable possessor, both 

*-e) to indicate the nature of the relation between source and goal. At the same time, it 

seems to indicate the telicity or boundedness of the predicate through the conjunction of 

igi and the *bi-√ prefix. Ultimately these differences seem to result—quite directly— 

from the basic word-order of the respective languages in that the SVO language is 

capable of forming a particle-verb construction by stranding a bare adposition to the right 

of the verb, whereas in an SOV language like Sumerian, this is not possible and the 

180

inalienability in combination with distinct types of raised possessor construction are used 

to accomplish the same morphosyntactic function. 

Den Dikken has noted that the element that corresponds to the particle in English 

particle-verb constructions seems to be represented by verbal affixes in other, more 

highly synthetic languages. He offers several examples in which an applicative use of an 

affix that attaches to the verb can be contrasted with a causative use of the same 

morpheme. 

(92) Dutch applicative in ver- (Den Dikken 1995, 229) 

a. Jan stuurde uitnodigingen voor het feest aan zijn vrienden. 

Jan sent invitations for the party to his friends. 

b. Jan verstuurde zijn vrienden uitnodigingen voor het feest. 

Jan VERsent his friends invitations for the party. 

(93) Dutch causative in ver- (Den Dikken 1995, 229) 

a. Jan maakte zijn positie op de arbeidsmarkt beter. 

Jan made his position on the job market better. 

b. Jan verbeterde zijn positie op de arbeidsmarkt. 

Jan VERbettered his position on the job market. 

Thus the (high) applicative in Dutch forms both an applicative construction in which the 

underlying prepositional phrase (“aan zijn vrienden” in [92a]) can be moved into direct 

181

object position if the applicative morpheme (“ver-“) is affixed to the verb. If, however, 

there is a resultative secondary predicate (“beter” in [93a]), the adjective that formed the 

resultative can move to the position of main verb in combination with the applicative 

morpheme to form a kind of causative construction (“verbeterde” in [93b]). This is 

clearly reminiscent of the bi 2.in.˙ul.le construction dealt with above (with [93a] 

corresponding in certain respects to the ßu ˙ul bi 2.in.du 11.ga construction). The 

interesting thing about Dutch, however, is that the “applicative” morpheme alternates 

with independent particles as in the following examples. 

(94) Dutch applicative in ver- and particle (Den Dikken 1995, 234) 

a. Jan verstuurde uitnodigingen voor het feest aan zijn vrienden. 

Jan VERsent invitations for the party to his friends 

b. Jan stuurde uitnodigingen voor het feest toe/op aan zijn vrienden. 

Jan sent invitations for the party out/up to his friends 

(95) Dutch applicative in ver- and particle + LOW APPLICATIVE (Den Dikken 

1995, 234) 

a. Jan verstuurde zijn vrienden uitnodigingen voor het feest. 

Jan VERsent his friends invitations for the party 

b. Jan stuurde zijn vrienden uitnodigingen voor het feest toe/op. 

Jan sent his friends invitations for the party out/up 

182

The construction in (94) parallels the English example in (90) quite closely, repeated as 

(96b) below. 

(96) a. Jan stuurde [ Indirect object zijn vrienden [ Direct object uitnodigingen … 

183 

[ Particle toe/op]]]. 

b. Hey, toss [ Indirect object me [ Direct object that wrench [ Particle up]]], will you? 

Another set of examples cited by Den Dikken provide an additional example of the 

complex interactions between particle-verb constructions, applicatives and causatives in 

Indonesian. 

(97) Indonesian applicative in -kan (Den Dikken 1995, 232) 

a. ali membuka pintu untuk bapak. 

Ali open the.door for father 

Ali opened the door for father 

b. ali membukakan bapak pintu 

Ali openKAN father the.door 

Ali opened the door for father

(98) Indonesian causative in -kan (Den Dikken 1995, 232) 

a. ratna tidur 

Ratna sleeps 

Ratna is sleeping 

b. Parto menidurkan ratna 

Parto sleepKAN Ratna 

Parto made Ratna fall asleep 

In (97) and (98), a (high) applicative is used to change the benefactee from an oblique, 

adpositionally-marked case to a bare noun phrase that immediately follows the verb. But 

when the applicative morpheme co-occurs with the adpositional phrase rather than 

replacing it, the result is a particle-verb construction as in (99). 

(99) Indonesian -kan as particle (Den Dikken 1995, 232) 

a. parto menulis nama saya 

Parto writes name my 

Parto writes my name. 

b. parto menuliskan nama saya di agendanya 

Parto writesKAN name my in his.agenda 

Parto writes my name up in his agenda. 

184

Den Dikken notes that Voskuil’s translation of (99b) as “Parto writes my name up in his 

agenda” suggests that “the affix -kan corresponds to a PARTICLE in languages like 

Dutch or English” (Den Dikken 1995, 233 citing Voskuil 1990, 87). 

The simplest way of describing BNBV inal predicates as particle-verb 

constructions—particularly given Den Dikken’s discussion of role of verbal affixes in 

forming particle-verb constructions in languages that do not exhibit prototypical particle- 

verb constructions—would be to interpret the *bi-√ prefix as a particle. But I think that 

the fact that *bi-√ prefix verbs regularly occur with either a locative or a BNBV 

construction, in which a locative-terminative argument is the raised possessor of the bare 

noun, suggests that the equivalent of the particle in such constructions is an actual 

postposition: either the locative postposition or the locative-terminative postposition that 

marks the raised possessor in a BNBV construction. The *bi-√ prefix simply indicates 

that the locative or locative-terminative phrase is to be interpreted along with a zero- 

marked noun as forming the equivalent of an existential “there” sentence (see section 3.4 

for details). Taking gig.e igi bi 2-√ as a paradigmatic example and adopting a 

conventionalized representation of it as [wheat-at eye there], the question is how does 

“at” function as a particle in a particle-verb construction in an SOV languages like 

Sumerian. Kayne has argued on the basis of the way in which hierarchically organized 

syntactic structure is linearized (see Kayne 1994 for explication) that the underlying order 

of this series of constituents is the opposite of the order in which they are found in the 

text-artifactual record. This would yield [there eye at wheat] as the underlying order of 

185

constituents, which happens to correspond to the sequence of constituents in an English 

existential sentence: “there is an eye on the wheat.” 

Koopman and Szabolcsi (2000) have argued on the basis of Kayne’s theoretical 

program as well as distributional facts in Hungarian, Dutch and German that SOV 

complex verbs form through a process known as “rolling-up”: the rightmost constituent 

moves to the left edge of the constituent to its left and forms a single constituent with it; 

the process is then repeated. 

(100) a. bi-√ igi e [gig 1] there eye at [wheat 1] 

b. bi-√ igi [[gig 1] e 2] there eye [[wheat 1] at 2] 

c. bi-√ [[[gig 1] e 2] igi 3] there [[[wheat 1] at 2] eye 3] 

d. [[[[gig 1] e 2] igi 3] *bi-√ 4] [[[[wheat 1] at 2] eye 3] there 4] 

In (100), complex verb formation in its simplest form is applied to our paradigmatic 

example: gig.e igi bi 2-√. The underlying order appears in (100a), but in (100b), the 

rightmost constituent [gig 1] has moved to the left of the constituent to its left, forming a 

new compound: [[gig 1] e 2]. This compound then functions as a unit and moves to the left 

of [igi 3] to form a three-term complex: [[[gig 1] e 2] igi 3]. Lastly, the complex made up of 

three terms moves to the left of [bi-√ 4], yielding the attested form of the construction in 

(100d). The interesting thing, at least in my view, about such a derivational process with 

respect to BNBV inal predicates is that at the stage represented in (100c), *bi-√ [[[gig 1] e 2] 

igi 3], the purely relational, non-referential quality of [. . . igi 3], particularly in that it is 

186

juxtaposed to the locative-terminative postposition [[. . . e 2] igi 3], could easily act as a 

particle, an adposition that lacks a nominal complement, since [. . . igi 3] lacks many of 

the features that would typically be associated with a noun, but has several features that 

might be associated with an adposition. 

187

2.7 The link between the BNBV construction and the rise of ergativity 

There are, in my view, two possible analyses of a BNBV construction: these two 

structures seem to correspond to the alienability of the absolutive noun that forms a 

component of the compound verb and they may also correspond to two distinct phases in 

the history of the development of ergativity in Sumerian. The older of these two 

structures, which corresponds to BNBV constructions in which the nominal component is 

inalienable, is analogous to an English existential sentence in which the post-copular 

noun is modified by a secondary/adjectival predicate. 

(101) [ ∃ (en aratta ki .ke 4) gig.e igi √du 8] √bi 

There is [ ∃ an eye (of the Lord of Aratta) opened up on the wheat] 

Within the structure outlined in (101), the ergative case-marked noun is the inalienable 

possessor of igi and is non-agentive, while the apparent main verb, √du 8, is actually an 

adjective modifying igi just as secondary predicates typically modify the direct object in 

secondary predicate environments. The actual main verb in such a construction is the 

verbal prefix *bi-√. As Pylkkänen predicted for low applicative morphemes, the low 

applicative takes three arguments: the lexical verb (√du 8), the direct object (igi) and the 

indirect object (gig.e). 

In contrast, the second of the two structures, in which the nominal component of the 

compound verb is alienable, would have resulted from the grammaticalization of the 

possessive construction as a new ergative construction (Trask 1979; see below, chapter 5, 

188

for an extended discussion of the different pathways through which ergativity in 

Sumerian might have arisen). 

(102) Lugalbanda II [1.8.2.2], l. 55a 

amar.e uzu suß 2 bi 2.in.gu 7 

da 

(*lugal.banda3 .ke4) [ ∃ amar.e uzu suß2] bi gu7 He (= Lugalbanda) had [ ∃ the young one (amar) eat salt-meat] 

Here in (102), the verbal root, √gu 7, is no longer an argument of the verbal prefix *bi-√; 

instead, the tables have turned. √gu 7 is the main verb and takes, in a rather informal 

sense, only two arguments: the causer (Lugalbanda), typically in an ergative case, and a 

complex made up of amar.e, uzu suß 2, and the *bi-√ prefix as the second absolutive 

argument. In formal terms, the most important change is the movement of the *bi-√ 

prefix to preradical position from its originally postradical position. This undoubtedly 

reflects the rearrangement of dependency relations between the *bi-√ prefix and the 

lexical root. Since the head of the clause (the main or highest predicate, in other words) in 

a strictly verb-final language is almost invariably the final morpheme, it is fairly 

reasonable to argue that in the older, non-ergative structure in (101), the *bi-√ prefix 

occurred at the end of the clause. We might, therefore, represent the argument structures 

of the two diachronic phases as follows: 

189

(103) a. BI (du 8, igi, gig) 

b. gu 7 (BI, uzu suß 2, amar) 

In the older of the two constructions, represented in (101) and (103a), the main verb was 

the *bi-√ prefix itself, acting as an existential verb along the lines of, say, a modal 

auxiliary in English; the primary difference was that the *bi-√ prefix and its fellow 

travelers presumably coded features such as quantification, definiteness and evidentiality, 

although these properties are quite difficult to recover from the later phases of the 

language are recorded in the excavated documentation. This resulted in an existentially 

quantified absolutive noun that was modified by an adjectival, secondary predicate in the 

form of √du 8. Note that the existential quantification coded by the *bi-√ prefix has scope 

over the entire clause on the basis of clause-final position. With the development of 

ergativity, however, the main verb is not longer the *bi-√ prefix, but rather a lexical verb 

such as √gu 7. Since √gu 7 is the main verb, it appears at the end of the clause and the 

scope of the existential quantification introduced by the *bi-√ prefix is limited to the 

causee (amar.e) and the absolutive noun ( uzu suß 2). Crucially, the ergative agent/causer 

stands outside the scope of the existential quantification. 

(104) a. [ ∃ (en aratta ki .ke 4) gig.e igi √du 8] √bi 

da 

b. (*lugal.banda3 .ke4) [ ∃ amar.e uzu suß2] bi gu7 190

The change in word-order (the inversion of the *bi-√ prefix and du 8/gu 7) and the change 

in the scope of the existential quantification as represented in (104) would reasonably 

follow from the reinterpretation of *-e as an ergative postposition and the consequent 

“verbalization” of the adjectival and secondary predicate that originally modified the 

absolutive noun. 

In my view, this is due to the fact that the ergative noun phrase is topical and, 

therefore, presupposed relative to the existentially quantified complex including the noun 

followed by the locative-terminative postposition, the bare noun and the *bi-√ prefix. In 

other words, the older structure in (104a) is thetic (a predication of an event such that the 

participants in the event are not differentiated from the event itself), whereas the later 

structure in (104b) is categorical (a predication where a “subject” is identified and a state 

of affairs of predicated of that particular participant in the event). The BNBV inal class is 

distinguished by the fact that the ergative case-marked noun and one component of the 

compound verb, namely the bare inalienable noun, are not referentially discrete and it is 

likely that this indiscretion prevented the conversion of the originally thetic predicate into 

a categorical one and preserved the original semantics of the construction. At some later 

point in the language, the particular configuration of the *bi-√ prefix and the lexical verb 

in (104a) was apparently brought into harmony with the other, non-BNBV inal predicates 

through an inversion of prefix and verbal root, but the semantics of the original 

construction remained intact (For additional discussion of the thetic/categorical 

opposition and its relation to head-internal relative clauses and verbs of perception, see 

below, section 3.5). 

191

3 

Diagnosing presupposition in Sumerian 

To review somewhat, having identified a distributional class, the BNBV inal predicates, on 

the basis of a relatively simple diagnostic in chapter 1 (including an identification of the 

BNBV inal construction as a low source applicative along with the “transfer of possession” 

model that underlies low applicative semantics), chapter 2 sought to identify the type of 

lexical aspect coded by a low source applicative such as the BNBV inal predicates. In the 

end, the BNBV inal predicates were tentatively classified as achievements, which also 

explained two seemingly unrelated properties: (a) the reduplication of *bi-√ prefix verbs 

(switching to a larger group of data due to the virtual non-existence of reduplicated 

BNBV inal predicates) which seem to function as adverbial clauses of cause or result (due 

to the point-like aspectual profile of achievement predicates and the consequent inability 

of another event to co-occur with the point of transition) and (b) the non-agentive 

character of the zero-marked, absolutive case-marked noun immediately preceding a *bi- 

√ prefix verb (including so-called “transitive” verbs in which the patient is not affected 

by the event that the verb refers to, e.g., “Mary reached the summit”). Two consequences 

of chapter 2, which were somewhat unexpected (though predicted by the theoretical 

literature to some degree) lead us to the subject matter of this chapter: the point-like 

192

aspectual profile of achievements as well as the use of the *bi-√ prefix to form 

constructions that are analogous to indefinite relative constructions in other languages 

suggest that one of the primary functions of the *bi-√ prefix is to indicate that the 

absolutive noun that immediately precedes the *bi-√ prefix is equivalent in a number of 

respects to an indefinite noun in languages that make use of an indefinite article, where 

the indefinite noun is quite often derived, in some fashion, from the number one. 

The only substantive way of testing whether what I have previously referred to as a 

“bare indefinite” is in fact indefinite or not is to identify what is often termed a 

“definiteness effect.” A morphosyntactic environment which exhibits a definiteness effect 

(at least of the most well-known type such as English existential “there” constructions: 

“There is a rat in the living room”) will generally exclude definite nouns from that 

environment unless the noun is focus-affected. I argue that BNBV inal predicates in the 

non-progressive/perfective aspect—the only grammatical aspectual category in which 

they regularly appear—offer an environment that exhibits a definiteness effect and that 

the zero-marked, absolutive noun that occurs in such environments must, therefore, be 

indefinite. I see the definiteness effect environment in question as defined by two 

essential features: the *bi-√ prefix codes a point-like achievement semantics, while the 

perfective (˙amt¬u) aspect ensures that there is only one such point-like event. In the 

progressive/imperfective aspect, however, the “oneness” of the predicate and hence its 

similarity to the indefinite article would break down. There is, however, one particular 

construction in which the restriction imposed by the definiteness effect weakens and the 

verb regularly appears in the progressive/imperfective aspect, namely the negative 

193

contrastive focus construction with the form *XP nam bi-√ (where XP refers to either a 

nominal or a postpositional phrase). 

On the basis of contextual investigation of all the known exemplars from the Old 

Babylonian period (in chapter 4), I interpret the *XP nam bi-√ construction as a negative 

contrastive focus construction (“It was not IN THE STREET that I hit the car, [but rather 

IN THE DRIVEWAY]”) and argue that the reason that the verbs in such constructions 

regularly appear in the progressive/imperfective is that the contrastive focus neutralizes 

the definiteness effect, allows a definite noun to occur in the construction, and forces the 

verb to appear in the progressive/imperfective so as to indicate that the noun is no longer 

indefinite. Since the indefinite interpretation of the BNBV construction (including 

alienable BNBV predicates) stems in part from the perfectivity of the verbal root, a 

progressive/imperfective verb presumably indicates that the absolutive noun is not 

indefinite (and by implication definite). Interestingly enough, the lone example of a 

BNBV inal predicate in the *XP nam bi-√ construction exhibits a number of eccentricities 

that differentiate it from both non-compound *bi-√ prefix verbs and other compound 

*bi-√ prefix verbs, the most striking of which is the absence of the marû-stem suffix that 

is found on all other, non-BNBV inal examples of the construction. 

The common theme that relates this chapter and the following one is presupposition. 

Indefinite nouns regularly lack presupposition in most contexts, but an indefinite noun is 

often used to introduce a new topic that is subsequently presupposed by subsequent 

discourse: in large measure, the meaning of any particular indefinite noun is a function of 

the relation that holds between presupposed and entailed propositions or parts thereof. 

194

Section 3.1 starts out with a brief definition of presupposition and juxtaposes it to 

entailment. The rest of the section notes the use of existential verb constructions to code 

indefiniteness in certain languages and argues that the same mechanism also operates in 

Sumerian. In the following section, section 3.2, a basic opposition is drawn between 

presupposition and focus, where the part of a sentence that is not under focus—namely 

the presupposition—remains true under negation in contrast to the focused constituent 

which is the only element that is negated. Section 3.3 looks at topicalized *bi-√ prefix 

verbs as represented by the form *mini-√: these are, I argue, to be interpreted as 

indefinite relative constructions that have been topicalized (and also made definite) 

through the addition of the third person animate possessive pronoun, *-ni. In other 

languages that make use of head-internal relative constructions, the head noun of the 

relative clause must be indefinite, but the clause itself can then be made definite as a unit; 

indefinite relatives in *bi-√ seem to conform to the pattern quite closely. Section 3.4 

defines the definiteness effect in some detail and section 3.5 returns to the question of 

head-internal relative constructions in Sumerian. Recent work on head-internal relative 

constructions in Korean, Japanese and other typologically similar languages has 

suggested that head-internal relative constructions regularly constitute definiteness effect 

environments, so section 3.5 also considers whether the same holds true in Sumerian. 

195

3.1 Presupposition and assertion, topic and focus 

Prior to delving into the correlation between topic, focus and presupposition, I take a 

moment here to review the basic meaning of the terms presupposition and assertion and 

also to establish some fairly broad correlations in their usage so as to orient the reader to 

subsequent discussion. In everyday use, the term presupposition is typically used to 

describe the part of an utterance that the speaker assumes his or her interlocutors would 

agree with without any argument (the following examples from Saeed 2003, 101): 

(1) a. Her husband is a fool. 

b. She has a husband. 

(2) a. I don’t regret leaving London. 

b. I left London. 

In each pair of examples, the assertion of the first sentence by the speaker presupposes 

the truth of the second sentence in each pair. In (1), the speaker and the interlocutor must 

mutually acknowledge that the referent of she has a husband—only when both speaker 

and interlocutor are in agreement about (1b), can the speaker go on to assert (1a). 

Likewise, it would be difficult for someone to regret doing something like leaving 

London as in (2a), if he or she had never left the city in the first place as in (2b). In 

languages in which topic and focus are explicitly marked, presupposition and assertion 

are, for the most part, mapped directly into topic and focus. The topic of a clause is quite 

196

egularly presupposed (and therefore usually definite), while the focus of a clause 

typically corresponds to the assertion. 

As segments of text follow in sequence, subsequent clauses regularly presuppose 

certain elements of previous clauses, but at the same time, certain assertions require that 

other assertions that may or may not follow in sequence also be true—this is entailment 

(The following examples are taken from Saeed 2003, 98-99, ex. 4.52, 4.56, and 4.57.) 

(3) a. The anarchist assassinated the emperor. 

b. The emperor died. 

(4) a. I bought a dog today. 

b. I bought an animal today. 

In (4), we have the simplest form of entailment, which involves a hyponym “dog” being 

replaced by a hypernym such as “animal.” Since, at least from the logical point of view, 

all dogs are animals, the sentence in (4b) “includes” the sentence in (4a) and any 

assertion of (4a) necessarily “includes” an assertion of (4b). Although they are somewhat 

more complex, the same goes for verbs, hence the entailment that holds between (3a) and 

(3b). Since the verb “assassinate” includes death as one of its semantic components and 

the referent of the nominal argument that undergoes assassination in (3a) is the same one 

that experiences death in (3b), namely the emperor, any assertion of (3a) necessarily 

includes an assertion of (3b). 

197

The easiest way of differentiating presupposition and entailment is through the 

different responses of the two phenomena to negation. If the sentences in (3) are negated, 

we end up with the following. 

(5) a. The anarchist did not assassinate the emperor. 

b. The emperor did not die. 

The tricky part is that the negative form of the sentence in (5a) does not entail the 

negative sentence in (5b) since it is entirely possible that, although the anarchist tried his 

or her best to end the emperor’s life and failed, the emperor nonetheless died of gout or 

something like that. That is to say, the negation can apply to either the part of the 

assertion in (5a) that is equivalent to (5b), in which case (5b) would be true, or the 

negation can apply to the manner in which the emperor died, in which case (5b) would be 

false. If one asserts, however, that (5b) is true, then (5a) must also be true, since (5b) is an 

essential component of (5a). 

Presupposition, however, works quite differently and if negation is introduced into 

one of the examples of presupposition dealt with above, the difference should be 

abundantly clear. 

(6) a. Her husband is not a fool. 

b. She does not have a husband. 

198

The sentence in (6a) does not have (6b) as its presupposition; instead, it has the original, 

non-negative version of (6b) as its presupposition, namely (1b): “She has a husband.” 

Basically, a presupposition remains true even when the sentence in which it occurs is 

negated. Unlike the example of a negated entailment in (5b), however, the negation of the 

presupposition as in (6b) does not so much result in a negative sentence such as (6a) as 

the feeling that the assertion in (6a) no longer makes any sense in the first place; the 

negation of presuppositions has recently been discussed by Herburger (2000) under the 

heading of “backgrounded focal entailment” and we return to her discussion of the matter 

briefly in the following section (for a case-study in the use of negation to pick apart 

presupposition and entailment in an archaeologically recovered language like Sumerian, 

see Loprieno 1991). 

If someone asserts that something is the case in a particular sentence, however, the 

part of the sentence that is being asserted will not, as a rule, be part of the presupposition: 

clearly one cannot presuppose and assert the same thing at the same time. In (1a), for 

example, the part of the sentence that the speaker is asserting is that someone is a fool, 

whereas the part of a sentence that is being presupposed is that there is someone who 

meets the criterion of being her husband. In English, this contrast is particularly evident 

in existential sentences with there in “subject” position as in (8) because the usual 

location of presuppositional elements is in the subject position, here occupied by “there.” 

(7) There is a rat in your soup. 

199

In the case of an existential sentence, the speaker does not include him- or herself in the 

utterance at all, since the claim is that the state of affairs exists quite independently of the 

person who is uttering the statement. In a certain sense, existential sentences are the 

inversion in pragmatic terms of so-called explicit primary performatives as in (8). 

(8) I promise to bring the book to class tomorrow. 

Although both (7) and (8) necessarily involve a speaker who is making an assertion of 

one kind or another, in (7), there is no element of the sentence that refers to the speaker 

and in fact there is no referential noun phrase in the nominative position before the verb. 

The presence of expletive there in the position in a sentence usually reserved for the 

person who takes responsibility for the truth of the state of affairs referred to by the rest 

of the sentence prevents any kind of responsibility or intention from being attributed to 

anyone (known as illocutionary force in some circles, abstracting away from culturally 

contingent notions such as intentionality, see Searle [1969] for the emic view and 

Rosaldo [1982] for the anthropological critique, cf. Duranti [1997, 231]). In precisely the 

opposite way, the first person pronoun in (8) results in a situation in which the speaker’s 

intention is presupposed and the speaker is held responsible for the promise in question 

(see Austin 1975 [1962]; Silverstein 1993b, 46). 

There are, furthermore, similarities between existential sentences, perception reports 

and (although not dealt with in this paper) reported speech in that they correspond to a 

segment of the implicational hierarchy for systems of evidentiality. 

200

(9) direct experience > visual > heard > inference > hearsay 

existential sentence > perception reports > direct speech > indirect speech . . . 

Thus the morphosyntactic form of expressions in English that are used to report 

perception in all its variety systematically exclude presupposition and regularly occur 

with indefinite perceived objects. At the same time, in a description of “languages 

without indefinite pronouns, Heine describes the use of existential sentence constructions 

to form indefinite pronouns in Tagalog and a Nilo-Saharan language named Lango 

(Heine 1997, 54-57, ex. 120, 130). Lipinski has also commented on definiteness in 

Nilotic (a branch of Nilo-Saharan) as follows: 

like Nilotic languages, Old and Middle Egyptian dispense, as a rule, with any 

equivalent of a definite or indefinite article, but an important feature of several 

Nilotic languages consists in showing definiteness by the use of verbal forms 

involving an internal vowel change, viz. the “qualitative” (indefinite) and the 

“applicative” (definite) (Lipinski 2001, 26) 

Presumably a significant portion of the morphosyntactic complexity of Egyptian arises 

from the need, in the earlier phases of the language, to code (in)definiteness through 

syntactic means such as topic and focus, a situation that parallels that of Sumerian in a 

number of respects. 

In terms of the language internal resources of Sumerian, clauses that report direct 

speech and/or direct perception are the best candidates if our goal is the identification of a 

morphosyntactic environment in which indefinite nouns regularly occur, and it just so 

happens that the verbal prefix that usually occurs with both verbs of direct speech and 

201

verbs of direct perception in Sumerian is the *bi-√ prefix. Much of the initial impetus for 

the investigation of the hypothesis proposed in this dissertation stems from this rather 

general observation. Note as well that, whereas English uses a verb of creation to form 

one type of light verb construction for verbs of speaking, namely “make,” Sumerian 

actually uses the verb du 11, “to say” as an auxiliary verb in the formation of a variety of 

constructions including the light verb constructions mentioned above (section 2.5). It is 

likewise apparent that not only the *bi-√ prefix, but also other members of the set of 

deictic elements that include it (*-ni-, *-bi-, and *-ri-) migrated to the main verb and were 

grammaticalized as auxiliary verbs coding not only an expletive subject in either 

presentational or existential mode (and thereby, at least in the case of the distal forms, 

*-bi- and *-ri-, indefiniteness), but also the relative distance in time and/or space from the 

event referred to by the predicate (a kind of evidentiality). Woods argues for two distinct 

sets of deictics: what I will term set A (*-e, *-ne, *-be) and set B (*-e, *-ße, *-re). 

Although Woods’ model has much to recommend it, I would argue for a slight 

reconfiguration if only to capture the fact that set B clearly includes a case-marking 

component (the generally allative/dative cast of the members of set B) and seems to be 

used predominantly in the grammatical texts—for the most part to modify 

inalienable/orientational nouns such as gu 2 (see Woods 2001, 167, for an insightful 

discussion of the riverine system of deixis and its exclusive use of set B). If nothing else, 

the use of set B in a somewhat restricted distribution but quite systematically with gu 2 

provides quite a bit of justification for distinguishing set B from the other deictics as a 

distinct series. Woods’ reconstruction of set A, however, is in need of revision in my 

202

view. Although it is one of the central presuppositions of the Woods’ hypothesis about 

Sumerian deixis, I find it untenable that Old Babylonian scribes would have glossed both 

*-e and *-ne with annûm, the Akkadian proximal demonstrative, if they were perceived 

as part of the same deictic series. 25 Given the fact that there appear to be at least two 

deictic series, the simplest hypothesis would be that *-e and *-ne each head a distinct 

series; since it is fairly clear that *-e is the proximal member of set B, I would argue that 

*-ne heads set A. If we further assume that *-be is a distal member of set A, the resulting 

table of deictics would be something like the following. 

Set A Set B Deictic value 

*-ne *-e PROXIMAL 

*-be *-ße DISTAL.VISIBLE 

(*-re) *-re DISTAL.NON-VISIBLE 

Woods (2001, 95-130) has shown that the VISIBLE/NON-VISIBLE opposition is valid 

in at least some cases, and, furthermore, that in the literary tradition *-re in conjunction 

with temporal nouns was typically used to refer to the mythic, a-historical time period 

typically associated with cosmogony and mythology. Woods has also shown that the 

visibility-driven system of set B can be mapped into person-driven system as represented 

in set A (Woods 2001, 146, 158, 168). I would argue, further, that the primary difference 

between set A and set B is that set A seems to be more closely associated with 

25 Even if *-e specifically referred to speech act participants to the exclusion of non-participants, as suggested by 

Woods, the absorptive character of deictics (Agha 1996, 647-648) would nearly ensure that, on at least some occasions, 

*-ne, which is “less proximal” in Woods’ reconstruction would have been equated with ullûm, the distal member of the 

Akkadian deictic series. In other words, a reconstruction in which *-e is proximal to speaker, *-ne is proximal to 

addressee, and *-be is distal from both is not unreasonable, but we might expect to find *-ne occasionally equated to 

ullûm if that were the case. 

203

demonstrative reference, whereas the members of set B all code a locative or allative case 

of some kind. Although *-re is clearly a member of set B as evidenced by the OBGT 

paradigms, the parentheses around *-re in set A are meant to indicate that presence of 

*-re in set A is somewhat uncertain; nonetheless, some of the examples of *-re cited by 

Woods (2001, 118-123) are presumably members of set A rather than set B due to the 

fact that they are clearly demonstrative in use with little or no case-marking value and 

show no obvious connection to the riverine system of deixis or visual perception. 26 

Woods also cites several examples from the OBGT and MBGT (Woods 2001, 133- 

139) that are—as Woods acknowledges—contradictory as they exemplify two distinct 

paradigmatic series: set B (*-e, *-ße, *-re) and what I will term set C (*-ne, *-ße, *-re). 

Woods argues that set C represents an aberrant form of set B, but I think that a more 

parsimonious explanation can be found if we begin with the deictic series as outlined 

above. Given the widespread grammaticalization of the set A distal demonstrative, *-be, 

both as the default possessive pronoun in the possessive pronoun topicalization system 

and as the verbal prefix *bi-, its replacement by the corresponding member of set B 

should not come a much of a surprise. Woods himself has argued (2001, 168, 220) that 

the most likely site of grammaticalization within the deictic system is the element that 

serves as an anaphoric pronoun most readily, namely *-be, the distal member of set A. 

Once *-be had been grammaticalized as a component of the verbal prefix, its position 

26 I should emphasize that such a reconfiguration of the demonstrative series need not call in question the overall tenor 

of Woods’ argument that matters of deixis play a central role in the grammar of Sumerian. Some form of obviation (see 

Aissen 1997), in which *-e may have functioned as a topic marker of some kind (and consequently as the marker of a 

proximal argument in an obviational system) presumably played a part in Sumerian grammar at some point in its 

history, but I would suspect that such an obviational system antedates even the development of case-marking in 

Sumerian, which I take as a fairly early development. 

204

within set A was simply replaced by the corresponding member of set B, yielding the 

attested form of set C. This seems to be demonstrated fairly well by the “aberrant” 

paradigms to be found in the OBGT materials cited by Woods (2001, 137). The early 

grammaticalization of *-be would also provide some explanation of the fact that *-be 

never exhibits the full or stressed form, **-ben, in the text-artifactual record (Woods 

2001, 179). 

205

3.2 Focus and focal presupposition 

Before turning to topic-marking constructions that involve the *bi-√ prefix, it is probably 

best to devote a few pages to focal presupposition. Just as negation plays a critical role in 

differentiating presupposition and entailment, it can also serve as a useful context for the 

elucidation of focus and focal presupposition. Take, for example, a relatively simple set 

of three variations on the same sentence in which the focal intonation (as indicated by 

capital letters) occurs on a different constituent in each variant. 

(12) a. Mary PUNCHED the mayor on Wilshire Boulevard. 

b. Mary punched THE MAYOR on Wilshire Boulevard. 

c. Mary punched the mayor ON WILSHIRE BOULEVARD. 

Each of these three intonational variants serve as the response to three distinct wh- 

questions. 27 

(13) a. What did Mary do to the mayor on Wilshire Boulevard? 

b. Who did Mary punch on Wilshire Boulevard? 

c. Where did Mary punch the mayor? 

27 The three intonational variants in (12) also have explicit syntactic variants that make use of the cleft (“It is the mayor 

that . . .”) and pseudo-cleft (“What Mary did was . . .”) constructions, but in the interest of brevity I have not gone into 

the matter here. See below in section 4.6 for a discussion of the difference between informational focus (typically 

coded through intonation) and identificational focus (usually coded by some kind of movement or morphosyntactic 

configuation). 

206

The same three variations under negation also yield an interesting result: it is only the 

part of the sentence that has focal intonation that is negated, the rest of the sentence 

remains in force. The import of the negation of a particular focal constituent is often best 

highlighted through the use of a subsequent contrastive sentence. 

(14) a. Mary didn’t PUNCH the mayor on Wilshire Boulevard; (she BIT him) 

b. Mary didn’t punch THE MAYOR on Wilshire Boulevard; (she punched 

THE GOVERNOR) 

c. Mary didn’t punch the mayor ON WILSHIRE BOULEVARD, (she 

punched him IN A CAB COMING FROM THE AIRPORT) 

The point is that focal intonation dramatically changes not only the pragmatic 

appropriateness of a particular utterance, but also the truth conditions and hence the 

semantics of any particular assertion. 

The traditional view holds that the part of a sentence that is not focused is 

presuppositional just as “her husband” is presuppositional in “her husband is a fool.” This 

is basically correct in the vast majority of cases. 

(15) a. Her husband is not A FOOL. 

b. Mary didn’t punch the mayor ON WILSHIRE BOULEVARD. 

207

In both (15a) and (15b), the focused part of the sentence is negated, but the presupposed 

part of the sentence is not: thus, even if (15a) and (15b) are true, someone in (15a) still 

has a husband, and Mary will still likely go to jail for assault in (15b). But specialists in 

semantics have actually expended quite a bit of energy attempting to differentiate focal 

presuppositions as in (15) from the logical presupposition involved in some of the 

examples at the beginning of the chapter such as (2), repeated below as (16). 

(16) a. I don’t regret leaving London. 

b. I left London. 

The sentence in (16b) would only be a focal presupposition if the focus structure in (16a) 

were as in (17a) below. 

(17) a. I don’t REGRET leaving London. 

b. I don’t regret LEAVING LONDON. 

The example in (17a) carries the implication that there is some other emotion that the 

speaker feels toward his or her departure from London, which is not the case in (16a), 

where the somewhat artificial omission of focal intonation forces the reader to fall back 

on other interpretive principles such as the factive character of “regret” (Kiparsky and 

Kiparsky 1970). The contrastive focus in (17b) explicitly excludes (16b) as 

presupposition. Clearly focus introduces an orthogonal dimension to questions of 

208

presupposition, and presupposition in the logical sense differs in certain fundamental 

ways from focal presupposition. Herburger (2000, §§2.3-2.5) has argued, quite 

persuasively in my view, that focal presupposition is not, in fact, presupposition at all, but 

rather what she terms a “backgrounded focal entailment.” Nonetheless, if only so as to 

preserve the title of this chapter, I use presupposition in its older somewhat less precise 

meaning in the balance of the chapter. 

Of perhaps more importance for the investigation of Sumerian morphosyntax than 

subtleties of presupposition and entailment in focal sentences is the basic opposition 

between topic and focus. Although presupposition and assertion, as a pair of related 

concepts, bear a great deal of similarity to topic and focus, respectively, it is probably 

best to keep them separate as analytical terms, even if the correlation between them does 

provide a good deal of the scaffolding, linking together different parts of my 

investigation. Definitions of and theories concerning topic and focus are, as a rule, quite 

involved and not immediately applicable to Sumerian at the present state of our 

knowledge, but I would like to characterize these phenomena in a preliminary way. To 

oversimplify in the direction—I hope—of truth, three features can be used to differentiate 

topic and focus: (a) definiteness, (b) pronominalization and (c) sentence-initial/sentence- 

final position. Topics are regularly definite, are capable of being resumed by pronouns in 

the core of the sentence and can typically be found in sentence-initial position, whereas 

focused constituents can be either definite or indefinite, are not resumed by pronouns in 

the core of the sentence and are typically found as close as possible to the end of a 

209

sentence (taking into consideration other typological factors such as verb-final word 

order). 

(18) a. Gudea, Cyl. A 5:4 [ETCSL 2.1.7] (Zólyomi 1993, 35) 

[ Topic e 2.a œiß.˙ur.bi] im.œa 2.œa 2 

[ Topic As for the plan of the temple] i, he was setting it i out. 

b. Gilgamesh and the Netherworld [1.8.1.4], l. 179 

[ Topic œiß e.ke 4.ma.zu] ganzer.ta [ Focus œa 2.e] ga mu.ra.ab.e 11.de 3 

[ Topic As for your ekema] i, let [ Focus me] retrieve it i for you from Ganzer, 

In the topic construction in (18a), the topic appears in sentence-initial position and bears 

a possessive pronoun, here made possible through the use of an anticipatory genitive 

construction (“Of the house, its plan”) and I have translated it accordingly using an “As 

for ...” topicalization and a resumptive pronoun (the coreferential elements in the 

translation share the subscript “i”). As noted above, the definiteness (and hence its 

suitability as topic) of the topical phrase e 2.a œiß.˙ur.bi is indicated by the possessive 

pronoun. The focus construction in (18b), however, is in answer to a wh-question, “who 

will retrieve my ekema from Ganzer,” and as expected in the answer to the wh-question, 

the first person singular pronoun œa 2.e, occurs as far to the right as possible in the 

sentence and can be juxtaposed to the topical noun phrase on the extreme left edge of the 

sentence. 

210

My working hypothesis, which has been made more-or-less explicit from time to 

time in the chapters leading up to this point, has taken nouns bearing a possessive 

pronoun in sentence-initial position (alongside ergative case-marked nouns and 

independent pronouns) as topical noun phrases due to the fact that they satisfy all three of 

the features associated with topics: the possessive pronoun construction is necessarily 

definite, I have assumed that they are resumed by null pronouns in the core of the 

sentence and have translated accordingly although I know of no independent evidence for 

such resumptive pronominalization, and they regularly occur in sentence-initial position. 

The identification of focus constructions is somewhat more difficult since they regularly 

lack possessive pronouns (unless they are also followed by the copula, which is 

presumably introduced in focus precisely so as to allow for possessive pronouns and 

other modifications of what would otherwise be a bare noun). Clearly the absence of 

resumptive pronominalization is difficult to identify, but the fact that they regularly occur 

in the position immediately before the sentence-final verb presumably indicates that there 

is no subsequent position within the sentence in which a resumptive pronoun could occur. 

The two following sections (as well as the entirety of chapter 4) speak to certain 

aspects of topic constructions and focus constructions respectively: section 3.3 offers a 

reinterpretation of verbs with the prefix *mini-√ as derived from *bi-√ prefix verbs that 

have undergone topicalization through the addition of the third person animate possessive 

pronoun, *-ni, as in the case of simple nouns above. This section also goes into the rather 

unusual structure of certain head-internal relative clauses in Sumerian that corresponds to 

211

similar structures in other languages and at the same time raises the issue of definiteness 

effects that preoccupy section 3.4. 

212

3.3 Topicalized and identificational *bi-√ prefix verbs in *mini-√ 

Before turning to the various issues surrounding definiteness effect environments and 

their relation to head-internal relatives clauses, this section represents a brief tangent so 

as to clarify both the difference between presupposition and focus (including the 

association between presupposition and topicalization) and an important point of 

Sumerian morphology: the derivation of verbs bearing the *mini-√ prefix from 

underlying *bi-√ verbs through the addition of a third person animate possessive 

pronoun, *-ni (the derivation of *mini-√ from *bi-ni-√ was first proposed by Postgate 

[1974, 21-22], see Attinger 1993, 273-275 and Rubio forthcoming, §3.9.6 for recent 

bibliography, but the interpretation of *-ni- as the possessive pronoun has not been 

suggested previously). On the basis of superficial formal as well as more substantial 

semantic similarities, I will assume for the time being that verbs of the form *mini-√ are 

definite versions of the indefinite relative construction dealt with in the previous chapter. 

Note, however, that the examples of the indefinite relative construction cited earlier 

regularly lacked pre-root pronominal agreement, *bi-Ø-√, whereas the topicalized form 

regularly includes it, *mini-n-√. 

Several authors have argued against Postgate’s derivation of *mini-√ from *bi-ni-√ 

on both formal/evidential grounds as well as semantic ones. Rubio has noted (personal 

communication, August 2004) that if *mini-√ were derived from *bi-ni-√, we might 

reasonably expect that in one or two cases the underlying form (sometimes known as a 

“morphographemic” orthography) might be attested in the text-artifactual record. All 

things being equal, such an evidential requirement would be entirely valid, but in the case 

213

of *mini-√ < *bi-ni-√, the situation is complicated by the fact that the prefix *bi-ni-√ is 

attested, but it is regularly and—in my view—correctly interpreted as an orthographic 

representation of bi 2.i 3-√ (= *bi-i-√), that is the *bi-√ prefix followed by a vocalic or glide 

preverbal pronominal agreement marker that is often associated with the directive case 

(Rubio forthcoming, §3.7.3; Karahashi forthcoming, 5 and 7; on the morphological form 

of the directive infix /y/, see Zólyomi 2000). 28 If bi 2.i 3-√ were the conventional 

orthography for a different verbal prefix, this would explain why the underlying form of 

*mini-√, namely *bi-ni-√, never surfaces in the text-artifactual record. The other 

objection that has been raised to Postgate’s derivation is that the underlying form, namely 

*bi-ni-√, would include two morphemes that are each “locative” in meaning and that we 

should expect to be in complementary distribution (Rubio forthcoming, §3.9.6). But in 

my view, the *-ni- element is not “locative” in meaning and, therefore, shares no 

complementary distribution with the *bi-√ prefix. Besides, if *-ni- were “locative” in 

meaning, how is it possible that it can be replaced by *-ri- in the prefix *miri-√, when 

*-ri- is often associated with the second person dative infix, namely *-ra-: how, in other 

words, do a “locative” agreement marker, *-ni-, and a second person directive case 

marker, *-ri-, form a coherent paradigm (see my discussion of the directive case 

hypothesis in section 1.1 above). If, on the other hand, the *-ni- and *-ri- infixes originate 

as paradigmatic alternations in the deictic series *-ne, *-be, *-re, then it makes a good 

28 Although I agree with the growing consensus as to the morphological form of the “directive infix” as it is usually 

now called, the morphosyntactic interpretation of the morpheme remains, in my view, quite uncertain; my support of 

the Zólyomi’s morphological analysis should not, therefore, be taken as agreement as to the morphosyntactic function 

of the morpheme. 

214

deal of sense that two members of that paradigm might well migrate to the verb in the 

same morphosyntactic position. 




349. gu.ru.um na 4 

215 



As for the pile of stones that had been 

kur.ra mi.ni.in.ak accumulated in the mountains, 


Since it (pile of stone) stretched across them 

a 2 bi 2.in.su 3.su 3.ud (mountains) like floating clouds, 

351. bad 3 ma˙.gin 7 kalam.ma Blocking the view of the land like a huge 

igi.ba bi 2.in.tab.≠ba± wall, 

352. œiß bala ki.ßar 2.ra He is the one who installed a spigot (?) at 

mu.ni.in.œar the horizon, 

This is a nice example of a topicalized *bi-√ verb. The underlying form of line 349 is 

probably something like *[gu.ru.um na 4 kur.ra bi 2.in.√ak].ni “his (= Ninurta’s) pile of 

rocks which he caused to accumulate in the mountains,” which is marked as topic 

through the use of the possessive pronoun topic-marking construction. The problem is 

that the third person animate possessive pronoun, *-ni, apparently cannot directly follow 

a finite verb, when it is functioning as an indefinite relative construction, therefore I 

would argue that the third person animate possessive pronoun incorporates into the verbal 

prefix: /bin/ + /ni/ > /bi ... n/ + /ni/ > /bi-ni-n/-√ (LoAp-3AnPoss-3AnimAgr-√). I should

emphasize that in all likelihood this is not infixation, but rather underlying suffixation: 

[n-√]-bi 2 + -ni > [n-√]-bi 2-ni > bi 2-ni-[n-√] > mi.ni.in.√. Whether the same goes for forms 

like mu.ni.in.√ ([n-√]-mu + -ni > [n-√]-mu-ni > mu-ni-[n-√] > mu.ni.in.√) and ba.ni.in.√ 

([n-√]-ba + -ni > [n-√]-ba-ni > ba-ni-[n-√] > ba.ni.in.√) must remain an open question, 

but I hope to extend my investigation of topicalized verb forms to these other verbal 

prefixes in future. 

(20) A tigi to Enlil for Ur-Namma (Ur-Namma B) [2.4.1.2], ll. 2-5 

2. en nam.nun gal.[la.ni ...] The lord who ... his great princeship, 

3. d nu.nam.nir an ki lugal.[bi ...] As for Nunamnir, the king of heaven and 

216 

earth ..., 

4. uœ 3.œa 2 igi mi.ni.in.il 2 sipa zi Looked around among the people—the good 

ur. d ≠namma± shepherd, Ur-Namma, 

5. kur gal d en.lil 2.le uœ 3 ßar 2.ra.ni.a The Great Mountain, Enlil, chose 

? 

im.ma.ni.in.pa3 [(x)] him from among the multitude, 

The ETCSL translation in (20) attaches the first half of line 4 to the previous line as its 

main predicate, but given the conventions of line demarcation, in which a finite verb 

would generally end its line unless a sequence of short predicates followed (not the case 

here), it is unlikely that uœ 3.œa 2 igi mi.ni.in.il 2 is a finite main clause. Given its position at 

the beginning of a series of epithets, the most reasonable hypothesis is that it is a relative 

clause describing a characteristic of Ur-Namma, namely that he is “the one given 

prominence, lit. the one whose head was raised up over the people.” This allows the first

half of line 4 to be the first of a pair of epithets preceding the mention of Ur-Namma at 

the end of line 4, which also characterizes the relationship that holds between Enlil and 

Ur-Namma. The use of a topicalized indefinite relative as an epithet is not particularly 

surprising, but the same construction is also used to form relative clauses which cannot 

be characterized as epithets such as the following. 

(21) Lugalbanda II [ETCSL 1.8.2.2], ll. 207-208 


217 

As for the Bird, who was looking from 


208. [lugal].≠ban 3±.da ki.ta igi As for Lugalbanda, who was looking from 

mi.ni.in.il 2 sa˙ar erin 2.e du 8.a below, he saw the troops enveloped 

igi bi 2.in.du 8.ru in dust, 

A literal interpretation of line 207, for example, would need to include the causative 

semantics of non-BNBV inal *bi-√ prefix verbs, yielding “The bird, whose eye was raised 

up, saw the troops.” But since igi is an inalienable noun, we can reasonably infer that it 

was the bird who raised his own eyes and translate accordingly (note that the causative 

semantics of the non-BNBV inal predicate is preserved, however, indicating that there is an 

agent/causer of the action, which is not true of BNBV inal predicates—this corresponds to 

the fact that verbs of attempted perception such as “look” and “listen” tend to be agentive 

cross-linguistically, whereas verbs of achieved perception such as “see” and “hear” do 

not). Manuscript QQ (3 N-T 824), interestingly, offers a different form of the first verb in

line 207: instead of igi mi.ni.in.il 2 eren 2.e igi bi 2.in.du 8.ru, QQ has igi i 3.il 2 eren 2.na igi 

bi 2.ib.du 8. This raises the possibility that in QQ the redactor interpreted the first half of 

the line as a finite verb, i 3.il 2, followed by a second finite verb, but a second verb that is 

no longer a BNBV inal construction due to the locative postposition that follows eren 2 in 

manuscript QQ. Although I can do no more than speculate, one possibility is that a single 

ergative case-marked noun cannot function as an agent with respect to one predicate, 

i 3.il 2, and as an inalienable possessor with respect to another, igi bi 2.ib.du 8. This seems to 

be further supported by the fact that the majority version, which is a BNBV inal 

construction, agrees with the perceived object in number: bi 2.in.du 8.ru (< bi-n-dur-eß = 

LoAp-3AnimArg-see-Plur). Such an agreement pattern only seems to be available in the 

BNBV inal predicate, since the variant form of the line in manuscript QQ, bi 2.ib.du 8 (< bi- 

b-dur = LoAp-3InanAgr-see) either agrees with some other argument or treats erin 2 as a 

collective. The “collective” interpretation would seem to be favored by the presence of 

the locative postposition following erin 2 in manuscript QQ. Two further examples that 

are similar in a number of respects occur in Enmerkar and the Lord of Aratta, line 172 in 

(22) below, and the Exploits of Ninurta, line 349 in (23) below. 


170. ˙ur.saœ 5(diß) ˙ur.saœ 6(diß) He traversed five mountains, six mountains, 

˙ur.saœ 7(diß).e seven mountains, 

im.me.re.bala.bala 

171. [igi mu.un].≠il 2± aratta ki .aß ba.te He lifted his eyes as he approached Aratta. 

218

172. [kisal] ≠aratta± ki .ka œiri 3 hul 2.la As for the one who had been made to take 

mi.ni.in.gub the joyful path to the court of Aratta, 

173. nam.nir.œal 2 lugal.a.na mu.un.zu He made known the authority of his king, 

174. bur 2.ra.bi inim ßa 3.ga.na As for the solution (of the riddle), he recited 

bi 2.ib 2.be 2 

219 

the words he had memorized, lit. the 

words of his heart, 

As in the previous examples, there is a locative argument in the construction, namely 

kisal aratta ki .ka, “in the halls of Aratta,” but forms in which there is a locative or 

ablative argument such as this one seem to differ systematically from those in which a 

terminative postposition occurs in the same position such as in the following example 

from Lugalbanda. 

(23) Lugalbanda II [1.8.2.2], ll. 345-351 

345. œi 6.sa 9.a œiß banßur ku 3 d inanna.ke 4 At midnight, before they had brought the 

nu.um.ma.te.a.aß offering-table to pure Inanna, 

ki 

346. sig4 kul.aba4 .ße3 œiri3 ˙ul2.la As for the one who made him take the joyful 

mi.ni.in.gub path to brick-built Kulaba, 

347. nin.a.ni ku 3 d inanna.ke 4 

As for his lady, holy Inanna, 

348. ßa 3.da.ga.na mu.na.an.tuß She was seated 29 on her cushion, 

29 The translation of the line so as to render Inanna relatively non-agentive is due to the use of ku3 d inanna.ke 4, where 

the inclusion of the adjectival epithet seems to indicate that Inanna is in the locative-terminative rather than the ergative 

case.

349. i 3.gur 2 ki.a mu.na.ab.za He bowed and prostrated himself before her 

220 

on the ground, 

Here, in the famous presentation scene from the end of Lugalbanda II, the reinterpretation 

of mi.ni.in.gub as a topicalized indefinite relative construction in combination with the 

causative semantics of non-BNBV inal *bi-√ prefix verbs would, if it were equivalent to the 

four previous examples, result in the translation of line 346 as an epithet characterizing 

Lugalbanda as causee (“The one made to take the joyful path to brick-built Kulaba”). 

But such an interpretation (though in conformity with the general tenor of previous 

interpretations) makes little since in context since both the line preceding 346 and the 

two lines that follow line 346 are a description of Inanna in which the entrance of 

Lugalbanda is, at least from a literary point of view, intrusive. There is, however, an 

important formal difference between this example and the four previous examples: 

whereas the location in the previous examples was in either the locative or the ablative 

ki 

case, the location in line 346 is in the terminative case: sig4 kul.aba4 .ße3. Since there is 

no obvious semantic difference between the locative construction in, for example, line 

172 of (22) above and the terminative construction in line 346, I would guess that the 

opposition between locative and terminative is coding some other syntactic or semantic 

feature that has previously gone unnoticed. 

On the basis of contextual coherence and narrative sequence, I would hypothesize 

that, whereas *mini-√ constructions that occur with a locative are causative relative 

clauses that characterize the causee in some fashion, *mini-√ constructions that occur 

with a terminative postposition characterize the causer or agent. This would allow line

346 to function as a characterization of Inanna rather than Lugalbanda. At first such a 

reinterpretation may seem to disturb the narrative coherence of the passage, but in my 

view it improves it markedly: whereas previous interpretations place Lugalbanda’s 

entrance in the middle of a series of lines otherwise devoted to increasingly 

circumscribed contextual descriptions of Inanna, in the reinterpretation offered here, lines 

345 through 347 form a series of circumstantial clauses and epithets describing the 

goddess Inanna, culminating with the description of her sitting on her cushion in line 348. 

Line 349 then represents Lugalbanda’s formal entrance into her presence. 

There are several others examples in which a phrase in the terminative case occurs 

with a *mini-√ verb; the interpretation of these passages is quite difficult, but some of 

them would seem to allow for an analysis in which they can be characterized as relative 

clauses that refer to a causer rather than a causee as I suggest. 

(24) The Death of Gilgamesh, segment F [1.8.1.3], ll. 180-183 

(Kramer 1944; van Dijk 1967, 248-250; Wilcke 1970, 82-84; Klein 1990a, 64-65; 

Jacobson 1980, 19-20, 23; George 1999, 195-208; Cavigneaux and al-Rawi 2000a; 

Veldhuis 2001; George 2003, 14-17) 

180. ≠si±.si.ig dumu d utu.[ke 4] Sisig (a god of dreams), the son of Utu, 

181. ki.bi ku 3.ku 3.ga u 4.ße 3 

mi.ni.in.≠œar± darkness, 

221 

the one who enlightens a place of

182. niœ 2. ak.a “What my evil magic of humanity brings will 

nam.lu 2.ulu 3.œu 10 i.de 6 

ma.ra.an.≠tum 3± 

183. niœ 2 gi.dur ku 5.da.œu 10 

222 

now be brought to you, 

What my cutting of the umbilical cord brings 

i.de 6 ma.ra.an.≠tum 3± will now be brought to you,” 

My line numbering follows the reconstruction of Cavigneaux and Al-Rawi’s new edition, 

which includes the Meturan texts (Cavigneaux and Al-Rawi 2000a, 16, N 1 v 4-6; 31, 179- 

183); the passage corresponds to ETCSL, segment F, lines 141-144. Except for line 181 

(= ETCSL line 142), my translation conforms quite closely to that of Veldhuis (2001, 

144). Lines 182 and 183 are quite difficult to understand and very little has been written 

about these two lines (Veldhuis 2001, 137), even in Cavigneaux and Al-Rawi’s edition 

(2000a, 16, 20). But a comparison of the versions from Nippur and Meturan seems to 

shed some light on the matter. The most difficult part of line 182, for example, is the 

word i.de 6 (with the variant i.da), but it seems that the line has been misunderstood by the 

scribes of Meturan (the sigla follow Cavigneaux and Al-Rawi 2000a, N = Nippur, M = 

Meturan). 

N 1 v 17 niœ 2 ≠gig± ak nam.lu 2.ulu 3.ke 4 ne.en de 6.a ma.ra.du 11 

lu 

N2 v 17 niœ2 ≠gig± ak nam.lu2.ulu3 .ka ne.en de6.a ma.ra.du11 M 2 iv ≠niœ 2± ak.a nam.lu 2.ulu 3.œu 10 i.da ma.ra.an.[tum 3] 

? 

M6 iii niœ2 ≠ak±.[a] x x x x ≠de6 ± ma.ra […] 

M 1 iv niœ 2 ak.a nam.≠lu 2±.ulu 3.œu 10 i.de 6 ma.ra […]

In the two exemplars from Nippur, we find the phrase ne.en de 6.a, which consists of what 

Woods has described as a “demonstrative adverb of manner” followed by a nominalized 

form of the verb √de 6 (Woods 2001, 67). On the basis of Woods’ interpretation, the line 

should probably be translated as follows: “he (= Sisig) has told you that the bane of 

humanity is brought about in this way (i.e., by means of death).” The scribes at Meturan, 

however, arrived at the form i.da/de 6 instead of ne.en de 6.a, presumably because they 

were unfamiliar with the use of demonstrative manner adverbs in Sumerian. I suspect that 

Meturan scribes may have reinterpreted a later orthographic form of ne.en de 6.a as an 

Akkadian gloss of some kind, i-da and i-du respectively. That is to say i-du and i-da may 

have been interpreted as first person forms of the verb edû, “to know,” which may also 

explain the otherwise inexplicable presence of a first person possessive pronoun in the 

Sumerian of the witnesses from Meturan, when no such pronoun appears in the Nippur 

manuscripts. 

(25) Enlil and Sud [1.2.2], ll. 145-148 (Heimerdinger 1979, 1-2; Civil 1983) 

145. nam i.ri.tar.ra.œu 10 ˙e 2.em.ßi.œal 2 

223 

As for the fate I have determined for you, 

(var. niœ 2 nu.kur 2.ru) saœ il 2.la it should be fulfilled (not be altered)! 

e 2.ma˙.ße 3 œen.na Go to the Emah head held high, 

146. ßu.ni im.ma.an.dab 5 d a.ru.ru eß 3 

As for her hand, it was Aruru, the one who 

ma˙.ße 3 mi.ni.in.kar led her to the Eshmah, who grabbed it, 

147. e 2.kur e 2 d en.lil 2.ke 4 

mu.un ! .ßi.ib.ku 4 x [...] Enlil, and …, 

She brought her into the Ekur, the house of

The curious word-order in line 146 of (25) presumably results from a secondary 

topicalization of ßu.ni im.ma.an.dab 5 based on an underlying word-order such as 

* d a.ru.ru eß 3 ma˙.ße 3 mi.ni.in.kar [ßu.ni im.ma.an.dab 5] > [ßu.ni im.ma.an.dab 5] d a.ru.ru 

eß 3 ma˙.ße 3 mi.ni.in.kar. This change in word order probably reflects a contrast between 

two types of *mini-√ predicate: what I refer to below as topicalizing and identificational 

uses of the *mini-√ prefix. The topicalizing form corresponds to the underlying form 

* d a.ru.ru eß 3 ma˙.ße 3 mi.ni.in.kar [ßu.ni im.ma.an.dab 5], “As for Aruru, the one who led 

her into the Eshmah, . . .,” as opposed to the identificational use represented by line 146: 

“The one who grabbed her hand is Aruru, who led her to the Eshmah.” Other interpretive 

possibilities certainly exist, but these depend on the interpretation of im.ma.an.dab 5, 

which is far from secure. 

(26) Gilgamesh and the Netherworld [1.8.1.4], UET 6, 60, ll. 1-6 

(George 1999, 190-191; Cavigneaux and Al-Rawi 2000b, 1-19; George 2003, 743, 

fn. 4, 769-770, 777) 

1. [unug ki ] ≠im.mi.gi 4.gi 4±.[ne] They returned to Uruk, 

2. ≠iri±.bi im.mi.gi 4.gi 4.≠ne± As for their city, they returned to it, 

3. ≠œiß!± pa.a.ßu 

ßu.kar2 a2.kar2.ra.ke4 a 2.gid 2.[da] da.da.ra.ße 3 

224 

As for the one who had them girded with 

gear and equipment, 

mi.ni.in.≠ku 4± axe and spear (= Gilgamesh), 

4. e 2.gal.la.na ˙ul 2.≠˙ul 2±.la As for those made to celebrate in his palace 

mi.ni.in.œar (= his soldiers),

5. œuruß ki.sikil unug !ki .ga saœ.tuku The young men and the young women of 

≠bur.ßum 2±.ma ≠kul.[aba ki ] Uruk and the male and female elites 

225 

of Kulaba, 

6. alan.bi igi mu.un.bar.bar.re.ne As for their statue, as they were staring at it, 

im.ma.˙ul 2.˙ul 2.la they rejoiced, 

Likewise, an interpretation of this passage in which the terminative *mini-√ construction 

in line 3 referred to the causer, namely Gilgamesh, while the locative *mini-√ 

construction in line 4 referred to a causee such as Gilgamesh’s soldiers “made to rejoice” 

in Gilgamesh’s palace, would yield a nice sequence of the various constituencies who 

participate in the celebration of the new statue: Gilgamesh himself, his soldiers, the 

young men and women, and the male and female elites. 

(27) Winter and Summer [5.3.3], ll. 102-106 

(Gragg 1973a, 57, 72; Vanstiphout 1997, 586) 

102. u 2.kuß 8 en.[te.en.na x] ßa 3 sug 4 

ßeg 7 saœ.œa 2 x […] e 3.a 

… brought forth in the early rain, 

sar ? 

103. lu.ub2 gal.gal AGA gal.e Large turnips, large …, cut down with an 

30 sar 

ku5.da TUN3.LA2 ga.raß 

gid 2.da 

104. en.te.en.e gu 2.un œar.ra.bi.ße 3 

axe(?), long leeks, 

Since Winter was the one who brought the 

ni 2.ba mi.ni.in.œar.œar tribute he had collected himself, 

30 This line is relatively straightforward except for TUN3.LA 2, which could be interpreted as a cutting device or, 

alternatively as a depression in the ground (= Akk. mußpalu).

105. e 2 nam.ti.la e 2 d en.lil 2.la 2.ße 3 

106. e 2.me.eß en.te.en.bi.da maß 2 

226 

For the E’namtila, the house of Enlil, 

Summer and Winter set about organizing the 

kadra si ba.ni.in.sa 2.sa 2.eß animals and offerings, 

Here—I would argue—we see a distinct use of the *mini-√ construction that is probably 

not topicalizing, since there is no subsequent clause for which the *mini-√ clause can act 

as an argument. Lines 105 and 106 involve en.te.en, “Winter,” but only as accompanied 

by e 2.me.eß, “Summer,” so it is unlikely that the clause in line 104 acts as the subject of a 

plural verb in line 106, namely si ba.ni.in.sa 2.sa 2.eß. The reduplication of the verbal root, 

as suggested in the previous chapter, may serve as an indication that line 104 as a whole 

functions in an adjectival/adverbial capacity with respect to another clause, hence my use 

of the translation “since . . .” in line 104. Although the precise function of the *mini-√ 

prefix in line 104 and other cases like it remains unclear, in the following I classify such 

examples as “identificational” (like the example in [25] above) on the assumption that the 

primary function of such clauses is to introduce a property of topical noun such as 

en.te.en in line 104 without directly linking the descriptive clause itself to any nearby 

clause (as seems regularly the case with the topicalizing use of the *mini-√ prefix). In 

more formal or diagnostic terms, the characteristic feature of identificational *mini-√ 

predicates is that the line in which it occurs begins with a coreferential noun in the 

ergative case and that it cannot act as an argument of the following clause (in [25], the 

coreferential noun was implicit in the initial verb, im.ma.ab.dab 5).

(28) A tigi to Enlil for Ur-Namma (Ur-Namma B) [2.4.1.2], ll. 20-22 

(Castellino 1957; 1959, 106-118; Sjöberg 1961, ll. 51-70; Cavigneaux 1987, 49-51; 

Civil 1985, 27-37; Hurowitz 1992, 57-59; Klein 1989, 44-62) 

20. sipa d ur. d namma.ke 4 e 2.kur ma˙ The shepherd, Ur-Namma, is the one who 

dur.an.ki.a.ka an.ße 3 

mi.ni.in.mu 2 

227 

made the lofty E-kur grow high in 

Dur-an-ki, 

21. u 6 di.bi.ße 3 uœ 3 ßar 2.ra.ba He made it to be wondered at by the 

ßi.im.ma.gub multitude of people, 

22. ka 2.ma˙ ka 2.gal ka 2.silim.ma Of the Lofty Gate, the Great Gate, the Gate 

˙ur.saœ galam.ma ka 2 

of Peace—of the Hursan-galama and 

ße nu.ku 5.da the Grain Not Diverted Gate, 

23. su 3.ra 2.aœ 2 ku 3 me.a sig 7.igi.ba He is the one who decorated their eyebrow- 

ße.er.ka.an mu.ni.in.du 11 

shaped arches with Sudran-stones and 

refined silver, 

Here in (28), the use of the *mini-√ prefix is identificational as well. In terms of the 

semantics of causation, the example in (28) is among the clearest, since whatever the 

overall discursive structure of the passage, it is clear that Ur-Namma must be the causer 

rather than the causee of mi.ni.in.mu 2, hence “the one who made it grow” rather than 

“the one made to grow.”

(29) Lugalbanda I [1.8.2.1], l. 409-413 (Vanstiphout 2003, 126-127; Gadotti 2004) 

409. u 4 ≠ga± LAM niœ 2 giri 17.zal By day . . ., they promise, luxury, 

[im].ma.ab.ßum 2.mu.ne 

410. ≠œi 6±.u 3.na muß 3 za.gin 3 

im.≠ßi±.ib.ßir 3.re.de 3.ne 

228 

At night they chant over the gleaming plain, 

411. si.si.ig sim mußen .sim mußen d utu.ka As for those who are brought in on the tail 

ur 2.ba ßu mi.ni.in.te of the breezes, the swallows of Utu, 

412. e 2.e 2.a i.im.ku 4.ku 4.ne They enter house after house, 

413. e.sir 2.e.sir 2.ra Street after street they haunt, 

gu 2 mu.un.gid 2.gid 2.i.ne 

So, although it is not always possible to differentiate between them, there seem to be two 

distinct uses of the *mini-√ construction that have been mentioned up to this point, 

topicalizing and identificational. Although the topicalizing use is prototypical, the 

identificational use can be easily identified since it regularly occurs at the end of a textual 

passage. The following is a particularly good example of the phenomenon. 

(30) Lament over Sumer and Ur [2.2.3], ll. 81-82. 

81. u 4.ba an ba.dub 2 ki ba.sig 3 igi On that day, heaven rumbled, the earth 

u 4.da ba.lim shook, the storm did not sleep, 

82. an ba.su˙ 3.su˙ 3 gissu ba.an.la 2 

kur.ra ur 5 mi.ni.ib.ßa 4 

The heavens were confused, a shadow hung 

over them, and the mountain was made 

to roar,

In (30), the final verb in a series of verbs that describe the storm appears in the form 

*mini-√. The lines following line 82 treat other figures such as the sun-god Utu and the 

storm does not recur in these lines, so the final verb in line 82 cannot be acting as a 

topicalized phrase for a following predicate. In my view, the use of *mini-√ in line 82 

acts to demarcate a sequence of lines from an unrelated sequence to follow. 

(31) Nippur Lament 103-104 (Vanstiphout 1985, 3; Tinney 1996, 104-105, 150, 207) 

102. uru 2 œal 2.la.bi nu.œal 2.la As for their city, which once existed, it is 

mi.ni.in.ku 4.ra.am 3 

103. dim 2.ma.bi œiri 3 

ib 2.ta.an.kur 2.ra.am 3 

229 

the one that became, lit., was made, 

non-existent, 

As for their reason, it was what got lost, 

104. umuß.bi in.su˙ 3.am 3 lil 2.e As for their intelligence, he confused it and 

bi 2.in.sag 3.ga.am 3 

105. u 2.bi a.bi ib 2.ta.an.kar.ra.am 3 

made it haunted, 

As for their food, as for their water, it was 

what they were deprived of, 

106. ga zu kaß zu.a u 4.bi As for their time of familiarity with milk 

mi.ni.ib.til.la.am 3 

and beer, it is what was brought to an 

end, 

In (31), we see the combination of a topicalized phrase, uru 2 œal 2.la.bi, and a 

corresponding, presumably “identificational,” *mini-√ construction followed by the 

copula. The identificational character of the *mini-√ construction is fairly clear and

similar to the previous examples of identificational *mini-√ constructions, except for the 

fact that the topical noun phrase in this case is not in the ergative case, but rather a 

possessive pronoun topicalization construction. At the same time, since the lexical 

meaning of the presupposed predicate œal 2.la is precisely the opposite of the focused 

predicate in nu.œal 2.la, the role of the copula as a contrastive focus marker is particularly 

clear (see below, section 5.6, for an extended discussion of identificational focus). 

My suggestion that *mini-√ prefix verbs, and in particular verbs having the prefix 

*mini-b-√, act as textual demarcators along the lines of our own paragraph indentation is 

also supported, in part, by the overall distribution of *mini-b-√ prefix verbs in the Old 

Babylonian corpus. Unlike verbs of the form *mini-n-√, which occur in non-final 

positions relatively often, verbs of the form *mini-b-√ regularly occur at the end of their 

respective line with only a handful of exceptions, most of which occur in texts that I 

suspect were composed in the Isin-Larsa period (including, crucially, a large number of 

exceptions that occur in the laments). Although I hope to conduct an exhaustive 

comparison between verbs of the form *mini-n-√ as opposed to those of the form *mini- 

b-√ in future so as to further solidify the difference between topical and identificational 

uses of the *mini-√ prefix, 31 I have not attempted to do so for the time being. The most 

that can be said is that *mini-b-√ prefix verbs tend to occur in final position, whereas 

verbs of the form *mini-n-√ show no such restriction. Given such a distribution, one 

strong possibility that should be explored in future work is that the opposition between 

31 In morphological terms, the alternation is driven by difference of grammatical aspect, but the morphology itself does 

not explain the distributional differences between the two constructions. 

230

*-n- and *-b- in conjunction with the *bi-√ prefix is used to indicate “same” and “switch” 

reference respectively. The relatively few cases assembled here conform to such a model 

in that in the examples of topicalizing use with *mini-n-√, the “subject” of the relative 

will also typically be the subject of the main clause (hence “same”), whereas the 

examples with *-b- (not presented here in detail) tend to be identificational in use and to 

end a discursive unit, so that the subject of the following clause is different (hence 

“switch”). I do not attempt to investigate the matter further here. 

There are also a couple of examples of BNBV inal predicates that appear in *mini-√ 

constructions; both examples are of igi-√du 8 “to see” and occur in related texts. 

(32) Dumuzi’s Dream [1.4.3], ll. 138-142 

(Alster 1972, 68-69; forthcoming; Heimerdinger 1979, pl. 4, bottom right) 

138. nin 9 ki ßeß am 3.mu.un.pa 3.de 3.a The sister who reveals where her brother is, 

139. u 4 ul ? .la 2.ta a.ba.a Since the most ancient times who has seen 

igi mi.ni.in.du 8.a such a person, 

140. œa 2.nam.ma.an.ze 2.en ku.li.ni.ir Come! Let us go to his friend! 

ga.an.ßi.re 7.en.de 3.en 

141. u 4 ne ku.li.ni.ir On the day that, to his friend, 

142. i 7 a.ba mu.un.na.ba.e.ne they offer a river’s worth of water, he will 

ßu am 3.ma.gid 2.de 3.en accept it, 

The textual witnesses to line 139 show a great deal of variation in the form of the verb 

without even so much as an arbitrary statistical edge for mi.ni.in.du 8.a. 

231

(33) Textual witnesses to Dumuzi’s Dream, line 139 

Synthetic text u 4 ul ? .la 2.ta a.ba.a igi mi.ni.in.du 8.a 

a = UET 6, 18 (Ur) […] ≠igi± mu.ni.in.du 8 

e = UET 6, 187 (Ur) […] gim.mu.ni […] 

f = A 30256 (Nippur) u 4.≠ul.la 2± [...] 

i = UM 29-15-236+ (Nippur) [...] ŒA 2 ba.ta mu.ni.in [...] 

E' = 3N-T 908, 289 (Nippur) [...].a mi.ni.in.du 8.a 

In his original edition, Alster restores the beginning of the line on the basis of line 155 of 

Gilgamesh and Huwawa, version B [1.8.1.5.1] (corresponding to line 143 of Edzard’s 

edition [1993, 33]); lines 154 and 155 read as follows: gudu 4 dab 5.ba ˙i.li.ße 3 

gur.ra.am 3 / u 4 ul.li 2.a.ta a.ba.a igi im.mi.in.du 8, “A captured gudu-priest restored to his 

wig! Who has ever seen such a thing?” The verbal form igi im.mi.in.du 8 also appears in a 

passage from The Death of Gilgamesh in column 6, line 15a of Cavigneaux and Al- 

Rawi’s edition (2000a, 17): u 4.ul.li 2.a.ta [a].ba.a igi ≠im.mi±.in.du 8.a. Apparently the 

sequence igi im.mi was something of a tongue-twister: because of the prosody of the 

/iCi-iCi/ pattern as well as the alternation between /g/ and /b/ in Emesal and Emegir (in 

slightly different contexts), problems of dissimilation and resyllabification may have 

been in play. Note in particular the orthography of witness “e” of Dumuzi’s Dream, 

where igi has been, at least in phonological terms, incorporated into the verb, gim.mu.ni. 

At the same time, the subtle morphosyntactic distinction between a.ba.a igi im.mi.in.du 8 

and a.ba.a igi mi.ni.in.du 8 probably comes down to a difference between an indirect 

232

perception (“he saw that . . .”) and an indefinite relative that has, subsequently, been 

made definite (“he is the one who saw . . .”). In line 139, if we can base ourselves on the 

opposition between direct, bi 2.in.du 11, and indirect speech, im.mi.in.du 11, we might 

expect a form such as igi im.mi.in.du 8. But the fact that the statement is a rhetorical 

question that expects a negative answer, may have led the copyists to draw an analogy 

with a negated verb of perception such as igi nu mu.ni.du 8.am 3 (Gilgamesh and the 

Netherworld 302), which closely mirrors the form of the verb in several witnesses. All of 

these factors would seem to indicate that igi mi.ni.in.du 8.a is simply ungrammatical, the 

result of either phonological or grammatical misprision. 

(34) Dumuzi and Geshtinana [1.4.1.1], lines 52-55 (Sladek 1974, 229, 234-235; Wilcke 

1969b, 86) 

52. ßu.œar sa 6.ga nu.tuku.me.eß sa 6.ga Demons are never kind, they do not know 

˙ul nu.zu.me.eß good from evil, 

53. lu 2.u 3 e 2.nu ni 2.te.na zi.ni Who has ever seen a man, without a family, 

silim.ma 32 a.ba.a igi mi.ni.in.du 8 

233 

all alone, escape with his life? 

54. ki ku.li.bi nu.um.ßi.du ! .de 3.en ki We shall go neither to the dwelling of his 

mi 2.us 2.sa 2.bi nu.um.ßi.du.de 3.en friend nor his in-laws, 

32 The syntax of this line is quite interesting: the topic of the line, zi.ni, “his life,” forms the final component of the 

anticipatory genitive construction ni 2.te.na zi.ni. The reflexive pronoun /niten/ then acts as a pivot for and is 

coreferential with the clause initial lu 2.u 3, which acts as the subject of the initial phrase, lu 2.u 3 e 2 nu, “a man without a 

house, lit. the man, to/for whom there is no house.” The topical phrase, zi.ni, is the subject of silim.ma, “that it (= zi.ni) 

is well/whole,” and the adjectival predicate followed by the complementizer *-a, silim.ma, then serves as the 

“complement” of mi.ni.in.du 8. If my suggestion as to the derivation of *mini-√ from *bi-ni-√ is true, it may even be 

possible to interpret mi.ni.in.du8 as the head of an anticipatory genitive: “Of a man without …, who is his perceiver 

([*igi bi 2.in.du 8]-ni),” but the matter clearly requires further study.

55. su 8.ba.ra ki d œeßtin.an.na.ka.ße 3 

234 

Rather, for the shepherd, let us go to the 

ga.an.ßi.re 7.en.de 3.en dwelling of Geshtinana, 

The interesting thing about these two examples is that they conform to the BNBV inal 

construction in that the experiencer appears in the ergative case as an inalienable 

possessor of a body-part noun, a.ba.a < /aba-e/, and there is not even a hint of the 

causative semantics that we might expect with a non-BNBV inal predicate, but the subtle 

differences between bi 2.in.du 8, im.mi.in.du 8 and mi.ni.in.du 8, which all seem to come 

down to matters of quantification and definiteness are in need of further investigation 

(see the end of this chapter for one possible set of distinctions). 

The obvious question is why the alternation between a locative postposition and the 

terminative position should have anything to with whether the relative form in *mini-√ 

refers to a causer or a causee. But recall the discussion of progressive/imperfective 

achievements in chapter 2. There I argued that progressive/imperfective achievements are 

unusual in that, whereas the non-progressive/perfective form is regularly non-agentive, 

the progressive/imperfective form has the possibility of expressing agentivity due to the 

“activity” inherent in any progressive/imperfective verb. There is, in other words, an 

association between activity predicates (as well as the activity component of 

accomplishments) and agentivity. If, as I argued at the beginning of the section, 

*mini-n-√ predicates are derived from the *bi-√ prefix indefinite relative construction, 

which is in turn limited to non-progressive/perfective aspect by the indefinite head noun 

that it serves as the relativization of, it should come as no surprise that the topicalized

form of the indefinite relative in *mini-√, likewise, rarely if ever occurs in the 

progressive/imperfective aspect. But the opposition between the locative postposition, 

*-a, and the terminative postposition, *-ße 3, seems to mirror the opposition between non- 

progressive/perfective aspect and progressive/imperfective aspect: whereas the locative, 

*-a, indicates location in or at (and thereby implicitly the completion of the movement or 

event), the terminative, *-ße 3, indicates motion toward (and thereby the non-completion 

of the movement or event). So as to clarify the difference, the opposition between *-a and 

*-ße 3 is perhaps best understood as an opposition between a “virtual perfective aspect,” 

*-a, and a “virtual imperfective aspect,” *-ße 3: but only when all other avenues of 

marking the opposition between non-progressive/perfective and progressive/imperfective 

are neutralized as in the case of *mini -√ constructions does the aspectual opposition 

between virtual perfective and virtual imperfective aspect come into play. 

Linguists working with a number of different languages have repeatedly noted that 

so-called head-internal relative clauses regularly take an indefinite noun as the head of 

such a relative clause (Williamson 1987 on Lakhota; Hale 1976 on the languages of 

Australia; Watanabe 1992 [cf. Nishigauchi forthcoming]; cf. Aldridge 2004 for a 

discussion of internally-headed relatives from the point-of-view of antisymmetry). In 

other traditions of grammatical investigation such as Assyriology and Egyptology, these 

are often spoken of as indefinite or appositional relative constructions (for an overview of 

previous interpretations of relativization in Sumerian, see Alster 2002) and are built up on 

the model of the Semitic construct state, where the initial, head noun must necessarily be 

unmarked for definiteness (for recent discussions of the Semitic construct state, see 

235

Shlonsky 2004). If one were to translate one of these internally-headed relatives into 

English, the result is a kind of apposition as in “Susan, a present had been given to her by 

her father,” but such a sentence in English sounds odd, even if does highlight the fact that 

topicalization is often used to form such constructions: the referent of the relative 

construction being introduced as topical noun and the head-internal relative following as 

comment qualifying (in predicate-like fashion) the referent of the topicalized noun 

phrase. 

(35) [ Topic Susan] i [ Focus a present had been given to her i by her father] 

Constructions like this are well-known from head-marking languages (cf. Nichols 1986) 

and in topic-prominent languages such as Japanese and Korean, where a great deal of 

research into head-internal relatives has taken place. 

One of the most interesting and surprising features of head-internal relative 

constructions (hereafter HIRC) is that there is a subtle and not entirely well-understood 

interaction between the head-internal relative itself, which is necessarily indefinite, and 

various markers of definiteness that can be attached “outside” the HIRC to indicate the 

definiteness of the entire relative clause. On the basis of Williamson (1987), Lyons 

(1999, 251) provides a nice example of the subtle dance of (in)definiteness in head- 

internal relatives from Lakhota (note that I have neglected the retroflex diacritics in the 

following example, see the referenced sources for the original transcription). 

236

(36) [[Joe wowapi wa owa] ki] wac‘i 

Joe book a wrote the want-1Sg 

I want the book that Joe wrote 

In (36), the indefinite determiner “wa” occurs inside the relative clause and indicates 

which noun is being relativized. The definite determiner “ki” then modifies the entire 

relative clause (although the determiner that modifies the entire clause can be either 

definite or indefinite). Readers who have some familiarity with Semitic languages will 

immediately recognize this as the pattern of (in)definiteness that underlies the genitive 

construction known as the construct state, where the head (possessed) noun cannot bear a 

definite determiner (hence definiteness is neutralized) and the definiteness of the entire 

construction is determined by either the presence or the absence of the article on the 

possessor. This allows one to draw a parallel between the Semitic construct state and the 

HIRC constructions in Akkadian (the following examples, after Huehnergard 2000, 188). 

(37) a. [ DP [ Topic Ø] [ Focusb®t e¢puß-u]] imqut 

Ø house do.1Sg.Perf-Subord fall.3Sg.Perf 

A/the house I built collapsed 

b. [ DP [ Topic b®t-um] [ Focusßa e¢puß-u]] imqut 

house-Def Rel do.1Sg.Perf-Subord fall.3Sg.Perf 

As for the house, namely the one I built, it collapsed, 

237

Although the omission of the nominal suffix -um and the relative pronoun ßa is often 

taken as more-or-less arbitrary, the fact that ßa can be replaced by indefinite pronouns 

such as mamma(n) “someone” and mimma “something” (Huehnergard 2000, 188) 

indicates that the construction in (37a) is not so much indefinite as neutralized with 

respect to definiteness, expressing neither definiteness nor indefiniteness. In (37b), 

however, the pattern of the two constructions comes into view: the suffix -um, as 

elsewhere in Akkadian, indicates that the noun that it modifies is a topic (this topichood 

was subsequently assimilated to nominative case, but such assimilation is a secondary 

phenomenon [the association of -um with topic and -am with focus undoubtedly served as 

the basis for a later assimilation to a nominative/accusative rection along the same lines 

as, for example, -(n)un and -(l)ul in Korean, cf. the contrast between -e and -am 3 in 

Sumerian]). If b®tum is taken as a topic, the rest of the relative clause in (37b) can then be 

interpreted as a HIRC, where the frozen relative marker, ßa (at least in Old Babylonian), 

corresponds to a highly grammaticalized demonstrative indicating definiteness or 

specificity. The clause in (37a) can then be taken as a head-internal relative in which the 

noun that was topicalized in (37b) is incorporated into the relative and serves as the head 

of a relative that is unmarked for definiteness. The crucial difference between the two is 

topicalization: (37a) lacks a co-referent topic that the demonstrative ßa would have 

resumed, while (37b) has such a topic available and uses the demonstrative ßa as a 

resumptive pronominal. Although there are good reasons to recommend Deutscher’s 

description of the history of relativization in Akkadian (2002), he fails to mention “head- 

internal relatives” anywhere in the paper (even in the section on typology) which 

238

indicates that the concept was unfamiliar to him. This may explain Deutscher’s 

characterization of the Old Akkadian system of relativization as “dysfunctional” (often in 

conjunction with other pejoratives such as “maladaptive”) on four separate occasions in 

the space of twenty pages (Deutscher 2002, 86, 92, 95, 102). But the “dysfunctional” 

aspects of the Old Akkadian system—such as agreement between the external, 

topicalized “head” and the true, clause-internal head in terms of gender, number and 

case—follows quite directly from the fact that the external “head” is topicalized and 

must, therefore, be resumed by a head within the relative clause that carries all of the 

same inflectional information as the topicalized, external “head” of the construction. This 

“full” resumption in conjunction with the restriction on definiteness marking on the head 

of the Akkadian relative both point toward some type of HIRC as the starting point for 

not only the Akkadian relative construction, but also the early construct state as 

represented in early Semitic and Egyptian. In fact, I would argue that the appositional, 

head-internal relative in Old Akkadian conforms quite closely to the Afroasiatic avatar of 

head-internal relative clause formation (on the specificity-driven system in Middle 

Egyptian, see Loprieno 1995, 158). 

The same opposition between indefinite and definite HIRCs can, likewise, be seen in 

the opposition between *bi-√ prefix relatives and *mini-√ prefix relatives in Sumerian. 

239

(38) a. Enmerkar and the Lord of Aratta 554 (see above, chapter 2, section 2.1) 

en aratta ki .ke 4 gig.e igi bi 2.in.du 8 

The Lord of Aratta saw the wheat, lit., there was a perception of wheat by 

the Lord of Aratta, 

b. Dumuzi and Geshtinana 53 (example [26c] above) 

lu 2.u 3 e 2.nu ni 2.te.na zi.ni silim.ma a.ba.a igi mi.ni.in.du 8 

As for the life of a man who is without a home, that it (= his life) is safe, 

who is the one who has seen it? 

Thus, just as (37a) and (37b) differed only in terms of the presence or absence of the 

relative marker ßa and its consequences for topicalization of the head noun, the two 

examples in (38) are differentiated by the presence in (38b) of a third person animate 

possessive pronoun that was, presumably, affixed to the underlying infinitive root 

(*√-bi 2 + -ni [+ -n]), became part of the verbal prefix (*√-bi 2-ni [+ -n]), and moved along 

with it to preverbal position (*bi 2-ni-n-√ > *mini-n-√). The primary effect of the addition 

of the third person animate possessive pronoun is that the verb that it is affixed to 

becomes definite and can be topicalized. As such it parallels, in a number of respects, the 

Lakhota example noted below. 

240

(39) a. [[Joe wowapi wa owa] ki] wac‘i 

Joe book Indef wrote Def want-1Sg 

I want the book that Joe wrote 

b. [[… √ bi] ni] … 

Indef Def 

The difficulty, however, lies in discovering how definiteness plays out in the grammar of 

Sumerian, particularly in terms of topic and focus. 

The role of definiteness in the matter of *bi-√ and *mini-√ remains somewhat 

unclear, primarily because definiteness in Sumerian has not yet been differentiated from 

specificity and it seems to be very much tied up with the pragmatic opposition between 

topic and focus. 33 If the *mini-√ hypothesis proposed above is valid, then verbs of the 

form *mini-√ are formally identical to the possessive pronoun topicalization 

constructions mentioned earlier. 

(40) a. ama.ni < ama + ni (3Poss) 

As for his mother, she . . . 

b. igi mi.ni.in.du 8 < igi √du 8-bi 2 + ni (3Poss) + n (AnimAgr) 

As for the one who has seen . . ., he/she . . . 

33 The shades or ghosts mentioned in example (43) below, however, are specific indefinites, which suggests that 

definiteness rather than specificity is the criterion at work in Sumerian. 

241

If we are to pick apart the subtle differences between bi 2.in.du 8, im.mi.in.du 8 and 

mi.ni.in.du 8 (leaving aside the “negative” form in mu.ni.in.du 8), a more detailed 

understanding the definiteness effect and its interaction with focus constructions is 

clearly in order. 

242

3.4 The Definiteness Effect 

Since the only way of proving that a bare noun, particularly a bare noun that acts as the 

head of a relative clause, is indefinite is through the identification of a definiteness effect, 

this section describes the prototypical case of a definiteness effect in English (the 

existential “there” sentence) and highlights several other related definiteness effects 

which occur in both English and, as I argue here, Sumerian. 

The first extensive study of the phenomenon was by Milsark (1974), in which he 

begins with a series of sentences in English: 

(41) a. There is a Santa Claus. 

b. There might be a duck in the sink. 

c. There has been a man shot by a maniac. 

d. There were at least fifty people sick. 

Milsark (1974; 1977) sought to characterize the noun that immediately follows the copula 

in (41a) through (41d) and argued that a number of distinct morphosyntactic and 

semantic restrictions apply to the noun in question as well as any deverbal modifiers such 

as “shot” in (41c). Milsark noted that definite nouns do not, as a rule, occur in this 

position (hence the name of the constraint), but that if the post-copular noun is affected 

by a co-occurring focus construction, then a definite noun could occur, but only in a 

rather particular interpretation known as a “pair-list” reading (see below for 

exemplification). Milsark also noted that what are known as individual-level predicates, 

243

which refer to permanent properties of the referent, are excluded as modifiers of the noun 

under the definiteness effect (see below, section 3.5, for a discussion of stage- and 

individual-level predicates, topicality and the opposition between adjectival and verbal 

small clauses in recent work by Basilico [2003]). Later authors have also identified 

definiteness effects in relational “have” sentences and, as noted above, head-internal 

relative clauses (see, among numerous other available references, in particular, Davidson 

1967; Stowell 1981, 1983; Safir 1985; Reuland and Meulen 1987; Diesing 1992; 

Herburger 2000; Jäger 2001, passim non vidi; on the role of indefinites in certain kinds of 

complex verb formation, strongly resembling the Sumerian verbal complex, see Szabolcsi 

1986; Koopman and Szabolcsi 2000: 32, 222, fn. 2). 

Before turning to the exceptions to the definiteness effect, which are really the most 

interesting part about it, the precise definition of what an existential “there” sentence in 

English is, is in need of some clarification. The most important detail is that it must be 

kept separate from presentational “there” sentences. 

(42) There is a post office next to the library. 

In (42), we have a standard existential “there” sentence: note the indefinite article on the 

noun that immediately follows the copula. If the immediately post-copular noun is 

definite, “there” and the post-copular noun can alternate on either side of the “to be” verb 

without affecting the meaning much at all as in (43) below. 

244

(43) a. There is the post office next to the library. 

b. The post office is there next to the library. 

In both (43a) and (43b), the normal use for such sentences would be to ostend and refer to 

the post office with a pointing finger and emphatic stress on the word “there,” hence the 

characterization of such sentences as presentational. Now, if one inverts “there” and the 

post-copular noun in a sentence like (42), the result is something like (44). 

(44) A post office is there next to the library. 

Once again the presentational rather than the existential reading is the only one available, 

so there really is something special about existential “there” sentences. 

As noted above, the primary exception—an exception in certain respects at least—to 

the restriction on definite nouns in definiteness effect environments is that definite nouns 

can occur in definiteness effect restricted environments if they are affected by focus. 

When in focus, as for example in answer to a wh-question, the post-copular noun phrase 

in an existential sentence can be definite, but in such cases it assumes a rather special 

interpretation. 

(45) Joe: “Who is going to drive me back home?” 

Mary: “Well, there’s [ Focus my brother, or the guy standing on the corner].” 

245

Definite noun phrases such as “my brother” or “the guy standing on the corner” can only 

occur in definiteness effect environments as a list of alternatives, known as a pair-list 

reading. Even if only a single definite noun occurs, it is still interpreted as only one of 

several possible alternatives. The pair-list interpretation of definite nouns that occur in 

definiteness effect environments can be confirmed by looking at the behavior of what are 

known as strong quantifiers such as “each” and “every” in definiteness effect 

environments. McNally (1992 apud Chung and McCloskey 2002, 6) notes that strong 

quantifiers only appear in contexts affected by the definiteness effect when they quantify 

over semantic kinds as in the following example from Chung and McCloskey (2002, 6). 

(46) There was every kind of ice cream in the fridge. 

Here the quantifier “every” does not quantify over individuals, but rather over types as 

indicated by “every kind of . . . .” 

Definiteness effects such as the one identified in English existential “there” 

sentences are fairly common in other languages and have been studied in considerable 

detail for years. Of the three major analytical traditions in the investigation of 

definiteness effects, identified by Chung and McCloskey as the impersonal approach 

(Milsark 1974; Chomsky 1981; Safir 1985; Chomsky 1986; Reuland and Meulen 1987, 

passim non vidi), the locative approach (Lyons 1967; 1968; Clark 1978; Kuno 1971; 

Freeze 1992), and the pragmatics approach (Lumsden 1988, Prince 1992; Abbot 1992; 

1993; 1997; Ward and Birner 1995; 1997; Zucchi 1995, passim non vidi), I would like to 

246

turn to the locative approach as represented by Freeze (1992). Freeze begins by pointing 

out that the existential “there” sentence in English is somewhat exceptional; the most 

common pattern is one in which the basic structure consists of three parts: (i) a copula, 

which may or may not be of the be.at' type, (ii) a theme, typically in the unmarked or 

zero case, and (iii) a locative. The corresponding elements in Sumerian—at least with 

respect to BNBV inal predicates—are: (i) the *bi-√ prefix as existential predicate, (ii) the 

bare inalienable noun, and (iii) the locative-terminative case-marked noun. 

Freeze then notes that these three components are commonly found in what are 

generally thought of as three distinct constructions: the predicate locative construction 

(“The book is on the bench”), the existential construction (“There is a book on the 

bench”) and “have” constructions (“Lupe has a book”) (1992, 553). In general, the 

“have” constructions pattern with the existential constructions and both can be 

differentiated from the predicate locative on the basis of what is known as “locative 

inversion” (The following example is from Hindi after Freeze 1992, 555; note that nasal 

diacritics are omitted, see Freeze for the original form). 

(47) a. mai hindustaan-mee thaa 

I India-in Cop.Sg.Masc.Past 

THEME LOCATIVE VERB 

I was in India 

247

. kamree-mee aadmii hai 

room-in man Cop.3Sg.Masc.Pres 

LOCATIVE THEME VERB 

There is a man in the room 

If the order THEME, LOCATIVE, VERB in (47a) is taken as the standard order, then the 

order of the elements in (47b) can be spoken of as “locative inversion” since the locative 

expression precedes the theme rather than following it as it did in (47a). Freeze then 

posits a basic complementarity between predicate locatives, which regularly include a 

definite theme, and existential/have sentences, which typically include an indefinite 

theme. Although Freeze fails to take into consideration the focus-affected, pair-list 

readings of definite nouns in existential sentences, the overall scheme seems to be fairly 

sound in typological terms and it conforms quite precisely to the basic dichotomy to be 

found with the *bi-√ prefix verbs. As noted in chapter 1, throughout the history of the 

*bi-√ prefix, there was always a basic dichotomy between an ordinary *bi-√ prefix 

construction in which an absolutive theme was followed by a locative that usually 

immediately preceded the verb and the construction dealt with in this study where a noun 

in the locative-terminative case is followed by a bare absolutive noun that immediately 

precedes the verb. The two examples from Hindi in (47), therefore, correspond quite 

closely to this basic opposition in the *bi-√ prefix construction in Sumerian; the primary 

difference is that in Hindi, two distinct lexical items serve as copula in the two 

constructions, whereas Sumerian uses a single existential/“have” preverb, the *bi-√ 

248

prefix, and differentiates the various functions on the basis of definiteness, inalienability 

and animacy. Note as well that the opposition between (42) and (44) above, the 

existential and presentational “there” sentences in English, is also on the basis of locative 

inversion. 

The conflation of “existential” and “have” predicates in the *bi-√ prefix in BNBV inal 

constructions is interesting in that the transfer of possession model that underlies my 

earlier interpretation of the BNBV inal construction as a low source applicative seems to 

include both an existential “moment” and a possessive “moment.” At the beginning of the 

event, the bare inalienable noun is said to exist at a particular location identified by the 

locative-terminative noun, while, at the end of the event, it is inalienably possessed by the 

noun in the ergative case—this corresponds to Freeze’s “possessed theme” (Freeze 1992, 

589). 

In the Hindi examples provided by Freeze, three distinct forms stand out as typical 

exemplars: the predicate locative construction, the alienable “have” construction and the 

inalienable “have” construction, which correspond to the three following examples. 

(48) a. mai hindustaan-mee thaa 

I India-in Cop.Sg.Masc.Past 

I was in India 

249

. larkee-kee paas kuttaa hai 

boy.Obl-Gen proximity dog Cop.3Sg.Pres 

The boy has a dog, lit. by the boy is a dog 

c. baccee-kee daat safeed hai 

child.Obl-Gen teeth white Cop.3Pl 

The child has white teeth, lit. the child’s white teeth are 

As I have argued previously, the BNBV inal construction seems, therefore, to consist of 

two, overlapping possessive constructions: an inalienably possessed theme lugal.e igi 

“the king’s eye” and a theme gig.e igi “perception of wheat, lit. an eye to the wheat” with 

raised possessor, as if the two Hindi sentences in (48b) and (48c) could occur as a single 

clause. Freeze goes on to note (using the term “subject” in a rather special sense where it 

refers to either the possessor or the locative) that “[t]here is a tendency for [+ human] 

subjects to occur in expressions interpreted as ‘have’ constructions, while [– human] 

subjects are found in expressions interpreted as existentials” (Freeze 1992, 584). This is 

precisely the state of affairs in the BNBV inal construction. 

250

3.5 Head-internal relatives revisited 

Now that topic, focus, and the definiteness effect have been outlined in a preliminary 

way, I would like to return to my earlier discussion of head-internal relatives in Sumerian 

and Akkadian. When last we saw the Akkadian HIRC at the end of section 3.3, I had 

argued that b®tum in the following example was topicalized and that it was then resumed 

by the demonstrative relative marker ßa, which I take as the true head of the relative 

clause in this Akkadian example. 

(49) [ DP [ Topic b®t-um] [ Focusßa e¢puß-u]] imqut 

house-Def Rel do.1Sg.Perf-Subord fall.3Sg.Perf 

As for the house, namely the one that I built, it collapsed 

One of the most interesting features of HIRCs is the fact that they are definiteness effect 

environments. A skeptic might point out that the presence of the demonstrative/relative, 

ßa, as the head of the relative clause, ßa e¢pußu, would seem to militate against it as a 

definiteness effect environment since demonstratives are necessarily definite and definite 

noun phrases are regularly excluded from such environments, but we should remember 

that definite nouns can occur in such environments when they are focused. The “topic + 

focus” model of the relative clause that I have sketched in (49) would regularly provide 

the head noun with focus as long as it is preceded by a coreferential topic. Unlike the 

sentence in (49), where the topicalized noun phrase, b®tum, must be definite in order to be 

251

topicalized, the following example lacks both the topicalized, external “head” as well as 

the demonstrative/relative ßa: the bare noun b®t has become the head of the HIRC. 

(50) [ DP [ Topic Ø] [ Focus b®t e¢puß-u]] imqut 

Ø house do.1Sg.Perf-Subord fall.3Sg.Perf 

A/the house I built collapsed 

Since there is no topicalized noun within the determiner phrase (DP) in (50), there is 

likewise no reason for a resumptive pronoun like ßa; the internal head of the relative 

clause is b®t and in the absence of a topicalized noun, the expression need not be in focus. 

In semantic terms, the HIRC works in Akkadian by setting up an existentially 

quantified noun (“There is an x such that . . .”) and then identifying the existentially 

quantified noun (the internal head of the relative) with some other noun that is prominent 

in the discourse, typically a topic. Thus b®tum ßa e¢pußu might be translated into English 

so as to capture the quantification involved as follows: 

(51) As for the house (b®tum), it is something that I made (ßa e¢pußu). 

The reason for the use of the existential construction (and the consequent definiteness 

effect) is that in the absence of explicit quantization such as a cardinal number or plural 

morphology, an existentially quantified noun refers to a single entity. Such an entity can 

then be identified, or better equated, with another single entity. Since the head-internal 

252

elative may have to be equated with both definite and indefinite nouns, it needs to start 

out as the equivalent of a bare noun, unmarked for definiteness (or, for that matter, 

gender or number). In other words, the HIRC is a noun phrase in the narrow sense of the 

generative tradition rather than a determiner phrase. 

The process of one-to-one identification where one entity is the HIRC (ßa e¢pußu) and 

the other is the topicalized noun (b®tum) seems to be justified, at least to some degree, by 

restrictions on many-to-one possessive relations within the HIRC proper. Huehnergard 

notes three conditions (two that I deal with here) under which genitive constructions in ßa 

must be used rather than the “construct state” genitive, or, in terms of my description of 

the Akkadian HIRC up to this point: certain many-to-one possessive relations require that 

the head noun be topicalized and resumed by ßa inside the relative clause so that the 

HIRC can qualify each topical noun separately as in (52a) or that the possessors be 

included within the HIRC and coordinated, so that the topicalized noun can be equated 

with each coordinated HIRC independently (Huehnergard 2000, 363). 

(52) a. when there is more than one governing noun: 

ma¢rum u ma¢rtum [ HIRC ßa ßarrim] 

‘the son and the daughter of the king, [lit., as for the son and the daughter, 

the one of the king CJ]’ 

253

. when there is more than one governed noun: 

eqlum [ HIRC [ HIRC ßa dayya¢nim] u [ HIRC (ßa) a˙®ßu]] 

‘the field of the judge and his brother, [lit., as for the field, the one of the 

judge and (the one) of his brother CJ]’ 

(unless the governed nouns are seen as a unit by the writer: be¢l ßamê u 

er≈etim) 

Thus, constructions in which the possessed noun and the possessor cannot be treated as a 

single entity are blocked within the HIRC and a periphrastic construction involving 

topicalization is forced in which either the multiple head nouns are topicalized (ma¢rum u 

ma¢rtum) and resumed by a single head-internal relative, the head of which of unmarked 

for number (ßa ßarr®m), 34 or the single entity is topicalized (eqlum) and resumed by two 

distinct, coordinated HIRCs. In both cases each HIRC refers to a single entity that can 

then be equated with one or more topics. Huehnergard’s caveat to the rule in (52b) is the 

exception that proves the rule: where “the governed nouns are seen as a unit” 

(Huehnergard 2000, 363), the HIRC can be used since the cardinality and the definiteness 

of the possessed noun and the possessor is equivalent; the same goes for plurals that 

make use of the HIRC. The “topic-plus-focus” articulation of the HIRC that I describe 

here can, presumably, be assimilated to, for example, Basilico’s description of 

topicalization in small clauses (2003) on the assumption that the marker of subordination 

*-u originated as a resultative suffix that formed an adjectival small clause (see below). 

34 In Old Akkadian, however, we find an older form of the construction in which the head of the HIRC, an inflected 

form of ßa, agrees with the topical noun in terms of case, number and gender. 

254

Clearly, however, by the time Akkadian appears in the text-artifactual record, it has 

largely been assimilated to the familiar form of head-external relative clause that we find 

in English and other dependent-marking languages. 

With the foregoing discussion of the head-internal relative in Akkadian still in mind, 

I return to the *mini-√ construction. If the *bi-√ prefix in its existential (NOUN-e 

NOUN Bare bi-√) configuration indicates that the bare noun that immediately precedes the 

*bi-√ prefix is an indefinite noun in a definiteness effect environment, then verbs of the 

form *mini-√ can be interpreted as a head-internal relative (bi-√) that has been made 

definite through the suffixation of the third person animate possessive pronoun, namely 

*-ni, so as to form a topicalized phrase. Note, therefore, the several differences between 

the Akkadian HIRC and the Sumerian one: (a) the Akkadian HIRC can occur with a 

distinct, topicalized noun that is coreferential with the head of the HIRC, but the 

Sumerian HIRC cannot; (b) if the Sumerian HIRC needs to be topicalized or made 

definite for some other reason, the HIRC itself is made definite through the addition of 

the third person animate possessive pronoun, *-ni. At a fairly rough level of granularity, 

this allows us to draw a parallel between the HIRC in Akkadian and Sumerian due to the 

fact that, in both cases, topicalization plays a central role in the formation of HIRCs that 

are definite. 

(53) a. Ø b®t e¢pußu // igi bi.in.du 8 

b. (b®tum) ßa e¢pußu // igi mi.ni.in.du 8 

255

The contrast in (53), therefore, is between the indefinite HIRC in (53a) and the 

definite form of the same construction in (53b): note, in particular, the parallel between 

b®t in (53a), ßa in (53b) and igi in both examples in (53). While the similarities between 

the HIRC in Akkadian and Sumerian are intriguing, particularly in terms of the use of 

topicalization to code definiteness, it is the difference between the two constructions that 

I would like to turn to at this point. 

Basilico (2003) has recently reemphasized the contrast between two basic types of 

small clause (hereafter SC): the adjectival SC and the verbal SC—the classic examples of 

a small clause include, for example, the post-copular position in an existential sentence 

(“there is [ SC a dog in the street]”), the part of a direct perception verb that follows the 

verb of perception (“Mary saw [ SC John leave the party]”), and secondary predicates such 

as those described in chapter 2 (“Rob painted [ SC the house red]”). Basilico identifies a 

number of syntactic tests that differentiate the two types of SC, but I do not rehearse them 

here (the examples in [54] correspond to Basilico 2003, 4, ex. 8a and 7a respectively). 

(54) a. The guard considers [ Adjectival SC the prisoner intelligent] 

b. The burglar saw [ Verbal SC the prisoner escape] 

The crucial difference seems to be that verbal SCs are thetic, whereas adjectival SCs are 

categorical—recall my brief discussion of the opposition in section 2.7 above. To 

reiterate somewhat, thetic statements can be thought of as predications that adopt a 

holistic representation of some state of affairs without singling out any particular 

256

participant in the event in question. Categorical statements are quite the opposite: they 

identify an individual and then attribute a property to that individual—note that the 

semantic content of the state of affairs described by a categorical statement can be the 

same as that described by a thetic statement, but the two statements will usually differ 

syntactically. In syntactic terms, at least according to Basilico’s proposal, this amounts to 

the fact that in a categorical statement “the subject of an adjectival SC occupies a distinct 

topic position,” whereas, the subject of a verbal SC in a thetic statement “does not form a 

topic” (Basilico 2003, 8). Clearly the use of topicalized nouns in conjunction with the 

Akkadian HIRC can be equated with the adjectival SC in contrast to the absence of 

topicalized nouns in the verbal SC found in the Sumerian HIRC. 

Other factors such as the use of Sumerian HIRCs to form the complement of verbs of 

direct perception also favor interpreting it as a verbal SC. While topicality and/or 

definiteness can be difficult to evaluate in the text-artifactual record, the ability of a 

relative clause to act as the complement of a verb of direct perception should be readily 

identifiable. Min-Joo Kim, for example, has noted the nearly identical form of HIRCs and 

direct perception complements [DPCs] in Korean (M. Kim 2003, 2): 

257

(55) John-un [[totwuk-i tomangka]-nun kes]-ul cap-ess-ta 

PN-Top thief-Nom run.away-Rel.Imprf KES-Acc catch-Past-Decl 

John caught a/the thief running away 

(56) John-un [[totwuk-i tomangka]-nun kes]-ul po-ess-ta 

PN-Top thief-Nom run.away-Rel.Imprf KES-Acc see-Past-Decl 

John saw the event of a/the thief running away 

In (55), the HIRC consists of [[totwuk-i tomangka]-nun kes], “a/the thief running away,” 

and corresponds precisely with the direct perception complement in (56). Kim notes that 

these two constructions, the HIRC and the DPC, can be differentiated from other, similar 

constructions such as propositional attitude verbs like “know” or “believe” on the basis of 

syntactic tests in Korean. The contrast between HIRC + DPC and propositional attitude 

verbs seems to be represented in Sumerian by the opposition between BNBV inal 

predicates and non-BNBV inal predicates; if so, these oppositions can be tabulated as 

follows (neither of the two examples of bi 2.in.zu that I am aware of [Letter of Shulgi to 

Aradmu about Apillasha (3.1.02), line 30, and Proverbs Collection 7 [6.1.07], section C, 

line 21] is a BNBV inal predicate): 

258

Thetic Categorical 

Basilico Verbal SC Adjectival SC 

Kim HIRC + DPC Propositional attitude verbs 

BNBV inal 

259 

non-BNBV inal 

The Akkadian HIRC, in which a distinct, topical noun phrase is followed by a 

relative clause that bears a resultative suffix in *-u seems to correspond to a second type 

of relative clause in Sumerian, namely those relatives formed through the suffixation of 

the nominalizing/resultative suffix, *-a, to a verbal root. This type of Sumerian relative 

clause, which is generally thought of as a head-external relative in accord with the 

conventional interpretation of the Akkadian relative clause, also allows for a distinct, 

coreferential noun; it is, in other words, appositional like the Akkadian HIRC. In the case 

of Sumerian, however, the coreferential noun is not topicalized and occurs in the zero- 

marked absolutive/nominative case. This second type of HIRC in Sumerian (or head- 

external, if the Akkadian relative ultimately proves to be head-external instead of head- 

internal), need not concern us further here. Although the *-a suffix and the second type of 

HIRC in Sumerian in which it appears are beyond the scope of this dissertation, the 

association between bare complements and *bi-√ type HIRCs, on the one hand, and the 

relative formed through the suffixation of *-a, on the other, is certainly in need of further 

investigation, particularly in light of possible connections with the opposition between 

direct and indirect speech/perception, definiteness and different types of 

complementation in Sumerian.

These distinctions can also provide a model for a possible interpretation of closely 

related verbal forms such as igi bi 2.in.du 8, igi im.mi.in.du 8, and igi mi.ni.in.du 8: igi 

bi 2.in.du 8 would represent a basic, thetic statement, neutralized with respect to 

definiteness (“. . . is one who . . .”); igi mi.ni.in.du 8 would be a definite/topicalized 

version of the thetic statement represented by igi bi 2.in.du 8 (“the one who . . .”); while igi 

im.mi.in.du 8 would correspond to the kind of categorical statement typically associated 

with propositional attitude verbs and indirect speech. I hope to investigate these contrasts 

in greater detail in future work. 

260

4 

Focus-affected constituents in the *XP nam bi-√ construction 

At the end of the last chapter, I argued that one of the characteristic features of *mini-√ 

prefix verbs is that the introduction of the third person animate possessive pronoun makes 

the construction definite and the definiteness of the construction allows it to serve as a 

topic. Topics are regularly definite, since they are necessarily presuppositional. But in 

this chapter I attempt to isolate a rather different form of presupposition, the 

presupposition in a sentence in which a particular constituent is in focus. The focused 

constituent in such a construction constitutes the assertion within the clause, whereas the 

rest of the sentence is its presupposition (section 4.1 offers a brief review of focus and 

presupposition and isolates the basic oppositions within Sumerian through the use of 

robust diagnostic situations such as wh-questions, which regularly elicit responses 

structured on the basis of informational focus, and examples of explicit contrastive 

focus). 

This chapter looks at a particular construction of the form *XP nam bi-√ (where XP 

stands for a constituent of any phrasal type) and argues that it is to be interpreted as a 

negative contrastive focus construction (“it is not THE DOG that ate your cereal, but 

rather THE BIRD that did so”). The unusual thing about this construction is the “but 

261

ather THE BIRD that did so” counterpart to the negative contrastive focus is left 

unstated and only drawn as an implication from the context. Given the contextual nature 

of such an unstated contrast, obviously the first course of investigation is to look in detail 

at the contexts in which the construction occurs and confirm that this interpretation of the 

construction “fits”—this is the objective of sections 4.2, 4.3 and 4.4, the centerpiece of 

the chapter, which correspond to three subsets within the overall class of *XP nam bi-√ 

constructions: section 4.2 covers examples of the construction that do not involve 

compound verbs; section 4.3 deals with the compound verb subset, excluding BNBV inal 

predicates; section 4.4 looks at BNBV inal predicates in the *XP nam bi-√ construction. 

The use of the “It is not A . . ., but rather B” translational equivalency is, however, 

somewhat artificial in that the positive contrast never appears in the context, so, in section 

4.5, my hypothesis about of the *XP nam bi-√ construction is refined somewhat on the 

basis of Herburger’s work on double negation in Spanish (2000). In the final section of 

the chapter, section 4.6, I turn to the opposition between what Kiss (1998) has called 

informational and identificational focus and argue that the use of the copula to mark the 

focused constituent (whether as the ordinary copular construction *XP-am or the copula 

that forms a component of the *XP nam [< *XP nu/na-am] constituent) is limited to 

instances of contrastive/identificational focus. 

262

4.1 Focus in Sumerian 

Since focus constructions are important in that they can act as a heuristic tool in 

reconstructing the pragmatics (and the syntax for that matter) of archaeologically 

recovered languages, I would like to turn to some basic examples of focus in Sumerian. 

One might imagine that two contexts would be promising: (a) the answer given to a 

content- or wh-question (known as informational focus in some circles), and (b) contexts 

in which there is an explicit contrast between one referent and another (known as 

identificational or contrastive focus). 

(1) Gilgamesh and the Netherworld [1.8.1.4], ll. 174-179 (George 2003, 898-905) 

174. œiß ellag.œu 10 kur.ße 3 mu.da.ßub (Gilgamesh:) As for my ellag, it has fallen 

a.ba.a ma.ra.ab.e 11.de 3 

175. œiß e.ke 4.ma.œu 10 ganzer.ße 3 

mu.da.ßub a.ba.a ma.ra.ab.e 11.de 3 

177. lugal.œu 10 er 2 e.ne ba.ße 8.ße 8 

263 

into the netherworld, who will descend 

on my behalf? 

As for my ekema, it has fallen into Ganzer, 

who will descend on my behalf? 

(Enkidu:) My king, you are weeping, 

178. u 4.da œiß ellag.zu kur.ta œa 2.e At daybreak, as for your ellag, let me 

ga mu.ra.ab.e 11.de 3 

retrieve it from the netherworld for you, 

179. œiß e.ke 4.ma.zu ganzer.ta œa 2.e As for your ekema, let me retrieve it from 


Ganzer for you, 

In (1), in response to Gilgamesh’s question, Enkidu answers and the noun phrase that 

specifically answers that wh-question is œa 2.e “I” with the rest of the sentence merely

epeating the presupposition of the wh-question. The only evidence that œa 2.e is in focus 

is that it immediately precedes the verb rather than appearing in its usual position at the 

beginning of the clause. In (2), interestingly enough, a haruspex is having a conversation 

with a cow: 

(2) Enmerkar and Ensu˙girana [1.8.2.4], ll. 175-177 

(The haruspex asks:) 

175. ab 2 i 3.zu a.ba.a i 3.gu 7.e ga.zu Cow, as for your butter, who will eat it? 

a.ba.a i 3.na 8.na 8 

(The cow responds:) 

176. i 3.œu 10 i 3.gu 7.e d nisaba.ke 4 

177. [ga].œu 10 i 3.na 8.na 8 d nisaba.ke 4 

264 

As for your milk, who will drink it? 

As for my butter, it will be eaten by Nisaba, 

As for my milk, it will be drunk by Nisaba, 

The example in (2) is particularly interesting in that the response to the wh-question 

actually follows the verbal predicate, a nearly unique phenomenon in Sumerian grammar. 

But as in example (2), the phrase in focus occurs as close as possible to the end of the 

clause. In example (3), the contrastive use is particularly clear. 

(3) The Tale of Gudam, segment C [1.3.4], ll. 10-11 

10. i 3.gu 7.a.zu i 3.gu 7.a.zu As for what you have eaten, as for what you 

11. ninda nu.e.gu 7 uzu.zu.um i 3.gu 7 

have eaten, 

It was not bread that you have eaten, it was 

your flesh that you have eaten!

The copula -am 3 undergoes vowel harmony in (3); the underlying form is uzu.zu.am 3. 

Still, it occurs immediately before the verb as in the previous examples. Thus it is 

presumably reasonable, bearing in mind parallels from a number of verb-final languages, 

to postulate a focus position immediately before the verb. If this is the case, then we may 

want to interpret the i 3.na 8.na 8 in example (2) as a type of nomen agentis and argue that 

the syntactic form of (2) consists of an extended nominal as represented in (4). 

(4) [ Topic ga.œu 10 [ Focus i 3.na 8.na 8 d nisaba.ke 4 [ Presupposition am 3] 

Bearing these examples of focus constructions in Sumerian in mind, I would like, at this 

point, to turn to examples of focus-affected *bi-√ prefix verbs in the hope of clarifying 

the particular way in which the *bi-√ prefix forms a definiteness effect environment. But 

in order to resolve some of these issues, the focused and presupposed parts of any 

particular clause need to be clearly differentiated. 

Since the use of the copula, *-am, to mark contrastive focus is the clearest 

morphosyntactic indicator of a focused constituent in Sumerian, it must obviously serve 

as the primary diagnostic of any focus construction. But since focus can apply to 

constituents of any size, we are in need of some means of delimiting the parts of a clause 

that are not in focus as well. Take, for example, the following passage from Gilgamesh 

and the Netherworld. 

265

(5) Gilgamesh and the Netherworld, Version A [1.8.1.4], ll. 254-277 

254a. lu 2 dumu.ni 1.am 3 igi bi 2.du 8.am 3 (Gilgamesh:) Did you see a man who had 

254b. igi bi 2.du 8.am 3 

266 

one son? 

(Enkidu:) I saw such a man, 

254c. a.na.gin 7 an.ak (Gilgamesh:) How does he fare? 

255. œiß gag e 2.œar 8.a.na ab.du 3.a (Enkidu:) He weeps bitterly at the wooden 

gig.ga i.i peg which was driven into his wall, 

256a. lu 2 dumu.ni 2.am 3 igi bi 2.du 8.am 3 (Gilgamesh:) Did you see a man who had 

256b. igi bi 2.du 8.am 3 

two sons? 

(Enkidu:) I saw such a man. 

256c. a.na.gin 7 an.ak (Gilgamesh:) How does he fare? 

257. sig 4 2.a al.tuß ninda al.gu 7.e (Enkidu:) He sits on a couple of bricks, 

eating bread, 

This is certainly an interesting passage from a number of perspectives—note that the 

question and the answer in (254a) and (254b) as well as (256a) and (256b) are each, at 

least in orthographic terms, identical in form—but the frequent use of the copula in such 

a passage prevents any clear demarcation of focus and presupposition in the passage. In 

the following section, I turn to a particular morphosyntactic environment, namely the 

*XP nam bi-√ construction, so as to demarcate focus and presupposition with some 

precision. The relevance of this particular construction is that, in this environment, the 

focused constituent regularly occurs in the middle of two well-demarcated, 

presuppositional constituents: the possessive pronoun topic-marking construction to its

left and the *bi-√ prefix verb to its right. The fact that each of these components can be 

identified and distinguished makes this construction particularly useful in evaluating 

matters of focus and presupposition. 

267

4.2 Focus-affected *bi-√ prefix constructions 

In the Sumerian literary materials from the Old Babylonian schools, there are a dozen or 

so examples of a verb in the progressive/imperfective (marû) aspect bearing the verbal 

prefix *nam-bi 2-√. These verbal forms have conventionally been assimilated to a modal 

system in which the prefix *na- in conjunction with a perfective (˙amt¬u) verbal form 

indicates positive assertion, while co-occurrence with an progressive/imperfective (marû) 

verbal form indicates prohibition. I would like to propose a somewhat different 

interpretation of the forms in which *nam- immediately precedes the *bi-√ prefix: 

namely, that *nam- is a conventionalized orthography for a sequence made up of either 

the modal particle *na- or the simple negation *nu- followed by the copula as represented 

in (6). 

(6) nam-bi 2-√ nam bi 2-√ (< *na/nu-am 3 bi 2-√) 

That is, given conventional principles of scope under negation, that it represents a 

negation of the focused constituent in an otherwise positive assertion of the content of the 

presupposition, namely whatever topic is in play as well as the semantic content of the 

*bi-√ verbal form. The special feature of this construction as opposed to other 

constructions in which a focus constituent is differentiated from other presupposed 

constituents is that, in most of the following cases, explicit morphosyntactic evidence of 

one kind or another separates the focused material from the presupposed material: the 

constituents to the left of the focused phrase are demarcated by either the possessive 

268

pronoun topicalization system or the equative postposition, *-gin 7, while the presupposed 

material to the right of the focused constituent consists of a *bi-√ prefix verb. 

Before turning to the particular examples of this distributional class, it needs to be 

carefully differentiated from a distinct class that bears a superficial resemblance to it, 

namely predicates of the form *XP nam.mi-√. Whereas the *XP nam bi-√ construction is 

always written in the Old Babylonian period with the graphemes NAM and NE, yielding 

the sequence *nam bi 2-√, the members of the *XP nam.mi-√ class, in which there is no 

copula to the right of *na-/*nu-, are typically written with the graphemes NAM and MI, 

yielding *nam.mi-√. Of the 95 examples of *XP nam.mi-√ and the 25 examples of *XP 

nam bi-√ in the composite texts made available through ETCSL, only one simple verbal 

root or compound verb occurs in both constructions: √˙u.lu˙. The decisive piece of 

evidence in favor of differentiating the two constructions, however, is that fact that *XP 

nam bi-√ regularly occurs with a marû root, whereas *XP nam.mi-√ never occurs with a 

marû root. If some degree of variation were present, it might provide some evidence in 

favor of arguing that the two forms exist in a paradigmatic opposition, but strict 

differentiation of the two classes and the nature of the two orthographies would seem to 

suggest that they are fundamentally different. Given the usual orthographic conventions 

for cuneiform, I would suggest that *nam.mi-√ represents the morpheme *na- prefixed to 

a verbal prefix of the form *bi-√, whereas the difference between the final consonant of 

/nam/ and the initial consonant of /bi/ presumably indicates that no prefixation is present 

in the *XP nam bi-√ construction. 

269

In the first of my examples of focus-affected *bi-√ verbs, example (7), Gilgamesh 

has been put to sleep by Huwawa, a creature that he and Enkidu meet in the Cedar Forest 

and Enkidu is attempting to revive Gilgamesh. 

(7) Gilgamesh and Huwawa, version A [1.8.1.5], ll. 81-84 

[ Topic ama ugu.bi [ Focus tilla 2 iri ki .za.ka.ße 3 nam [ Presupposition bi 2.ib.sar.re]]] 

As for their mothers, it is not TO THE MARKETPLACE OF YOUR CITY 

that they should be running, (but rather TO THE SONS OF YOUR CITY) 

81. d bil 4.ga.mes en.ße 3 i 3.nu 2.de 3.en Gilgamesh, how long will you sleep for? 

82. dumu iri ki .za mu.un.de 3.re 7.eß.am 3 

270 

As for your city’s sons, they came with you, 

83. ur 2 ˙ur.saœ.œa 2.ka nam They should not be waiting at the foot of the 

ba.e.de 3.gub.bu.ne hills, 

84. ama ugu.bi tilla 2 iri ki .za.ka.ße 3 

As for their mothers, it is not to the 

nam bi 2.ib.sar.re marketplace of your city that they 

should be hurrying, 

(8) Enmerkar and the lord of Aratta [1.8.2.3], ll. 116 and 118 (cf. 188 and 190) 

[ Topic iri.bi] [ Presupposition mußen.gin 7 [ Focus gud 3 us 2.sa.bi.a nam] bi 2.ib.dal.en] 

As for their city (= the inhabitants), it is not FROM THE NEST ATTACHED 

TO IT (with Aratta as metaphorical tree) that I want them to fly like birds, (on the 

contrary, the inhabitants can stay)

[ Topic ganba œal 2.la] [ Presupposition iri gul.gul.lu.gin 7 [ Focus sa˙ar nam] bi 2.ib.˙a.za.en]]] 

As for what is in the marketplace, it is not THE DUST that I want to take 

possession of, as if (I were) someone who destroys cities, (but rather the precious 

objects) 

115. iri.bi ir 7.saœ mußen .gin 7 œiß.bi.ta As for their city (= the inhabitants), it is not 

na.an.na.ra.ab.dal.en from their tree that I want them fly like 

271 

a wild dove, 

116. mußen.gin 7 gud 3 us 2.sa.bi.a It is not from the nest attached to it that I 

nam bi 2.ib.dal.en want them to fly like a bird, 

117. ganba œal 2.la.gin 7 

It is rather (-gin 7) what is in the market that 

na.an.si.ig.en I want to load up, 

118. iri gul.gul.lu.gin 7 sa˙ar nam It is not the dust that I want to take 

bi 2.ib.˙a.za.en possession of, as if (I were) someone 

who destroys cities, 

119. aratta ki a 2.dam d en.ki.ke 4 nam Aratta, a settlement cursed by Enki, 

ba.an.ku 5 

120. ≠ki bi 2±.in.gul.la.gin 7 ki nam Is it not such as place, a place in ruins, that 

ga bi 2.ib.gul.en I should want to undermine, 

In (8), we find two examples of the construction in lines 116 and 118, separated by the 

*na- modal prefix plus perfective verb of assertion in line 117. The alternation between 

negative contrastive focus constructions and positive assertion in line 117 serves to

indicate that Enmerkar seeks neither the inhabitants of the city themselves, nor the 

destruction of the city, but rather simply the material wealth accumulated in Aratta. The 

entire passage is built up around the grapheme NAM which occurs in the *XP nam bi-√ 

construction, the compound verb nam-√ku 5 “to curse,” and an *XP nam bi-√ 

construction in which the cohortative prefix, *ga-, also occurs. 35 Note as well that, 

although not a single progressive/imperfective (marû) verbal root occurs in the entire 

passage, the temporal reference is to the future, so I have translated the modal forms as if 

they were imperfective so as to capture the temporal reference as well as its presumably 

modal associations. The only exception is the verb in line 117, which I interpret as 

equivalent to *na- plus perfective, yielding the assertive rather than prohibitive meaning 

in that line. Although one might be tempted to justify the differential treatment on the 

basis of the syllable closing grapheme in line 117, since one would have expected the 

orthographic sequence si.ge.en rather than si.ig.en had the root been in the 

progressive/imperfective (marû) aspect, the best evidence for the differential 

interpretation of line 117 as opposed to the other *bi-√ prefix lines is the variation in the 

pronominal agreement in the *bi-√ prefix verbs. Whereas the verb in line 117 shows *-n-, 

the *bi-√ prefix verbs elsewhere in the passage—lines 116, 118 and 120 as well as 115, 

which should probably be assimilated to the *bi-√ prefix verbs in this passage—have *-b- 

as the preverbal pronominal agreement marker. 

35 There is a great deal of confusion of ga.nam and nam.ga evident in the text artifactual record, so the proper 

interpretation of line 120 remains uncertain. 

272

(9) A praise poem of Shulgi (Shulgi E) [2.4.2.05], l. 58 (Ludwig 1990, 44-45) 

[ Topic en 3.du.œu 10 …] [ Focus ki.ßu.ke 4 lu 2 nam [ Presupposition bi 2.ib 2.da 13.da 13.a]] 

As for the songs about me …, it is not (only) THE PEOPLE ASSIGNED TO DO 

SO who should sing them, (but rather they should be sung on every important 

occasion) 

53. en 3.du.œu 10 a.da.ab ˙e 2.em tigi As for the songs about me, whether adab, 

ma.al.ga.tum ˙e 2.em whether they are tigi or malgatum, 

54. ßir 3.gid 2.da ar 2 nam.lugal.la Whether shirgida or praises of kingship, 

55. ßumun.ßa 4 kun.œar bal.bal.e Whether shumunsha, kunnar or balbale, 

˙e 2.em 

56. gi.gid 2 za.am.za.am ˙e 2.em Whether gigid or zamzam, 

57. œeßtu 2.ge nu.dib.be 2 ka.ta That they never pass out of memory and 

nu.ßub.bu.de 3 

273 

never lapse from people’s mouths, 

58. ki ßu.ke 4 lu 2 nam It is not (only) the people in the cultic 

bi 2.ib 2.da 13.da 13.a centers who should perform them, 

59. e 2.kur za.gin 3.na muß nam They should sung without end in the shining 

ba.an.tum 2.mu E-kur! 

60. d en.lil 2.ra eß 3 u 4 sakar.ra.ka.na so that they are played for Enlil in his 

˙e 2.na.du 12 

61. eß 3.eß 3 kaß gi.rin a.gin 7 

Shrine of the New Moon! 

(and) at the Esh’esh festival, where the 

su 3.su 3.u 3.da.bi girin-beer is endless like water,

62. d en.lil 2 d nin.lil 2.da tuß.a.ra May they be performed repeatedly before 

˙e 2.en.ne.œa 2.œa 2.œa 2 

274 

Enlil as he sits with Ninlil, 

Example (9) and example (10) below are particularly interesting in that the case of the 

focal constituent in each example is quite different: the focal constituent in (9) is in the 

absolutive case with a raised inanimate possessor, ki.ßu.ke 4 lu 2, whereas the focal 

constituent in (10), namely su iri.na.ke 4, is in the locative-terminative case coding the 

experiencer of the predicate bi 2.ib.du 10.ge. Elsewhere, in other examples of this 

construction, the focal constituent occurs in the terminative case as in (7), tilla 2 

iri ki .za.ka.ße 3, and the locative case as in (8), gud 3 us 2.sa.bi.a. The variety of 

postpositional phrases seen in the focal position in the *XP nam bi-√ construction is 

indicative of the fact that it is a pragmatic phenomenon that we are dealing with here. 

Note, however, that there are no cases in which an ergative noun occurs in focus; this is 

probably due to the fact that ergative nouns are usually topical and their presence in a 

focus position would be somewhat surprising. 

(10) A praise poem of Shulgi (Shulgi E) [2.4.2.05], l. 83 

[ Topic en 3.du.œu 10 [ Focus su iri.na.ke 4 nam [ Presupposition bi 2.ib.du 10.ge]]] 

As for my hymns, it will not be THE PEOPLE OF HIS OWN CITY who will 

enjoy them, (but rather THE PEOPLE OF MY CITY)

74. tukum.bi en 3.du.œu 10 

275 

But if, as for my hymns, 

75. mu.≠œu 10± [x] ≠ba±.ta.œa 2.ar mu.ni ... removes my name and fails to ... his name, 

. . . . 

la ba.ru.gu? 

81. niœ 2.erim 2 maßkim.ße 3 ˙e.en 6.ßum 2 Let him be given enmity as his companion, 

82. a 2.sag 3 nam.uß 2 dugud œal 2.la.gin 7 

May an asag demon, as causer of plague, 

83. su iri.na.ke 4 nam bi 2.ib.du 10.ge It is not the people of his own city who 

will enjoy them, 

84. gug.kal bar.ße 3 mu ße.œar.ra.ka igi Because of famine in the midst of hungry 

kalam.ma.ka na.ni.sa 6 

(11) A praise poem of Shulgi (Shulgi E) [2.4.2.05], l. 117 

years, he will not find favor in the 

countenance of the land, 

[ Topic lu 2 en 3.du.œu 10 …] [ Focus ki.lugal.gub ki dadag.ga nam [ Presupposition bi 2.gub.be 2]] 

As for my hymn-singer, it is not IN THE PLACE OF THE KING, IN THE 

SPOTLIGHT that he should stand, (but rather at the king’s side) 

111. lu 2 en 3.du.œu 10.a ßu i 3.[...] The man who ...... my hymns, 

. . . . 

115. bad 3 iri.na.ka na.am 3.≠bi 2?±.[x x] May he no longer ... within the walls of the 

city. 

116. a sa.ga nam.lugal.la.na May the criminal offspring of his kingship 

na.≠an ? ±.[x x] not...

117. ki lugal gub ki dadag.ga nam It is not in the place of the king, the shining 

bi 2.gub.[be 2] place that he should stand, 

118. ub.ßu unken.na.ta bar.ße 3 ˙e 2.[...] May he ... from the Ubshu’unkena, 

(12) A hymn to Inanna for Ur-Ninurta (Ur-Ninurta A) [2.5.6.1], l. 85 (Sjöberg 1977, 192, 

194) 

[ Topic x x ≠bara 2± mi 2 zi.de 3.eß du 11.ga.œu 10] [ Focus muß 3 nam [ Presuppposition bi 2.ga.ga.an]] 

As for the dais which I care for, it is WITHOUT END that you shall sit on it, (rather 

than a normal reign) 

83. ur 2 ku 3 nam.ti.la SI.A.œu 10 u 4.zu As for my holy, life-giving embrace, to 

su 3.su 3.de 3 

276 

prolong your days (by means of it), 

84. ßul d en.lil 2.le e 2.kur.ta a 2.bi Young Man Enlil has commanded me from 

mu.un.da.an.aœ 2 

the Ekur, 

85. x x ≠bara 2± mi 2 zi.de 3.eß As for the dais which I care for, it is without 

du 11.ga.œu 10 muß 3 nam end that you shall sit on it, 

bi 2.ga.ga.an 

86. d nanna an ku 3.ga pa e 3.a Like when Nanna, admired by the land (= its 

ka.na.aœ 2.e u 6 di.gin 7 

inhabitants), appears in clear sky, 

87. [d] ≠ur±. d nin.urta.œu 10 kur.kur.re ˙i.li As for my Ur-Ninurta, may every land 

˙u.mu.u 8.ßi.ak.e rejoice over you,

(13) The Instructions of Shuruppak [5.6.1], ll. 15 and 18 

(Alster 1974, 34-35, 79-80; Alster 1990; Alster forthcoming) 

[ Topic gana 2 [ Focus kaskal.[la] nam [ Presupposition bi 2.ib 2.œa 2.œa 2 ]] nam.silig gu 2 œa 2.am 3 

As for (your) fields, it is not BY THE ROAD that you should put them, it is only 

attention they will yield, 

[ Topic e 2 [ Focus sila daœal nam [ Presupposition bi 2.ib.la 2.e]]] KEÍ 2.da œal 2.la.am 3 

As for (your) house, it is not ONE ON A WIDE STREET that you should buy, it 

will always have someone singing (as the walk), 

15. gana 2 kaskal.[la] nam As for fields, near the road is not where 

bi 2.ib 2.œa 2.œa 2nam.silig you should put them, (you will) gather 

gu 2 œa 2.am 3 

277 

only attention, (not barley), 

16. a.ßa 3 ka.œiri 3.ka nam ba.e.ur 11.ru You should not plough a field at the 

zi bulug.ga.am 3 

entrance of a pathway, it is the 

uprooting of boundary stakes, 

17. gana 2.zu.am 3 pu 2 na.an.ni.du 3.e.en You should not build a well in your field, 

uœ 3.e ßa.ri.ib.˙ul.˙ul people will damage it, 

18. e 2 sila daœal nam bi 2.ib.la 2.e As for a house, next to a public square is not 

keß 2.da œal 2.la.am 3 

where to buy one, there are always 

groups (of people) there,

Alster (1974, 34-35) reconstructs line 17 as gan 2 ≠ßa 3± [...] pu 2 na.an.du 3.un a.e 

ßa.re.eb.˙ul.˙ul, which treats the water, a, from the well as deleterious agent rather than 

people (UN > un, uœ 3) who might assemble there. Alster’s suggestion that keß 2.da is to be 

translated as “taboo,” while valid in other contexts (Alster 1974, 79-80), does not seem to 

make much sense here, so I have taken it as a headless relative (people) ‘tied up/together’ 

meaning small groups of people. Such an interpretation also fits the overall theme of the 

passage that human traffic in various forms is harmful to one’s property. 

(14) The Instructions of Shuruppak [5.6.1], l. 22 (Alster 1974, 34-35, 81) 

[ Topic (za.e) [ Focus ki du 14.da.ka nam [ Presupposition bi 2.du.x.de 3]]] 

As for you, it is not NEAR A QUARREL that you should walk, (but rather AWAY 

FROM ONE) 

22. ki du 14.da.ka nam bi 2.du.x.de 3 

278 

It is not near a quarrel that you should walk, 

23. du 14.de 3 lu 2.ki.inim.ma nam Don’t let the quarrel make you a witness, 

ba.e.ku 4.ku 4 

24. du 14.de 3 ni 2.zu nam ≠ba±.[...] Don’t let yourself be ... in a quarrel, 

25. du 14 ≠nam.ak ± .de 3.en x [...] You should not cause a quarrel, ... 

(15) The Instructions of Shuruppak [5.6.1], l. 54-59 (Alster 1974, 36-38, 88) 

[ Topic gi.sig.ga œiß kiri 6.ka [ Focus da.ga nam [ Presupposition bi 2.du 8.e.en]]]

54. sun 7.na.da e 2 na.an.da.œa 2.œa 2.≠an± You should not establish a home with an 

279 

arrogant man, 

55. gi 4.in.ße 3 du.de 3 ßi.me.ßi.ib.ßub.ßub He will make your life that of a slave girl, 

56. ki.tuß lu 2.ka na.ab.ta.bala.e.de 3 

You’ll not be able to pass by anyone’s place, 

57. ßi.du.un ßi.du.un “Restore it! Restore it!” they will tell you, 

ßi.me.ßi.ib 2.be 2.e.ne 

58. gi.sig.ga œiß kiri 6.ka da.ga nam Inside the garden’s reed fence, you should 

bi 2.du 8.e.en not hold a social gathering, 

59. su.ga.ab su.ga.ab “Restore it! Restore it!” they will tell you, 

ßi.me.ßi.ib 2.be 2.ne 

Line 58 corresponds to line 63 in Alster’s edition (1974, 38). The context of line 58 is 

particularly difficult: the meaning of the preceding lines, for example, is entirely up for 

grabs. The translation above is close to that of ETCSL, but Alster translates it quite 

differently: “(55) When you go to the slave girl, she will throw it back to you, (56) “Do 

not transgress people’s dwelling places (Alster 1974, 37). The proper interpretation of 

line 58 is equally unclear. 

(16) The Instructions of Shuruppak [5.6.1], ll. 245-249 (Alster 1974, 82-83; Cooper 1983, 

248) 

œiß 

[ Topic (dumu.œu10)] [ Presupposition mi.si.IÍ.a [Focus lag nam] bi2.ßub.be2.en] My son, it is not A CLOD that you should throw into the strongbox, (but rather 

GOLD that you should throw in it),

245. d ezina 2.ra You should not ... grain, its ...... are 

na.an.keß 2.keß 2.re.de 3.en numerous, 

œiß a2.bi i 3.ßar 2 

246. kir 11.e aß 2 nam.en dumu.munus You should not abuse a female lamb, 

in.u 3.tu.un lest you give birth to a daughter, 

247. œiß mi.si.IÍ.a lag nam It is not a clod that you should throw into 

bi 2.ßub.be 2.en the strongbox, if you want to beget a 

dumu.nita˙ in.u 3.tu.un son, 

248. dam nam.mu.un.kar.re.en You should not abduct a wife, you should 

gu 3 KA na.an.œa 2.œa 2 

280 

not make her cry (?), 

249. ki dam kar.re nam.silig The place where the wife was abducted … 

gum.œa 2.am 3 […] 

(17) The Instructions of Shuruppak [5.6.1], l. 256 

[ Topic dumu.œu 10] 

[ Focus inim ama.za inim diœir.za.ka.ße 3 nam [ Presupposition bi 2.ib.dib 2.be 2.en ]] 

My son, it is not THE WORDS OF YOUR MOTHER AND YOUR GOD that 

you should transgress, (but rather THE WORDS OF STRANGERS that you should 

pass by [and ignore]), 

254. emeda.ga.la 2 ama 5.a.ke 4 

The wet-nurses in the women’s quarters 

lugal.bi.ir nam determine the fate of their lord, 

ßi.im.mi.ib.tar.re

255. ama.zu.ur 2 inim diri nam You should not tell your mother 

ba.na.ab.be 2.en ˙ul exaggerations, that causes hatred 

ßa.ba.ra.gig.ga.am 3 

281 

against you, 

256. inim ama.za inim diœir.za.ka.ße 3 It is not the words of your mother and your 

nam bi 2.ib.dib 2.be 2.en god that you should transgress, 

257. ama d utu.≠am 3± lu 2 mu.un.u 3.tu The mother, like Utu, gives birth to the man, 

258. ab.ba diœir.ra.≠am 3± [x] The father, like a god, makes him bright, 

mu.un.dadag.ge 

(18) Letter from Ibbi-Suen to Puzur-Shulgi about Ishbi-Erra [3.1.20], Version A 33 

[ Topic ßu.ni] [ Focus iri ki .a nam [ Presupposition bi 2.ib.sa 2.sa 2]] 

As for his activities, it is not IN THE CITY that you should oppose them, (but 

rather IN THE COUNTRYSIDE), 

29. i 3.ne.eß 2 inim du 10.ga gi 4.gi 4.de 3 

Now, that good words are restored, 

30. lul du˙.du˙.u 3.da and falsehoods are cleared up, lit., loosened, 

˙e 2.ni.ib.tum 2.tum 2.mu you are to bring it about, 

31. uœ 3.ba buru 14.ba ˙e 2.ak.e.de 3 

32. za.e na.an.gur.ru ugu.œu 10.ße 3 

You are to let him (= Ishbi-Erra) participate 

in the harvest among the people, 

You yourself should not contest this, and do 

nam.ma.ßi.du.un not come back to me! 

33. ßu.ni iri ki .a nam bi 2.ib.sa 2.sa 2 

As for his hand, it is not (your) city that 

it will capture,

ki 

34. lu2 ma2.ri2 .ke4 œalga ur.re nam.en This man from Mari, with dog-like 

na.an.na.ak.e understanding, should not rule! 

(19) Proverb Collection 1.40, line 53 

[ Topic ninda.ni] [ Focus lu 2 nam [ Presupposition bi 2.in.ßu 2.ßu 2]] 

As for his bread, no one should COVER IT UP, (but rather EXPOSE IT TO 

VIEW) 

52. ninda.ni niœ 2.u 2.zug 4 ˙e 2.a As for his bread, even if it is rotten, 

53. lu 2 nam bi 2.in.ßu 2.ßu 2 

282 

No one should cover it up, 

(20) Lipit-Ishtar A (Karahashi forthcoming, 8; Römer 1965, 29-38) 

[ Topic kalag.ga me.en [ Focus niœ 2 nam [ Presupposition bi 2.i.e 3]] 

As for me, the one of strength, it is not ONE THING that I made come forth, (but 

rather EVERYTHING), 

93. niœ 2.gi.na.œu 10.uß a.na mu.da.ßub What of my truthful things can be thrown 

away? 

94. nun d li.pi 2.it.eß 4.tar 2 me.en uœ 3.e I, prince Lipit-Eshtar, keep the people on a 

si bi 2.sa 2 

straight path. 

95. si sa 2.œu 10.uß a.na œa 2.la As regards my integrity: in what respect 

im.mi.dag have I ever been idle?

96. kalag.ga me.en niœ 2 nam I am a strong one, it is not just one thing I 

pa bi 2.i.e 3 

(21) Proverb Collection 17.b2, ll. 8-10 

[ Topic (za.e) [ Focus lu 2 nam [ Presupposition bi 2.ib.til.e]] 

283 

have made appear, 

As for you, It is not A MAN who brings about your end, (but rather THE GODS), 

8. ≠piriœ±.gin 7 ne he 2.gub May you establish yourself like a lion, 

9. x.zu œiß kiri 6 a ˙e 2.eb ? .gub . . . 

10. lu 2 nam bi 2.ib.til.e It is not a man who brings about your end, 

Overall, the foregoing examples demonstrate the relatively wide applicability of the 

construction in different syntactic conditions, but a concentration of the exemplars in the 

epic and wisdom literatures, both of which derive from the Ur III period at the latest—an 

early version of Shuruppak’s Instructions, for example, dates to the ED IIIa period—may 

suggest that the use of the construction was already archaic and poorly understood prior 

to the Ur III period. The absence of exemplars from texts composed in the Old 

Babylonian period would suggest that the Old Babylonian scholars no longer had 

productive use of the construction.

(22) The Young Scribe and his Father, ll. 114-117, manuscript Z 

(Sjöberg 1973, 112, 117-118, 128; Karahashi 2000, 87) 

[ Topic nam d en.lil 2.la 2] [ Focus lu 2.u 18.lu.a nam [ Presupposition bi 2.ib.tar.re]] 

As for the fate (assigned) by Enlil, it is not AMONG THE PEOPLE that it is 

decided, (but rather AMONG THE GODS), 

114. œiß gid 2 ad.da.œu 10.ße 3 

geßtu 2 ga.œar nu.ub.be 2 

284 

You didn’t say, “I want to listen to the ... for 

my father,” 

115. nam d en.lil 2.la 2 lu 2.u 18.lu.a It is not among humanity that the fate of 

nam bi 2.ib.tar.re Enlil is determined, 

116. dumu kiœ 2.gi 4.a ad.da.na.ka That a son should follow in the occupation 

ib 2.us 2.e of his father, 

117. a.ra 2 na.me.ka ka.tar Haven’t you heard these precious statements 

du 11.du 11 œiß li.bi 2.in.tuku.am 3 

made innumerable times? 

Sjöberg notes that his version of this manuscript is based on a handcopy by Kaneva, but I 

have not yet been unable to locate the publication: the foregoing transliteration is based 

on Sjöberg’s composite text in conjunction with his list of variants. This manuscript 

essentially changes the mood from interrogative to indicative, so, for example, in line 

117, what is a question in the majority text becomes an assertion in this manuscript, “you 

have heard . . . innumerable times!” The meaning of the passage is still quite uncertain 

and I have taken the absence of an explicit agreement marker before the verbal root in

combination with the copula following the verb as an indication that line 117 is a 

question (for a similar construction, see the series of igi bi 2.du 8.am 3 [“did you see . . . ?”] 

predicates at the end of Gilgamesh and the Netherworld). The final verb in line 115 also 

shows a great deal of formal variation: nam.bi tar.ra (K and EE), nam bi 2.ib.tar.re (Z), 

nam.bi in.tar.re (AA), nam.bi in.tar.ra.am 3 (FF), nam.bi tar.re (II), nam bi 2 [...] (JJ). 

It seems fairly clear that the variation revolves around the sequence /nam bi tar/ and one 

interpretation, nam bi 2.ib.tar.re (Z), belongs to a grammatical class dealt with in this 

chapter, namely the *XP nam bi-√ distributional class. Otherwise, the differences 

between manuscript Z and the majority text are more difficult to evaluate: œiß gid.da 

ad.da.œu 10.ße 3 instead of œiß gid 2.da ad.da.œa 2.ße 3; nu.ub.be 2 instead of nu.ub.be 2.en; and 

ib 2.us 2.e instead of i 3.ib 2.us 2. 

285

4.3 Focus-affected compound verb constructions (excluding BNBV inal) 

The interpretation of the four following examples is particularly difficult due to the fact 

that they are compound verbs with relatively opaque compositionality. 

(23) Lugalbanda [1.8.2.2], ll. 213-217 

[ Topic gu 3 i.ri.de 2.a nam mu.ni.ib.be 2.en 

[ Presupposition ßeß.zu.ne.er [ Focus pa nam] bi 2.ib.e 3.en] 

As for what I have told you, the fate I have fixed for you, you should not MAKE IT 

KNOWN to your brothers, (but rather you should TAKE CARE OF THEM [cf. 

a 2-√e 3 “to take care of”], lit. it is not THE BRANCH that you should cause to go 

forth [so as to inform] your brothers, but rather THE ARM that you should cause to 

go forth. 

213. gu 3 i.ri.de 2.a nam i.ri.tar.ra As for what I told you, your fate I ordained, 

214. ku.li.zu.ne.er nam mu.ni.ib.be 2.en It is not your friends whom you should tell, 

215. ßeß.zu.ne.er pa nam bi 2.ib.e 3.en It is not to your brothers that you should 

216. sa 6.ga ˙ul ßa 3.ga œal 2.la ur 5 

˙e 2.na.nam.ma 

286 

make it known, 

Fair fortune may conceal foul, isn’t it true? 

217. œa 2.e gud 3.œu 10.ße 3 ˙e 2.me.en za.e Leave me to my nest, you should be with 

erin 2.zu ˙e 2.me.en your troops,”

Though the translation I offer in (23) is, in my view, roughly correct, a rather subtle pun 

may also be at work in the passage: if the contrast in (23) is in fact between pa-√e 3 “to 

appear” and a 2-√e 3 “to take care of,” then a literal translation, taking the bird imagery into 

consideration might be something like “it is not YOUR WINGS, pa, that you should 

show your brothers, but rather YOUR ARMS, a 2. In itself the contrast is rather 

innocuous, but since pa “wing” and a 2 “arm” as body-part nouns are emblematic of the 

contrast between bird and man and by extension, in the language of the epic, the contrast 

between the divine realm of the bird-man and that of an ordinary mortal, what Anzu 

seems to be telling Lugalbanda is that the time of his rite of passage is over (cf. 

Michalowski 1980b) and that it is time to return to human company. 

(24) The Instructions of Shuruppak [5.6.1], ll. 9-12 

[ Topic na.ri.ga.œu 10] [ Focus ßu nam [ Presupposition bi 2.bar.re]] 

As for my instructions, you should not LOSE TRACK of them, (but rather follow 

them closely [cf. igi-√bar “to keep an eye on something”], lit. it is not THE HAND 

that you should open [so as to release], but rather THE EYE that you should (keep) 

open [so as to keep an eye on something]) 

9. dumu.œu 10 na ga.ri na ri.œu 10 

he 2.dab 5 

287 

My son, let me instruct you, remember my 

advice, 

10. zi.u 4.su 3.ra 2 inim ga.ra.ab.≠du 11± Zi’usudra, let me speak a word to you, you 

œizzal ˙e 2.em.ßi.ak should pay attention!

11. na ri.ga.œu 10 ßu nam bi 2.bar.re As for my instructions, you should not 

12. inim du 11.ga.œu 10 

na.ab.ta.bala.e.de 3 

288 

lose track of them, 

As for my words, you should not turn away 

from them, 

This passage is repeated more-or-less verbatim in lines 84-87 and lines 149-152. ETCSL 

omits ßu from line 151 (equivalent to line 156 in Alster’s edition [1974, 42]), but all four 

of the exemplars that preserve the relevant part of the line in Alster’s edition include the 

ßu (Alster 1974, 62). 

(25) Instructions of Shuruppak [5.6.1], ll. 32-34 (Karahashi 2000, 82) 

[ Topic dumu.œu 10] [ Presupposition daggan.na lu 2 dam tuku.da [ Focus dur 2 nam] bi 2.œa 2.œa 2] 

My son, you should not SIT DOWN in a chamber with a married woman, (but 

rather you should REMAIN STANDING) 

32. nita˙ niœir.si na.an.ak ni 2.zu You should not make a young man best man 

na.an x x …, 

33. ki.sikil dam tuku.≠da± e.≠ne± You should not play around with a young 

nam.mu.un.ne ? .e inim sig.ga married woman, (her) weak words will 

≠mah±.[am 3] be weighty, 

34. dumu.[œu 10] daggan.na lu 2 dam My son, in a chamber with a married 

≠tuku.da± dur 2 nam woman, lit. married person, you should 

≠bi 2±.e.œa 2.œa 2 

not sit down,

(26) The Instructions of Shuruppak [5.6.1], ll. 29-31 (Alster 1974, 36-37, 83; 

nam 

[ Topic ni2.zu] [ Focus tun3.am3 [Presupposition bi2.ib.bar.≠re±.e]] As for fearsomeness, it is not AN AXE that you should hold on to, (but rather A 

TRUE EYE (igi zi(d) [cf. igi zi √bar]), 

29. e 2 na.a.an.ni.buru 3.e.en You should not break into a house, you 

mi.si.IÍ.ra al nam.me should not wish for the money chest, 

30. ni 2.zu ≠piriœ na.nam± ul.dab 5 saœ As for your fearsomeness, it is that of a lion, 

na.nam but once caught, it is that of a slave, 

31. dumu.œu 10 sa gaz nam.mu.u 3.ak.e My son, you should not commit robbery—as 

nam 

ni2.zu tun3.am3 bi 2.ib.bar.≠re±.e hold, 

289 

for fearsomeness, it is not an axe that 

Line 31 in the ETCSL version, which I adopt as conventional, is equivalent to lines 35-36 

in Alster’s edition (1974, 36-37). In his commentary (1974, 83), Alster takes ni 2.zu as a 

phonetic writing of ni 2.zu˙ “thief,” but translates it in his edition as ‘yourself’. I also take 

nam 

the NAM in line 31 as a gloss on tun3.am3 so as to clarify that it is /tun-nam/ rather 

than /tun-am/: the orthography is interesting in that if the author had wanted to indicate a 

simple copula, the expected orthography would have been tun 3.na.am 3. Note, therefore, 

that, apparently, one of the reasons for the highly conventionalized orthography of NAM 

for *na-/*nu-am might be that, if the underlying form is *na-am, a syllabic orthography 

would have been systematically confused with the third person animate possessive 

pronoun followed by the copula *XP-ni-am > XP-na.am 3—the occurrence of NAM

following a word that ends in /n/ in this case represents a similar problem for the 

orthographic system. At the same time, given the conventionalized orthography of NAM 

in the negative contrastive focus constructions dealt with in this section, it could just as 

well be taken as a kind of semantic, or more precisely, a morphosyntactic determinative, 

indicating the particular morphological segmentation of the sequence in question. The 

contrast in this example is somewhat unclear to me, but I suspect that it is with *igi zi 

√bar “to look at something/one with a clear/just eye.” 

290

4.4 Focus-affected BNBV inal predicates 

At present, I know of just one example of a focus-affected BNBV inal predicate in line 410 

of the Ur Lament. Due to the fact that there is only a single example of the construction, 

that it is formally anomalous in various ways and that it appears in a composition that 

was, in all likelihood, composed in the Isin-Larsa period, I cannot exclude that possibility 

that the postulation of a focus-affected BNBV inal class in the *XP nam bi-√ construction 

is mistaken. Nonetheless, if the single example is valid to some degree, it is a very 

interesting example and the following analysis proceeds on the assumption that it 

represents some stage of Sumerian linguistic reality. The line numbers used here are 

consistent with ETCSL’s numbering, but Römer’s new edition numbers my line 410 as 

line 408. 

(27) Ur Lament [2.2.2], ll. 407-410 

(Kramer 1940; Jacobsen 1987, 473; Römer 2004, 82 and 166) 

[ Topic uœ 3 saœ gi 6.zu] [ Presupposition [ Focus igi. ? nam] bi 2.ib.du 8] 

As for your black-headed people, it is not YOU WHO SEE them (after the storm 

has cleared, but rather THEY WHO SEE you, the moon-god, Nanna, in a night sky 

free of storms) 

407. u 4.de 3 u 4 kalam.ma The storm that causes the light of the land to 

u 2.gu bi 2.ib.de 2.a.re disappear, 

291

408. u 4 ˙ul gig du 11.ga d en.lil 2.la 2.ta The evil storm that sweeps through, ordered 

im.mi.in.zal.la.re in hate by Enlil, 

409. a.a d nanna u 4.bi uru 2.zu.ta Father Nanna, may that storm swoop down 

ki nam ba.œa 2.œa 2 

410. uœ 3 saœ gi 6.zu igi. ? 

36 

nam bi2.ib.du8 292 

no more on your city, 

As for your black-headed people, it is not 

you who see them, 

The reconstruction of line 410 used here is in conformity with ETCSL’s presentation, but 

the evidence in the text-artifactual record is quite slim. Römer’s score transliteration, 

which I have altered in some typographic details, reads as follows (Römer 2004, 82): 

(28) Ur Lament [2.2.2], l. 410 (l. 408 in Römer 2004, 82) 

A V 408 uœ 3 saœ gi 6.zu igi ? .zu ?? nam bi 2.ib.du 8 

CC OMITS 

HH 19 ≠uœ 3±* saœ gi 6.ga.zu igi.bi nam* bi 2.ib.du 8 

R 4 IV 14 […] ≠ib± !? .du 8 

Da rev. II 9 […] igi.zu nam ≠bi 2.ib !? .du 8± 

C VI 3 […] ≠bi 2±.ib 2.du 8 

S 2 […] ≠gi 6±.zu ≠igi± […] 

36 The pronominal agreement marker *-b- in bi2.ib.du 8, like the sequence of similar forms in example (8), seems to 

indicate future/modal reference, see the discussion following example (8).

Römer notes the “conspicuous alternation” between igi.zu (A V 408; Da, rev. II 9) and 

igi.bi (HH19),” and goes on to contrast Jacobsen’s interpretation of the line on the basis 

of √du 8 as a simple verb with Falkenstein’s suggestion that it is a compound verb 

igi-√du 8, but offers little insight into the matter (2004, 166). I suspect that the relatively 

unusual syntactic form of the line has resulted in the two versions, which, interestingly 

enough, present two different perspectives on the perception of Nanna. Since the 

possessive pronoun presumably refers to the perceiver rather than the one who is 

perceived, the version with igi.zu could be translated “as for your black-headed people, it 

is not YOU WHO CAN SEE them, (but rather THEY WHO CAN SEE you).” The 

alternative in manuscript HH has only igi.bi, but there is also an additional locative 

postposition inserted into the topicalized phrase, ≠uœ 3± saœ.gi 6.ga.zu, in order to form an 

anticipatory genitive within the topicalized phrase (with *-bi as the topic-marking 

possessor rather than *-zu). This makes the black-headed people of Nanna the possessor 

of igi and the perceiver of whatever is to be seen: “As for your black-headed people, it is 

not THEY WHO SAW … (?).” This interpretation seems, however, to collapse under its 

own weight. 

I would submit that the original form is not preserved in any of the extant witnesses: 

namely, uœ 3 saœ.gi 6.zu igi nam bi 2.ib 2.du 8, “As for your black-headed people, it is not 

THE STORM that they saw, (but rather YOU, NANNA, THE MOON GOD, whom they 

saw in the clear, storm-free sky).” The confusion seems to arise from the fact that the 

perceived object, namely u 4 … re, was originally in the underlying form of the line: *uœ 3 

saœ gi 6.zu u 4…re igi nam bi 2.ib 2.du 8. But I looks like it has been moved out of line 410 

293

and made the subject of a lengthy “appellation poem” (Michalowski 1996, 146), each line 

of which is followed by the demonstrative *-re. Woods, as part of a larger effort to make 

sense of deixis in Sumerian, has collected 42 examples of similar uses of *-re as a 

demonstrative, 18 of which occur in the very same passage from the Ur Lament. Nearly 

every line in the series of appellations that precede line 410 in description of the storm 

ends with the demonstrative *-re (Woods 2001, 119-123). Woods has argued, quite 

persuasively in my view, that the *-re demonstrative is the distal, non-visible member of 

a ternary set of demonstratives (Woods 2001, 167 and 171): 

*-e PROXIMAL 

*-ße DISTAL, VISIBLE 

*-re DISTAL, NON-VISIBLE 

Although I find Woods’ interpretation of *-re quite appealing, many of his proposed 

semantic/pragmatic features (from disjoint ontic realms [Woods 2001, 157-148] to an 

iconic model of its discourse functions that maps the distal feature of the demonstrative to 

the “end [or] final position [in] a relative clause [Woods 2001, 116]) can safely be 

disregarded. As the examples that Woods has assembled make clear, the pragmatics of 

the demonstrative follow quite directly from its semantic features: the demonstrative *-re 

marks nominals (including relative clauses) that refer to an entity that is no longer visible 

at the point in time when the verb of perception takes place. Woods indicates this in some 

294

of his translations through the use of “once” and in others the addition of “used to” or 

past tense on the main verb in the translation. 

(29) Lament for Sumer and Ur 338 

(Michalowski 1989, 56-57, 98; Woods 2001, 120, ex. 133) 

ne.mur dugud.gin 7 i 3.ra.a.re i.bi 2.bi ba.gul 

As for the fire that used to burn like a heavy coal, its smoke is now extinguished, 

(30) Nippur Lament 38-39 

(Woods 2001, 120, ex. 135) 

kuß ßem5 kuß a.la 2 mu.un.du 12.a.re i.lu gig.ga.a a.na.ße 3 u 4 mi.ni.ib 2.zal 

As for those who used to play the shem and ala drums, why do they pass the day in 

bitter lament? 

These examples as well as the vast majority of the other examples occur—as we might 

have expected for a syntactic construction that refers to entities that no longer exist—in 

laments. I am even tempted to say that the frequent occurrence of *-re and the semantics 

associated with it could probably be used as a criterion for the identification of laments 

and elegies as generic categories. Interestingly enough, this example may represent an 

exception to my earlier statement that the composers of OB texts no longer understood 

the *XP nam bi-√ construction, but this is the only lament in which I have identified the 

construction. 

295

It is interesting that the construction differs from the other examples of the 

*XP nam bi-√ construction in a couple of ways: not only is the inalienable noun modified 

by a possessive suffix that is not followed by any subsequent case-marking postposition, 

which is not the case with any other examples, but the verb itself is not in the 

progressive/imperfective (marû) aspect even though it refers to a future event. It may be 

that the syntax of BNBV inal predicates was simply different from the other examples, but 

it is also entirely possible that the composers were attempting to use two different 

morphosyntactic/rhetorical devices so as to achieve the trope at work in the passage: the 

use of *-re to indicate something no longer visible and the use of the *XP nam 

bi-√ construction to code negative contrastive focus. Although the artistry is quite 

advanced in my opinion, the grammaticality of the passage—if we can judge on the basis 

of comparison to other *XP nam bi-√ constructions—is either marginal or no longer 

familiar to its tradents and consequently, the manuscripts exhibit a great deal of variation 

and orthographic uncertainty. 

Now, to return to the passage from the Ur Lament, (27) above, the real poetic beauty 

of the passage lies in its juxtaposition of the terrifying darkness of the storm and the 

calm, tranquil darkness of a clear, moonlit night. In the sections that lead up to lines 388- 

414, a litany of destruction is attributed to the storm, but in the passage itself, the lines 

leading up to line 410 draw a contrast between the storm (u 4) and the light of the land (u 4 

kalam.ma), namely the moon-god Nanna. This opposition is epitomized by line 407. 

296

(31) Ur Lament, l. 407 

u 4.de 3 u 4 kalam.ma u 2.gu bi 2.ib.de 2.a.re 

The storm that causes the light of the land to disappear, 

The storm (u 4) can, therefore, be thought of as an utter darkness, whereas the light of the 

land (as made clear by lines 409-410) is to be equated with Nanna, the moon-god. The 

series of lines that end in *-re all seem to refer to the storm and the use of the 

demonstrative *-re indicates the hope of those singing the lament that the storm will no 

longer be visible in future. Clearly this use of *-re inverts the standard trope in which the 

entities destroyed by the storm bear the *-re demonstrative. The inversion reaches its 

culmination in line 410 and my reconstruction of the line without any possessive pronoun 

on igi completes the contrast between the no longer visible storm and—as a direct 

consequence of the disappearance of the storm—the visibility of Nanna, the moon-god, in 

a clear, nighttime sky: uœ 3 saœ.gi 6.zu u 4…re igi nam bi 2.ib 2.du 8 “As for your black- 

headed people, it is not THE STORM (u 4) that they will see, (but rather YOU, NANNA, 

THE MOON-GOD, that they will see).” Hence the addition of the second person 

possessive pronoun to igi to form igi.zu in some of the witnesses is apparently an attempt 

to indicate the meaning of the passage in a more direct fashion than the use of the 

contrast implicit in the negative contrastive focus construction. 

Although this phenomenon can be thought of as a form of evidentiality (Chafe and 

Nichols 1986), the contexts in which *-re occurs point to a more specific “problem,” if 

you will, that the use of *-re solves: there is a fundamental difference between verbs such 

297

as “to look at,” which need not specify an object of perception and are regularly agentive, 

and verbs such as “to see,” which must occur with a perceived object and are regularly 

non-agentive. Whereas a statement like “Mary is looking at nothing” can be interpreted to 

mean that Mary is looking at an entity that happens to be empty space, when “to look at” 

is replaced by “to see,” this is no longer the case. But the use of the demonstrative *-re 

seems to allow for the direct perception of an entity that is no longer visible even when it 

occurs with the verb that means “to see” in Sumerian. 

298

4.5 Seeing something that is no longer there: The demonstrative *-re and 

visual perception 

The somewhat unusual characterization of contrastive focus in Sumerian in the foregoing 

pages has a number of features to recommend it, notably the degree to which it makes 

sense of otherwise recalcitrant textual passages. But the major flaw it embodies is that in 

no case—not even one—is the positive contrastive alternative to the negated focal phrase 

present in the text-artifactual record. In light of such a seemingly problematic empirical 

gap, I would like to offer a redescription of the phenomena in the previous section that 

captures the contrastive force of the construction without necessarily positing a distinct, 

positive contrastive sentence that is regularly deleted under relevance. At the same time, 

the alternative proposed in this section—based on Herburger’s analysis of double 

negation in Spanish—also helps us to make sense of the igi nam bi 2.ib 2.du 8 construction 

in the previous section. I should emphasize that the interpretation of the *XP nam bi-√ 

construction in this section is not necessarily incommensurable with that of the previous 

section: given the different levels at which a particular grammatical construction can be 

evaluated (morphological, semantic, pragmatic, generic), the previous section can be 

thought of as an investigation of the construction in pragmatic or generic terms, whereas 

this section looks more directly at its morphosyntactic and semantic details. 

As part of a quite involved argument on the nature of focal presupposition that need 

not detain us here, Herburger reviews certain interesting double negative constructions in 

Spanish (Herburger 2000, 23-26). The following examples provide a useful starting point 

(Herburger 2000, 24, ex. 19 and 20). 

299

(32) María se fué con nada 

Maria left with nothing 

“Maria left with nothing” 

(33) María no se fué con nada 

Maria Neg left with nothing 

a. “Maria didn’t leave with anything (simple negation)” 

b. “Maria didn’t leave with nothing, i.e., she left with something (double 

negation)” 

Whereas the example in (32) with its negative indefinite pronoun in the prepositional 

phrase con nada “with nothing” has only one interpretation: “that Maria left, but that she 

did not take anything with her,” the sentence in (33) has two diametrically opposed 

interpretations. The interpretation in (33a) is basically equivalent to the meaning of the 

sentence in (32), where nada “nothing” is what is known as a negative polarity item 

(NPI) such as “any” in English. Hence the proper translation of (32) and (33a) in English 

would be the two following sentences. 

(34) a. Maria left with nothing. 

b. Maria didn’t leave with anything. 

300

In English, the NPI allows the negative particle to move to the main verb, while being 

interpreted as negating “thing” rather than “left”: hence “Maria didn’t leave with 

anything” still entails that Maria actually left due to the presence of the NPI. The double 

negative interpretation in (33b), however, does not interpret nada “nothing” as a NPI, but 

rather as an independent referential term. In other words, according to the interpretation 

in (33b), nada refers to an entity that can be described as “nothing, thin air, empty space” 

and only after nada’s act of reference is the sentence as a whole negated. This double 

negation then leads the addressee or reader to draw the opposite implication: if she did 

not leave with nothing at all, then she must have left with something. Note that regardless 

of which interpretation of (33) that we choose, the fact that Maria left is presupposed and 

is not affected by the negation. 

Herburger then goes on to verbs of visual perception in Spanish and contrasts two 

versions of the following sentence: 

(35) Spanish (Herburger 2000, 25, ex. 22) 

El pobre Juan se ha melto loco. Se pasa los días mirando a nada / **viendo nada 

Poor Juan has gone insane. He spends his days looking at nothing / **seeing nothing. 

As Herburger writes, “[i]t is possible to look and to look at nothing (even if a sane person 

might not spend his days doing that). It is not possible, though, to see and to see nothing” 

(Herburger 2000, 25). Herburger is arguing that as long as nada is part of a prepositional 

phrase rather than an argument of the verb, it can have narrow scope (the negation only 

301

applying to nada as in [32b]) and “refer” in some sense to nothingness. The difference 

between mirando a “to look at” and viendo “to see” also exists in Sumerian, thus in the 

two following examples, which have already been presented earlier in the study, igi-√il 2 

“to look at” does not require a perceived object, but igi-√du 8 “to see” does, even when it 

is negated. 

(36) Lugalbanda II (Wilcke 1969a, 110-111, 184) 


302 

As for the Bird, who was looking from 


208. [lugal].≠banda 3± da ki.ta As for Lugalbanda, who was looking from 

igi mi.ni.in.il 2 sa˙ar erin 2.e below, he saw the troops 

du 8.a igi bi 2.in.du 8.ru enveloped in dust, 

(37) Gilgamesh and the Netherworld, ll. 300-303 

(Karahashi 2000, 121, ex. 25; Shaffer 1963, 95, 119) 

300. niœar x(NIGIN 3) œar tur.tur.œu 10 

ni 2.ba nu.zu igi bi 2.du 8.am 3 

(Gilgamesh:) As for my little newborns, 

who never knew themselves, did 

igi bi 2.du 8.am 3 a.na.gin 7 an.ak you see them? (Enkidu:) I saw them. 

(Gilgamesh:) What happened to them? 

301. œiß banßur ku 3.sig 17 ku 3.babbar (Enkidu:) They play on a table of gold and 

lal 3 i 3.nun.ta e.ne im.di.e.ne silver, (laden with) honey and ghee,

302. lu 2 izi.la 2 igi bi 2.du 8.am 3 

igi nu mu.ni.du 8.am 3 

(var. mu.de 3.du 8.am 3) 

303 

(Gilgamesh:) Did you see a man who’d 

burned? (Enkidu:) I didn’t see one. 

303. gidim.a.ni nu.œal 2 i.bi 2.ni As for his ghost, it is not there. As for his 

an.na ba.e.e 11 

smoke, it went up into the sky, 

In (36), neither of the two occurrences of igi mi.ni.in.il 2 take a perceived object as an 

argument: igi refers to the eye of the beholder rather than that which is perceived and if a 

target of the “looking at” does occur, it is never followed by the locative-terminative 

postposition *-e: in the set of attestations collected by Karahashi (2000, 125-126), the 

object being “looked at” is followed by the dative (*-ra), the locative (*-a) or the 

terminative (*-ße 3). The main verb in each line, igi bi 2.ib 2.du 8.ru, does occur with the 

perceived object in the locative-terminative case and agrees with it in number, *-eß, as 

well. In (37), the point of interest is the contrast between Enkidu’s report of successful 

perception in line 300 and the report of unsuccessful perception in line 302, igi nu 

mu.ni.du 8.am 3. This type of alternation apparently provided the impetus for a number of 

forerunners to the directive case hypothesis (Poebel 1933; Falkenstein 1978, 188; Gragg 

1973a, 72), so that, for instance, just a few lines after arguing for a distinctively 

“locative” interpretation of *-ni-, Gragg cautions that “[t]his position which we have 

taken should not however be allowed to obscure the close relationship that exists between 

- n i -, i m – m i -, (and also b í -)” (Gragg 1973a, 72). In my view, the differences in form 

result from the need for a direct object with a verb meaning “to see” even when the verb

is negated. The phrase igi nu mu.ni.du 8, “I didn’t see anyone (like that),” therefore, 

corresponds to an NPI in English in certain respects, but I do not go into the matter 

further here. 

The question, or rather the textual passage, looming in the background of this entire 

discussion is the igi nam bi 2.ib.du 8 construction in line 410 of the Ur Lament, example 

(27) above, which I repeat below. 

(38) Ur Lament [2.2.2], ll. 407-410 

407. u 4.de 3 u 4 kalam.ma The storm that causes the light of the land to 

u 2.gu bi 2.ib.de 2.a.re disappear, 

408. u 4 hul gig du 11.ga d en.lil 2.la 2.ta The storm that sweeps through, ordered in 

im.mi.in.zal.la.re hate by Enlil, 

409. a.a d nanna u 4.bi uru 2.zu.ta Father Nanna, may that storm swoop down 

ki nam ba.œa 2.œa 2 

410. uœ 3 saœ gi 6.zu igi. ? 

nam bi 2.ib.du 8 

304 

no more on your city, 

As for your black-headed people, it is not 

the storm that they will see, 

In (38), the verbal prefix *bi-√ indicates that there should be a raised possessor of igi as 

the perceived object and as I have argued above, the raised possessor that we are looking 

for is represented by the series of epithets and appellations applied to the storm (u 4) in 

lines 392, 395-398, 402-408, each of which ends with the distal, non-visible 

demonstrative *-re.

The double negation implicit in the juxtaposition of the negative contrastive focus 

particle, *nam, and the non-visible demonstrative, *-re, i.e., that which is no longer 

visible, complicates matters. I understand the basic underlying form of the line to be the 

following: 

(39) **u 4…re igi bi 2.ib 2.du 8 

They see/saw the storm that is/was no longer there. 

Regardless of variation in tense or other factors, the English translation is problematic. If, 

however, the negative contrastive focus particle, *nam, is introduced into (39), and the 

new formulation translated in line with the non-BNBV inal examples in sections 4.2 and 

4.3, the result might be something like this: 

(40) *u 4…re igi nam bi 2.ib 2.du 8 

It is not the storm (u 4) that is/was no longer visible that they saw, (but rather the 

light of the land (u 4 kalam.ma) that is/was visible = Nanna) 

The contrast, therefore, consists of two parts: the contrast between that which is not 

visible and that which is visible and the contrast between the storm (u 4) and the light of 

the land (u 4 kalam.ma = Nanna). If the parallel with the double negative construction in 

Spanish holds, then each of the two possible interpretations of the double negative 

construction in Spanish corresponds to half of the translation in (40): the interpretation in 

305

(33a), “Maria didn’t leave with anything,” corresponds to “it is not the storm that is no 

longer there (that they see),” whereas the interpretation in (33b), “Maria didn’t leave with 

nothing, i.e., she left with something,” corresponds to “it is the light (of the land) (that 

they see).” 

In all of the examples of the *XP nam bi-√ construction including the example from 

line 410 of the Ur Lament, the main verb occurs to the right of the copula and is 

presupposed, while the constituent immediately to the left of the copula is negated under 

contrastive focus. Here, however, in the example from line 410 of the Ur Lament, the 

extensional meaning of the two contrastive interpretations, namely “it is not the storm 

(u 4) that is no longer there (that they see)” and “it is the light of the land (u 4 kalam.ma) 

(that they see),” is the same. Due to the peculiar nature of verbs of direct perception such 

as igi-√du 8 and the context at issue, the act of “not seeing the storm” is equivalent to the 

act of “seeing Nanna in the clear evening sky.” Recall the discussion of the thetic 

character of direct perception complements [DPCs] and the similarities between DPCs 

and HIRCs outlined in section 3.5. If DPCs are thetic, it should not be possible, under 

ordinary circumstances, to distinguish two entities in a thetic statement and bring them 

into contrast. But here in line 410 of the Ur Lament, we find a rather unordinary situation 

in which two contrastive intensions are equivalent in extensional terms, thereby allowing 

contrastive focus and the thetic semantics of a DPC to coexist in a single construction. 

306

4.6 Contrastive/identificational focus and the thetic/categorical 

opposition 

Given the fact that the example in the previous section is unique among *XP nam bi-√ 

constructions in a number of respects, it obviously cannot serve as a model for 

interpreting the other exemplars, but it does at least raise the issue of the 

thetic/categorical opposition and its relation to contrastive focus. Since all of the 

examples of the *XP nam bi-√ construction necessarily represent examples of contrastive 

focus, I turn initially to a further explication of contrastive focus, followed by a 

discussion of the interactions between contrastive focus and the thetic/categorical 

opposition. 

Kiss (1998) argues that there are two basic types of focus: identificational focus and 

informational focus. Identificational focus, which I have generally termed contrastive 

focus above, can be distinguished from informational focus on the basis of both 

morphosyntactic and semantic criteria. The semantic criteria have been outlined in some 

detail above, but in terms of morphosyntax, identificational focus regularly moves the 

focused constituent to the left periphery or puts it into some other distinctive syntactic 

configuration such as the copular construction in Sumerian. In the two pairs of examples 

that follow, from Greek (41) and Finnish (42) respectively, constituents under 

identificational focus appear in bold, while items bearing informational focus are in 

capitals (Kiss 1998, 246, ex. 2 and 3). 

307

(41) a. ston petro dhanisan to vivlio 

to.the Petro lent.3Pl the.Acc book 

It was to Petro that they lent the book 

b. dhanisann to vivlio STON PETRO 

lent.3Pl the.Acc book to.the Petro 

They lent the book TO PETRO 

(42) a. annalle mikko antoi kukkia 

Anna.Adess Mikko gave flowers 

It was to Anna that Mikko gave flowers 

b. mikko antoi kukkia ANNALLE 

Mikko gave flowers Anna.Adess 

Mikko gave flowers TO ANNA 

In each pair of sentences, the first member, (41a) and (42a), is an example of 

identificational focus as evidenced by the fact that the focused constituent has moved to 

clause-initial position, whereas the examples of informational focus, in (41b) and (42b), 

show no such movement, merely distinctive intonation. The English glosses in (41) and 

(42) are, in this regard, instructive as well: identificational focus in English is indicated 

308

through the use of the cleft sentence, whereas a constituent that bears informational focus 

remains in situ and has a distinctive intonation. 

The difference between identificational and informational focus in Sumerian (as 

briefly noted in relation to examples (1) and (3) in section 4.2 above) is that 

identificational focus regularly makes use of the copula, whereas informational focus 

does not make special use of the copula. 

(43) The Tale of Gudam [1.3.4], ll. 10-11 

10. i 3.gu 7.a.zu i 3.gu 7.a.zu As for what you have eaten, as for what you 

11. ninda nu.e.gu 7 uzu.zu.um i 3.gu 7 

309 

have eaten, 

(44) Gilgamesh and the Netherworld [1.8.1.4], ll. 175 and 179 

175. œiß e.ke 4.ma.œu 10 ganzer.ße 3 

… 

mu.da.ßub a.ba.a ma.ra.ab.e 11.de 3 

It was not bread that you have eaten, it was 

your flesh that you have eaten! 

As for my ekema, it has fallen into Ganzer, 

who will descend on my behalf? 

179. œiß e.ke 4.ma.zu ganzer.ta œa 2.e As for your ekema, I shall retrieve it from 


Ganzer for you, 

In (43), it seems clear that uzu.zu.um (< *uzu.zu.am 3) contrasts with ninda, since ninda 

is followed by the negative form of the same verb that follows uzu.zu.um. Although both

ninda and uzu.zu appear in the same line in (43), the first half of the line, in which ninda 

occurs with a negative form of the verb is a simple statement of fact; the second half of 

the line, however, draws a contrast with the first half of the line and it is uzu.zu.um in 

contrastive focus that bears the copula. In (44), however, œa 2.e “I” under informational 

focus is not contrastive and is not followed by the copula. Although both types of focused 

phrase appear immediately before the verb, it seems that they can be distinguished on the 

basis of the presence or absence of the copula. If my analysis of nam as derived from 

*na/nu-am is valid, then the *XP nam bi-√ construction would represent a particular kind 

of identificational focus construction, namely what I have previously described as a 

negative contrastive focus construction. 

Kiss describes two semantic features so as to characterize identificational focus: 

± exhaustive and ± contrastive and describes the semantics of identificational focus as 

follows (Kiss 1998, 245): 

An identificational focus represents a subset of the set of contextually or situationally 

given elements for which the predicate phrase can potentially hold; it is identified as 

the exhaustive subset of this set for which the predicate phrase actually holds. 

The import of this definition is that thetic statements and contrastive focus should be 

incompatible: since contrastive focus typically identifies an individual or group of entities 

and predicates something of them in contrast to some other entity or entities, it would 

seem to call for reference to individuals—so as to draw the contrast—and, consequently, 

categorical rather than thetic predication. Since I have not investigated alienable BNBV 

predicates or *bi-√ prefix verbs that do not form compound verbs, I can do little more 

310

than offer a tentative suggestion. But if the *bi-√ prefix refers to a single, bounded event 

(as might be suggested by my discussion of the particle-verb constructions in 2.6), the 

formation of the marû stem may well indicate a plurality of possible events. I do not 

mean to suggest that the marû of a *bi-√ prefix verb is equivalent in some way to ˙amt≥u 

reduplication as a sign of absolutive/nominative plurality, nor that it is iterative in the 

usual sense. When I speak of a plurality of possible events, what I have in mind is the 

pair-list reading of a focus-affected phrase in a definiteness effect environment. Recall 

example (45) from section 3.4, repeated below: 

(45) Joe: “Who is going to drive me back home?” 

Mary: “Well, there’s [ Focus my brother, or the guy standing on the corner].” 

In (45), when the definiteness effect environment (to the left of the copula in an 

existential sentence) is focus-affected, as in answer to a wh-question for example, any 

phrase that occurs in that environment—even if there is only one—is interpreted as one 

member of a list of possibilities. I would suggest that the marû form of the verb in *XP 

nam bi-√ constructions is due to the pair-list reading of the focused constituent to its left. 

In other words, under contrastive focus (as in the *XP nam bi-√ construction) a bare noun 

to the left of the verb can no longer be interpreted as an indefinite noun that only 

contributes to the specification of the meaning of the verb (“Restrict” in the opposition 

described by Chung and Ladusaw 2004); instead, the introduction of the copula that acts 

as a contrastive focus marker converts the bare noun into a specific, referential noun 

311

(“Specify” in Chung and Ladusaw’s opposition), and, at the same time, the change to a 

marû verbal root invokes a pair-list reading of the bare noun that is now referentially 

specific. 

Such a scenario has several things that recommend it: (a) it fits nicely with the 

contrastive focus construction (for non-BNBV inal predicates) in that a noun needs to be 

individualized and made referential if is it is to serve as a point of contrast for some other 

entity; (b) the non-occurrence of the marû form of the verb in the one example of a 

BNBV inal predicate in the *XP nam bi-√ construction makes a certain amount of sense in 

that, unlike all of the other examples of the *XP nam bi-√ construction, the example from 

line 410 of the Ur Lament refers to what is, from an extensional point of view, a single 

state of affairs rather than two entities that are mutually exclusive; and (c) the nominal 

component of a compound verb is treated as, alternatively, non-referential (“Restrict”) or 

referential (“Specify”) depending on the morphosyntactic environment in which it occurs. 

If the predicate in question is a non-BNBV inal predicate, the bare noun is presumably non- 

referential with a ˙amt≥u *bi-√ prefix verb, but in the complex morphosyntactic 

environment of the *XP nam bi-√ construction, it becomes referentially specific and 

acquires a pair-list reading. In terms of BNBV inal predicates, however, the semantic 

interpretation of the bare noun is much more highly constrained: the non-referential 

character of the bare inalienable noun clearly makes the transfer of possession model 

advocated above possible, and in all likelihood, it also contributes to the thetic character 

of verbs of perception. But the restriction on BNBV inal predicates in the *XP nam bi- 

√ construction are equally clear: there is only a single, questionable example of the 

312

construction in a nearly unique context in spite of the fact that igi-√du 8 occurs quite 

frequently in the text-artifactual record. 

The inability of BNBV inal predicates to form the *XP nam bi-√ construction more 

readily seems, therefore, to follow from the highly thetic character of verbs of perception 

and experience and the transfer of possession model that underlies these constructions in 

Sumerian. This line of argument also allows for a harmonization of the general findings 

of Sumerologists who have investigated the matter (particularly Karahashi’s discussion of 

non-referentiality and the set of devices that introduce referential specificity (Karahashi 

2000, 42-45) and a number of different research traditions in formal semantics that draw 

a distinction between events or portions thereof (stage-level / thetic / “Restrict”) and 

individuals (individual-level / categorical / “Specify”). 37 

37 The named contrasts are, for our purposes, best associated with Carlson 1977, Basilico 2003, and Chung and 

Ladusaw 2004, respectively. 

313

5 

Possession, animacy and the rise of ergativity in Sumerian 

In the final chapter, I attempt to draw together a few strands of the discussion up to now, 

primarily in terms of the diachronic dimension of some of the foregoing proposals. In 

section 5.1, I look at the tendency for inalienable constructions to be older—in terms of 

the diachronic changes that are constantly taking place in a grammar—than constructions 

involving alienable nouns and what this might tell us about the history of the BNBV inal 

construction. In section 5.2, I evaluate several different proposals to explain the 

development of ergativity in general and, in particular, the split of what I will term the 

old locative postposition *-e into two distinct postpositions in the text-artifactual record: 

the locative-terminative postposition and the ergative postposition. Section 5.3 offers a 

brief conclusion and highlights several prospects for future research. 

314

5.1 Inalienability and the historicity of the case-marking schemata 

Since inalienability is the distinctive quality of the BNBV inal lexical class and the 

morphosyntactic environment that defines it as a lexical class, not to mention the fact that 

a number of morphosyntactic and semantic distinctions seem to be based on whether or 

not the nominal component of a *bi-√ prefix compound verb is inalienable, the 

typological literature on inalienability may be of some help in elucidating the diachronic 

background to the BNBV inal construction in Sumerian. It seems fairly clear—at least 

according to Nichols’ survey of inalienable possession in North America (1988)—that 

the type of inalienability represented by the BNBV inal construction is among the most 

common ways of marking inalienability. Nichols describes a similar case as follows: 

… languages of the Caddoan and Iroquoian families systematically deprive kin terms 

of the possibility of displaying an alienability opposition, in that those nouns 

obligatorily take clausal rather than phrasal possession: one says, roughly, ‘she is 

mother to me’ rather than ‘my mother’. Since the verb is the head of the clause and 

hence dominates mother, these constructions amount to placement of the pronominal 

marker of possession not on the head of the phrase but even higher, on the head of 

the clause. The same can be said of constructions involving body-part terms, known 

variously as the ethical dative, possessor ascension, etc. (Nichols 1988, 577-578). 

Although “possessor raising,” which is the same thing as what Nichols calls “possessor 

ascension” and describes in the foregoing paragraph, has been previously documented in 

Sumerian (Zólyomi 1999, 231-237), I would submit that the kind of possessor raising 

described by Zólyomi as an argument for the directive case hypothesis is better 

315

understood as a single component of a broader and older possessor raising regime in 

which the BNBV inal construction plays a crucial role. Zólyomi notes, for example, that the 

*bi-√ prefix “agrees” with a nominal phrase of the form *NP-a after the possessor raising 

has taken place (Zólyomi 1999, 236), but this is just the familiar statement of locative 

“agreement” and makes no clear connection between the existence of possessor raising 

and the contribution of the *bi-√ prefix to either the possessor raising itself or the 

semantics of the predicate. 

I would argue, however, that the low source applicative hypothesis proposed in 

chapter 1 explains both the presence of possessor raising (including *NP-a bi-√ 

environments as well as BNBV environments) and the distinctive causer/experiencer 

semantics of the construction as a whole. Zólyomi’s causative analysis of alienable 

BNBV predicates can be subsumed, in other words, as one component of a broader (low 

source) applicative hypothesis alongside the experiential semantics of the BNBV inal class: 

the *bi-√ prefix in conjunction with a bare inalienable noun acts as a low source 

applicative morpheme, while in the presence of an alienable noun, it becomes causative 

in function. The same pattern is also found in the examples of applicative/causative 

alternation at the end of section 2.6. More importantly, it brings ongoing work on the 

morphosyntax of Sumerian in line with descriptions of the same phenomena in other 

languages. 

Nichols does not describe a particular historical scenario for the type of clausal 

possessive pattern that seems to be manifest in the BNBV inal construction, but what she 

describes as “Pattern 6” is very close to the clausal pattern seen in Sumerian. 

316

In languages with Pattern 6, whose ‘inalienable’ possession involves no marking 

while ‘alienable’ possession is dependent-marked, the lack of marking for the 

inalienable type is due to one of two historical factors: either the fixation of 

inalienable constructions antedates the rise of case in the language, or the possessor 

loses its case marker due to its incorporation into the head in a form of compounding 

(Nichols 1988, 579). 

Since the second of the two scenarios described by Nichols is clearly not the case in 

Sumerian, I take the first of the two descriptions (“the fixation of inalienable 

constructions antedates the rise of case in the language”) as a plausible scenario for 

Sumerian. This would suggest that the BNBV inal construction is a kind of historical 

artifact embedded in the grammar of Sumerian and that it may well provide a window 

into the earlier history of the language. In particular, the nominative/dative, or 

alternatively, the absolutive/oblique case marking system found in the BNBV inal 

construction ([ absolutive/nominative ergative possessor + bare inalienable noun] vs. [ dative/oblique 

object of perception]) corresponds quite closely to the “normalform” conjugation 

discussed by Silverstein as the point of departure for the development of case-marking 

schemata in other ergative languages (Silverstein 1976, 161; cf. Silverstein 1993a). This 

raises the strong possibility that the different types of *bi-√ prefix constructions 

(BNBV inal, alienable BNBV, *NP-a bi-√, etc.) can actually be thought of as 

morphosyntactic primitives registering distinct combinations of particular case-marking 

relations (or thematic structure) and inherent lexical content such as distinctions based on 

inalienability and animacy. The following section describes two successive (and 

317

increasingly specific) models of the rise of ergativity in Sumerian and attempts to 

determine what, if any, correlations hold between particular models of the development 

of ergativity and features such as inherent lexical content and/or possession. 

318

5.2 Possession and the development of ergativity 

Theorists have long posited relations of one kind or another between ergative case- 

marking and possession. One of the most insightful comments actually occurs in the 

midst of a debate between Silverstein and Hale on the nature of ergative case-marking 

systems: Hale had reasserted an ergativity-as-obligatory-passivization model that 

harmonized in some respects with the aspirations of early generative theory (Hale 1970) 

and Silverstein responded with one of the most important papers ever written about 

ergativity (Silverstein 1976). As part of an argument to the effect that the early 

generativists did not have a theory that could encompass the attested data, Silverstein 

describes the ergative “type” as follows: 

. . . possessors and ergators (or apparent agents) are frequently identified at the 

surface at least (Eskimo [including the West Greenlandic examples (70), (71), and 

(72) in section 1.8, CJ], Chinook, Tsotsil, Quiche); non-ergators (or apparent 

patients) are incorporated into verb-complexes in the same way, whether they are in 

transitive or in intransitive structures (Iroquoian, Tsimshian, Wichita); 

mediopassives and reflexives are identical in syntax and sometimes in form (Dyirbal, 

Chinook, Bandjalang); the ‘antipassive’ forms an ‘active intransitive’—in Sapir’s 

(1922: 150-151, 153-154) felicitous phrase—with the underlying agent of the 

transitive in nominative case (Chinook, Aleut), or it forms a nominative-dative 

schema for inflectional purposes (Dyirbal, Georgian). (Silverstein 1976, 115) 

Thus the relation between possession and ergative case-marking seems to have been—at 

least in descriptive terms—quite clear. But Silverstein is hesitant to address the 

319

diachronic origins (as opposed to the typological variation) of ergativity, since it was 

precisely the question of origins that had given rise—a century before—to the obligatory- 

passive hypothesis as well as it corollary to the effect that all languages “start out” as 

nominative/accusative languages and change into ergative languages under certain 

circumstances. 

Several years later, Trask (1979) would attempt to split the difference in historical 

terms, arguing that the typological framework proposed by Silverstein can actually come 

into being in historical terms in at least two different ways: Type A ergativity develops 

from an obligatory passivization, whereas Type B ergativity results from the 

grammaticalization of a possessive construction. Trask describes Type B ergativity, 

which bears a certain resemblance to the situation in Sumerian, as follows: 

Now in a language lacking a verb ‘have’, possessive predications are commonly 

made by putting the possessor into an oblique case, most often the dative, 

locative, or genitive—carrying an overt mark. And re-interpretation of such a 

possessor as an agent would automatically bring about ergative case-marking. 

Thus a sentence of the general form To me / Of me / At me (is) a window broken, 

on being re-interpreted to mean ‘I have broken a window’, would yield a typical 

Type B pattern, with the agent overtly marked, the patient unmarked, the verb 

agreeing with the patient in number and gender but not in person, the verb not 

agreeing with the agent at all, and the whole thing confined to the perfect (Trask 

1979, 398). 

320

Trask’s work is early in the rather extensive literature on ergativity (see Dixon 1994 for a 

general treatment), but it does capture a basic distinction between ergative cases that 

derive from the agentive adjunct of a passive verb and ergative cases that derive from 

other, possessive postpositions, while the agent of the verb in question is either 

impersonal or unexpressed. At a fairly rough level of description, Trask’s type B 

ergativity explains the development of ergativity in Sumerian in that, as discussed in 

chapter 1, one of the characteristic features of low source applicative constructions is that 

they make use of possessor raising in the form of clausal possessive constructions in 

which the possessor bears a dative or locative case and is also an argument of the verb. 

Clearly such a configuration is very close to the one described by Trask and would be 

ripe for grammaticalization as a ergative case. 

While Trask’s model is descriptively adequate to a certain degree, a more precise 

and detailed explanation can be formulated on the basis of Garrett’s discussion of the 

development of ergative case-marking in Hittite (Garrett 1990). Like Trask, Garrett has 

also argued that there are at least two primary ways in which ergative case-marking 

develops in diachronic terms. Whereas Trask differentiated between obligatory 

passivization (Type A) and the grammaticalization of adpositionally-marked possessors 

(Type B), Garrett juxtaposes cases in which tense/aspect oppositions in conjunction with 

passivization have yielded ergative case-marking (= Trask’s Type A), and cases in which 

noun phrase animacy contrasts (known as “inherent lexical content of noun phrases,” in 

Silverstein’s work) interacts with instrumental case-marking patterns in impersonal 

causative constructions to produce ergative case-marking, a pattern that I will term “Type 

321

C” ergativity. Whether Type C ergativity actually corresponds to Type B ergativity to 

some degree is actually an interesting question in light of Pylkkänen’s low source 

applicative hypothesis: since a low source applicative is equivalent in certain semantic 

respects to an ablative postposition and necessarily involves a possessive relation, there 

might be the possibility of an encompassing theory, but I will not attempt to come up 

with such a theory here (for Garrett’s argument against relating the Hittite possessive 

adjective and the ablative endings, see Garrett 1990, 272-273). 

Coghill and Deutscher (2002) have recently proposed that ergativity arises in 

Sumerian through a Type A process in which demoted agents, i.e., agentive adjuncts, 

were marked by the locative-terminative postposition in passive clauses and that the 

locative-terminative postposition was subsequently reinterpreted as an ergative 

postposition. They then go on to argue that the “inverse” 38 character of the pronominal 

agreement markers on the verb in the perfective aspect derives from a reinterpretation of 

the locative-terminative as an ergative case-marking postposition. But there are several 

problematic components of their proposal that I would like to clarify, and, once the 

problematic items are identified, I would like to propose an alternative account that relies 

on the Type C grammaticalization process outlined above. The most serious problem 

with Coghill and Deutscher’s proposal is their interpretation of the locative-terminative 

postposition as regularly corresponding to a demoted agent. As Coghill and Deutscher 

freely admit (2002, 282-283), the use of the locative-terminative postposition to refer to a 

(demoted) agent is only attested in causative constructions in which the noun phrase 

38 Note that Coghill and Deutscher are using “inverse” in a descriptive way that has no connection to what is known as 

better known inverse case-marking, see Van Valin and LaPolla 1997, 288-289, 373-376. 

322

earing the locative-terminative is the causee, a morphosyntactic environment that is 

much more reminiscent of both Garrett’s Type C ergativity and Zólyomi’s reconstruction 

of the history of four-participant verbs (Zólyomi 1999, 244-251) than the Type A process 

that Coghill and Deutscher propose. There are no attestations of the locative-terminative 

postposition marking a demoted agent anywhere else in the language (leaving open, for 

the moment, the interpretation of the mes.an.ne 2.pa 3.da construction below) and the most 

common uses of the locative-terminative postposition are (a) as the indicator of a goal in 

certain morphosyntactic environments and (b) as marker of demoted patient in certain 

compound verb constructions. In Michalowski’s recent study of “Enmebaragesi” 

(Michalowski 2003), he identifies a construction that can be exemplified in contrast with 

the mes.an.ne 2.pa 3.da construction as follows (both examples are personal names). 

(73) ME.bara 2.ge 4.e.si 

ißib-Ø barag-e si-Ø 

priest-Nom/Abs dias-LocTerm fill-ActivePrt 

“the priest who fills the dias” 

(74) mes.an.ne 2.pa 3.da 

mes an-e pad-a 

youth An-LocTerm call-PerfPrt 

“the youth called by An” 

323

Here in the realm of participial constructions and nominalized verbs, we do find the kind 

of system of demoted agents that Coghill and Deutscher imagine. In (73), the locative- 

terminative marks either the original goal argument (the most likely interpretation), or 

represents the demoted patient within an antipassive rection (Michalowski 2003, 200; cf. 

Johnson 2000 and Rubio forthcoming, §3.13.2), whereas in (74), the locative-terminative 

marks what might be termed a demoted agent. The contrast in meaning between the 

locative-terminative in (73) and (74) seems to be determined by whether the participle at 

the end of each phrase is active, as in (73), or passive (and perfective), as in (74): this 

type of tense/aspect split, in which a locative adpositional phrase acts as agentive adjunct 

with a transitive perfective participle would correspond precisely to the examples of a 

tense/aspect split from Middle Indic and Iranian as presented by Garrett (1990, 263-264) 

except for the fact that there appears to be no evidence that these perfective/passive 

participial constructions were ever reanalyzed as active finite verbs in Sumerian (one of 

the primary requirements for a subsequent reinterpretation of the agentive adjunct as 

ergative). The opposition between (73) and (74), in other words, would have provided an 

ideal environment for the type of tense/aspect-driven split ergativity envisioned by 

Coghill and Deutscher, but there is no evidence that such a grammaticalization ever took 

place. 

Within a Type C model of the development of ergativity as outlined by Garrett for 

Hittite, there are three necessary components of the morphosyntactic environment in 

which ergativity arises: (a) a transitive verb, (b) a null (or impersonal) agent, and (c) an 

324

inanimate noun in an ablative/instrumental case. Garrett describes the morphosyntactic 

environment as well as its justification as a locus of change as follows: 

According to this scheme, instrumental NPs in null-subject transitive clauses . . . 

are reanalyzed diachronically as subjects. Such a reanalysis is permitted because 

of the functional overlap between instrument and agent in clauses with transitive 

predicates: cf. e.g. John opened the door with the key and the key opened the 

door, or John extinguished the fire with water and water extinguished the fire. A 

similar reanalysis cannot occur in clauses with intransitive predicates because 

thematic instruments are rare or absent altogether in the subject position of 

intransitive clauses. (Garrett 1990, 265) 

I would argue that that [+locus] component of the ablative/instrumental case that 

Garrett’s model calls for is coded by the old locative postposition, *-e, whereas the 

directionality of the ablative/instrumental case is coded by the low source applicative, 

*bi-. 

Given these constraints, it may be useful at this point to imagine two templates for 

such an environment that differ only in terms of the (in)alienability of the bare noun. 

(75) NOUN animate-e (GOAL) NOUN inanimate-e (SOURCE) NOUN inalienable bi-√ 

(76) NOUN animate-e (GOAL) NOUN inanimate-e (SOURCE) NOUN alienable bi-√ 

In the old system, in which both perceived object (SOURCE) and perceiver (GOAL) 

appeared in the same old locative case, *-e, the transfer of possession model was fairly 

325

straightforward: the perception began at one location and ended up at another 

location—both locations appearing in the same case, *-e. Since the two locations were 

indicated, as locations, by the same postpositional morphology, namely *-e, the thematic 

roles of the two locations must have been assigned by some other component of the 

clause. This is where the *bi-, *ba-, *ni-, *na- system would have stepped in: the *bi- and 

*ba- prefixes may have identified the topical (or, in an obviational system [see Aissen 

1997 for a recent overview], the proximal) locative noun as a source, whereas the *ni- 

and *na- prefixes identified the topical or proximal locative noun as a goal. Such a 

scenario would also explain the fact that in later phases of the language *ba- indicated 

movement away from speaker (if the proximal locative noun—proximal to speaker that 

is—is a source, then the corresponding goal will necessarily represent movement away 

from speaker), whereas the derived form *imma-√ indicates motion toward speaker. As 

predicted by Nichols’ diachronic model (see above), the BNBV inal construction seems to 

be the only part of the grammar in which the full system is preserved: the proximal 

locative, gig.e in the paradigmatic example, is identified as the source of the perception, 

while the non-proximal locative, en aratta ki .ke 4, by implication, must be the goal of the 

perception. I would imagine that the *bi-, *ba-, *ni-, *na- system was formed through 

grammaticalization of the set A deictics as low applicative heads, which originated as 

auxiliary verbs in clause final position. 

326

*bi- SOURCE, INDEFINITE (“There is a … in …”) 

*ba- SOURCE, DEFINITE (“The … is there”) 

*ni- GOAL, INDEFINITE (“Here is a … for …) 

*na- GOAL, DEFINITE (“The … is here”) 

But, for whatever reason, the transfer of possession model began to break down and, as a 

result, the two templates sketched out above, (75) and (76), began to diverge. 

The BNBV inal predicate in (75) presumably maintained the possessive relation 

between NOUN animate-e (GOAL) and the NOUN inalienable due to the requirement that an 

inalienable noun must have a possessor. So the possessive meaning of the old locative 

(GOAL) postposition in clauses such as (75) was presumably fairly stable due to its 

inalienability. The template in (76), however, is another story entirely: with the 

dissolution of the transfer of possession model in those cases in which the bare alienable 

noun was indefinite (as in the alienable BNBV construction in [76]), the low source 

applicative that had previously functioned as an applicative morpheme became causative 

and the two locative arguments of the applicative were reinterpreted as participants in a 

causative construction. The old locative goal argument was reinterpreted as causer 

(presumably due to the semantic overlap between possessors and ergators), while the old 

locative source argument was reinterpreted as causee due to the similarity between the 

privative and adversely affected semantics of the source argument (BNBV inal) and the 

characterization of a causee as the coerced or controlled participant in some situation, or 

alternatively the association between a source or ablative thematic relation and 

327

instrumentality. In other words, the semantics associated with the source argument 

bearing the old locative postposition, *-e, split into two subfields on the basis of the 

inalienability of the noun: the source argument in BNBV inal predicates was associated 

with deprivation and adversity, whereas the source argument in alienable BNBV 

predicates was identified as coerced and/or instrumental. 

Once the split between causative *bi-√ and applicative *bi-√ constructions (on the 

basis of the (in)alienability of the bare noun in the construction) had taken place, the 

subsequent reshuffling of functions assigned to the old locative postposition set the stage 

for the Type C grammaticalization suggested by Garrett. In a causative construction with 

null or impersonal causer and a causee in the equivalent of an ablative/instrumental case 

(due to the co-occurrence of the old locative postposition, *-e, and the low source 

applicative, *bi-), the resulting configuration would look something like the following. 

(77) Ø animate.causer NP inanimate-e causee NP alienable bi-Ø-√ 

Since the inanimate causee is very similar to the animate causer in that they both share 

the same morphological form with old locative postposition and nearly the same function, 

the reinterpretation of the postposition that marks the causee as an ergative case-marking 

on the basis of these similarities is in conformity with the Type C model described by 

Garrett and it also incorporates certain components of the Coghill/Deutscher hypothesis. 

Thus, any noun bearing the old locative postposition in a transitive clause with an 

alienable noun would—via two distinct grammaticalization pathways—have been 

328

einterpreted as an ergative postposition. At the same time, the low transitivity and 

coreferentiality of possessor/ergator and bare noun in BNBV inal constructions would have 

blocked the development of ergativity in BNBV inal clauses. The three possible functions 

of the old locative postposition can, therefore, be represented as follows: 

Form Environment Function 

*NP animate/inanimate-e BNBV alienable ergative 

*NP inanimate-e BNBV inal ablative 

*NP animate-e BNBV inal possessor/allative 

But a second component of the Coghill/Deutscher hypothesis also seems to play a role in 

the formation of the Sumerian preverbal agreement system. 

One of the most interesting components of the Coghill/Deutscher hypothesis is their 

suggestion that the preverbal pronominal markers of ergative case in ˙amt¬u clauses 

originate from the *bi- and *ni- verbal prefixes in verbs in which the original preverbal 

pronominal element is absent. The suggestion that verbs of the form …n-√gu 7 might 

derive from an underlying form such as …ni-Ø-√gu 7 was first proposed by Krecher 

(1979) and subsequently adopted by a number of authors (see references in Coghill and 

Deutscher 2002, 285, fn. 52). Coghill and Deutscher argue that the form that has 

undergone vowel syncope was subsequently reanalyzed as an ergative agreement marker, 

but their argument actually makes much more sense in a Type C model of the 

development of ergativity, particularly if we assume the null or impersonal agent of the 

329

causative environment described by Garrett would have been represented by a zero 

element in the immediately preverbal slot. If this scenario is valid, then we have to 

acknowledge the possibility that two, homophonous systems of preverbal “agreement” 

may either coexist within the Sumerian of the text-artifactual record, or at least in one of 

the historical phases that immediately precedes the attested documentation. If so, then we 

should expect that BNBV inal predicates would preserve the older system of pronominal 

agreement—whatever that might be (for my suggestion that it was a switch-reference 

system, see the discussion following example [31] in section 3.3). At the same time, 

causative alienable BNBV predicates would exhibit a distinct system of agreement 

involving some form of agreement with the causee. Since this is not, however, a study of 

the alienable BNBV class of predicates, I do not go into the matter further here, but it 

should be emphasized that such a historical scenario would lead us to expect that the 

preverbal pronominal agreement markers may well function quite differently depending 

on the lexical class of the main verb and the morphosyntactic environment in which they 

occur. 

330

5.3 Conclusions and prospects 

Of the many hypotheses, proposals and suggestions included in this dissertation, two 

strike me as particularly important: (a) the interpretation of the *bi-√ verbal prefix in the 

BNBV inal construction as a (low source) applicative in that it incorporates and extends 

previous work on the directive case hypothesis in a model that allows for typological 

comparison, and (b) the opposition between topic-marking using the possessive pronoun 

topic-marking construction (*NP-bi) and the role of focus in the *XP nam bi-√ 

construction. The further extension of recent work on applicative typology and 

interpretation (particularly within the Distributed Morphology framework) to some of the 

other recalcitrant verbal prefixes in Sumerian as well as an investigation of the 

quantificational properties of the different verbal prefixes would, in my view, be quite 

fruitful and I hope to continue with the investigation of the some of the other classes of 

the *bi-√ prefix and other related prefixes in future. But it is the second of the two 

conclusions, namely the opposition between topic and focus in Sumerian, that deserves 

further attention and research by all those who work on Sumerian morphosyntax and the 

literature that it encodes. 

The most productive avenue to substantive progress in our understanding of 

Sumerian morphosyntax is undoubtedly further investigation of topic and focus in. In 

particular, the connections between possession, definiteness and topicalization, which I 

have attempted to elucidate in this study, are in need of a great deal of additional 

research, particularly if such research manages to identify further connections between 

these features and certain components of the verbal prefix. At the same time, additional 

331

work on bare nouns, indefiniteness and head-internal relativization will undoubtedly shed 

light on Sumerian forms that have resisted coherent interpretation for decades. Thus the 

prospects for an advance in our understanding of Sumerian morphosyntax and 

consequently the language itself are auspicious and I hope to be able to continue working 

on these problems in future. 

332

References 

Abbot, Barbara. 1992. Definiteness, existentials, and the ‘list’ interpretation. In Chris 

Barker and David Dowty, SALT II, pp. 1-16. Columbus, Ohio: Ohio State University. 

––––––. 1993. A pragmatic account of the definiteness effect in existential sentences. 

Journal of Pragmatics 19: 39-55. 

––––––. 1997. Definiteness and existentials. Language 73: 103-108. 

Agha, Asif. 1993. Structural Form and Utterance Context in Lhasa Tibetan: Grammar 

and Indexicality in a Non-configurational Language. Monographs in Linguistics and 

the Philosophy of Language 2. New York: Peter Lang. 

––––––. 1996. Schema and Superposition in Spatial Deixis. Anthropological Linguistics 

38(4): 643-682. 

Aissen, Judith. 1997. On the syntax of obviation. Language 73: 705-750. 

Aldridge, Edith. 2004. Internally Headed Relative Clauses in Austronesian Languages. 

Language and Linguistics 5(1): 99-129. 

Al-Fouadi, Abdul-Hadi A. 1969. Enki's Journey to Nippur: The Journeys of the Gods… 

University of Pennsylvania. 

Alster, Bendt. 1972. Dumuzi’s Dream: Aspects of Oral Poetry in a Sumerian Myth. 

Mesopotamia: Copenhagen Studies in Assyriology 1. Copenhagen: Akademisk 

Forlag. 

––––––. 1974. The Instructions of Íuruppak. Mesopotamia: Copenhagen Studies in 

Assyriology 10. Copenhagen: Akademisk Forlag. 

––––––. 1997. Proverbs of Ancient Sumer. Bethesda, Maryland: CDL Press. 

––––––. 2002. Relative clauses and case relations in Sumerian. Wiener Zeitschrift für die 

Kunde des Morgenlandes 92: 7-31. 

––––––. Forthcoming. New Sources for Dumuzi’s Dream. 

Attinger, Pascal. 1993. Eléments de linguistique sumérienne: La construction de du 11/e/di 

. Göttingen: Universitätsverlag, Freiburg, Vandenhoeck and Ruprecht. 

Austin, John. 1975 [1962]. How to Do Things with Words. Harvard University Press. 

333

Baker, Mark. 1988. Incorporation: A Theory of Grammatical Function Changing. 

Chicago: University of Chicago Press. 

Basilico, David. 2003. The Topic of Small Clauses. Linguistic Inquiry 34(1): 1-35. 

Bauer, Josef, Robert K. Englund and Manfred Krebernik. 1998. Mesopotamien: 

Späturuk-Zeit und Frühdynastische Zeit. OBO 160/1. Göttingen: Vandenhoeck and 

Ruprecht. 

Black, Jeremy. 1998. Reading Sumerian Poetry. Cornell University Press. 

Borer, Hagit, and Josef Grodzinksy. 1986. Lexical cliticization vs. syntactic cliticization. 

In Hagit Borer, ed., The Syntax of Pronominal Clitics. Syntax and Semantics 19. 

New York: Academic Press. 

Brisch, Nicole. 2003. Tradition and the Poetics of Innovation: Sumerian Court Literature 

of the Larsa Dynasty (2003-1763 BCE). Unpublished dissertation, University of 

Michigan. 

Carlson, Gregory. 1977. Reference to Kinds in English. Unpublished dissertation, 

University of California. 

Castellino, G. 1957. Urnammu: Three Religious Texts. Zeitschrift für Assyriologie 52: 

1-57. 

––––––. 1959. Urnammu: Three Religious Texts (continued). Zeitschrift für Assyriologie 

53: 106-132. 

Cavigneaux, Antoine. 1987. Notes Sumérologiques. Acta Sumerologica 9: 45-66. 

Cavigneaux, Antoine, and Farouk Al-Rawi. 2000a. Gilgameß et la Mort: Texts de Tell 

Haddad VI, avec un appendice sur les texts funéraires sumériens. Cuneiform 

Monographs 19. Groningen: Styx. 

––––––. 2000b. La fin de Gilgameß, Enkidu et les Enfres d’après les manuscripts d’Ur et 

de Meturan (Textes de Tell Haddad VIII). Iraq 62: 1-19. 

Cavigneaux, Antoine, Hans Güterbock and Martha Roth. 1985. MSL XVII: The Series 

Erim-˙uß = anantu and An-ta-gál = ßaqû. Pontifical Biblical Institute. 

Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris. 

––––––. 1986. Knowledge of Language—Its Nature, Origin and Use. New York: Praeger. 

334

Chung, Sandra, and William A. Ladusaw. 2004. Restriction and Saturation. Linguistic 

Inquiry Monographs 42. MIT Press. 

Chung, Sandra, and James McCloskey. 2001. Existentials at the Interface—NSF Project 

No. BCS-0131767. Unpublished manuscript. 

Cinque, Guglielmo. 1999. Adverbs and Functional Heads. Oxford University Press. 

Civil, Miguel. 1964. The “Message of Lú-dingir-ra to his mother” and a Group of 

Akkado-Hittite “Proverbs”. Journal of Near Eastern Studies 23(1): 1-11. 

––––––. 1971. Izi = ißatu, Ká-gal = abullu, Níg-ga = makkuru. MSL 13. Rome: 

Pontifical Biblical Institute. 

––––––. 1973. From Enki's Headaches to Phonology. Journal of Near Eastern Studies 32: 

57-61. 

––––––. 1983. Enlil and Ninlil: The Marriage of Sud. Journal of the American Oriental 

Society 103: 43-66. 

––––––. 1985. Sur les “livres d’écolier” à l’époque paléo-babylonienne. In Jean-Marie 

Durand and Jean-Robert Kupper, eds., Miscellanea Babyloniaca: Mélanges offerts a 

Maurice Birot, pp. 67-78. Paris: Éditions Recherche sur les Civilisations. 

Clark, Eve. 1978. Locationals: Existential, locative, and possessive constructions. In 

Joseph Greenberg, Charles Ferguson and Edith Moravcsik, eds., Universals of 

Human Language, vol. 4, pp. 85-126. Stanford, California: Stanford University 

Press. 

Coghill, Eleanor and Guy Deutscher. 2002. The origin of ergativity in Sumerian and the 

‘inversion’ in pronominal agreement: a historical explanation based on Neo-Aramaic 

parallels. Orientalia NS 71(3): 267-296. 

Cohen, Mark E. 1981. Sumerian Hymnology: The Erßemma. Cincinnati: Hebrew Union 

College Annual Supplement 2. 

Cohen, Sol. 1973. Enmerkar and the Lord of Aratta. PhD dissertation, University of 

Pennsylvania. 

Cooper, Jerrold. 1978. The Return of Ninurta to Nippur: an-gim dím-ma. Analecta 

Orientalia 52. Biblical Institute Press. 

335

Cooper, Jerrold, and Wolfgang Heimpel. 1983. The Sumerian Sargon Legend. Journal of 

the American Oriental Society 103: 67-82. 

Davidson, Donald. 1967. The logical form of action sentences. In Nicholas Rescher, ed., 

The Logic of Decision and Action. Pittsburgh, Pennsylvania: University of 

Pittsburgh Press. 

Den Dikken, Marcel. 1995. Particles: On the Syntax of Verb-Particle, Triadic, and 

Causative Constructions. Oxford Studies in Comparative Syntax. Oxford University 

Press. 

Deutscher, Guy. 2002. The Akkadian Relative Clause in Cross-Linguisic Perspective. 

Zeitschrift für Assyriologie 92: 86-105. 

Diesing, Molly. 1992. Indefinites. Cambridge: MIT Press. 

Dixon, R. M. W. 1994. Ergativity. Cambridge: Cambridge University Press. 

Dowty, David. 1979. Word Meaning and Montague Grammar. Dordrecht: Kluwer. 

Duranti, Alessandro. 1997. Linguistic Anthropology. Cambridge University Press. 

Edzard, Dietz. 1967. Das sumerische Verbalmorphem /ed/ in den alt- und neu- 

sumerischen Texten. HSAO 1, pp. 29-62. 

––––––. 1971. ˙amt¬u, marû und freie Reduplikation beim sumerischen Verbum. ZA 61: 

208-232. 


1-34. 


45-61. 

––––––. 1990. Gilgamesh und Huwawa A. I. Teil. Zeitschrift für Assyriologie 80: 

165-203 

––––––. 1991. Gilgamesh und Huwawa A. II. Teil. Zeitschrift für Assyriologie 81: 

165-233. 

336

––––––. 1993. “Gilgamesh und Huwawa”: Zwei Versionen der sumerischen 

Zedernwaldepisode nebst einer Edition von Version “B”. Sitzungberichte der 

Bayersichen Akademie der Wissenschaften. Philosophisch-historische Klasse, pp. 

1-61. 

Embick, David. 2004. On the Structure of Resultative Participles in English. Linguistic 

Inquiry 35(2): 355-392. 

Englund, Robert. 1998. Texts from the Late Uruk Period. In P. Attinger and M. Wäfler, 

eds., Mesopotamien: Späturuk-Zeit und Frühdynastische Zeit, pp. 15-233. Göttingen: 

Vandenhoeck and Ruprecht. 

Falkenstein, Adam. 1947. Der “Sohn des Tafelhauses”. Die Welt des Orients 1: 172-186. 

––––––. 1959. Sumerische Götterlieder, I. Teil. Heidelberg: Carl Winter. 

––––––. 1978. Grammatik der Sprache Gudeas von Lagaß, 2 vols. Analecta 

Orientalia 28-29. Rome: Biblical Institute Press. 

Filip, Hana. 2000. The Quantization Puzzle. In J. Pustejovsky and C. Tenny (eds.) Events 

as Grammatical Objects, pp. 3-60. CSLI Press. 

––––––. 2004. Prefixes and the delimitation of events. Journal of Slavic Linguistics 

12(1-2). 

Finkel, Irving. 1982. The Series SIG7.ALAN = Nabnitu. MSL 16. Pontifical Biblical 

Institute. 

Freeze, Ray. 1992. Existentials and other locatives. Language 68: 553-595. 

Gadotti, Alhena. 2004. Enkidu Revived. Handout for paper presented at the 214 th 

American Oriental Society, March 13, 2004. 

Garrett, Andrew. 1990. The Origin of NP Split Ergativity. Language 66(2): 261-296. 

Geenhoven, Veerle. 1998. Semantic Incorporation and Indefinite Descriptions. Stanford: 

Center for the Study of Language and Information. 

George, Andrew. 1999. The Epic of Gilgamesh. New York: Penguin. 

––––––. 2003. The Babylonian Gilgamesh Epic: Introduction, Critical Edition and 

Cuneiform Texts, 2 vols. Oxford: Oxford University Press. 

337

Givón, T. 1984. Syntax: A Functional-Typological Introduction, Volume 1. Amsterdam: 

John Benjamins. 

Gragg, Gene. 1973a. Sumerian Dimensional Infixes. AOATS 5. Neukirchen-Vluyn. 

––––––. 1973b. Linguistics, Method, and Extinct Languages: The Case of Sumerian. 

Orientalia NS 42: 78-96. 

Green, Margaret Whitney. 1975. Eridu in Sumerian Literature. Unpublished dissertation, 

University of Chicago. 

Haji-Abdolhosseini, Mohammad, Diane Massam and Kenji Oda. 2002. Number and 

Events: Verbal Reduplication in Niuean. Oceanic Linguistics 41(2): 475-492. 

Hale, Ken. 1970. The passive and ergative in language change: the Ausralian case. In S. 

Wurm and D. Kaycock, eds., Pacific linguistic studies in honor of Arthur Capell, pp. 

757-81. Canberra: Pacific Linguistics, Series C, no. 13. 

––––––. 1976. The adjoined relative clause in Australia. In R. M. W. Dixon, ed., 

Grammatical Categories in Australian Languages, pp. 78-105. Canberra: Australian 

Institute of Aboriginal Studies. 

Harley, Heidi. 2003. Possession and the double object construction. Yearbook of 

Linguistic Variation 2: 29-68. 

Heimerdinger, Jane. 1979. Sumerian Literary Fragments from Nippur. Occasional 

Publications of the Babylonian Fund 4. Philadelphia, Pennsylvania: The University 

Museum. 

Heimpel, Wolfgang. 1968. Tierbilder in der sumerischen Literatur. Studia Pohl 2. Rome: 

Biblical Institute Press. 

Heine, Bernd. 1997. Possession: Cognitive sources, forces, and grammaticalization. 

Cambridge: Cambridge University Press. 

Herburger, Elena. 2000. What Counts: Focus and Quantification. Cambridge, 

Massachusetts: MIT Press. 

Huehnergard, John. 2000. A Grammar of Akkadian, revised edition. Harvard Semitic 

Studies 45. Atlanta, Georgia: Scholars Press. 

338

Hurowitz, Victor. 1992. Building Stories in Sumerian and Old Babylonian Literature. In I 

Have Built You an Exalted House: Temple Building in the Bible in Light of 

Mespotamian and Northwest Semitic Writings, pp. 32-67. JSOT Supplement Series 

115. Sheffield. 

Jacobsen, Thorkild. 1980. Death in Mesopotamia. In Alster, Bendt, ed., Death in 

Mesopotamia: Papers Read at the XXVIe Rencontre assyriologique internationale, 

pp. 19-24. Copenhagen: Akademisk Forlag. 

––––––. 1987. The Harps That Once . . . Sumerian Poetry in Translation. New Haven: 

Yale University Press. 

Jagersma, Bram. Forthcoming. A Descriptive Grammar of Sumerian. Unpublished 

manuscript. 

Jäger, Gerhard. 2001. Topic-Comment Structure and the Contrast Between Stage Level 

and Individual Level Predicates. Journal of Semantics 18: 83-126. 

Johnson, Cale. 2000. Evidence of Antipassivization in Sumerian. The Bulletin of the 

International Institute for Linguistic Sciences, Kyoto Sangyo University 21: 205-240. 

Julien, Marit. 2002. Syntactic Heads and Word Formation. Oxford: Oxford University 

Press. 

Kaiser, Stafan, Yasuko Ichikawa, Noriko Kobayashi, and Hilofumi Yamamoto. 2001. 

Japanese: A Comprehensive Grammar. Routledge. 

Kapp, A. 1955. Ein Lied auf Enlilbani von Isin. Zeitschrift für Assyriologie 51: 76-87. 

Karahashi, Fumi. 2000. Sumerian Compound Verbs with Body-Part Terms. PhD 

dissertation, University of Chicago. 

––––––. 2003. Towards a Typology of Compound Verbs. Paper presented at AOS. 

––––––. Forthcoming. The Locative-Terminative Verbal Infix in Sumerian. Acta 

Sumerologica 22. 

Kayne, Richard. 1994. The Antisymmetry of Syntax. MIT Press. 

Kienast, Burkhart. 2001. Historische semitische Sprachwissenschaft. Wiesbaden: 

Harrassowitz. 

339

Kim, Ji-yung. 2003. Verb Reduplication in Mandarin. Unpublished Manuscript, 

Unversity of Massachusetts Amherst. 

Kim, Min-Joo. 2003. Internally Headed Relatives Parallel Direct Perception 

Complements. Manuscript, University of Massachusetts-Amherst. 

Kiparsky, Paul, and Carol Kiparsky. 1970. Fact. In M. Bierwisch and K. E. Heidolph, 

eds., Progress in Linguistics: A Collection of Papers, pp. 143-173. The Hague: 

Mouton. 

Kiss, Katalin. 1998. Identificational focus versus informational focus. Language 74(2): 

245-273. 

Klein, Jacob. 1989. Building and Dedication Hymns in Sumerian Literature. Acta 

Sumerologica 11: 27-67. 

––––––. 1990a. Shulgi and Ißmedagan: Originality and Dependency in Sumerian 

Royal Hymnology. In Jacob Klein and Aaron Skaist, eds., Bar-Ilan Studies in 

Assyriology Dedicated to Pinhas Artzi. Bar-Ilan Studies in Near Eastern Languages 

and Culture. Ramat-Gan: Bar-Ilan University Press. 

––––––. 1990b. The “Bane” of Humanity: A Lifespan of One Hundred Twenty Years. 

Acta Sumerologica 12: 57-70. 

Koopman, Hilda, and Anna Szabolcsi. 2000. Verbal Complexes. MIT Press. 

Kowaki, Mitsuo. 1982. bunpo tekisuto-ni okeru siekikoubun-ni tuite (“On the Sumerian 

causative construction in the Old Babylonian Grammatical Texts”). Bulletin of the 

Faculty of Education Hiroshima University 2, no. 31, pp. 25-31. 

Kramer, Samuel Noah. 1940. Lamentation Over the Destruction of Ur. Assyriological 

Studies 12. Chicago: Chicago University Press. 

––––––. 1944. The Death of Gilgamesh. Bulletin of the American Schools of 

Oriental Research 94: 2-12. 

Krecher, Joachim. 1979. Zu einigen Ausdrücken der neusumerischen Urkundensprache. 


––––––. 1985. Die /m/-Präfixe des sumerischen Verbums. Orientalia 54(1-2): 

133-181. 

340

––––––. 1987. Morphemeless Syntax in Sumerian as Seen on the Background of 

Word-Composition in Chukchee. Acta Sumerologica 9: 67-88. 

––––––. 1993. The Suffix of Determination -/a/. Acta Sumerologica 15: 81-98. 

––––––. 1994. Über einige ‘zusamengesetzte Verben’ im Sumerischen. In A. F. Rainey, 

ed., kinattu¢tu ßa da¢râti: Raphael Kutsche Memorial Volume, pp. 107-118. Tel Aviv 

University Institute of Archaeology. 

––––––. 1995. Die marû-Formen des sumerischen Verbums. In M. Dietrich and O. 

Loretz, eds., Vom Alten Orient zum Alten Testament: Festschrift für Wolfram 

Freiherrn von Soden zum 85. Geburtstag am 19. Juni 1993, pp. 141-200. 

Krifka, Manfred. 1989. Nominalreferenz und Zeitkonstitution: Zur Semantik von 

Massentermen, Individualtermen, Aspektklassen. München. 

Kubo. Miori. 1992. Japanese Passives. Working Papers of the Department of Languages 

and Cultures 23. University of Hokkaido. 

Kuno, Susumo. 1971. The position of locatives in existential sentences. Linguistic 

Inquiry 11: 333-378. 

Landau, Idan. 1999. Possessor raising and the structure of the VP. Lingua 107: 1-37. 

Larson, Richard. 1988. On the Double Object Construction. Linguistic Inquiry 19(3): 

335-391. 

––––––. 2004. Sentence-final adverbs and “scope”. In M. Wolf and K. Moulton (eds.), 

Proceedings of NELS 34. UMASS: GLSA. 

Lee, Benjamin. 1997. Talking Heads: Language, Metalanguage, and the Semiotics of 

Subjectivity. Durham: Duke University Press. 

Levinson, Bernard. 1997. Deuteronomy and the Hermeneutics of Legal Innovation. 

Oxford: Oxford University Press. 

Li, Charles, and Sandra Thompson. 1981. Mandarin Chinese: A Functional Reference 

Grammar. Berkeley: University of California Press. 

Lipinski, Edward. 2001. Semitic languages: Outline of a comparative grammar, second 

edition. Leuven: Peeters. 

341

Loprieno, Antonio. 1991. Focus, Mood, and the Negative forms: Towards a History of 

the Formation of Middle Egyptian Syntactic Paradigms. In Proceedings of the 

Second International Conference on Egyptian Linguistics (Crossroads II). LingAeg 

1: 201-226. 

––––––. 1995. Ancient Egyptian: A Linguistic Introduction. Cambridge University Press. 

Ludwig, Marie-Christine. 1990. Untersuchungen zu den Hymnen des Ißme-Dagan von 

Isin. Wiesbaden: Harrassowitz. 

Lumsden, Michael. 1988. Existential Sentences: Their Structure and Meaning. 

Beckenham: Croom Helm. 

Lyons, Christopher. 1999. Definiteness. Cambridge Textbooks in Linguistics. 

Cambridge: Cambridge University Press. 

Lyons, John. 1967. A note on possessive, existential, and locative sentences. Foundations 

of Language 3: 390-396. 

––––––. 1968. Existence, location, possession, and transitivity. In B. van Rootselaar and 

T. F. Staal, eds., Logic, Methodology, and Philosophy of Science 3, pp. 495-509. 

Marantz, Alec. 1998. No Escape from Syntax: Don’t Try Morphological Analysis in the 

Privacy of Your Own Lexicon. In Alexis Dimitriadis, ed., Proceedings of the 1998 

Penn Linguistic Colloquium. University of Pennsylvania Working Papers in 

Linguistics 4(2): 201-224. 

McNally, Louise. 1992. An interpretation for the English existential construction. PhD 

dissertation, University of California, Santa Cruz. 

Michalowski, Piotr. 1980a. Sumerian as an Ergative Language I. Journal of Cuneiform 

Studies 32: 86-103. 

––––––. 1980b. Adapa and the Ritual Process. Pocznik Orientalistyczny 41: 77-82. 

––––––. 1989. The Lamentation over the Destruction of Sumer and Ur. Mesopotamian 

Civilizations 1. Winona Lake, Indiana: Eisenbrauns. 

––––––. 1993. The Torch and the Censer. In M. Cohen et al., eds., The Tablet and the 

Scroll: Near Eastern Studies in Honor of William W. Hallo. Bethesda, Maryland: 

CDL Press. 

342

––––––. 2003. A Man Called Enmebaragesi. In W. Sallaberger et al., eds., Literatur, 

Politik und Recht in Mesopotamien: Festschrift für Claus Wilcke, pp. 195-208. 

Wiesbaden: Harrassowtiz. 

––––––. 2004. Sumerian. In R. Woodard, ed., The Cambridge Encyclopedia of the 

World’s Ancient Languages, pp. 19-59. Cambridge: Cambridge University Press. 

––––––. Forthcoming. The Life and Death of the Sumerian Language in Comparative 

Perspective. Acta Sumerologica 22. 

Milsark, Gary. 1974. Existential Sentences in English. PhD dissertation, MIT. 

Molina, Manuel, and Marcos Such-Gutiérrez. 2004. On Terms for Cutting Plants and 

Noses in Ancient Sumer. Journal of Near Eastern Studies 63(1): 1-16. 

Newman, Paul. 1990. Nominal and Verbal Plurality in Chadic. Dordrecht: Foris 

Publications. 

––––––. 2000. The Hausa Language: An Encyclopedic Reference Grammar. New Haven: 

Yale University Press. 

Nichols, Johanna. 1984. Functional Theories of Grammar. Annual Review of 

Anthropology 13: 97-117. 

––––––. 1986. Head-marking and dependent-marking grammar. Language 62: 

56-119. 

––––––. 1988. On alienable and inalienable possession. In: Shipley, William ed., In 

William Shipley, ed., In Honor of Mary Haas: From the Haas Festival Conference 

on Native American Linguistics. New York: Mouton de Gruyter. 

Nishigauchi, Taisuke. Forthcoming. Head-Internal Relative Clauses in Japanese and the 

Interpretation of Indefinite NPs. Manuscript, Dobe Shoin Graduate School. 

Pesetsky, David. 1995. Zero Syntax: Experiencers and Cascades. MIT Press. 

Poebel, Arno. 1933. Eine sumerische Inschrfit Samsuilunas über die Erbauung der 

Festung Dur-Samsuiluna. Archiv für Orientforschung 9: 241-292. 

Postgate, Nicholas. 1974. Two Points of Grammar in Gudea. Journal of Cuneiform 

Studies 26: 16-54. 

343

Prince, Ellen. 1992. The ZPG Letter: Subjects, Definiteness, and Information-Status. In 

Sandra Thompson and William Mann, eds., Discourse Description: Diverse 

Analyses of a Fundraising Text, pp. 295-325. Amsterdam: John Benjamins. 

Pylkkänen, Mariliina. 2002. Introducing Arguments. PhD dissertation, MIT. 

Reuland, Eric, and Alice ter Meulen, eds. 1987. The Representation of (In)definiteness. 

Cambridge, Massachusetts: MIT Press. 

Rischel, J. 1971. Some Characteristics of Noun Phrases in Greenlandic. In F. Kiefer, ed., 

Derivational Processes. Stockholm: Research Group for Quantitative Linguistics. 

Rizzi, Luigi. 1997. On the fine structure of the left-periphery. In L. Haegeman, ed., 

Elements of Grammar, pp. 281-337. Dordrecht: Kluwer Academic Publishers. 

Römer, W. H. Ph. 1965. Sumerische ‘Königshymnen’ der Isin-Zeit. Leiden: Brill. 

––––––. 2004. Die Klage über die Zerstörung von Ur. AOAT 309. Ugarit-Verlag. 

Rosaldo, Michelle Z. 1982. The Things We Do With Words: Ilongot Speech Acts and 

Speech Act Theory in Philosophy. Language in Society 11: 203-237. 

Rothstein, Susan. 2004. Structuring Events: A Study in the Semantics of Lexical Aspect. 

Blackwell. 

Rubio, Gonzalo. Forthcoming. A Sketch of Sumerian Morphology. 

Sadock, Jerrold. 1991. Autolexical Syntax: A Theory of Parallel Grammatical 

Representations. Chicago: University of Chicago Press. 

Saeed, J.I. 2003. Semantics, second edition. Blackwell. 

Safir, Kenneth. 1985. Syntactic Chains. Cambridge: Cambridge University Press. 

Sapir, E. 1922. The Takelma Language of Southwestern Oregon. In F. Boas, ed., 

Handbook of American Indian Languages, vol. 2, pp. 1-296. Washington D.C.: 

Bureau of American Ethnology, Bulletin 40, part 2. 

Sawada, Miyuki and Richard Larson. 2004. Presupposition and root transforms in adjunct 

clauses. In M. Wolf and K. Moulton (eds.), Proceedings of NELS 34. UMASS: 

GLSA. 

Searle, John R. 1969. Speech Acts: An Essay in the Philosophy of Language. Cambridge: 

Cambridge University Press. 

344

Shaffer, Aaron. 1963. Sumerian Sources of the Tablet XII of the Epic of Gilgamesh. PhD 

dissertation, University of Pennsylvania. 

Shlonsky, Ur. 2004. The form of Semitic noun phrases. Lingua 114: 1465-1526. 

Silverstein, Michael. 1976. Hierarchy of features and ergativity. In R. M. W. Dixon, ed., 

Grammatical Categories in Australian Languages, pp. 112-171. Canberra: 

Australian Institute of Aboriginal Studies. 

––––––. 1981. The Limits of Awareness. Working Papers in Sociolinguistics 84. Austin: 

Southwestern Educational Laboratory. [Reprinted in A. Duranti (ed.), Linguistic 

Anthropology: A Reader, pp. 382-401. Blackwell.] 

––––––. 1993a. Of Nominatives and Datives: Universal Grammar From the Bottom Up. 

In Robert Van Valin, ed., Advances in Role and Reference Grammar, 465-498. 

Amsterdam: Benjamins. 

––––––. 1993b. Metapragmatic Discourse and Metapragmatic Function. In J. Lucy, ed., 

Reflexive Language, pp. 33-58. New York: Cambridge University Press. 

Sjöberg, Åke. 1961. Ein syllabisch geschriebener Ur-Nammu-Text. Orientalia Suecana 

10: 3-12. 

––––––. 1973. Der Vater und sein missratener Sohn. Journal of Cuneiform Studies 

25: 105-169. 

––––––. 1975. in-nin ßà-gur 4-ra: A Hymn to the Goddess Inanna by the en-Priestess 

Enheduana. Zeitschrift für Assyriologie 65: 161-253. 

––––––. 1977. A blessing of King Urninurta. In M. Ellis, ed., Essays on the Ancient Near 

East in Memory of Jacob Joel Finkelstein, pp. 189-195. Memoirs of the Connecticut 

Academy of Sciences. Hamden, Connecticut: Archon Books. 

Sjöberg, Ake, and E. Bergmann. 1969. The Collection of the Sumerian Temple Hymns. 

Locust Valley, New York: J. J. Augustin. 

Sladek, William. 1974. Inanna’s Descent to the Netherworld. PhD dissertation, Johns 

Hopkins University. 

Stowell, Tim. 1981. The Origins of Phrase Structure. PhD Dissertation, MIT. 

––––––. 1983. Subjects Across Categories. The Linguistic Review 2(3): 285-312. 

345

––––––. 1991. The Alignment of Arguments in Adjective Phrases. In S. D. Rothstein, ed., 

Syntax and Perspectives on Phrase Structure: Heads and Licensing, pp. 105-135. 

Semantics 25. San Diego: Academic Press. 

––––––. 1995. Remarks on clause structure. In Anna Cardinaletti and Maria T. Guasti 

(eds.), Syntax and Semantics 28, 77-107. San Diego: Academic Press. 

Streck, Michael. 1998. The Tense Systems in the Sumerian-Akkadian Linguisic Area. 

Acta Sumerologica 20: 181-199. 

––––––. 1999. Das “Perfekt” iptaras im Altbabylonischen der Hammurapi-Briefe. In 

Nebes, Norbert (ed.), Tempus und Aspekt in den semitischen Sprachen: Jenaer 

Kolloquium zur semitischen Sprachwissenschaft, pp. 101-126. Harrassowitz. 

––––––. 2003. Die akkadischen Verbalstämme mit ta-Infix. AOAT 303. Münster: Ugarit- 

Verlag. 

Szabolcsi, Anna. 1986. Indefinites in Complex Predicates. Theoretical Linguistic 

Research 2: 47-83. 

Tenny, Carol. 2000. Core events and adverbial modification. In C. Tenny and J. 

Pustejovsky (eds.), Events as Grammatical Objects: The Converging Perspectives of 

Lexical Semantics and Syntax, pp. 285-334. Stanford, California: CSLI Publications. 

Thomsen, Marie-Louise. 2001 [1984]. The Sumerian Language. Copenhagen: Akademisk 

Forlag. 

Tinney, Steve. 1996. The Nippur Lament. Occasional Publications of the Samuel Noah 

Kramer Fund 16. The University Of Pennsylvania Museum: Philadelphia. 

Trask, R. L. 1979. On the origins of ergativity. In F. Plank, ed., Ergativity: Towards a 

theory of grammatical relations, pp. 385-404. New York: Academic Press. 

Van Dijk, J. 1967. VAT 8382: Ein zweisprachiges Königsritual. In Heidelberger Studien 

zum Alten Orient: Adam Falkenstein zum 17. September 1966, pp. 233-268. 

Wiesbaden: Harrassowtiz. 

––––––. 1983. Lugal ud me-lám-bi nir-gál: La récit épique et didactique des 

Travaux de Ninurta du Déluge et de la nouvelle Création, Texte, traduction et 

introduction, 2 vols. Leiden: Brill. 

346

Vanstiphout, Herman. 1985. On the Verbal Prefix /i/ in Standard Sumerian. Revue 

d’Assyriologie 79(1): 1-15. 

––––––. 1997. Sumerian Canonical Compositions. C. Individual Focus. 5. Disputations. 

In William Hallo, ed., The Context of Scripture, I: Canonical Compositions from the 

Biblical World, pp. 575-588. New York: Brill. 

––––––. 2003. Epics of the Sumerian Kings: The Matter of Aratta. Writings from the 

Ancient World 20. Atlanta: Society of Biblical Literature. 

Van Valin, Robert, Jr. 1993. Advances in Role and Reference Grammar. Philadelphia: 

John Benjamins. 

––––––. 2001. An Introduction to Syntax. Cambridge: Cambridge University Press. 

Van Valin, Robert, Jr., and Randy LaPolla. 1997. Syntax: Structure, Meaning and 

Function. Cambridge: Cambridge University Press. 

Veldhuis, Niek. 2001. A Solution of the Dream: A New Interpretation of Bilgames’ 

Death. Journal of Cuneiform Studies 53: 133-148. 

Vendler, Zeno. 1967. Linguistics in Philosophy. Cornell University Press. 

Voskuil, J. 1990. Some Transitivization Processes in Malay. MA thesis, University of 

Leiden. 

Ward, Gregory, and Betty Birner. 1995. Definiteness and the English existential. 

Language 71: 722-742. 

––––––. 1997. Response to Abbot. Language 73: 109-112. 

Wasow, T. 1977. Transformations and the Lexicon. In P. Culicover et al., eds., Formal 

Syntax, pp. 327-360. New York: Academic Press. 

Watanabe, Akira. 1992. Subjacency and s-structure movement of wh-in-situ. Journal of 

East Asian Linguistics 1: 255-291. 

Wilcke, Claus. 1969a. Das Lugalbandaepos. Wiesbaden: Harrassowitz. 

––––––. 1969b. ku-li. Zeitschrift für Assyriologie 59: 65-99. 

347

––––––. 1970. Eine Schicksalsentscheidung für den toten UrNammu. In André Finet, ed., 

Actes de la XVIIe Rencontre Assyriologique Internationale. Université Libre de 

Bruxelles, 30 juin - 4 juillet 1969. Ham-sur-Heure: Comité belge de recherches en 

Mésopotamie. 

Williamson, Janis. 1987. An indefiniteness restriction on relative clauses in Lakhota. In 

E. Reuland and A. G. B. ter Meulen, eds., The Representation of Indefiniteness, pp. 

168-233. Cambridge: MIT Press. 

Woods, Christopher. 2001. The Deictic Foundations of the Sumerian Language. PhD 

Dissertation, Harvard University. 

Yoshikawa, Mamoru. 1968a. On the Grammatical Function of -e- of the Sumerian Verbal 

Suffix -e-dè / -e-da(m). Journal of Near Eastern Studies 27: 251-261. 

––––––. 1968b. The Marû and Óamt≥u Aspects in the Sumerian Verbal System. 


––––––. 1974. The Marû-Conjugation in the Sumerian Verbal System. Orientalia NS 43: 

17-39. 

––––––. 1979. The Sumerian Verbal Prefixes mu-, ì- and Topicality. 


––––––. 1981. Plural Expressions in Sumerian Verbs. Acta Sumerologica 3: 111-124. 

––––––. 1988. Telicity and Momentariness in the Sumerian Verb. 


––––––. 1993. Studies in the Sumerian Verbal System. Acta Sumerologica 

Supplementary Series 1. Middle Eastern Culture Center in Japan. 

Zólyomi, Gábor. 1993. Voice and Topicalization. Unpublished dissertation, Eötvös 

Loránd University. 

––––––. 1996. Review of Eléments de linguistique sumérienne: La construction 

de du 11/e/di by Pascal Attinger. Bibliotheca Orientalis 53(1/2): 95-107. 

––––––. 1999. Directive infix and oblique object in Sumerian: An account of the history 

of their relationship. Orientalia 68: 215-253. 

348

––––––. 2000. About a found donkey and the « local » prefix. Nouvelles assyriologiques 

breves et utilitaires no. 34. 

Zucchi, Alessandro. 1995. The ingredients of the definiteness effect. Natural Language 

Semantics 3: 33-78. 

––––––. 1999. Incomplete events, intensionality and imperfective aspect. 

Natural Language Semantics 7:179-215. 

349

Johnson 2004 - CDLI - UCLA

Create successful ePaper yourself

Delete template?

Save as template?