Чтобы посмотреть этот PDF файл с форматированием и разметкой, скачайте его и откройте на своем компьютере.
The Grammar of Perspective
Cuneiform Monographs
The Grammar of Perspective
The Sumerian Conjugation Pre
xes as a
System of Voice
Christopher Woods
This book is printed on acid-free paper.
Library of Congress Cataloging-in-Publication Data
Woods, Christopher.
The grammar of perspective : the Sumerian conjugation pre
xes as a system of
voice / by Christopher Woods.
p. cm. — (Cuneiform monographs ; v. 31)
Partly based on the author’s dissertation (doctoral—Harvard University).
ISBN 978-90-04-14804-8 (hardcover : alk. paper) 1. Sumerian language—Af
2. Sumerian language—Morphology. 3. Sumerian language—Voices. I. Title.
PJ4019.W66 2008
For Jennie
Preface ......................................................................................... ix
List of Figures ............................................................................. xiii
List of Tables .............................................................................. xv
Abbreviations—Linguistic Terms ............................................... xvii
Abbreviations—Texts Cited ........................................................ xviii
Chapter One Introduction ...................................................... 1
1.1 Prospectus—Voice and Sumerian ................................. 3
3.2.1 Other High Transitivity Events ......................... 118
3.2.2 Low Transitivity Events ..................................... 121
3.3 Contexts in Which
May Replace
........................ 134
3.4 Animacy, Empathy, and the Dative Case (I) ................. 143
This book has its origins in my doctoral dissertation from Harvard
The Deictic Foundations of the Sumerian Language
chapter four, which is concerned with the pre
gests, my primary concern there was with the spatial and directional
associations of the pre
x, although I did discuss what I understood to be
its middle and passive voice functions. Grammatical voice, an abstract
notion, was described in spatial terms and given structure as a function
of location. This was the seed of the current work. Here I incorporate
the other basic conjugation pre
—into this
conceptual framework, understanding their functions as constituting a
grammatical system. Having considered these other pre
xes and the
in April of 2007. I would like to express my deep gratitude to Nicole
Brisch, Jerry Cooper, Bram Jagersma, Fumi Karahashi, Piotr Micha-
lowski, Gonzalo Rubio, Walter Sallaberger, Niek Veldhuis, Konrad
Volk, and Claus Wilcke for their insightful comments and astute sug-
gestions for improvement. These have ranged from pointing out typos
and inconsistencies of transliteration, to providing further evidence
Oriental Institute. In particular, I would like to thank Gertrud Farber
and Walter Farber for their painstakingly close reading of the April
2007 draft, which resulted in many improvements both large and
small. Miguel Civil, Dennis Pardee, Rebecca Hasselbach, Theo van
den Hout, and Matt Stolper read various portions of the manuscript
and provided valuable comments and answered an endless barrage
of questions on topics of all sorts over the years. On a more personal
note, the un
agging support and encouragement of my colleagues have
meant a great deal to me and the writing of this book would have been
an immeasurably more dif
cult task without it.
For their support, enthusiasm, and—importantly—patience, I am
grateful to Mark Geller and Michiel Swormink and the editorial board
of the Cuneiform Monographs series at Brill. I am indebted to Doug
Frayne for sending me a pre-publication version of his
Royal Inscriptions
of Mesopotamia 1: Presargonic Period
and to Gábor Zólyomi for providing
me with a copy of his unpublished dissertation. Likewise, Bram Jagersma
graciously gave me an early draft version of his Sumerian grammar.
Maria Yakubovich and, particularly, Monica Crews did much of the
tedious checking of references and compiling of indices, and I thank
them for all of their hard work.
I would also like to express my gratitude to those colleagues who
have devoted so much of their time and energy to making the textual
evidence from early Mesopotamia accessible to others. This book has
ted greatly from the ability to search the databases compiled and
maintained by the Cuneiform Digital Library Initiative (http://cdli.ucla.
edu/index.html), the Electronic Corpus of Sumerian Literature (http://
It is no exaggeration to state that this book would have been impossible
without her. From consulting on everything from subtleties of Sumerian
grammar to the nuances of English usage, to editing the manuscript
and proofreading a seemingly in
nite parade of drafts, she has had a
hand in all that is contained in these pages. This she has done while
managing her own demanding career and raising our two young sons
Tristan and Sebastian, and all with a grace and sel
essness that are
continually a source of amazement for me. With great love and affec-
tion this book is dedicated to her.
Figure 1. Basic Case-Marking Strategies ................................... 49
Figure 2. Nominal Hierarchy ..................................................... 50
Figure 3. Action-Chain ............................................................... 65
Figure 4. Action-Chain of the Prototypical Transitive Event ... 67
Figure 5. Coincident Structures Relating to the Prototypical
Transitive Event ...................................................................... 68
Figure 6-A. Unmarked View of the Event: Agent primary
focus, Object secondary focus ................................................ 70
Figure 6-B. Marked View of the Event: Object primary focus,
Agent secondary focus ............................................................ 70
Figure 7-A. Active/Direct Perspective ....................................... 85
Figure 7-B. Antipassive Perspective ............................................ 85
Figure 7-C. Inverse Perspective .................................................. 85
Figure 7-D. Passive Perspective .................................................. 85
Figure 8. Relative Distinguishability of Participants for the
Basic Event Types ................................................................... 101
Figure 9. Middle Perspective ...................................................... 109
Figure 10-A. Two-Participant Mental Event ............................. 186
Figure 10-B. One-Participant Mental Event ............................. 186
Figure 11. Organization of the Primary Pre
xes According
to Prototypical Usage ............................................................. 308
Linguistic Terms
+ a morpheme boundary in which there may or may not be
intervening elements, e.g.,
Lat. Latin
l.e. lower edge
middle marker
direct object
o.s. oneself
nominative case
NP nominal phrase
past tense
or pl. plural
PN personal name
rev. reverse
exive marker
RN royal name
SAP Speech Act Participant
s.o. someone
Sp. Spanish
ana (royal inscriptions)
Flood Flood Story
FmInst Farmer’s Instructions
GgAk Gilgame
GgBH Gilgame
& the Bull of Heaven
GgDth Death of Gilgame
GgEN Gilgame
NSJN Nanna-Suen’s Journey to Nippur 1.5.1
Nungal A Nungal A
OmYg The Old Man & the Young Girl Alster 2005
RCU 1 Royal Correspondence of Ur 1 3.1.01
RCU 2 Royal Correspondence of Ur 2 3.1.02
RCU 10 Royal Correspondence of Ur 10 3.1.08
RCU 17 Royal Correspondence of Ur 17 3.1.15
RCU 19 Royal Correspondence of Ur 19 3.1.17
Rim-Sin (royal inscriptions)
(royal inscriptions)
Samsuiluna (royal inscriptions)
Sargon (royal inscriptions)
ShGr Sheep & Grain
Sin-Iddinam (royal inscriptions)
SKL Sumerian King List
Ur-Namma (royal inscriptions) RIME 3/2
UrNin A Ur-Ninurta A
UrNin C Ur-Ninurta C
Ur-Ningirsu II (royal inscriptions) RIME 3/1
UrNm A Ur-Namma A
UrNm B Ur-Namma B
Utu-hegal (royal inscriptions) RIME 3/2
Warad-Sin (royal inscriptions) RIME 4
WnSm Winter & Summer 5.3.3
YN year name
Horsnell 1999, Sigrist and
Gomi 1991
It must be obvious to any one who has thought about
the question [of the general form of a language] at
been possible to discover what principle, if any, governs the use of the
various forms, or in what respect, e.g.
, ‘he built’, differs from
, ‘he reigned’. For merely practical purposes it may be assumed that
no material divergence of meaning is indicated by the use of one form
in preference to another” (1924: 32–33).
In the early seventies, Sollberger was not alone in his resignation.
That same year, M. Lambert would write in stark terms, “On explique
xes, assez bien les suf
xes; on n’explique pas les pré
(1972–1973: 13; see also Gragg 1968: 107 n. 8). And the sense of
helplessness before the problem would be stated more stridently the
following year by Postgate, who, again with regard to the pre
, observed, “our failure to de
ne the difference in a satisfactory
way has epitomized our helplessness before Sumerian grammar as a
whole” (1974: 24). Such was the progress that could be boasted on the
sind danach als verschiedene ‘Konjugationen’ zu bezeichnen” (Scholtz
1934: 2). Although the term originates with Landsberger, the idea has
its roots in the very beginnings of the
eld, as these morphemes were
always tightly bound up with notions of what constitutes a
nite verb
in Sumerian. Poebel, for instance, considered
the one primary and
indivisible pre
x that made verbs
nite, suggesting that even the pre
Fifty years earlier still, the pioneers of the
eld viewed these elements
as pronouns that lent person and number to the verbal root, and which
linguistic meaning is not based on ‘objective’ properties of the denoted
object, but on a subjective perspective that represents the interaction
of the speaker’s mind with the world around her or him” (1994: 89).
The pre
is written m&#xm19u;u;
/i-ma�, im-mai1m;-1m;ᨐ, e-maǡ-;m1a;က (Pre-
먀ba (rarely ba
� (Fara, Abu Sal
bikh, and Ebla); and
: i
�, also &#xi000;i in Old Babylonian and e in the third millennium. Regardless of
period or place, I will refer to these pre
xes as
as the conjugation pre
xes. However, the
weight of scholarly opinion—early scholarship in particular, but many
recent studies as well—favors a very different understanding of what
constitutes a conjugation pre
x, or at least a primary element of this
class (see §1.3). Traditionally,
are considered primary, although
chapter one
the pre
x may have acquired the latter meaning secondarily in some
dialects (Attinger 1993: 268–269; Edzard 2003: 111–112; Westenholz
1975: 8; in particular, see Jagersma 2007).
Thus, I regard the conjugation pre
stituting a distinct morphological class. They may belong to a broader
focus system that encompasses all of the conjugation pre
xes and that
expresses, additionally, various types of locative focus (i.e.,
) and aspectual distinctions (i.e.,
- vs.
)-), but these four are the
primary grammatical means of expressing voice in Sumerian. As such,
voice in Sumerian should take its place alongside other verbal categories
recognized in the language, such as aspect (i.e., the
tion) and mood. That Sumerian would exhibit a voice system is by no
means surprising, since the majority of the world’s languages possess
a means of expressing voice (Payne 1997: 172; Shibatani 1988: 3). In
this book voice is understood to be, essentially, a semantic-pragmatic
phenomenon. Broadly described, grammatical voice systems provide
speakers with a series of linguistic options for expressing distinct
conceptualizations of experience and reality. This is precisely the role
played by the Sumerian conjugation pre
xes: providing the Sumerian
speaker with the linguistic means to express alternative perspectives
on events.
The theory presented in these pages is based upon the simple, but
or English
—and that valid cross-linguistic generalizations can be
drawn about these events on the basis of comparison. As a corollary, it
follows that in a semantic analysis of this type verbs are taken in their
primary idiomatic meanings regardless of historical origins or literal
, for instance, represents the
event in Sumerian,
the analytical meaning in terms of the component parts,
to approach
or the like, is irrelevant for our purposes.
Prototypical combinations of the type exempli
ed above notwith-
and their derived passive counterparts, with the Object of the basic
active clause being “remapped” to the subject position of the non-
basic passive clause. Hence, the label “derived voice” (Klaiman 1991)
tone to emphasize an argument that is of particular importance, voice
or focus markers signal alternations in the salience or importance that
speakers wish to attribute to an argument (§2.9.4). Inverse systems
may be similarly pragmatically based, with alternations in verbal
shape encoding the relative topicality of arguments according to what
is canonically expected (active or direct voice) or unexpected (inverse
voice) in discourse. Or they may be semantically based, in which case
changes in verbal shape signal
Of these assorted voice categories, Sumerian best
ts the basic, active-
middle, voice mold, although it clearly shares features with the derived
voice and, particularly, with the pragmatic voice types. A cross- or bleed-
ing-over comes as no surprise, since these categories merely represent
prototypes, proposed with the goal of providing a convenient classi
tion, rather than suggesting distinct and exclusive groups. Rarely can
syntax, semantics, and pragmatics be neatly teased apart; even in the
case of derived voice, for instance, there are semantic and pragmatic
motivations prompting the choice of the active or passive voice. Indeed,
the existence of a medio-passive category, in which the middle and
passive voices are not formally distinguished, again shows how elusive
the classi
cation of voice phenomena can be. Functional studies of
voice phenomena across a wide swath of languages (e.g., Givón 1994,
2001a, 2001b; Haspelmath 1994; Kemmer 1993a; Fox and Hopper
1994; Shibatani 1988) have shown the porous quality of these types,
revealing these labels to be more like tick marks along a continuum
than representative of rigid boundaries. It is in this spirit that the label
basic voice should be understood when applied to Sumerian.
Voice in Sumerian is most clearly elucidated by beginning with the
prototypical case of a transitive verb with two arguments: a volitional
destroyed Urbilum’ ~ mu Ur-bi
‘The year: Urbilum was
destroyed’ (Amar-Sin YN 2).
As this description suggests, an event may be described in terms
of its terminal points: a starting point, represented by an initiator or
actor, who brings about and controls the action, and an end point, an
entity that is affected by the action. These notions can be extended to
arguments they generally take, and in certain pragmatic contexts more
than others. The extension from these more prototypical usages to the
less typical or peripheral, owing to a perceived similarity, represents a
basic organizing principle in language, one commonly discussed under
the designation of prototype theory (§2.3). As my comments above imply,
at the center of the semantic range of the pre
—its prototypical
usage—is the canonical or highly transitive event. More broadly, the
focuses upon what may be regarded as the
rst half of the
event, the Initiator of the event—often the Agent of a transitive clause
or a controlling subject of an intransitive clause—and the action or
state denoted by the verb itself. The pre
other hand, focus upon the second half of the event, its Endpoint—an
affected Object or affected subject—and the resultant state that is a
consequence of the verbal action.
the prototypical usages are middle-voice
that is absent when these same events are expressed with
This notion is critical to the description of the middle voice and will
be discussed under the rubrics of “granularity” and the “elaboration
of events” (Kemmer 1993a; §2.9.6).
In terms of the traditional labels given to voice markers,
, as
an Initiator or Actor focusing device, is an active voice marker, while
, as Endpoint- or Undergoer-oriented notions, are middle
voice markers. The two can be distinguished on the basis of their
phonological shapes. Re
ecting its greater phonological substance, its
greater phonological bulk,
may be described as a
marker and
middle marker. Alternatively, to acknowledge
the difference in the situations with which each tends to occur,
be said to represent the middle voice and
the medio-passive. The
is neutral to these voice distinctions. However, this de
in terms of a negative does not imply that
is merely a
marker, devoid of any meaningful function. Rather, by neutralizing the
voice opposition,
functions as a defocusing device. As such, it often
stands pragmatically in opposition to the voice pre
xes, de-emphasizing
the functional thrust of each. Most commonly, h
x replaces
, an unsurprising fact given that the active voice is
the unmarked member of any voice system. In this sense,
particular emphasis on the Initiator of the event and its close association
with prototypical transitivity, may be regarded as signaling the
voice vis-à-vis
, which may be said to represent the unmarked
or basic active voice when used in these same contexts.
The conjugation pre
xes are not limited to the expression of gram-
matical voice, but express a number of grammatical categories that are
either directly related to voice or are iconically associated with voice
by virtue of sharing a similar structure. Among these are included
other hand—can be explained in terms of the Nominal Hierarchy,
particularly when it is understood as a graded scale of
, the
likelihood with which a speaker identi
es with various persons and
entities in the context of discourse (§2.2). But these correlations also
the pre
xes constitute a
paradigm with verbs of motion. As expected, the pre
to this directional contrast.
The purpose of cross-linguistic comparison is not to assert or prove
the existence of a grammatical category or phenomenon for a lan-
chapter one
to claim that this book would not have been possible twenty years
ago. Contributions that have had a major impact upon the theory put
forth here include: Barber 1975; Croft 2003; Fox and Hopper 1994;
Givón 1984, 1990, 1994, 2001a, 2001b; Hopper and Thompson 1980;
Langacker 1987, 1991; Kemmer 1993a; Klaiman 1991; and Shibatani
1985, 1988. Many others are referred to in the pages that follow.
My goal in this book is to describe the basic functions of the conju-
gation pre
, within the framework of gram-
Thus, the pre-radical pronouns
, which index or cross-reference Agent
beyond the scope of the present work. Also beyond consideration is the
stative pre
(see §1.1). Finally, the pre
treated—with respect to voice phenomena—as functionally equivalent
respectively. Although the distinction was largely
abandoned by the Old Babylonian period,
represented, in origin at least, the stative counterparts to
(e.g., see [493]–[494]). However, in terms of voice, there is no
in function, to future research.
It is implicitly understood, if not always admitted, by all Sumerologists
that virtually any proposition regarding the language may be proven,
or at least bolstered, if one chooses one’s evidence carefully. To avoid
this pitfall I have perhaps erred on the side of providing too much,
rather than too little, data, particularly for those functions that have
not been described previously. I have placed particular stress on paral-
lel contexts, which form minimal pairs of a type—passages that are
identical, or nearly so, but differ with respect to pre
x. Because there is
no consensus on the precise meaning of many of the modals (see Civil
2000, which contrasts with the traditional theory as voiced, for instance,
are best illustrated by way of direct comparison. Consequently, there is
considerable degree of cross-referencing of examples and some bleed-
ing over of the descriptions of one pre
x into the chapters on others.
ts of this structure outweighed the
disadvantages, as it avoided some of the more deep-rooted problems
that would arise if, say, the book were organized on the basis of verb
type or verbal semantics. The
chapter (§5) is the longest of the pre
chapters, re
ecting the fact that this pre
x has the broadest functional
range of the four. Also included here are the temporal associations of
the medio-passive domain; much of what is discussed also pertains, to
a degree, to
. In the
nal chapter (§6), the
ndings of §§2–5 are
Overviews of previous theories can be found in Jestin 1935: 21–30; Jo 1991: 2–13;
M. Lambert 1972–1973: 17–19; Pallis 1956: 253–65; Römer 1999: 105–116; Sollberger
1952: 107–176; Steiner 1994: 33–35; Thomsen 1984: 182–183; Vanstiphout 1985:
1–2; Zólyomi 1993: 99–107.
Lenormant, in the
rst grammar of the
language, wrote of “la voix passive” and “les différentes voix actives”
(1873: 149). But he did not see this opposition as the functional domain
of the conjugation pre
xes. Rather, the pre
xes were thought to rep-
resent pronominal elements, speci
cally, subject and object markers.
This basic idea persevered in various forms through the early decades
The designation “Akkadian” continued to be used by Delitzsch, Lenormant,
Sayce, Schrader, and others despite Oppert’s 1869 proposal of “Sumerian” (see Pallis
perspective and argued, on the basis of third-millennium texts, that
they are essentially semantic modi
ers that impart a deictic force to
the verb. The proposal laid the foundation of what for many years was
referred to as the
directional theory
. In Thureau-Dangin’s view,
Thureau-Dangin distinguished this pre
x from
, i.e.,
, which he tentatively
absent, as is well documented in variations of the Ur III year-names
A particularly prescient aspect of Thureau-Dangin’s conception is the
notion that the directional sense of
anticipates or implies a medio-
passive meaning. And although he did not describe
cally in
terms of the active voice, this function is implicit in his meaning: “
e ‘celui-là apporta ici (telle chose)’,
e ‘celui-là
, and even, curiously,
Langdon (1911: 134–135), writing shortly after Thureau-Dangin, arrived at a
similar conclusion, suggesting that both
can express the passive and
middle voices.

(see Poebel 1923: 226, 249, 253; cf. Falkenstein 1978a: 180
n. 3). Moreover, his rejection of the deictic theory had the consequence,
as is commonly encountered in year names beginning in the Sargonic
period, is very much a secondary development: “Über die Gründe,
aus denen in späterer Zeit das dimensional-re
exive Intransitivum vor-
Scholtz followed Poebel’s pairing of
—“e- und mu- ständen
sich nahe” (1934: 2)—but placed the pair in a conceptual framework
that was metaphorically based upon Thureau-Dangin’s directional
model, stipulating that the verbal event is oriented around two focal
points: a starting-point (place of the subject) and goal-point (place of the
represents the event “mehr vom Ausgangspunkt”
, “mehr dem Zielpunkt zu” (Scholtz 1934: V).
Grammatik der Sprache Gudeas von Laga
) represents
a watershed in the understanding of the pre
xes, and most subsequent
theories to this day ride in the wake of this authoritative work. The
uence of Poebel is clear, although this is not to say that Falkenstein
adopted Poebel’s ideas wholesale. Gone are the medio-passive func-
tions that Poebel assigned to
as well as the latter’s understanding
as representing a temporal distinction (1978b: 161 n. 2).
The pre
are, however, re
garded, after Poebel, as an
) are con-
sidered to be conjugation pre
xes in Falkenstein’s conception;
, are labeled simply as “Prä
xe” (1978a: 180–193;
see also 1959a: 45–47). Falkenstein’s basic approach can be described
as syntactic to the extent that he stresses indexation, speci
cally, the
co-occurrence of the conjugation pre
xes with various in
xes of the
verbal chain, and, with respect to the secondary “Prä
xe,” what he
saw as their essential dimensional or case cross-referencing character.
Without essential meanings of their own, these elements indexed nouns
and their syntactic roles in the Sumerian sentence. For Falkenstein, the
semantic properties of the pre
xes were of considerably less interest,
a fact re
ected in his decision to relegate this discussion to the second
volume entitled
. Furthermore, Falkenstein
made the important observation, which would be incorporated into most
subsequent theories, that
, additionally, can serve as a neutral pre
, “i- ist das Konjugationsprä
x der neutralen Diktion. Der
u ba-
—which Thureau-Dangin saw as essentially
deictic, and Poebel as re
exive—Falkenstein saw as displaying “einen
zweifachen Richtungsbezug [i.e., ba- and -
i-] . . . ‘er brachte die Hand
zu ihm daran heran’ = ‘er nahm es von ihm an’” (1978b: 184).
Of all of Falkenstein’s ideas concerning the pre
xes, scholars have
remained most faithful to his conception of
. This is certainly true in
terms of form—most scholars to this day accept Falkenstein’s morpho-
logical analysis of
as comprising an inanimate pronominal element
and a locative case element
And here assumptions about form
ect perceptions of function. Edzard, for instance, in his study of the
behavior of the verb
‘give’, comes to the basic conclusion that the
conventional theory, according to which
, above all else, expresses a
dative-locative inanimate co-reference, is essentially con
rmed by the
verb (1976: 170). The pre
x is used with the verb
nds, when
the recipient belongs to the class of
, an abstract concept,
or a group of
, and is marked, most often, in the locative-ter-
minative, locative, or terminative cases (see §5.3). Similarly, Römer, in
his grammar, basically maintains Falkenstein’s view that “die Prä
] beziehen sich auf Lokative, Lokativ-Terminative
und Terminative der Sachklasse im nominalen Satzteil” (1999: 113).
A nearly identical position, but more nuanced in admitting semantic
functions, is taken by Attinger (1993: 204; also 281–284), who accepts
Falkenstein’s morphology and observes that the pre
x primarily serves
equivalent to the dative
, referring to a noun in the
locative-terminative, locative (rare), or the absolutive (i.e., an unmarked
noun). And although Thomsen sees this dative function as a secondary
development, her description of the pre
x is, nevertheless, essentially
a restatement of Falkenstein’s position: “It seems as if the choice of
either /mu-/ or /ba-/ is primarily decided by the element immediately
following,” and so
is associated with animates,
with inanimates;
tends to occur with transitive
forms, and
with intransitive forms (1984: 178–179).
It comes as no surprise that the directional theory, semantic in its
See, for instance, Attinger 1993: 204, 281; Edzard 2003a: 94; Thomsen 1984:
183; Yoshikawa 1978: 481, 1992a: 397; cf. Yoshikawa 1979: 206. In a similar vein,
note also Jacobsen 1965: 75.
years following the publication of
until the present, a number
of scholars have pursued and modi
ed the theory
rst put forth a
century ago by Thureau-Dangin. Interest in the idea appears to have
been reawakened with the publication of the collected Old Babylonian
grammatical texts in
IV. Underscoring a correlation known from
bilinguals, in these texts
are frequently equated with the
Akkadian verbs bearing the ventive morpheme, while
have been largely discounted (one exce
ption being Jo [1991]).
the directional theory,
mped with the inclusion of bilingual
evidence bolstering the traditional argument, became in
uential once
again with the publication of
IV, as recent descriptions of the
xes and their functions attest. Foxvog (1974; brief summary in
1975: 400–401 n. 17), for instance, introduced the term
ing Landsberger 1923) as a descriptive label for the allative function of
. Black (1991) similarly described Sumerian
as a ventive element and attributed to
a separative meaning
on the basis of equations with Akkadian forms in the grammatical
texts. Yoshikawa (1978), in meshing spatial deixis with social deixis,
(i.e., that contained in the pre
but, notably, not
, which, in his view, is concerned with topicality)
as a ventive morpheme, indicating “spatial and emotional movement
) as an ientive morpheme, denoting
“spatial and emotional movement away from the speaker” (1978: 461).
Similar ideas have been put forth by Krecher (1985), who sees
all its combinations, including
, as expressing a constellation of
meanings that center on a basic spatial one, the ventive.
The result is that some aspect of the directional theory has been
accepted by most scholars, even if deixis is not considered a primary
function of
). Attinger acknowledges a ventive function
and a separative one for
, noting,
, that
these functions are naturally rare with the verb
279–280, 283–284). Thomsen (1984: 173) accepts a ventive function
, following Yoshikawa, but is silent on this account regarding
, while Edzard (2003a: 92–109) goes much further, describ-
ing the conjugation pre
xes and in
felicitously captured by the terms
but extends to
as well, for in the view of most scholars
the morphological linchpin of
with only a shade of difference in
The main proposals have been
, following Falkenstein, and
, which
has found wide acceptance; Krecher, however, has proposed
280–281, for discussion and previous scholarship.
perfect and can be traced back to Scholtz (1934), who believed that
intransitiven und passiven Verben zur Bildung von Formen wie ba-ug
‘ist verstorben’
verwandt worden” (1978a: 188 n. 2).
bestätigen” (1965: 106–107). Although von Soden’s position is often
cited, the basic notion that
convey a perfect meaning
has not been subsequently pursued and most descriptions of the pre
are silent on this account.
Thomsen speaks for many when she joins
with Falkenstein and writes with some skepticism about the primacy
of this function: “How such a function may harmonize with the other
functions of ba- is not evident, and I follow Falkenstein in the opinion
that this is a later development which cannot be observed in the Gudea
texts” (1984: 183; see also Streck 1998: 190–191).
As for the so-called passive uses of
, for Falkenstein this was “ein
Sonderfall.” But the early attestations of this use compelled him to
approach the problem with some hesitation. While Falkenstein admitted
that the pre
x is encountered in the Gudea texts, similar to the Ur III
date formulae, with this function, he was nevertheless certain that this
was a secondary development despite the date of the evidence. For him,
One exception is Oberhuber (1982: 132) who believed that the pre
x, because it
contains a locative element, has an inherent inessive value that makes it ideally suited
to express a terminative or resultative-punctual (i.e., perfective) nuance, as well as an
ingressive sense.
and so, if only implicitly, understood a voice-like function to be inher-
the reassignment of many of
Poebel’s Akkadian
forms from reciprocals to perfects dealt an early
blow to the idea, while Falkenstein’s syntactic theory of inanimate-
locative indexation could hardly accommodate voice phenomena. But
the notion was abandoned by many for other reasons, the foremost
being that
does not conform to a rigid system of active-passive
alternations, and so, it has been reasoned, the pre
x cannot be labeled
a passive marker and, more generally, the pre
xes cannot be said to
signal voice in a strict sense.
Already Jestin, seeking and not
does not represent an active-passive opposition since “it is now certain
that the verbal pre
xes do not of themselves denote active or passive
voice, which are instead indicated by the tense formations” (1977:
283). Thus, Horsnell, arriving at the same conclusion as Christian but
by different means, claimed that all year names should be translated
Oberhuber (1982), in discussing the passive in Sumerian, took a view
of the pre
that intermingled the ideas of Falkenstein with those
of Poebel. He agreed with Falkenstein’s morphological understanding of
the prefix, that
consists of the inanimate
element and an
, originally meaning ‘darin’ (1982: 132). And like
Falkenstein, Oberhuber was convinced that the passive function of
the pre
x was secondary—although the passive may have
rst made
its appearance in Pre-Sargonic Sumerian, it was fully developed only
in the Ur III period (1982: 129, 132). In terms of meaning, however,
Oberhuber sided more with Poebel, arguing that the pre
x initially
imparted a re
exive meaning. It was Oberhuber’s view that the notion
of a
, that is, agentless, passive is foreign to Sumerian as an erga-
tive language. With this claim, he echoed an opinion commonly held
during the infancy of ergativity studies, namely, that the passive voice
is incompatible with ergativity—it is an opinion that is not maintained
by most linguists today. Oberhuber proposed that Sumerian, again in
origin, could only express a passive with an agent, as indicated by
(lit.: ‘young man selected by An’) and PN
‘PN given strength by DN’) constructions. Thus, it was his contention
that the agentless passive, as expressed by
, came about secondarily,
in all likelihood through Akkadian in
uence given that agentless passives
are a regular feature of that language (1982: 132–133).
Black, while eschewing the term ergative, came to a similar conclusion
and the passive voice, at least in terms of nomenclature.
In addition to assigning a separative function to the pre
x, Black also
attributed to it a
value, according to which the pre
x served to
describe a state or condition—a label he preferred over passive, which
he claimed is not appropriate for Sumerian (1991: 37). Finally, in this
connection, we may quote Thomsen, who once again summarized the
scholarly opinion of the time when she wrote, “/ba-/ has been called
a ‘passive pre
x’ because of its frequent occurrence in one-participant
forms . . . this use of /ba-/ depends on its inanimate/non-agentive refer-
ence, and it has nothing to do with the category ‘passive’” (1984: 183;
see also Postgate 1974: 27).
formulation. Some have stressed more semantically oriented functions,
even if not speci
cally introducing, or resurrecting, the notion of voice
in Sumerian. Christian, as we have seen, broke ranks with Falkenstein
early on and rejected nearly every aspect of his theory concerning the
functions of
er, his ideas, unorthodox for their day as they
were, can hardly be said to have had an enduring impact on our under-
standing of the language. In general, it is more recent scholarship that
has looked with more favorable eyes upon semantic considerations.
The most serious challenge to Falkenstein’s conception in recent
times has come from Postgate (1974), who pointedly rejects the notion
has a locative cross-referencing function, the very crux of
Falkenstein’s theory—a position that Edzard would later describe as
radical (1976: 177). Having refuted the theory on functional grounds,
Postgate concludes, “In view of these objections, it seems preferable
to avoid positing a dimensional (and speci
cally locative) sense for
ba- in the prehistory of Sumerian, and to admit that ba- is
express in the verbal complex a relationship represented (outside it) by
x” (1974: 18–19). Given this understand-
ing of function, it comes as no surprise that Postgate likewise rejects
Falkenstein’s morphemic analysis of the pre
Rather, Postgate leans
toward agreeing with Civil’s assessment that “the pre
x ba- has no con-
nection with a locative element /a/” (1974: 20 n. 11)—a proposal that,
although embraced by some, remains very much a minority opinion
to this day. In terms of the function, Postgate, as I have noted, sees
the primary opposition among the conjugation pre
xes to be between
requires a person (other than the verb’s subject) on
whom the action has an effect, whereas
is used in cases where no
such person is involved.
Yoshikawa, it will be recalled, maintains the traditional opposi-
tion of
. But breaking with Falkenstein’s syntactic theory,
he understands this pair to be governed by topicality, which is, most
prominently, a function of the relative social standing of the agent
vis-à-vis the dative-bene
is a mark of topicality, signaling
the higher social status (i.e., topicality) of the bene
ciary (referent
of the dative case);
is a mark of non-topicality, signaling the lower
social status (i.e., non-topicality) of the bene
ciary. As this description
suggests, Yoshikawa believes that the fundamental function of both
“is to show that the action expressed with them is for the
passive function of the pre
x was therefore a secondary development:
“We were naturally inclined to correlate the non-occurrence of the
agentive suf
x -e with the occurrence of the pre
x ba- and this led to
the conclusion that the pre
x ba- denotes the (middle-) passive. Actu-
ally, h
the pre
x ba- . . . indicates the reduction of the agentive
-e, and, as a natural result, of the bene
ciary/indirect object” (1992a:
398). In this, his latest contribution to the subject of the conjugation
made by Falkenstein, he describes
as positively focused for person
and negatively for locus;
clause—the pre
x ba- is used” (2004: 44; see also Rubio 2007: 1346 n.
30, 1347–1348). In what represents a rare description of the functional
range of
, Michalowski argues that this pre
x expresses, essentially,
cation of
: “When the focus is intensi
ed, as with verbs
denoting movement towards the agent, or the agent manipulates an
object, such as a tool, the pre
x imma- is often used” (2004: 44). In
accord with this perception of function, Michalowski would see this
x not as a compound of
, as it is commonly understood,
but rather as representing “a form of reduplication of mu-, in which
the initial consonant is copied and the cluster is reinforced by an initial
vowel” (2004: 44; see also Rubio 2007: 1346, 1363). As in other recent
descriptions of the conjugation pre
is considered to be a neutral
x, used “when focus is not speci
ed” (2004: 44).
Lastly, there are those recent treatments of the pre
xes that revisit
the voice hypothesis, considering this to be a—if not the—basic func-
tion of these elements. The analyses of Jagersma (2007) and Zólyomi
(1993) take what may be described as a formal, morphosyntactic
approach to Sumerian grammar, but view voice and the conjugation
xes, as this study does, primarily from a typological perspective,
citing well-known typological investigations of voice phenomena (e.g.,
Kemmer 1993a, Klaiman 1991). As is true of so much previous work
on voice and the conjugation pre
takes center stage. Zólyomi
(1993: 108; see also 2005: 31–32) accepts three basic functions for the
x: (1) indexing an NP in the locative-terminative, or more rarely
the locative case, following Falkenstein; (2) serving as a separative or
directional element; and (3) marking the middle voice. The pre
in this last role may, by extension, mark the passive, since “the function
of middle ba- is naturally compatible with the meaning of passivity”
Jagersma’s study (2007) remains a work in progress at the time of this
writing. But as it stands at present, the author accepts a middle-voice
function for
and it is anticipated that, on this account at least, his
work will agree on many points with the present one. As does Zóly-
omi, Jagersma (2007, 2006) accepts
as a marker of the passive in
Sumerian and sees this as an extension of its middle-voice function.
This position differs from the one put forth by Attinger, who perceives
also Wilcke 1990: 488–498; Schulze and Sallaberger 2007; cf. Jagersma
2006). As noted above, more recently Attinger (1998b) has speculated
expresses the re
exive, a category that, naturally, belongs to
the middle domain. Finally, in my 2001 dissertation I described the
middle and related passive functions of
and to a lesser extent
Serving as the framework for the present study, voice was described as
a series of alternative perspectives on an event. The event itself, it was
argued, could be conceptualized in spatial terms, consisting of Initiator
and Endpoint nodes. The analysis sought to reconcile and unite the
deictic and voice functions of the pre
x by demonstrating the inherent
to the entirely syntactic. There is a great deal of ambiguity in the lit-
erature concerning these three terms, particularly with regard to the
transitive clause like the
of the intransitive clause, hence, the
grammatical rela
tion, or case, of subject in English. H
: the participant, either animate or inanimate, that
either is in a state or registers a change-of-state as
a result of an event.
: a typically animate participant, for whose bene
the action is performed.
: a typically animate participant into whose posses-
sion an object is transferred.
: a participant who engages in a mental activity,
either intellectual, emotive, or perceptual.
an entity that stimulates a change-of-state in or
reaction from an
: the location from which an entity moves.
: the location toward which an entity moves.
I will also refer to the following basic grammatical roles:
[3] Agent (A), transitive subject:
The king
destroyed the city
Subject (S), intransitive subject:
The king
Object (O), transitive direct object:
The king destroyed
The use of
, as a designation of a semantic role, and Agent (or
A), as a designation of a grammatical relation, is an unfortunate, but
deeply entrenched, terminological convention in the linguistic litera-
Rather, there is always an overlap in which two are treated identically
The Nominal Hierarchy
The notion that Agents are prototypically associated with control and
Objects with affectedness can be traced in modern scholarship back
to Silverstein (1976). In a celebrated study, Silverstein demonstrated
that some nouns, based on their inherent lexical semantics, are more
Agent-like—are more likely to control an action—than others, and
that a scale of all possible noun types could be established. This scale,
known as the Animacy or Nominal Hierarchy, ranges from nouns hav-
ing prototypically, or probabilistically, the greatest potential for serving
as Agents to those having the greatest potential for serving as Objects.
Silverstein’s primary interest lay in explaining mixed accusative-ergative
case-marking systems, that is, split-ergativity, and in accounting for the
rationale behind the identity of S with A and S with O. This led to
the critical observation that “noun phrases at the top of the hierarchy
manifest (nominative-)accusative case-marking, while those at the bottom
manifest ergative(-absolutive) case-marking” (Silverstein 1976: 113).
1st person 2nd person Demonstratives, Proper Human Animate Inanimate
pronouns pronouns 3rd person nouns
More likely to be in A than in O function
Figure 2.
Nominal Hierarchy
(after Dixon 1994: 85).
The guiding principal behind the Nominal Hierarchy is that the
higher a noun phrase (NP) in S function is on the scale, the more
likely it is to be marked like A; conversely, the lower an NP in S func-
tion, the more likely it is to be marked like O. As a result, accusativity
dominates the left end, or top, of the spectrum in Fig. 2 and ergativity,
the right end, or bottom. For languages that mix ergative and accusa-
tive case-marking strategies, at a certain language-speci
c point along
the hierarchy, everything to the left will exhibit accusative marking and
everything to the right, ergative marking. For some languages this point
might be second-person pronouns, in which case third-person pronouns
and all nouns are ergative, and
rst- and second-person pronouns are
accusative. Other languages might take human common nouns or even
rst-person pronouns as their critical point.
, by far the most
commonly attested pattern, cross-linguistically, is the one in which the
rst and second persons—the speech act participants (SAPs), the pro-
totypical agents, the agents
par excellence
—exhibit accusative marking,
while the third-person pronoun and nouns demonstrate ergative mark-
ing (cf. the marking pattern in the Sumerian
aspect; see Woods
The basic prediction of the Nominal Hierarchy is that in the typical
transitive event, nouns that rank higher on the spectrum in Fig. 2 are
more likely to act upon nouns of lower rank, rather than vice versa.
Although in principle any NP can ful
ll either the A or O roles, it is
clear, cross-linguistically, that in actual discourse there is a tendency for
NPs in the role of A to rank higher on the Nominal Hierarchy than
those in O. When couched in these terms, the Nominal Hierarchy
can be seen to have applicability beyond case marking strategies. Case
marking is simply one aspect of language for which it is relevant. In its
essentials, the Nominal Hierarchy, or perhaps better the natural ordering
Note the comments of Wierzbicka regarding the special status of the
rst and
second persons, “1st and 2nd personal pronouns seem to favour, universally, situations
where their central role, their topicality and prominence in terms of the speaker’s inter-
est, are OVERTLY MARKED. They have privileged status, and they like to SHOW
IT. They tend to command agreement, they tend to occupy the most topical leftmost
position in the verbal complex, they like to collect as many subject-like properties as
possible. It is only to be expected that they will also favour case marking which would
ect their privileged status” (1981: 70; see further Woods 2000).
A more precise de
nition of topic is given by Lambrecht: “A referent is interpreted
as the topic of a proposition if in a given situation the proposition is construed as
being about this referent, i.e. as expressing information which is relevant to and which
increases the addressee’s knowledge of this referent” (1994: 131).
For Kuno, the natural ordering of NPs is an Empathy Hierarchy,
representing the degree to which speakers identify or share common
concerns with the people and things under discussion (Kuno 1976,
1987; Kuno and Kaburaki 1977; see also DeLancey 1981; Langacker
1991: 306–307). A speaker naturally empathizes most with himself. He
is concerned and identi
es most with himself and, secondarily, with the
addressee, his partner in discourse. The speaker displays progressively
Nominal Hierarchy, are the prototypical Agents, possessing the most
Clustering of members around the prototype
. Because properties tend
to correlate, most members of a category will cluster around
the prototype; peripheral and ambiguous members tend to be
a relatively small minority of the total population.
An often-cited example from Rosch’s research draws immediate par-
allels to our notions of prototypical Agents and Objects—the category
suggested by the label
(see Rosch 1973: 135–139, 1975: 199–206;
see also 1977: 25, 1978: 39; discussed by Allan 2003: 31; Croft 2003:
162–163; and Lakoff 1987: 39–57). As Rosch’s experiments have shown,
cognitively, the category is based on the co-occurrence of a number
linguistic background
Scalar Transitivity
chapter two
action non-action
telic atelic
punctual non-punctual
realis irrealis
These properties are described by Hopper and Thomson (1980: 252–
253) as follows:
. In order for an activity to be transferred there must
be at least two participants (A and O) involved. H
property of
, alone, does not de
ne the category as a
whole. Since transitivity is a prototype category with graded mem-
bership, it is possible to speak of degrees of intransitivity as well
as transitivity. A clause may have an Agent and an Object, but
still be marked syntactically as intransitive (e.g., instances of noun
incorporation). Conversely, in some languages a clause may have
linguistic background
represents an entity that is distinct from its own background.
is actually a cluster of grammatical properties that
are identical to those discussed in connection with the constituent
with inanimate Objects that differ in terms of count. Compare
drank the bottle
of water
I drank some water
; the implication that
the water is
. The Object is naturally not affected when an action
does not take place.
refers to propositions that are true,
to unreal or
of adapting a limited inventory of conventional units to the unending,
ever-varying parade of situations requiring linguistic expression” (1991:
295). Cross-linguistically, the prototypical transitive event is taken as a
model to structure other, less prototypical, event types.
This result, of course, is predicted by prototype theory, which advo-
cates that members of a category will tend to cluster around the pro-
totype. In terms of morphosyntactic coding, this means that deviations
from the norm will be coded like the norm, that there will be verbs
that do not describe canonically transitive events—which verbs will
that elaborates or modi
es the basic narrative line without carrying it
forward” (Croft 2003: 182).
Cross-linguistically, there is a strong tendency for certain gram-
matical categories to occur in foregrounded portions of discourse, and,
similarly, for other categories to be disproportionately represented in
backgrounded portions (Hopper and Thompson 1980: 280–295). Hop-
per and Thompson observe that these grammatical categories are the
very same that describe semantic transitivity: “The likelihood that a
the organization of any differentiated visual
eld involves, in its most
. Where Givón’s account differs from Hopper’s and Thompson’s is
1968: 275). These are the elements of language which relate an utter-
ance to its spatio-temporal context and whose meanings are relative to
that context. As linguistic
of the coordinates of person, time,
and space, coordinates that have an existence beyond the world of
language, deictic expressions allow the speaker “to anchor utterances
in the extra-linguistic reality” (Kryk 1987: 1). Here we include, for
instance, demonstrative pronouns and adverbs (
this, that, here, there
well as personal pronouns (
I, you
), and temporal adverbs (
today, yesterday
Given the sentence
the speaker places himself, as Greenberg (1985: 272) describes, at the
reference point or, more precisely, at the origin of a space described by
spatial polar coordinates, where the most salient feature is the relative
being completely absorbed by a
nal, terminal object (Fig. 3), e.g.,
swung the
, which broke the
cases, the transfer of energy that characterizes the billiard-ball model
events most often in terms of the notions of Initiator and Endpoint.
, these notions are not the only macroroles capable of describ-
linguistic background
Initiator = Agent Endpoint = Obj
are inanimates that are naturally uninvolved with the discourse and
are conceptualized as
(see Woods 2000). Thus, there is DeLancey’s
observation that the transitive event can be represented as a motion
from Agent to Object, and that this motion, prototypically, can be
interpreted—extending the spatial metaphor still further—as movement
away from the speech situation, the location of the SAPs (DeLancey
1981; see also Croft 2003: 182).
High Animacy
Low Animacy
1st/2nd pers. =
3rd non-pers. =
High Empathy
Low Empathy
High Salience
Low Salience
The Stage Model
The second cognitive model with which we will be concerned relates
Perspective, as I have explained, is a deictic quantity as it is a func-
tion of spatial orientation with respect to a reference point. In the stage
model, perspective is dependent upon the viewer’s position relative to
that of the actors, and the
xed stage serves as the reference point.
Much in language can be explained, or at least described, with reference
to a deictic category of viewpoint or perspective (see DeLancey 1982).
In the following section I describe voice in these terms; speci
cally, I
portray the category as the grammatical expression of perspective on
the clausal event.
Grammatical Voice as Alternative Perspectives on the Event
Figure. 6-A.
Unmarked View of the Event: Agent primary focus, Object secondary focus.
and far-reaching. The director corresponds to the speaker, the audience
chapter two
The Active and Passive Voices
are generally animate,
inanimate, and so on—this
hierarchy can be related directly to the Nominal Hierarchy (§2.2).
have the greatest probability of serving as subjects, while the closely
related roles of
outrank all others with the
exception of
. From a pragmatics perspective, this scale may also
be seen as a topic accession hierarchy. As such, it re
ects the likelihood
of a speci
ed semantic role to serve as a sustained topic in discourse.
is at the top of the hierarchy as the most natural topic.
[9] ���agent beneficiary/recipient patient source/goal,
What this suggests is that the active voice, which typically takes
as its subject, represents the unmarked—the basic or most
natural—perspective on the event. Cognitively, the
starting point of the event and, in the active voice, the starting point
of the clause itself, assuming the initial or left-most position. The
naturalness of this correlation stems from the broader observation that
humans tend to organize structures around starting points—starting
points typically have greater cognitive salience than end or mid-points
(MacWhinney 1977).
In the sentence corresponding to the second rendition of the scene,
[8], the action is perceived from Anzu’s perspective—the passive voice.
The focus is now upon the affected Object, while the Agent, Ninurta,
having been relegated to a peripheral case, is de-emphasized. The
logical sequence of action is no longer re
ected in the structure of the
sentence. From a formal point of view then, the passive may be said to
select some non-Agent clausal participant, one that is less topical and
less likely to control the action than the
the subject. In the passive, the Object does not become the Agent; it
maintains its role as Object—it has merely been promoted, or perhaps
more complex morpho-syntax of passive constructions, as well as the
distribution and frequency of the passive vis-à-vis the active (see Croft
1991: 254, with references).
In terms of perspective, the passive may similarly be understood
as representing a marked perspective on the event. The passive views
the event from its Endpoint, taking for its vantage point some non-
dynamic, non-controlling, and affected Object. The direction in which
the event logically proceeds con
icts with the perspective from which
it is viewed. With respect to factual content, however, there is no fun-
no longer provides a suitable vantage point from which to observe the
event; the Object now ful
lls this function by default, being the most
salient remaining argument in the clause. Passives, therefore, may be
described, somewhat ironically, as revolving around Agents as “their
fundamental function,” in Shibatani’s conception, “has to do with the
defocusing of agents” (1985: 831).
Shibatani’s understanding of the passive
nds support in the cross-
linguistic observation that languages, if not outrightly prohibiting its
expression altogether, generally avoid the Agent in passive constructions.
English, for instance, allows for the expression of the Agent in passive
sentences, as in our sentence [8], but agentless passives nevertheless
outnumber passives with overt Agents by a wide margin (Shibatani
1985: 830–831).
If the passive were primarily a means of promoting
the Object, agentless passives would be left without an explanation.
Further, Shibatani’s conception of the passive accounts for so-called
impersonal passives, e.g., Eng.
One should exercise
, Gr.
So etwas tut man nicht
‘That’s not done’, Fr.
peut le dire
‘One can say that’. Such sentences
have an Agent, but the Agent is low in salience and topicality; being
non-referential or generic, it is little more than a dummy placeholder.
In fact, the Agent may be so unimportant to the context that it can
agentless) passives by virtue of being compatible with a generic Agent
(e.g., Eng.
one, they
, Fr.
, Gr.
, Eg.
appear at some intermediate point and suggest what is indeed known to
Cf. also in Semitic the use of 3
m. verbs with inde
nite subjects corresponding
functionally to the passive.
Givón 2001b: 168–169). As with the passive, topicality is the critical
dimension for the antipassive. In the active voice, both the Agent and
the Object are topical, but the Agent is more topical—more central
[12] a. Para baihu -bisita si Rosa gi espitat
‘I’m going to visit Rosa in the hospital’
b. Para baihu
-bisita gi espitat
‘I’m going to visit somebody in the hospital’
(Cooreman 1987: 133–135; Cooreman 1988: 585–587).
A speaker chooses to use the antipassive, and so de-emphasize the
Object, because he deems the Object to be of only marginal interest
or, perhaps, even irrelevant to the message. Analogous to the pragmatic
motivations behind the use of the passive [10], the speaker may consider
the Object to be unimportant, obvious, or generically predictable.
assessment of the Object corresponds to various semantic properties
of the Object itself, properties that are essentially the same as several
of those outlined in [5], which were described in connection with
and its effects on transitivity. Speci
cally, a less individu-
ated Object—one that is inde
nite or non-referential, or represents a
plural, abstract, or mass quantity—is predictably a less topical Object
and one that may trigger the use of the antipassive voice (Givón 2001b:
168–169). Clearly, with such Objects the transmittal of the action from
Agent to Object, or energy in the billiard ball conception, is hindered.
The Object is inherently less affected and so the transitivity of the clause
is correspondingly decreased.
It follows that there are certain aspectual correlates to the antipassive
given that high transitivity corresponds to punctual, telic events and
low transitivity to non-punctual, atelic events. Because of its Endpoint
defocusing character and its proclivity “to emphasize the action or
state-of-affairs depicted by the predicate” (Cooreman 1988: 585), the
Cooreman (1987: 132–133; 1988: 584) points out the corresponding difference
Endpoint (Cooreman 1988: 584, 587–588). Observe, for instance, the
following use of the antipassive in Quiché, a Mayan language [13]. In
rst sentence, [13]a, the antipassive, with its focus on the Agent and
discourse pressure to use the passive voice (note the second of three
discourse factors promoting the use of the passive [10]b) and to view
the event from the perspective of the more salient and topical partici-
pant. Compare the naturalness of the sentence
John was hit by a car
with the stilted
A car hit John this morning
, or the preferable
was killed by lightning
. In both cases, the Agent,
lacking animacy and so volition, represents a considerable departure
when the action proceeds in the opposite direction, with a non-SAP
referent acting upon a SAP, then the inverse voice is triggered [14]b
and the verb must be marked with the suf
[14] a. Ni-s
n atim
1 -scare-
‘We scare the dog’
b. Ni-s
n atim
1 -scare-
‘The dog scares us’ (Klaiman 1991: 32, 162).
Similarly, note the direct-inverse contrast in Nocte (Naga) [15], a Tibeto-
Burman language. The direct voice [15]a is unmarked, while the inverse
voice [15]b is marked by the suf
. Observe that in both cases, the
verb agrees (i.e., the suf
rst-person argument; in [15]a
Direct-inverse voice systems may also be pragmatically organized
and dependent upon the relative topicality of the Agent and Object.
chapter two
As this characterization suggests, for many linguists, including Givón
(1994), the inverse is principally a pragmatic notion. Topicality, in this
view, is the common denominator behind pragmatic and semantic
And, as we have seen, topicality underlies the active, passive,
and antipassive voices as well. Upon the basis of this primary pragmatic
notion, the active/direct and these three detransitive voices may be
schematically described as follows:
Table 2.
Relative Topicality of Agent and Object in the Active/Direct, Antipassive, Inverse,
and Passive Voice Constructions
(Givón 2001b: 155, after Cooreman 1987).
Active/Direct �Agt Obj
Antipassive ��Agt Obj
Table 2 balances the relative topicality of the Agent and the Object
in the four voice types discussed in §§2.9.1–2.9.3 for the semantically
transitive clause. Pragmatically, the unmarked active/direct can be
described as a voice construction in which both the Agent and the
Object are topical, but the Agent is more topical than the Object.
The antipassive magni
linguistic background
Figure 7-A.
Active/Direct Perspective.
Figure 7-B.
Antipassive Perspective.
Figure 7-C.
Inverse Perspective.
Figure 7-D.
Passive Perspective.
the relative topicality of the Object that distinguishes the passive from
the inverse, Fig. 7-C is more appropriately matched to the inverse.
In the inverse, the Agent is topical; it has a certain on-stage presence
even though it is surpassed in topicality by the Object, which occupies
center stage. The passive, then, with its extreme suppression of the
ed with Fig. 7-D, a
in which the
Agent is only marginally visible.
Focus Systems as a Pragmatic Voice Phenomenon
The passive, antipassive, and inverse voices have been presented as
essentially pragmatic phenomena, although each has a considerable
semantic component in that the pragmatic options closely correlate
with the “ontological semantics” of the arguments involved. This was
particularly clear in the case of the semantically motivated inverse
voice. Similarly, the pragmatic choice of the passive or antipassive
often corresponds to the semantic properties of the Agent and Object.
There exists,
, a system of voice oppositions that, like the
pragmatic inverse, may, more or less, be described as driven by purely
pragmatic factors, being largely independent of semantic criteria. Such
systems, commonly referred to as focus systems, such as those famously
known from Philippine languages, signal an argument’s “centrality or
noncentrality to speaker/hearer concerns, interests and expectations”
(Klaiman 1991: 227). Often coexisting with other voice constructions,
focus systems, as the name implies, are a means by which a speaker
may highlight or emphasize a particular nominal, bringing it to center
stage in the context of discourse.
Cebuano [17], a Philippine language, for instance, attests a series of
prepositional particles that mark nouns re
ecting their semantic role
in the clause. The various types of focus or voice—Actor (A), Goal
(G), Directional (D), and Instrumental (I)—are indicated by indexing
the role of the nominal that occurs with the focus particle (Klaiman
1991: 247; Shibatiani 1988: 88–89); each of the
rst three sentences
[17]a–c have the same meaning content, but differ in the nominal that
is emphasized or focused. Sentence [17]d exempli
es the focus of the
instrumental argument.
Ni- hatag
sa libro sa bata
G book D child

gave the book to the child’
hatag ni Juan
ang libro
sa bata
-give A Juan
D child
‘Juan gave
to the child’
ang bata
ni Juan sa libro
A Juan G book
‘Juan gave
ang kutsilyo
sa mangga ni Maria
G mango A Maria
‘Maria cut the mango with
(Klaiman 1991: 247; Shibatani 1988: 88–89).
The Middle Voice
If grammatical voice is “the most complex grammar-coded functional
domain in language” (Givón 1994: 3), then by this same criterion pride
of place within this domain must go to the middle, the most intricate
of the voices. I begin the discussion of this elusive category with Lyons’
description, which, because of its clarity and concision, remains, after
forty years, the conventional springboard for investigating the middle
voice. Lyons characterizes the middle as an indication that “the ‘action’
or ‘state’ affects the subject of the verb or his interests” (Lyons 1968:
373). In this, the middle stands in contrast to the active, which views
or doer of the action. Similar characterizations
have been put forth by others. Benveniste, for instance, describes verbs
in the active as denoting “a process that is accomplished outside the
subject,” while in the middle, “the verb indicates a process centering in
the subject, the subject being inside the process” (Benveniste 1971: 148).
Indeed, it is a characterization that is already implicit in the work of
signify action, like the active, or the resultant state like the passive
(Lyons 1968: 373).
What is common to all characterizations of the middle voice is that
the subject is perceived as being, mentally or physically, affected by the
action he initiates. In recent linguistic parlance this semantic property is
subject affectedness
(Barber 1975; Kemmer 1993a, 1994;
Klaiman 1988, 1991). The active, however, does not necessarily signal
the absence of this property. Many events construed as active cross-
linguistically can be seen as impinging in some way upon the subject.
This is certainly true of many intransitives that are marked as actives,
such as the Classical Greek
manner of motion
‘creep’, among others, which occur only in the active voice
(Allan 2003: 243–247; see §3.2.2). And it is also true of many transitive
verbs, such as the Classical Greek verbs of eating and drinking, which
are mostly construed as actives (Allan 2003: 26). Thus, rather than being
understood as an express indication that the subject is unaffected by
the action, the active—in opposition to the middle—is more accurately
described as neutral to the notion of subject affectedness.
The de
nitions of the middle voice given above, broadly stated as
they are, allow for a spectrum of functions, ranging from the expression
of the re
exive, the reciprocal, and the passive to those that are more
are middle marked since the subject bene
ts from
here are verbs that are intrinsically re
exive, such as body-action verbs
of the type
and verbs of emotion, for instance,
cry, laugh
Affectedness in these cases is an umbrella term that only acquires a
c meaning when combined with a given verb in a particular
context (Bakker 1994: 24–25; Rice 2000: 191–192). In this function, the
middle represents a pragmatic option that is a function of the speaker’s
desire to construe the event as affecting the subject.
In terms of orientation and perspective, the middle voice is an End-
point oriented category. It does not emphasize the subject’s qualities as
initiator and controller of the action, for this is the functional domain of
the active voice, but the subject’s affectedness as a result of the action.
While it is true that in the middle “the action notionally devolves from
the standpoint of the most dynamic or Agent-like participant in the
depicted situation,” as Klaiman observes, it is also true that this “same
participant has Patient-like characteristics as well, in that it sustains the
action’s principal effects” (1991: 3). It is this latter characteristic that is
Consider the Classical Greek sentences given in [18]. The active
‘I am washing (s.th.)’ contrasts with the forms with the middle
marker (in bold). The middle voice verb can correspond to the re
[18]b or to the reciprocal [18]c, but it may also take an object that is
distinct from the subject [18]d. This last function is referred to vari-
self-benefactive middle
indirect middle
indirect re
Agent’s conscious participation, either physically or mentally, in the
event, e.g.,
‘plan’ vs.
‘look at’
‘choose’ (Smyth
1956: 392–393). Similarly, with verbs belonging to a common semantic
class, the middle-marked verb may denote the subject’s more intense
involvement in the event, a “more vigorous participation on the part
of the subject” (Smyth 1956: 391), compared to its active counterpart,
‘dart’ vs.
middle to the passive is the common property of subject affectedness.
Of course, in many languages the passive and middle voices are not
formally distinguished (e.g., Akkadian), but instead a portmanteau cat-
egory, the medio-passive, does the duty of both. Indeed, for many of
these languages, the passive is a secondary development, a diachronic
outgrowth of the middle voice, as has often been claimed for Indo-
European. The primary voice opposition in these cases is, or was
linguistic background
b. T
antaiyai a
‘The mother, embracing the child, engulfed it’
(Klaiman 1991: 77–78).
Similarly, the Na-Dene family, which includes Athapaskan languages
such as Navajo, possesses a
referred to as the D-element, which,
xo res -ii
he deposit-
Kemmer (1993a: 20, 83–84) also includes here what she refers to as the logophoric
middle, a relatively rare use of middle markers occurring with reportive verbs in some
languages. In this function the middle serves to overtly mark co-reference between
participants in main and dependent clauses. Speci
cally, in languages that exhibit this
function, in a clause of the type,
will quit his job tomorrow
, there would be
linguistic background
Table 3.
Typical Middle-Marked Situation Types
(after Kemmer 1993a [16–20]; 1994 [182–183]).
1. Grooming or body ‘wash’
Djola (Niger-Congo)

chapter two

Old Norse
Mohave (Hokan)
linguistic background
10. Naturally reciprocal ‘embrace’
Bahasa Indonesia
to describe this basic characteristic of cognition, and so, of natural
language (see Croft 1991: 163–164; Kemmer 1993a: 209–120). When
a speaker conceptualizes an event, he does so with a certain level of
linguistic background
participants. Clearly, if the effects of the action are perceived to recoil
back to the subject, as they do in the middle, then the transmittal is
less effective than if the Object absorbs the effects of the action in their
totality. In this way, the middle marks a departure from the prototypical
Agent and Object are physically and conceptually distinct to a high
degree, as exempli
ed by the prototypical transitive event, to a mini-
mum, the one-participant event, in which the physical and conceptual
Two-participant Re
exive Middle One-participant
Degree of distinguishability of participants
Figure 8.
Relative Distinguishability of Participants for the Basic Event Types
(Kemmer 1993a: 73).
take an object that is distinct from the subject, e.g.,
She dressed the child,
He shaved the wood
, while many middle verbs cannot. The liminal status
of the re
exive event, its proximity to the transitive event, explains why
exives vary in their conceptualization cross-linguistically, in some
languages being coded as transitives, with the subject “‘separated’ into
cause and effect,” as when a re
exive pronoun serves as the Object of
the clause, but in others as intransitives. The middle voice, on the other
hand, is cross-linguistically closely associated with intransitives (Croft
The difference in the relative conceptual separability of the re
and the middle is re
ected iconically in morphology (Haiman 1983).
Many languages possess both middle and re
exive markers. What is
universally true of all such languages is that the middle marker has less
in terms of the number of segments than its re
ive counterpart (see Kemmer 1993a: 24–28). Signi
cantly, this holds
true even in cases in which the middle and the re
exive markers are
historically and morphologically unrelated (Haiman 1983: 797–798;
linguistic background
therefore presumed to be one and the same. And the middle marker
ects this fact. The lack of complexity of the subject from the per-
spective of the middle—its “lower degree of ‘conceptual weight’”—is
iconically re
ected in the lesser phonological heft that is afforded to it
(Kemmer 1993a: 65–66).
Much of what has been written in this section about participant
distinguishability as a graded quantity can be inferred from the list of
typical middle-marked situations presented in §2.9.5. In fact, Table 3
is organized, from situation type 1 through 9, according to a rough,
intuitive judgment of relative participant distinguishability (see Kemmer
1993a: 67–81; 1994: 210). That is, since participant distinguishability
is a scalar notion, there are gradations within the middle itself with
respect to the conceptual separability of participants. For the tenth
situation type in Table 3, naturally reciprocal events, distinguishability
of participants is not relevant; rather, this event type is discussed below
in connection with the relative elaboration of events.
With the
rst class of verbs in Table 3, grooming and body-care
events, it is possible to conceive of the event in terms of an Initiator
and an Endpoint that are on some level distinct. Although obviously
belonging to a common entity, there are distinguishable parts of that
entity, e.g.,
wash one’s hair, shave one’s beard
, as “the lack of differentiation of Initiator and
Endpoint is part of the semantics of the verb” (Kemmer 1994: 210).
Also at the extreme low-end of the spectrum are spontaneous events,
grow, die, sink, shine
, events that by de
nition code only one partici-
pant and so are necessarily intransitive. Absent a distinct Initiator, as
the event is portrayed as occurring independently of a speci
ed cause,
it is impossible to speak of distinguishability—it is, perforce, zero.
* * *
The relative distinguishability of participants, as I have noted, is but
Jack and Joe arm-wrestled for ten minutes
. Jack and Joe arm-wrestled with each other for ten minutes
In [26]a the event is presented as a single, unitary action, without any
suggestion of internal structure. The implication is that there was a
single arm wrestling bout in which Jack and Joe struggled for ten min-
simultaneously. Sentence [27]b, in contrast, lends itself, in the opinion
of many native speakers, to an alternative conceptualization of the
event. Speci
cally, the sentence suggests there was a sequence of kisses,
Among the native German speakers with whom I have discussed [27]a–b, some
Collective events share a common semantic basis with reciprocal events
to the extent that each participant can be viewed as ful
lling two
roles, that of Initiator and co-Initiator or “companion” of the other
participants (Lichtenberk 1985: 28; see also Kemmer 1993a: 123–125;
Kemmer 1993b). Often reciprocal or middle marking is used to express
collective events that designate group actions with multiple participants.
For instance, in Shona there is
Initiator and Endpoint identity—subject-affectedness, in more traditional
terms—and it subtends the property of the relative distinguishability of
participants. The degree to which an event is elaborated is essentially
a matter of perspective and granularity. It depends upon the perceived
linguistic background
revolves around high agentivity and high animacy, with
all uses of the pre
x being explainable, directly or indirectly, in terms
of these two interrelated properties (§2.2). The pre
represents the
Initiator or Actor perspective on the event. It is a perspective that may
active voice, particularly when contrasted
, a pre
x that commonly occurs in similar environments as
but that neutralizes much of the functional thrust of the latter and so
may be said to express the
active voice in these same contexts.
When a Sumerian speaker chose to pre
x the verb of the simple one-
and two-participant clause with
as the initiator and controller, literally or
guratively, of the action
denoted by the verb. In comparison with
does not
expressly indicate the subject’s affectedness by the action. This is not to
say that the pre
x signals that the subject is unaffected by the action.
Certainly with many intransitive events, which occur commonly with
, the subject is affected. Rather,
is neutral to the property of
subject affectedness. The pre
x signals that the effects of the action are
conceived as accumulating outside the
this basis, these low transitive events are modeled after the prototype,
representing assimilations to it.
The particular stress that
, or more broadly,
the starting point or Initiator of the event, at the expense of the
Endpoint, has some predictable consequences. The pre
x can express
situations in which the Endpoint of the event is severely de-empha-
sized and the focus is entirely on the starting point and the initiation
of the event. Similarly, in contexts in which the speaker is interested in
the action itself, rather than its Endpoint, the pre
x can emphasize the
of the action. In grammar and in cognition
are notions that relate to the Initiator role, they belong to the
rst half
of the action chain (§§2.6, 2.7), the half with which
concerned (see Langacker 1991: 321–322, citing the psycho-linguistic
semantic rating experiments of Osgood, Suci, and Tannenbaum 1957).
It also follows that the pre
x, in low transitivity clauses, is associated
with atelic events, that is, events without a natural end point. The cor-
relation of
with atelic events is naturally strongest when the subject
The Prototypical Transitive Event
I begin with what I have described as the canonical environment for
, the exemplar to which all other uses assimilate: the prototypical
transitive event. The idea is a re
nement of the position that Gragg
rst to voice, namely, that
“represents the focus
on an explicit or implicit animate agent” (1972b: 210; similarly, see
Michalowski 2004: 44). In the prototypical transitive event, the primary
focus is upon the Agent, who controls the action—who is volitional,
animate, and highly topical—and upon the action itself. The Object,
on the other hand, is out-ranked by the Agent in terms of topicality,
is secondary in terms of focus, and is non-controlling, non-volitional,
verbs, when construed with
, represent situations that
deviate signi
cantly from the prototypical transitive event.
In the following passages, the verbs
in the sense of ‘build (s.th.)’,
‘fashion (a statue)’, ‘give birth (to s.o.)’,
‘create s.o./s.th.’,
‘excavate (a canal)’ are coupled with the pre
, exemplifying
prototypical transitive events with an object of result in Sumerian.
The pre
x is so well attested with these verbs that few examples are
necessary to make the point.
[28] Ur-
e . . . e
‘Ur-Nanshe . . . built the temple of Nanshe, built the
sanctuary of Girsu, built the Ibgal, built the Kinir, built the
Edam, built the Bagara, and built Abzueg’ (Ur-Nan
š3-ir mu-tud
fashioned (the statue of) Eshir . . . built the Aeden, built the Nin-
gar, built the Egidru, built the wall of Lagash, and fashioned
(the statue of) Lugaluru’ (Ur-Nan
e 17 iii 1–v 2). Lugal-uru is
possibly for Lugal-urub
[30] sipad-me e
‘I, the shepherd, built the temple’ (Gudea
Cyl. B ii 5). Similarly, e
‘(Gudea) built the temple’
[31] ama Laga
-bi kur-ku
‘The mother of Lagash, holy Gatumdug, gave birth to the
bricks among the waves (of amniotic
uid)’ (Gudea Cyl. A xx
4-bi lu2 hul-gal
prison’s) brick walls crush evil men, but give rebirth to honest
men’ (Nungal A 56).
Šul-gi a-zu [šag4 k]ug-ga ud ba-an-ri-a ama ugu
en3] dingir-zu kug
en3] ‘Shepherd Shulgi, when your seed was
poured into the holy womb, your birth mother, Ninsumun, gave
birth to you. Your (personal) god, holy Lugalbanda, created you’
sag gi
‘After An, Enlil, Enki, and Ninhursag had created the
black-headed people’ (Flood A11–12).
[35] ud e
-gal mu-bi id
‘When (Ur-Namma) built the temple of Enlil,
he excavated the canal named “Nanna-gugal,” the boundary
canal’ (Ur-Namma 28: 8–13).
2 E2-babbar e
‘(Sin-iddinam,) the one who built the Ebab-
bar, the temple of Utu, and who excavated the Tigris, the broad
river’ (Steinkeller 2004: 141 i 15–18).
At the pinnacle of highly transitive situations, the prototype among
the prototypes, stands
, the most transitive of events. With respect
to perfectivity, one of Givón’s three criterial properties [6], the event is
“inherently instantaneous and punctual.” And there can be no question
of Object affectedness nor of the ef
ciency with which the action is
transmitted from Agent to Object, as it “produces the ultimate effect”
on an Object (Bakker 1994: 40; see also Tsunoda 1985: 387). Exhibit-
ing a similar degree of transitivity is
(and like events), although
it is not as quintessentially transitive as the
event, since the lexical
frame of
does not require that the action be instantaneous or
These events in Sumerian, commonly represented by the verbs
‘kill’, appear overwhelmingly
with the pre
in the active construal [37]–[47]. In the case of
the verb with
attestations of
, which universally denotes the
event (§5.1.5;
see [503]–[507]). The association raises the issue of the relationship
and the semantics of causation. Since
correlates with
high transitivity, one might expect it to be used whenever there is an
increase in valence—that is, to appear regularly in causative contexts—
which is not the case. The reason for this is that not all causatives are
alike. The verb
in English is a lexical causative, which differs from
analytic causatives of the type
. Lexical causatives, e.g.,
the spider, I felled the tree
, express direct causation in which there is direct,
chapter three
erim2-du di erim2-e ba-ab-sum-mu hul-gal
‘(Inana) delivers a malevolent judgment for the evil and she
destroys the wicked’ (IdDgn A 120).
[44] uru-a nin
-mu mu-un-til
-le-en ku-li-mu
the city my sister gave me life, but my friend killed me!’ (DzD
153). Note that
, is a direct
3 kug-ga ma
gaba-na i
Silver too held against his chest the goats, oxen, and sheep he
slaughtered’ (CpSv A91).
[46] ur-sag-e A
‘Because the hero had killed
Asag’ (Lugal-e 698).
e-sir2 Unug
‘Gudam crushed a multitude on the
chapter three
In some contexts an otherwise highly transitive event has self-bene-
factive overtones. The use of
in these cases indicates the speaker’s
wish to focus upon the Agent’s role as Initiator rather than Endpoint.
In [52] and [53] the action is undertaken in the subject’s interests;
similarly, [54] is a re
exive event (consequently, Initiator and Endpoint
are coreferential; see also [105]). An analogous use of the active is
found in Classical Greek, where the active may be used in lieu of the
middle “when it is not of practical importance to mark the interest of
the subject in the action” (Smyth 1956: 393). Consistent with the notion
that the active is neutral to subject-affectedness, the active may alternate
with the middle, “if it is inferable from the context that the action is
performed in the interest of the subject” (Allan 2003: 25).
š2] mu m[u-(na-)sa
] ‘(Lugal-Anda) fashioned his (own) statue
and named it “Lugal-Anda-nuhunga never tires (in his efforts)
for the Girnun”’ (Lugal-Anda 2 iii
Šul-gi nita kalag-ga lugal Urim
-ma lugal Ki-en-gi Ki-uri-ke
-hur-sag e
of Ur, king of Sumer and Akkad, built the Ehursag, his (own)
beloved house’ (
šag4 gur4-ra ni2-bi ak-a-ba ni2-bi mu-un-gul-gul
‘(Your) arrogant
heart will destroy itself by its own deeds!’ (BdFh 128).
The pre
x also commonly occurs with the causatives of verbs that
are typically univalent (or bivalent in the case of compound verbs) and
semantically middle in the sense of the situation types enumerated in
Table 3, e.g.,
‘shine’ [55]–[66]. As such, these verbs com-
monly occur with the pre
when construed as transitives,
, underscores the speaker’s percep-
tion of the event as an act of direct causation, stressing the conceptual
še3 giri3-ni-še3 mu-nu
(var. i
2 kug-sig17-gim mu-e-zalag-e
‘You make justice shine like
gold’ (Enlbni A 76–77). Kapp (1955: 78) restores [lu
just man’. Cf.
–3.2.2 Low Transitivity Events
At the far-end of the spectrum,
is encountered with events that
are low in semantic transitivity, many of which are, in fact, syntacti-
cally intransitive. Again, the use of the pre
x—that is, the rationale
for conceptualizing events that are ostensibly far removed from the
prototypical transitive event on the pattern of this prototype—has its
basis in the speaker’s perception that the subject exercises some volition
or control over the action or state denoted by the verb. The focus is on
the subject as Initiator, although the subject may occupy the Endpoint
role as well. This use of the pre
x conforms to the cross-linguistically
observed pattern whereby the active voice expresses low transitivity (per-
haps syntactically intransitive) events that are perceived by the speaker
to be neutral to the property of subject affectedness (§2.9.5).
Even when there is a distinct Object, the Agent’s control, or
of force
chapter three
Klaiman (1988: 66) describes a similar classi
cation on the basis of control for the
Oto-Manguean language Chocho, which attests what are referred to as active intransi-
The following examples [67]–[69] are syntactically transitive (with
compound verbs in [68] and [69]). Semantically
er, these are
low transitive events. It is impossible to speak of affectedness or a
tives (verbs marked in the ergative case), e.g., ‘arrive’, ‘dance’, ‘go’, and inactive intransi-
tives (verbs marked in the absolutive case), e.g., ‘die’, ‘tremble’, ‘slip’, ‘sneeze’.
in abilative function; see also
‘do, act’ in [472];
I thank Claus Wilcke for pointing out the frequency of this combination and for
providing me with dozens of references (including MVN 5, 26: 5–6 [72]) from Ur III
economic texts.
dNa-ra-am-dSuen mu 7-am
‘Naram-Sin was
immobile for seven years! (lit.
xed [a period of] seven years)’
-le-na mu-un-til
-en ‘You must swear to me that you live there, that you
live there! Brother, you must swear to me that it is just that you
live in the outskirts of the city!’ (DzIn B 13–14).
The verb
‘know’ nicely captures the meaning of
with low
transitive situations of various types, illustrating the common semantic
denominator that these share with the prototypical transitive event.
does not simply express a state of knowing, as it does,
for instance, with the pre
. Rather the union
focuses on
the subject and his ability, if perhaps only
guratively, to control what
he knows (cf.
[645]–[653]). The subject is credited with having
certain powers of discernment, exhibiting a mastery over some knowl-
edge, or possessing some skill or expertise [80]–[89] (see also [137],
[138]). The verb may take an Object in which case it readily lends
[83] lugal uru
‘(A man) should know
how to revere the lord of his city’ (InUrN 66). Similarly,
gal-zu ni
(var. [h]e
-en-zu) ‘You should
know how to revere your older brother’ (CnWd 80).
[84] mu
‘Bird, with variegated plumage and variegated face, was con-
vinced of his own beauty’ (BdFh 52).
dumu ur-sag
-da nir mu-un-gal
‘Summer, heroic son of Enlil,
was convinced of his own might, and therefore trusted in himself’
š-bar dug4-ga-ni ki-bi-
you understand how to establish and carry out the decisions
rmly pronounced (by Enlil)’ (UrNin C 15).
[87] ur
mu-ti-in ud-te-en-
-a ‘I know how (lit. where) to give your loins pleasure: Sleep,
man, in our house till morning! I know how (lit. where) to make
your heart rejoice: Sleep, lover, in our house till morning!’ (
[88] sipad sag gi
‘Shepherd, you understand how to keep a check on the black-
headed (people). Ewes along with their lambs come to seek you
out, and you understand how to wield the scepter over the goats
, and the highly transitive action, the most salient of actions, an
action that yields a perceptible, lasting effect in some Object.
The pre
x’s close bond with agency is corroborated by its interaction
with the pre-radical agent pronouns. Speci
cally, there is a tendency,
particularly in third-millennium texts, to omit these pronouns when
they would appear immediately after the
x. As most scholars
today would acknowledge, this is primarily a graphic phenomenon.
Remarkably, however, the omission of these pronouns in early texts is
not randomly distributed with respect to the pre
xes, suggesting that
the exclusion of these morphemes in writing, and perhaps even to a
certain degree in speech, is partly conditioned by the pre
xes themselves.
The omission of the pronouns is most conspicuous when the pre
immediately preceding the verbal root is
Indeed, the avoidance of
the agent pronouns with
led Yoshikawa to the noteworthy conclu-
sion, with respect to Ur III texts at least, that “the in
x -n- is basically
not compatible with the pre
x mu-” (1991: 389). In contrast, in these
same texts there is the relatively commonplace attestation of
; for
, see p. 136 n. 8). One
Early attestations of the agent pronoun appearing immediately after
Šu-Sin 4]) and mu-un-GAM (Ibbi-Sin YN 14)—references
courtesy of C. Wilcke.
If the pre-radical pronouns could express the agency of the subject or assumed this
function secondarily (a broader use than simply indicating the presence of an Agent
in the transitive clause), this could account, in some cases, for the otherwise dif
presence of these pronouns in intransitive contexts. This possibility, h
problem more broadly would naturally require systematic study.
you lied! You have plotted (against me even though) you had
sworn an oath by the lives of your birth mother, Ninsumun, and
your father, holy Lugalbanda!’ (GgHw-B 137–138).
[91] a-gim
-zu nam-gur
-ra im-de
en for sun
; note the var. mu-un-sun
) ‘How has your
heart become so arrogant, while you, yourself, are so lowly?’
[92] ur-sag hu
me kur-ra an-gim dugud-da-am
me Eridug
ga ki-gim mah-am
l2] ‘O Þ erce warrior, you have taken
’s, which are like heaven! Son of Enlil, you have taken
’s, which are like earth! You have taken up the
’s of
the mountains, which are as heavy as heaven! You have taken
up the
’s of Eridu, which are as massive as the earth!’ (NinRN
Civil reaches a similar conclusion when he writes of the form
in [93] that “the omission of the dative and object in
xes, as well as
the pre
, topicalize the agent and, secondarily, the action itself”
(1999–2000: 187 ad 104; see also Edzard 1976: 165).
3 zi mu-e-sum
(var. ma-an-sum) Ak-ka
(var. ma-an-sum) ‘Aka, you give breath! Aka, you give life!’
Similarly, stressing agency at the expense of the expected expression
of the dative:
dEn-lil2 nam-kug-gal
(var. mu-de
) ‘Father Enlil, you gave control of irrigation! You
brought plentiful water!’ (WnSm 289).
nam-mu sum-ma-zu Dilmun
nam-mu sum-ma-zu) ‘You have
given a city! You have given a city! But what is your giving to
me? You have given a city, Dilmun! You have given a city! But
what is your giving to me?’ (EnkNh 30–31).
The claim that
stresses the verbally denoted action itself, which
share this focus: verbs with reduplicated or triplicated stems and, more
conspicuously, verbs that are accompanied by adverbial expressions of
manner. The pre
x occurs in both contexts with predicates that lexi-
cally express middle or low transitivity events, which shows that the
x in these particular instances is not triggered by the semantics of
high transitivity. The iteration of the verbal root, among other things,
may iconically express an intensi
pens shriek continuously like the one who wanders the desert,
[102] nig
-e gu
2-e] eden Ke
‘The magpie calls out and wails
continuously in the steppe of Kesh’ (Nan
gi-zi di
-al il
-la sag si
-ba mu
ad mu-ga
‘In the midst of
your carefully tended small
-reeds with erect offshoots,
of sparrows twitter away as in a holy swamp’ (
Similarly, with the af
rmative verbal forms in [104], reduplication
stresses the intensity of the actions in which the
subject engages—continuous, atelic body-action middle-type events (cf.
[91]). The actions are further underscored by the respective adverbial
expressions (see below).
-bi dal-la-gim
(var. a
-mu hu-
igi il
(var. dub
-mu hu-bad-bad) ‘Like a
dove frantically
eeing from a
-snake, I
apped my arms.
Like Anzu lifting its gaze to the mountains, I
exed my knees’
)sag il
‘walk proudly’, ‘walk eagerly’ (lit. with head
–gub ‘stand’ and gen ‘go’, as predicates
that lexically denote middle domain events, appear frequently with
in other contexts (see §§ 4.1, 5.1; for
contexts, see §3.5). H
owever, wh
en they occur in these adverbial idioms,
which describe the nature of the action, they are regularly joined with
. The phenomenon
nds a parallel in the active, as opposed to
middle, construal of many
manner of motion
verbs in Classical Greek. As
described previously (§2.9.5; Allan 2003: 243–247), these predicates tend
to describe atelic events in which the focus is on the action and how
it unfolds rather than on subject affectedness. Of course, this is not to
say that the subject is unaffected—on the contrary, subject affectedness
is implicit to many of these events. Rather the active voice in Greek,
like the verb joined with
in Sumerian, is neutral with respect to
chapter three
to the context of discourse. Consider, in this light, the verb
increase’, which typically takes a direct object, ‘add s.th. (to s.th.)’ (see
pp. 280–281 n. 20). In the Ur III personal names of the type
[121], often associated with onion cultivation, occur exclusively with
in these texts (see Yoshikawa 1979: 200 for references and a dif-
ferent explanation for this correlation). As in [119], the Object, the
produce, is not directly part of the structure of the clause, but is enu-
merated at the beginning of the text.
However, what
an antipassive meaning in these instances is the nature of the action.
As is true of many agricultural activities, these are not punctual events,
ing—being unconstrained to the beginning of the event as is
to its end as are
—the pre
x most often alternates with
as representative of the active, unmarked voice category. The pre
is used when there is a discourse requirement to defocus, or more
cally to background, certain information, decreasing its topicality
and salience vis-à-vis
is also used for stylistic reasons, when the
highlighting of the agency or animacy of an argument, as indicated by
, does not serve the broader discourse agenda. As such, the pre
x is
a pragmatic option for the neutral reporting of information, for relating
a message without particular emphasis on any one part or participant.
The pre
x, then, is not—as is often claimed—without function. Rather,
its function derives from its neutrality, allowing it to operate as a back-
grounding device (see Vanstiphout 1985: 14, who arrives at a similar
conclusion). Thus, while
c purpose of
altering transitivity, the decrease in topicality that
pragmatically correlate with a decrease in semantic transitivity vis-à-
. This result is anticipated by our discussion of topicality and
chapter three
In connection with what was discussed in §3.2.2 regarding
and agentivity
and the suggestion that the pre
x renders the pronoun to a certain extent redundant,
observe that the Agent pronoun
regularly occurs with the pre
in the verbal
). That is, there is a greater pressure to express the pronoun in this case,
, unlike
3-de2-a ensi2 Laga
‘Gudea, ruler of Lagash, the one who built the Eninnu of
Ningirsu’ (Gudea St. G i 5–10).
[124] Ur-
-su . . . lu
alan-na-ni mu-tud ‘Ur-Ningirsu . . . the man who built the
Eninnu of Ningirsu, fashioned his (own) statue’ (Ur-Ningirsu
3-de2-a ensi2 Laga
(var. in-du
-a) ‘Gudea, ruler of Lagash, the one who built the
Eninnu of Ningirsu’ (Gudea 82).
[126] ud e
‘When (Gudea) built the temple
of Ningirsu’ (Gudea St. B v 21–22).
Relative clauses are not the only representatives of background
information; they are simply one of the most obvious. In the follow-
ing example, the building events denoted by
represent secondary,
occurs with events that deviate from the transitivity prototype.
Negative clauses present another environment in which the force
is occasionally neutralized, particularly when contrasted with
built for Ningirsu—but (Gudea) did in fact build it!’ (Gudea
[137] ur
‘A dog
understands “Take it!” but it does not understand “Put it
down!”’ (SP 5.81).
[138] An gal dug
-ga-zu nig
nam gal tar-ra-zu sag
nam-sipad zi-gal
]-en ‘Great An,
your word is preeminent—who can countermand it? Father
Enlil, no one knows how to undo the great destinies that you
[134] and similar attestations of
in the law codes), these verbs
commonly occur with
Consider the following passage [139] from
a legal proceeding. The form
appears in the neutral presenta-
tion of the background facts of the case. The verb stands in contrast
to the more emphatic
occurring in the reported
testimony, which is naturally disengaged from the legal idiom. Here
the defendant, understandably, seeks to stress agentivity and volition,
and so culpability.
I.gišDur2-gar-ni Ga-li
-gar-ni en
killed Gali. Durgarni was interrogated. (Durgarni) declared “It
was Gali (who
rst) struck at my face with his
st”’ (NG 202:
i- in lieu of expected
(see §3.1 for
) may be seen as
doubly motivated, appearing both in a subordinate clause and within
a legal text:
[140] [Ku-li dumu U]r-e
4] [Ba-ba]-mu nar
še3 ba-gi-in ‘It has been established before the
chief minister that Kuli, the son of Ur-Eana, killed Babamu,
the musician’ (NG 41: 2–4).
And the pre
x functioning as a defocusing device is encountered
, for instance, is often the pre
x of choice in background
narrative sequences (see Vanstiphout 1985), where the goal is, similarly,
to state the facts, to describe a sequence of actions as unitary events
without the semantic thrust that other pre
xes would convey. Relat-
edly, by virtue of being unattached to the starting or ending point of
the event, the pre
x often occurs in static descriptions, situations that
are unchanging over time and for which the expression of agency and
control are not of paramount importance. In this use the pre
x is
Naturally, these events do not occur exclusively with
. Although administrative
texts display a preference for
, in certain contexts and for certain verbs,
is never-
theless well attested in this genre (cf.
‘live’ and
‘be present’ [70]–[73],
as well as
‘grub’ [120]–[121]). A more extensive study of
this phenomenon will ideally analyze the distribution of
on a verb-by-verb
basis across various text genres, controlling for syntactic and semantic contexts.
found in personal names, e.g.,
‘Sara is favorable’,
‘Enlil knows’,
A well-attested waterway in the Ur III period—I thank Jacob Dahl for providing
me with references.
reports is, typically, to convey the basic facts, to report these events
of this pre
with a necessarily animate referent and the most marked of the in
cases with distinct allomorphs for each person. Consequently, in this
representing a tendency, it nevertheless shows that the two functions of
the pre
x—relating to the event and the core participants, on the one
hand, and relating to the animacy of a peripheral participant, on the
other—operate largely independently of one another. This is borne out
in these cases is presumably the animate peripheral argument. Observe,
in this connection, the standard Ur III brick inscription, which typi-
, where the dative refers to a deity, highly
topical and occupying the left-most position, for whom the activity is
Evidence that the voice and animacy marking functions share a com-
mon basis presents itself in the form of the distribution of the pre
with the dative allomorphs. This distribution describes a continuum
that mirrors the Nominal Hierarchy. That is, the frequency with which
occurs with the dative is directly proportional to its referent’s posi-
tion on the Nominal Hierarchy. As Gragg has observed,
is obliga-
tory with the
rst-person dative,
derives from
Þ rst-person dative]), overwhelmingly frequent with the second,
, and is very frequent with the third,
. These correlations
complement the distribution of
-, which is incompatible with the
person, infrequent with the second,
, but common with the third,
, as well as that of
, which is incompatible with the
rst- and
second-person dative and infrequent with the third,
Although not attested in the same numbers, a similar spectrum
of values could be demonstrated for comitative and terminative cases.
Naturally, what is at issue is not the grammatical case marker itself, but
the semantics, or animacy, of the referents. And it is on the basis of the
quality of the referents that the correlation with
is predictable from
the Nominal Hierarchy.
is most closely associated with the
rst and
second persons, the Agents
par excellence
, so much so that the pre
x is
obligatory with the former. In a progressively graded correlation that
mimics the Nominal Hierarchy, the pre
x is all but mandatory with the
second person and optional with the third. As I will describe in §5.3
this graded scale does not end with
and third-person animate NPs,
but continues and ends with
and its correlation with
of the
class, inanimate nominals that belong
to the lowest rungs of the Nominal Hierarchy.
The most intriguing feature of this distribution is the optional nature
with the third person. While
rst- and second-per-
son is
de rigeur
, or nearly so, the speaker has a legitimate choice in the
case of the third person. Thus, the equivalent of the clause
He gave
can be rendered in Sumerian alternatively as
this is a Girsu text: “
RTC 19 is also discussed in connection with the pre
by Scholtz
(1934: 27–32) and Falkenstein (1978b: 164 n. 3). Marchesi (2004: 179) has recently
discussed the content of i 1–iii 2.
What is actually at issue with our text, and with the
more broadly, is
as a function of the Nominal Hierarchy. It will
be recalled from §2.2 that the Nominal Hierarchy, the natural ordering
of nominals, can assume various guises. From one perspective it is a
hierarchy of animacy, from another, of topicality, and from still another,
of agency. The ability of the Nominal Hierarchy to assume various
labels speaks to the interrelatedness of these categories, each focusing
arguments of the second person. With third persons, the hypothesis
predicts that when a speaker or writer uses
expressing a greater degree of empathy for, or identity with, that par-
ticipant than when he uses
is neutral or unmarked with
respect to speaker empathy, while
low speaker empathy for the
. Recall that the
Nominal Hierarchy is a continuum, on which the number of points is
nite, corresponding to all the possible nominals in language. With
the exception of two, the
rst- and second-person pronouns, this spec-
trum consists of third-person nominals. The choice of
be seen as a means of making a dichotomous pragmatic distinction
Although Malgasud and Anedanumea are nowhere in the text explicitly indenti-
ed as belonging to the respective Lagash and Adab courts, this may be safely inferred
based on the fact that a Ma-al-ga(-sud) is well attested in the Lagash e
texts, while
the PN A-ne-da-nu-me-a is, apparently, known only from this text.
Nominal Hierarchy. Further, empathy accounts for many of those
uses of
that Yoshikawa’s topicality and social status theory is at
pains to explain. An example in point is the
nal clause of [151], i.e.,
‘Ningishkimtil gave (one garment)
to Malga(sud)’, for which Yoshikawa is forced to propose, “This usage
may suggest the superiority of Laga
over Adab or topicality on the
side of Laga
[155] nu-siki lu
‘I did not expose the orphan to the wealthy, nor the
widow to the in
uential’ (Gudea St. B vii 42–43; Gudea Cyl.
B xviii 6–7 [broken context]).
accept /
/ as a re
ex of /
/, in some, or even many, cases. If
the morpheme had a limited functional range, much of the presumed
ambiguity surrounding a surface (
would be obviated.
It may indeed be the case that there is a distinct pre
-, but it is
restricted to certain environments. This is suggested by attestations of
the pre
x with the writing
in the Old Babylonian grammatical
texts where it occurs exclusively with the verb
‘go’ and is equated with
. It is a correlation that is corroborated by Ur III attesta-
tions, in which the pre
x occurs, again, primarily with
with the motion verb
‘carry’, and only sporadically with other verbs
(see Yoshikawa 1977a: 230–232; 1978: 463–465, 480).
reserved for verbs of motion in third-millennium texts, to judge from the
vast majority of attestations,
has a broader range of uses in all periods.
, it too conveys a ventive meaning with verbs of motion.
The verb
Babylonian period), indicates motion
, toward the speech event,
I see little evidence, in the Ur III attestations at least, that suggests that
contains a locative element (cf. Karahashi 2000b: 122; Yoshikawa 1977a: 236).
A closer de
nition of the ventive in Sumerian is certainly warranted, but must
await a more extensive study of the phenomenon; for recent discussion of the ventive
in Akkadian, see Hirsch 2002 (particularly, pp. 3–21) and Kouwenberg 2002.
-containers of onions for Ishtup-
Irra—Enlil-Anzu was the commissioner—(when) the king came
from the upper country; 2 (
-containers of onions), when
the king came from the lower country, again for Ishtup-Irra’
(Westenholz OSP 2, 135: 1–8 [Nippur]).
[160] lugal Nibru
(STTI 26 l.e. 2–3 [Girsu]).
This meaning is also discernible in the Old Babylonian period (it is
particularly common in the expression
[163] referring to the
break [lit. the coming
] of day):
-mu mu-ra-tar-r[e] za-e
ki-mu nam-mu-ni-in-pad
‘Your lady, Ninlil, will be com-
ing—if she asks you about me, you must not reveal where I
dDumu]-zid gal
(var. mu-e-re
) ‘Dumuzi,
your demons are coming here (for you)!’ (DzD 90).
ud im-zal dUtu im-ta-e3-a-ra ‘When day had broken and
Utu had come forth’ (In
But often in the Old Babylonian period the pre
x appears to be bleached
of its original deictic meaning, suggesting motion but not necessarily
[164] ga
-e Ak-ka
(vars. ga-gen, ga-an-
i-gen) ‘I
will go to Aka’ (GgAk 57, similarly 54).
[165] me 7-bi
zag mu-ni-in-ke
me mu-un-ur
(var. um-mi-in-[. . .]) ‘(Inana)
bundled up the seven divine powers, collected the divine powers
non-motion verbs), I leave as an open question. The important point is
that I consider the ventive function of this pre
x to apply
verbs of motion and doubt that we are to understand Old Babylonian
attestations of (
- with non-motion verbs as representing—in all
carp—Udu, the fresh-water
sherman, brought them here;
(subsequently,) they were taken to Nina’ (VS 14, 19 i 1–iii 1
[= Bauer AWL no. 132]).
[168] 130 sa-numun
Ambar-kam E
‘130 containers of carp—these are
sh of Ambar-
eld. Eigarasu, the
sherman, brought them
here’(VS 14, 139 i 1–ii 1 [= Bauer AWL no. 142]).
[169] (totals of dates, apples,
gs, and grapes) nu-kiri
‘The gardeners of Bau brought (the dates, apples,
and grapes) here’ (DP 107 vii 1–3).
bar še-ba-ka lu2 he2-ši-gi4-gi4-a-ka še-mu
-a-ka ‘Because of that barley, (Enanatum I) sent envoys
to (Urluma), having them say to him, “You must deliver my
barley!”’ (Uruinimgina 3 iv 1
2-inim-ma-bi nu-mu-da-de
‘He was not able to bring forth
the witnesses’ (NG 82: 6). Similarly, NG 132: 10, 169: 12.
1NE-da-ti A-gu-za nu-banda
‘They said to NEdati and Aguza, the
captain, “Bring him here!” Aguza declared, “I’ll bring him”’
(NG 121: 10–13). Note the use of the deictically neutral
the second clause, for which the deictic context is established.
[173] ud
(Shu-ilishu) brought back (the statue of) Nanna from Anshan
id2Idigna id2Buranuna
‘Water, which the Tigris and Euphrates had brought
since ancient times’ (Rim-Sin 15: 9–10). Note the erroneous
placement of the pronoun,
; the text is a school copy from
2-kur-re e
‘I brought the arts from the Ekur, the temple of Enlil,
to my Abzu in Eridu’ (EnkWO 66–67).
[176] a-a-mu
‘My father Enlil brought
me (i.e., Anzu) here (to the mountain)’ (LgB 2: 101).
2 da gal-la-
] [ki-s]ur-ra [
2-ka-ke4 [b]a-ra-mu-bala-e
‘Forever and evermore, I shall not transgress the
boundary of Ningirsu!’ (Eanatum 1 rev. i 16–19).
The parallelism between space and person, between
and the
location of the SAPs, which underlies the functions of
, is given its
Where the pre
(i.e., ventive (
) and
differ, as the
above comments anticipate, is with respect to transitivity. The former
is largely limited to the intransitive verb
‘go’ in early texts, whereas
, in ventive function, is typi
ed by its high correlation with the
transitive verb
The reading of DU as
, rather than DU as
, is based on context and
Note, in this connection, the corroborating claim of Attinger (1993: 270–271)
and Krecher (1985: 149–152) that
usually implies the presence of an Agent
pronoun. More broadly, Attinger (1993: 270–280), taking a more formal morphological
chapter three
guruš ud 1-še3 ki-su7 Igi-E2-mah-‹še3-›ta e2-duru5 dAmar-
‘71 men for one day, who will bring the
grain from the Igi-Emahshe threshing
oor to Amar-Sin-village’
[180] 1 ma
e ma
‘One grain boat, which
our boat (and) brought here’ (Watson Bir-
[181] ki
‘The sealed document of
Abbasaga was brought forth’ (YOS 4, 318: 3).
speaks to the greater functional range of
, which includes non-allative as well as
allative uses.
The pre
ed by events in which the subject is affected
by the action that he brings about, events in which the Initiator and
the Endpoint are the same entity. As such, the pre
x is a middle-
voice marker, a claim that
nds support in the pre
x’s high degree of
correlation with situations that are commonly middle-marked cross-
linguistically (see Table 3), particularly body-action, translational motion,
self-benefactive, and mental situations. With its primary focus on the
affectedness of the subject,
views the event from its Endpoint. In
this way the pre
x differs critically from
, which is neutral to subject
affectedness and revolves around the Initiator-Actor role. There has
long been a broad, if vague, consensus that the pre
x is functionally
, even regarded by some as, essentially, an allomorph of
. It is an assessment that is mirrored in the morphological derivation
that most scholars subscribe to, even if differing in
See, in particular, Attinger 1993: 280–281; Falkenstein 1978a: 184; Jacobsen 1965:
83–84; Postgate 1974: 19–22; Shaffer 1969: 437–438; cf. Michalowski 2004: 44, who
to be a reduplicated form of
(see also Rubio 2007: 1346, 1363).
chapter four
with these predicates, may, more often than
, add a re
exive reading
to verbs that are not naturally middle, e.g.,
‘build s.th.
for o.s.
cing [for himself, as the one making
the offering]’—see §2.9.5). But there are also differences in the types
of middle situations to which each gravitates. With
the subject
is more likely to exercise some degree of control and volition over the
action, while
correlates most highly with middle events in which
the subject is a non-volitional
, e.g.,
introduces the notion of subject as Actor, which is largely absent from
the expression with
. In other cases, the greater complexity that
implies is associated with a certain plurality, either in terms of the
participants or the event itself. In general, where there is a pragmatic
chapter four
this way from
, with which the subject more often plays a
-like, or Undergoer role in the event. Both pre
xes stress, in
typical middle voice fashion, the subject’s affectedness by the action he
initiates. H
as noted above, with
there is often a complexity
to the event that is not suggested when
is the pre
A parallel that may speak broadly to the semantic difference between
may be found in the use of the re
exive pronoun in Eng-
lish as opposed to the unmarked verb in certain contexts (see Kemmer
1994: 202–209). In pairs such as
She got herself out of bed
of bed
The child dressed himself
The child (got) dressed
that include
exive pronouns lend themselves to an emphatic
or contrastive meaning. There is a suggestion of greater accomplish-
subevents, there is a degree of elaboration of events (Kemmer 1993a:
58). These are the situations that correlate highly with
x serving to
pull apart
the event, exposing internal structure with
respect to the subject or the action itself. Often this amounts to an
emphatic or contrastive nuance, with the pre
x stressing the distinction
chapter four
dative in
in one of the variant texts, which prompts the
use of
(see p. 277 n. 16).
2 kug-ga-am
munus-e id
‹a› im-ma-ni-tu
) ‘The river is pure—the woman bathed in
the river that is pure’ (EnlNl 23).
dEn-ki-ke4 igi-ni im-ma-an-sig7-sig7 gidru šu bi2-in-du8 dEn-ki-ke4 dUttu-ra giri3 im-ma-an-gub ‘Enki (in the guise of a
gardener) made his face beautiful and took a staff in his hand.
collective events, collectivity sharing a common semantic basis with the
middle (see Lichtenberk 1985: 29; Kemmer 1993a: 123–125; Kemmer
1993b). What the heavy marker
presumably expresses here is
the notion of
Note that the inclusion of the dative (here as well as the variants to ll. 8 and 9)
and the accompanying separation of the pre
x from the verbal root prompts the use
(see p. 277 n. 16).
chapter four
, mu-un-na-sug
) ‘The Anunaki
2 dNin-gir2-su gu3-di-de3 im-ma-gub ‘Ningirsu stood up in
order to address the temple’ (Gudea Cylfrgm. 2 iii
[202] ud i
‘I go to it by day, I go to it under cover
of night—I always stand by the lamentation drum!’ (SP 7.22;
similiarly, SP 1.167).
[203] en nig
u nu-gi
(var. am
-gub) ‘The lord
who does not hold back his goods stands to be admired’
(TmpHym 138). Cf. [394]–[395].
In the case of events that are conceptualized as indirect causatives, the
x may re
chapter four
And demonstrating varying degrees of overlap with the emotion middle
category (§§4.1.3, 5.1.3):
Note, once again (see §3.1) that instances of indirect causation often have no effect
on the construal of the event as a middle action on the part of the causee:
-ra ninda mu-un-ku
(var. im-mi-in-tu
) An-ra
(var. im-ni-in-us
Nin-tu zag gal-la
(var. im-ma-an-tu
A-nun-na ki-us
) ‘In the shrine Nippur, Enki provided a meal for
Enlil, his father. He had An take a seat at the lofty place and Enlil sit next to An. He
‘Gilgamesh craned
his neck over the wall’ (GgAk 89, similarly, 66).
3-da gu
-mu im-ma-an-la
‘I have craned my neck over
[214] nin
(var. im-ma-an-
-ga) e-ne
(var. bi
‘(Ereshkigal,) her sister, arose from her throne, and instead
(Inana) seated herself on her throne’ (InD 165–166).
[215] abul Ganzir igi kur-ra-ka
‘At the
chapter four
[221] kur-kur-re
sag im-ma-da-sag
(vars. im-da-(ab-)sag
‘All the lands tremble with fear’ (Enlil A 72).
As is true of virtually every use of the pre
xes, pragmatic factors
may take precedence in the construal of an event, dictating the choice
of pre
Note the appearance of
rst clause and
chapter four
have sex with her, a man would very much like to kiss her’
[229] lug
al-mu MES-ga-e dirig-ga-am
-bi na-mu-un-dug
ne-bi na-mu-un-su-ub
MES-ga-e dirig-ga-am
-bi na-mu-un-dug
ne-bi na-mu-un-su-ub
(var. ne-bi
) ‘My master, . . .,
oating downstream—he was
really going to have sex with her, he was really going to kiss her!
Father Enlil, . . .,
oating downstream—he was actually going
to have sex with her, he was actually going to kiss her!’ (EnlNl
In the following example, note the use of
rather than
subordinate clauses (the second clause is reconstructed on the basis
of the
rst). The backgrounded quality of these clauses suggests that
the events represented are viewed as unitary wholes, lacking internal
complexity and intensity that is signaled by
(see §5.2.1 for the backgrounding functions of
; cf. [288]).
ni-in-su-ub] gi
a ne ba-ni-in-su-ub-ba] zag sar-ra-ka-ni im-
The exeception in [232] is the last clause, i-
Utu kur-ra
, lit. ‘The
chapter four
primary functions of middle markers cross-linguistically (cf. [18]–[21]).
Remarkably, the earlier parallel passage in this text, which frames these
very same events as an admonition from Gilgamesh to Enkidu not to
undertake these very actions (GgEN 184–198), avoids
affectedness is not at issue. Many of these events in GgEN 184–198 are
expressed with
merely associated with motion, but also with direction. Moreover, the
deictic associations of the pre
chapter four
suggested by such common locutions as
asked by a guest
on his host’s doorstep. In asking the question, the guest adopts his host’s
perspective and rejects his own (note the host’s answer,
Please, Come in
Idiosyncratically, deictic transfers of this type often provide the most
natural construal of the event (compare, for instance, substituting
). It is always possible with motion events to
replace the location of the speaker, or the speech act, for some other
spatial reference point.
But there are more profound reasons for believing that this apparent
the motion, then, is itself quite secondary. Rather, it is the inchoative
phase of
Commenting on
cally, Kemmer, for instance, notes (based on Hatcher
1942) that “in
s’en voler
y away’, the focus is on the fact that the
ying entity is no
longer in its initial state; it has
own, and is gone from at least the immediate vicinity.
er, refers to the act of
ying without any speci
c reference to the change-
of-state, or to the beginning or ending point. In other words the morphologically
unmarked form is associated with the ‘semantically unmarked’ meaning” (1993a: 157;
chapter four
observations have been made by Allan (2003: 245) for Classical Greek
and, outside of the Indo-European group,
by Klaiman (1991: 58) for
Fula—in both, the middle voice correlates with telic motion events and
the active with atelic events. Of course, we need not go so far from
home to corroborate this phenomenon. The separative sense of the
Akkadian Gt-stem presumably grew out of the
-stem’s primary re
and reciprocal—that is middle marking—function precisely along these
same lines (see Kouwenberg 2005). Indeed, observing the regular cor-
Also outside of the Indo-European languages, note Hungarian
ee’, and Lingala
‘run (away), hurry’ (Kemmer
I would not go as far as Kouwenberg (2005) does in his enlightening article on the
tive—but not without quali
cation. In many of these cases the ventive
meaning of the pre
x is underscored by the absence in the clause of an
of the motion is implicitly understood to
by employing this pre
x. Howev
er, there are also many instances,
particularly from later literary texts, in which a ventive meaning is
doubtful or is at odds with the broader context, and the pre
x appears
merely to stress the inchoative aspect of the motion event without any
particular suggestion of direction.
Rather than presenting a contradiction, h
these cases under-
score that
is foremost a middle marker, conveying an inchoative
or ingressive meaning, and that motion events are not fundamentally
different from other middle situations. Although the pre
x may impart
a ventive sense with verbs of motion in certain contexts, this is very
much a secondary function. In all cases the pre
x emphasizes subject
affectedness, which in the particular case of motion events is associated
chapter four
took away 30
of grain on the
(system) to Saggatur-
eld as fodder for the plowing donkeys’ (VS 14, 16 iii 1–iv 3
[= Bauer AWL no. 39]).
[235] 1 udu nita Il
1 udu nita Lugal-sa-
-gal [eger
udu nig
. . . [x udu nita] Ur-
1 udu nita Lugal-sa-
-gal udu siki-
‘1 ram from Il and 1 ram from
Lugalsashushgal, after being sheared, were led to the fattened
‘dig up [and bring from
mng. 5;
[241] mu-bi-e an-zag-ta kur-kur-re
-gan Me-
luh-ha kur-bi-ta
‘Because of the (Eninnu’s)
chapter four
, represents a more emphatic or intensive portrayal is sug-
Similarly, compare
rst clause) vs. asseverative
(uttered) by Enlil, picked up the tree in her hand and brought it
into Uruk. She herself brought it into Inana’s luxuriant garden’
(GgEN 32–35, similarly 75–78, 119–122).
Mental Events
Mental events, which include events of cognition, emotion (and emotive
speech), as well as perception, are often middle-marked cross-linguisti-
cally, as the salient participant—the sentient being in whose mind the
event unfolds—is affected by the action or state predicated by the verb,
thereby serving as both the Initiator and the Endpoint.
is an
and the mental event that he experiences
is, in certain instances, induced by a
. The semantic roles
[2] are re-introduced here to account
for the cross-linguistic, indeed often language-internal, variation, by
which the
may be coded as subject or object,
depending on language or context (Croft 1991: 213–214). When an
reacts to a
(Fig. 10–A), there are two participants
and a two-way causal process is involved. The
rst makes
—a turn of phrase that seeks to liken
the mental event to the prototypical transitive event and its
of energy
(§§2.6, 2.7)—initiating the event by directing, volitionally or
non-volitionally, his attentions to the stimulus. The
, in turn,
induces a mental state in the
, e.g.,
I’m afraid of spiders, I
like pizza
, and affectedness is naturally of the mental, rather than of
For the semantics of mental events and the middle voice, see the discussions of
Croft 1991: 213–225 and Kemmer 1993a: 127–142, upon which this description is
chapter four
Figure 10-A.
Two-Participant Mental Event
(Kemmer 1993a: 128).
Figure 10-B.
One-Participant Mental Event
(Kemmer 1993a: 128).
Similarly, implicit to the semantics of many cognition verbs is the
expectation that the
is to some degree a volitional par-
ticipant, who exerts some control over the event. Consequently, these
verbs often appear as active intransitives (Kemmer 1993a: 135, who
points to Eng.
, Lat.
, Fr.

; cf.
[§3.2.2]). H
chapter four
the former may be found in the different types of
verbs attested
According to Karahashi (2000a: 113),
‘look’ expresses inten-
tional visual activity, while
‘see’ expresses visual experience. This
distinction was already observed by Gragg, who suggested the existence
of semantic subclasses among verbs of visual perception, with verbs of
class being marked for “intentional apprehension, direc-
tion of attention” and differing from
as English
does from
(1973c: 22). Although one would not expect either verb to occur
frequently with
, particularly when the verbal chain includes case
elements with animate referents, our theory of voice would predict
to occur more frequently with the experiential
the more agentive
(see [69]). This is in fact the case. In the 36
examples of
collected by Karahashi (2000a: 113–117), there is
only one certain attestation of
, with the verb commonly occur-
ring with the pre
. Remarkably, in those examples that
ttingly rendered by Cooper in his translation of the text (1980: 185).
dªNinº-lil2-le gu2 id2-Nun-bi-ir-du-ka i-im-du-de
igi kug-ga-am
lugal-e igi kug-ga-am
[kur ga]l
-le igi kug-ga-am
2 tar-tar-re igi kug-ga-am
[lugal-e gi
‘As Ninlil walked along the banks of
the Nunbirdu-canal, the bright-eyed one, the master, the bright-
eyed one, eyed her there. The great mountain, father Enlil, the
When this episode is repeated, with Enkidu scaling the wall, note that the pre
employed in this second rendition of the event is
, i.e.,
the phonologically and semantically lighter of the two markers (see [230] and p. 174
chapter four
And, similarly, with
with the sense of
)[256] a-na-aš-am3 [x]-gim lu
-ra ugu-mu-
-e u
ub ga-am
‘Why did you send me someone (saying,) “Ishbi-Erra is study-
In a broken context, but with perhaps similar intensive or experiential meaning:
‘He inspected the troops’ (InGud C8).
I thank Miguel Civil for discussing this passage with me—‘contemplate’ as an apt
translation of
in this passage stems from our conversation.
denotes an intensive or active
-event, one that is
by the
subject, prompting him to further action:
[259] [
dEn-ki-ra e2-a ba-an-
-e [ud]ug hul edin-na sug
‘Asaluhi took notice (of the evil wrought by the demon). He
entered the temple before his father, Enki, and said to him,
“The evil
-demon is standing in the desert”’ (FAOS 12, p.
76: 808–810, similarly, pp. 30: 180, 58: 655, 68: 721). Cf. from
rst millennium,
‘Marduk took notice’ (
urpu V/VI 17–18,
similarly VII 37); note that the equation of
the medio-passive N-stem of
underscores the middle
semantics of the event.
Interestingly, the pre
commonly occurs with
torical questions.
in these cases refers to personal observation or
—these are acts of
that have the potential of
impressing the subject. A certain complexity is implicit to these events
chapter four
True cognitive events in general represent a relatively small class
and are dif
cult to identify in Sumerian, particularly when compared
Ningirsu’ (Gudea St. B vii 38–41). Similarly, Gudea Cyl. B xviii
[268] bi
‘He paid
particular attention to the cultic rites of Bau, his lady’ (Gudea
2ki dagal-e-de
mu mah tuk-tuk-de
‘In order to enlarge Ur, in order that it acquire
a exalted reputation, I pray humbly (to Nanna)’ (Warad-Sin 18:
And overlapping with the speech-act middle:
2 Eridugki-ga-ke
im-ma-kar-ra (vars. um-ma-kar-ra/re) id
ad im-mi-ib-gi
(var. ad he
[. . .])
amar-ra gu
‘When (Enki) departs
from the temple of Eridu, the river resounds (lit. deliberates) for
its king—its sound is a calf’s mooing (in response) to the moo-
ing of a good cow’ (EnkJN 90–92). See Gragg 1973c: 62 (also
Thomsen 1984: 184) for the “re
ective sense” of
the comitative in
x is absent. For the translation “resounds,” cf.
rigmu ap
(see CAD A/2 s.v.
A mng. 2e); for
Compare the alternating subject-
ments that some languages use to express nuances of control and volition (see Croft
chapter four
[271] kur gal
‘After the great mountain, Enlil, made his heart fearsome,
after he frowned upon the foreign land, he cursed that rebel
land. My father, Enlil, dispatched me to the rebel land, the
foreign land at which he had frowned’ (Ninisina A 105–109).
[272] ud an-na-gim sig
‘May the people become as fearful at the cry as at a storm in
the heavens!’ (
Consistent with the pre
x’s ability to open up an event, revealing an
internal structure, emotion events expressed with
may be marked
for their intensity, the pre
x stressing the
’s involvement
in the action. As in [260]–[262], complexity may take the form of the
Agent or Object consisting of a proposition or situation rather than an
able entity. A common emotion-middle event of this type that
regularly employs
(i.e., the situation) made s.o. very happy’ (lit. ‘It happened to him as a
joyous occurrence’). Here, affectedness does not, naturally, involve the
syntactic subject, but the
, the topic, in the dative case (see
)18 The assignment of the
to the
indirect, rather than direct, object role, a phenomenon well attested
cross-linguistically, re
ects the fact that this is a low transitivity event in
which the
-sag šul dUtu en dGilgame
7-bi-e-ne mu-na-ra-an-sum
(var. hul
-la-gim im-mu-
na-ni-ib-gar) ‘The warrior, the youth Utu, gave Gilgamesh
these seven (warriors)—the feller of cedars was very happy,
lord Gilgamesh was very happy’ (GgHw-A 45B–47).
[275] en-ra
(var. hul
mu-na-ni-ib-gar) 5 ma-na kug-sig
mu-na-ab-sum-mu ‘(The
speech of the minister) made the lord so very happy that he
gave him
ve minas of gold’ (EmkEsg A163–164).
Another middle event, notable for its impact upon the subject and its
frequent occurrence with
‘weep, grieve, lament’.
This is a naturally intensive, iterative activity, an event of emotion that
intersects with the body-action and emotive speech categories (note the
appearance of the neutral pre
in the subsequent backgrounded
§)[276] abul Ganzir igi kur-ra-ka dur
‘At the gate of
chapter four
-nam mu-un-kin
‘Summer considered everything in his head and was able to
stay calm’ (WnSm 297).
[281] nig
al na-an-ga-am
1993: 409–410). The phonologically heavier forms
on the other hand, are often connected with speech actions that are
emotionally charged, that have greater semantic
than those events
expressed with
) (see also discussion of [230] and p. 174 n. 6).
Already in the Pre-Sargonic royal inscriptions there is the suggestion of a
[288] Ur
An-ta-sur-ra ga
‘Urluma spoke arrogantly, stating “Antasura
is mine! It is my territory!”’ (Uruinimgina 3 iv 5
). Similarly,
leader of Umma spoke arrogantly’ (Eanatum 1 ii 24–26).
In the following passage, the combination
may express more
than simply ‘speak’. The pre
x appears to add an intensive meaning,
the coupling of pre
x and verb connoting an emotive speech event
along the lines of
boast, brag, exclaim
(see Table 3):
2 ensi
kar-kar nig
e ga
An-ta-sur-ra-ta E
‘Il, ruler of Umma, the
eld thief, speak-
ing hostilely, boasted, “The boundary dike of Ningirsu and
the boundary dike of Nanshe are mine!” He bragged, “I will
, see Cooper 1986: 38 n. 2; a reading
), syllabic for
‘angry’, is also possible, yielding the emotive speech event
speak angrily
, i.e., ‘Urluma/the
leader of Umma spoke angrily’—see Steiner 1986: 253 ad 35 and RIME 1 comm.
to E1.9.3.1, with previous literature. Although somewhat obscured by lexical/graphic
chapter four
critically from those adverbial clauses discussed in connection with
, i.e., [106]–[115], is with respect to the notion of subject affect-
edness. Whereas
is neutral to this quality and the focus is on the
of the action, with
with an adverbial expression serves to emphasize the subject as the
Endpoint of the action. More than simply describing the manner of
the action, the particular adverbs that occur with
to stress the intensity of the action undertaken by the subject—again,
subject affectedness takes center stage. Common among the adverbs that
‘bitterly’, the combination frequently
occurring in laments:
[290] a uru
out bitterly, “O the destroyed city, my destroyed temple!”’ (LSU
[291] ad-da-bi e
(var. em-me)
‘May (Akade’s) old man cry out bitterly in the house of his slain
šu-mu a giri
‘He cries out bitterly,
tions or adverbial expressions that describe the nature of the speech
act, the context nevertheless often suggests an exclamatory manner
(e.g., [302]–[305], the last representing a string of middle situations).
And in those instances in which the subject is not emotionally affected,
e.g., [306], there is likewise an intensity to the event with the subject
vigorously participating in it. The action is perceived to emanate from
-subject, which ful
lls the roles of both the Initiator and
[294] eger-bi
‘In the wake (of the devastation), Inana arose, shrieked,
and yelled out aloud’ (EmkLA 193–194).
[295] ama dumu-ni-
hul til
-la-e ama lugal-la kug
‘The mother, who is miserable on account
of her son, the mother of the king, holy Ninsumun, was crying
out, “O my heart!”’ (UrNm A 15–16).
šag4-mu lipi
‘He cries out feverishly, “O
my insides! O my belly!”’ (Ninisina A 35).
2 amar-bi nu-ub-da-la
(var. i-
im-me) ‘Like a cow, whose calf is not with her, I utter pitiful
chapter four
-a-ni mu-un-
nu im-me
‘The father turned away
from his wife exclaiming, “This is not my wife!” The mother
turned away from her child exclaiming, “This is not my child!”
He who had a well-ordered household neglected his household
exclaiming, “This is not my estate!”’ (LSU 95–97).
kug dInana-ke4 E2-an-na gu3 im-me
‘O holy Inana, the Eana cries out! . . . O holy Inana,
emma no. 165: 5–7).
dEn-ki [n]u-zu-gim a-na-am3 ne-e im-me
‘Enki, as if oblivious,
cried out, “What is this!”’ (NinTrtl B42).
[305] An-ra a i-bi
me-e he
-mu nam-ma-gul-lu
) ug
-bi nam-ma-til-e
‘I shed my tears before An! I personally made supplication
part of the verbal semantics; and, secondly, those verbs that are not
innately self-benefactive, for which the pre
x serves to
self- benefactive meaning, overlaying a self-benefactive meaning on top
of what is lexically expressed by these verbs (see §2.9.5). The latter
comprises a particularly large, diffuse class of events that speak directly
to the fundamental meaning of the middle voice—these are events in
which a subject performs
action in his own self-interest with the
x explicitly signaling this quality.
One difference between
is that
appears to be
limited, generally, to the
rst category of events. With verbs that are
not naturally self-benefactive, other functions of
evidence, such as Object and resultant-state focus. Indeed, this result is
See Kemmer 1993a: 78–81, who distinguishes what she labels indirect re
verbs that are not intrinsically self-benefactive, from the category that she describes
as the indirect middle, “actions that one
performs for one’s own
t” (1993a: 78).
chapter four
the object transferred, which at least changes locations). With
subject is very much a
and the combination
See Karahashi 2000a: 168, who observes this same difference in meaning
u ba+ti
, but does not offer an explanation or rationale for
the distinction.
had not eaten and each of his weapons one by one’ (LgB 2:
[311] nu-siki nu-mu-un-su
)2-mu im-de
ne ‘The orphans, the widows, and the destitute pick up their
chapter four
ªkadraºa-bi dA-nun-na-ke
u nu-um-ma-gid
Anunaki no longer reach out for his gifts’ (UrNm A B15
)[321] id2 a-ba mu-un-na-ba-e-ne
e-ba mu-un-na-ba-e-ne
‘They offer (his
friend) a river of water—he takes it! They offer him a
eld of
grain—he takes it!’ (DzD 142–143).
[322] Lugal-banda
numun-e ki ag
u nu-um-ma-gid
‘Lugalbanda who loves seed would not reach out for (this
[323] lugal nig
-ga ku
king, who used to eat
ne food, grasped at a (mere) ration’
[324] a-da-al-la-bi Mar-tu du
gal-gal didli-bi
‘Now all of the
chapter four
in the resuming clause (
), which follows the message—as a
substitute for himself in the Eninnu. The statue, as intercessor, conveys
messages to Ningirsu on Gudea’s behalf. Similarly, in [329] the collective
event is obviously undertaken in the subject’s best self-interest.
3-de2-a alan-e inim im-ma-sum-mu
alan lugal-mu u
. . . alan na inim-
‘Gudea gives word to
the statue, “Statue tell my master: . . .” Thus, (Gudea) used the
stone statue for the message’ (Gudea St. B vii 21–48).
š] šu me3-ta im-ma-ta-Šub-bu-
ªe3º ‘The
men who quit the battle(-
eld) and who, like birds, saved their
lives (by
eeing) to their cities, did not escape (Shu-Sin’s) hand’
u-Sin 3 iv 2–7). For
, motion events
with self-benefactive overtones, see §5.1.2.
More striking examples of this phenomenon are found with those
verbs, which, by virtue of their lexical semantics, lend themselves to
the expression of prototypical transitivity. Representative of these
predicates is
‘build’, which is regularly coupled with
when the
event is viewed from the perspective of the Agent, the Initiator (§3.1;
in particular, see [28]–[30], [36]).
, when this same event
is viewed from the perspective of the subject, not as
but as
—when it is the contrastive self-benefactive quality of the
event that is being stressed—the pre
x is frequently
. Each of the
following examples involves a deity building his or her own cult center,
events for which there is a high degree of self-interest:
dEn-lil2 a2-dam kug ki-a hur-ra-za Nibru
) ‘Enlil, when you designed the holy
Note the remarkable contrast in perspective provided by [333], as well
as the sequence of self-benefactive events in [334].
[333] dumu-mu e
(var. im-an-du
-da ‘My son has built a temple!
King Enki has made Eridu come out from the earth like a
mountain range! He has built himself a temple in a pleasant
place, in Eridu, the pure place, where no one is to enter’ (EnkJN
[334] ur
-bi-a mu
pa-bi-a mu
‘In (the tree’s) roots, the Snake-that-
See Croft 2003: 130 for the intermediate categorization of animals in terms of
animacy in some languages.
chapter four
šag4-ba buru
(var. gud
) nunuz-bi ba-ab-gar da-da-ba mu
(var. gud
) amar-bi
‘In the midst (of the mountains), the
-bird made
its nest and there hatched its eggs; nearby, Anzu had established
this event is
. However, the pragmatic focus in this context,
I would suggest, is not on Huwawa as the locus of the action’s effects,
on killing as an act of agency, but on Enkidu’s emotional
the action (note the discussion of [19]–[21])—Enkidu kills Huwawa in
t of sudden and unrestrained rage, an act of killing that intersects
with the spontaneous and emotional middle types (§§4.1.3, 4.2). In what
chapter four
meaning ‘woman’s quarters’ and the reading
‘womb’ (see Michalowski 1989: 76–78 ad 60).
And when Ninhursag removes Enki’s semen from Uttu’s body, planting
it in the earth and growing (or
) the plants that will allow her
to exact her revenge upon him, the pre
x is consistently
unique context bringing about the rare combination
(for the
spontaneous events represented by
–)[344] [u2aš-tal
u2x x x i]m-ma-an-mu
‘She grew the
(for her use), she grew the . . . plant (for her use), she grew the
-plant (for her use)’ (EnkNh 195–197).
Finally, when Ninhursag, in a change of heart, cures Enki by giving
birth to eight deities from his pains—a powerful demonstration of her
role, and prerogative, as birth goddess—
‘give birth’ is invariably
paired with
, e.g.,
šeš-mu a-na-zu a-ra-gig ugu-dili
-mu ma-gig
‘“My brother, what hurts you?” “The top of
my head hurts me”—(Ninhursag) gave birth to Abu out of it’
This broad self-benefactive meaning is supported by earlier texts as
well. In [346] and [347] the recruiting or levying of troops is under-
taken in the self-interest of the subject, namely, Urluma, the ruler of
-su-ka-ka e-ma-ta-bala
‘(Urluma) recruited foreigners (as mercenaries) and transgressed
as when he inscribes his own name on the standard that he fashions
for his master [349]:
[348] na Ka
-sur-ra bi
-a lugal a-ma-ru
nu-tuku Gu
igi zid mu-
Note that
is commonly attested with this self-benefactive meaning in Old
Babylonian royal inscriptions (particularly those from Larsa) often occurring in contexts
where we might otherwise expect
, e.g., bad
gal Bad
-ma-ni-ta gal-bi
‘By means of his triumph he grandly built the great wall of Badtibira’
(Sin-Iddinam 14: 28–31), temen mu-pad
-da n[am]-lugal-mu ud ul-
put there, for posterity, a foundation inscription proclaiming my kingship’ (Rim-Sin
13: 36–37; similarly, Rim-Sin 16: 27–28, 17: 44–46).
Likely to be included among events of this type is that represented by the idiom
chapter four
(vars. u
‘When I (Hoe) have worked on the canals and the dams, put
the paths in order, and built shelters there on the banks’ (HoPl
151–152). Note that in the case of the variant,
event is viewed as agentive.
-la uru
‘The shepherd Ur-
Namma destroys all evil like a
ood. He sweeps away the great
oppression like a wind’ (UrNm B 60–61, similarly, 58–59).
[352] [kur-ra] ur
-ra nu
da nu
on the roofs those of the foreign lands who lie on the roofs.
I will slaughter on the walls those who lie on the walls!’ (
Spontaneous Events, the Passive, and Related Low Transitivity Situations
The discussion in the preceding section does not mean to imply that
all occurrences of
can be attributed to a self-benefactive mean-
ing when the accompanying verb does not lexically express a typical
middle situation as outlined in Table 3. There are contexts in which
the pre
x appears that do not correspond to any of the situations
described in §4.1 but which share with these situations the underlying
property of low or reduced transitivity. The events discussed in this
section range from the syntactically transitive to the intransitive, from
those events that are commonly marked as middles cross-linguistically
to those that are passive. In all of these environments, as is true of
the typical middle situations described previously,
ects the
speaker’s focus on the perceived Endpoint of the event as the locus of
the action’s principal effects.
With the events described in this section,
functions as a general
Cross-linguistically, middle markers are frequently called upon to ful
is motivated by a self-benefactive meaning in the sense
exhibited in [330]–[334]. It is also unlikely that in the stripped-down
register of administrative records it was deemed necessary to stress this
quality of the action, or that Eniggal has greater self-interest in this
particular event than he does in the many others that he supervises
and that are not expressed with this pre
x. Rather, what is conspicuous
about the clause is that this lexically highly transitive and bivalent verb,
‘build’, is missing its expected (de
nite) Object (note that causation
is indirect and the clause is semantically equivalent to ‘the occasional
chapter four
of identical
carving out
subevents (referring to the cutting down and pre-
paring of wood).
Agency, like transitivity, is a scalar notion. The mere
presence of an Agent (frequently in Object position in administrative
texts) does not necessarily make the Agent the focus of the clause.
[354] 1
‘1 (dead) wool-bearing sheep was
separated (from the herd)’ (DP 260 i 1–2).
[355] (numer
ous objects made of pine) kiri
‘From the garden of Urki, Enshu, the steward, carved out (these
wooden objects) (i.e., he cut down the timber and prepared
the objects) and brought them to the palace, to Bau’s personal
garden’ (DP 416 ii 1–iii 3). See also DP 412, 417, 447.
Similarly, in the Old Babylonian example [356], it is not the affected-
ness of the Agent that is at issue, but that of the Object—the Object is
more topical and
more animate
than the Agent, which naturally has the
effect of reducing the transitivity of the clause (cf. the inverse §2.9.3).
The pre
x in these transitive clauses (similarly,
)does not indicate subject affectedness in the sense of the self- benefactive
More commonly the verb
(see Sallaberger 2005a) occurs with
- (and
in Pre-Sargonic administrative texts, as does
in the expression PN ma
‘PN loaded (the objects) into boats’, e.g., (wooden objects) gi
ki mu
En-ig-gal nu-banda
‘The wood was inventoried where
it grew; Eniggal, the captain, carved out (the wooden objects) and loaded them on
(a) boat(s)’ (DP 470 iii 1–5). The pre
-, I would suggest, more than relating to the
transitivity of the clause, focuses in these contexts on
the action took place, i.e.,
middle, but topic affectedness, the topic being, as in the impersonal
passive, the affected Object (the rendering of these clauses as passives
with explicit Agents attempts to capture this nuance). As the topic of
a sentence typically coincides with the subject, this use of the middle
presumably arose secondarily from its primary subject-affectedness
En-ki-dug3 e11-de3 i3-gi4-en kur-re
(var. nu-un-dab
) kur-re
Nergal sag
u nu-du
-a nu-un-
ub kur-
‘Enkidu went down (to retrieve his
chapter four
Naturally, the clearest evidence for the Agent defocusing function of
the pre
x are those relatively rare instances in which the Agent is omit-
chapter four
with spontaneous events, a category that is, like the passive, more
frequently expressed with
. Indeed, spontaneous and passive events
[372] unu
gal-bi ka
wine, and honey ceased (to
ow) in the great dining hall’ (LSU
2 ma2-gur8-ra Kar-za-gin
and barges ceased (docking) at the Shining Quay’ (LSU 322).
2-ma kalam-ma u2-gu im-ma-an-de2! (vars. u
) ug
e am
‘The judgment of
the land vanished—the people moan’ (LUr 230).
3-bi Šu suh3-ha im-ma-ab-dug4 ‘(Sumer’s) people became
confused’ (LUk E56).
d]Udug eden-na dLamma
‘(Bilulu) and her son
The pre
Observe in this connection that there is no obvious semantic distinction governing
the distribution of writings
- in the Ebla lexical texts (cf. Edzard 2003a:
94–95). For instance, the writing with
- is used regularly with
‘receive’ (self-
benefactive middle, “reverse” verb [see §5.1.4]), but is also coupled with DU and
translated by
(likely related to
and so a motion(separative)-middle [§5.1.2]).
The more common writing with
occurs with
and is equated with
‘fugitive’ [CAD A/2 sub
lex.], as well as with
and DU, and so
also appears to have a separative meaning with verbs of motion. The
latter pre
x also occurs with other verbs that denote typical middle situation types,
) ‘hear’ (perception-middle [§5.1.3 (§4.1.3)]),
) ‘fall, drop’ (spontaneous event [§5.1.5])—see D’Agostino 1990: 77–87
and 1991 with citations.
marker. More than giving a nod to the pre
x’s well-known, if at times
disputed, passive function, this label underscores the one semantic
role that is most representative of the pre
x, namely, that of
Where the pre
diverge is, again, in degree.
the phonologically lighter of the two. And, in an elegant display of the
iconicity of form and meaning, it is also the semantically lighter of the
two, conveying less meaning than its heavier counterpart. As a general
tendency, particularly evident when there is no case element interven-
That the pre
x primarily re
ects rather than imparts middle meaning is again
suggested by its appearance in regular re
exive situations (i.e., those in which re
is not intrinsic to the meaning of the verb). Here re
in terms of the types of middle situations with which each tends to
occur and the way in which each views the event.
correlates with the typical middle-situation types
described in Table 3, but it is weighted toward the low end of the spec-
trum, where the elaboration of events is minimal. The pre
x does com-
monly occur with events in which Initiator and Endpoint can be viewed
as conceptually distinct, such as grooming and body-action events, but
the speaker’s choice of
, re
ects his perception
of the subject as an undifferentiated entity and the event itself as a
unitary whole without internal complexity. And, whereas the subject of
the event construed with
tends to exercise some control or volition
over the action, the prototypical environment of
are those events
in which the subject is a
—predicates that Perlmutter (1978:
162–163) has aptly described as “unaccusative.” Control and volition
are often not in consideration and subject-affectedness is paramount.
Hence, the close association of
with the extreme manifestation of
Endpoint focus as represented by certain spontaneous events such as
‘die’ (§5.1.5), instinctual emotion events in which the subject
is adversely affected such as
‘fear’ (§5.1.3), as well as self-
benefactive events in which the subject plays a non-active role such as
‘receive’ (§5.1.4). From middle events in which the subject is
, it is but a small step to that other category with which the
x has close ties: the passive, constructions in which the
the Object of the transitive clause) is promoted to the subject position
obligatory with the
rst-person in this respect, nearly so with the second,
while its occurrence with third-person animates is a matter of speaker
empathy (§3.4). The pre
resumes this scale, appearing with NPs
on the lower rungs of the Nominal Hierarchy:
correlates with
, as well as with third-person
animates which solicit little empathy—those that the speaker regards
as object-like—such as collectives and inde
nites. Also to be included
here are instances of
—clauses that lack a
participant, but whose verbs are typically
construed with one, e.g.,
‘give (to s.o.)’,
‘address (s.o.)’.
The Endpoint or Undergoer focus that typi
es the pre
is, with the phenomena that constitute grammatical voice.
Rather, this
x expresses a locative focus, which is exploited to ful
ll a number
of further grammatical functions, the most obvious being the marking
of the semantic object of certain compound verbs. Remarkably, there
appears to be greater
Consider, for example, the following clause: e
-gim ki
‘(Gudea) built Ningirsu’s temple on a place that was as pure as Eridu’
(Gudea St. B iv 7– 9). The focus here is not on
did the building or
was built,
but on
the building event takes place, a fact that is underscored by the inclusion
of the modifying adjective
and the positioning of the locative clause immediately
before the verb, iconically signaling its semantic
to the verbal meaning. On the
morpho-syntax of
-, see Karahashi 2000a, 2000b, Zólyomi 1999, and, in particular,
Johnson 2004.
Body-Action Events
As in §4.1.1, I include under this designation grooming, change of body
posture, and non-translational motion events, reserving translational
motion events, because of their particular deictic associations, for the
following section. Body-action events typically occupy the highest rung
of the middle voice domain in terms of volition, control, and the abil-
ity, at least in theory with grooming events, conceptually to separate
Similarly, with other verbs denoting dressing and donning events:
tug2nig2-lam2 banda3da-mu ib
(vars. ba-a-du
) ‘I fastened to my hips my short
‘Great mother Ninlil
in (Nippur) clothed herself in priestly linens’ (NSJN 36).
[382] lu
-priest no longer dons priestly linens in your holy
(LUr 352).
[385] gala-e bid
‘A lamentation priest
Change of body posture and non-translational motion events with
are exempli
ed below. Again
‘sit’ may be
taken as representative of this class; these are predicates that typically
denote uncomplicated actions that lack structure, qualities that promote
the choice of
. Further, when paired with
frequently in evidence contextually in the case of
—these predicates
tend to express dynamic, punctual, and telic actions.
‘lie (down)’ (see also [226]); cf.
‘make lie’ [55] (for [386]
[503]–[507] as well as
)[386] 1 ma
-da Munus-sag
-ga Ad-da-
‘1 small goat was brought for Munussaga and
Addashusikil, the minister, who had “lain down” (i.e., died)’
(Nik 153 i 6–ii 4 [= FAOS 15/1, pp. 374–375]).
[387] 2 har kug-babbar 5 gin
-ta Ur-
ud ba-nu
‘2 silver rings (weighing) 5 shekels
each—for Ur-Nanna, the minister, brother of Arad-Nanna, the
head minister, when he “lay down” (i.e., died)’ (JCS 10 [1956]:
30 no. 9: 1–3).
dNu-ªdim2º-[m]ud ªitimaºma-ka ba-an-ku4 šag4-ka-tab-ba ba-an-nu2 ‘Nudimmud entered the bedchamber and lay down fasting’
še3 dEn-lil2 itima kug ba-an-ku4 šag4-ka-tab-ba ba-an-nu2 ‘Because of this, Enlil entered his holy bedchamber and
lay down fasting’ (CAk 209). Similarly, EmkLA 390.
[390] lugal u
(var. la-ba-an-nu
-e) ma-mu
‘The king did not lie down to sleep, but lay down to
dream’ (LgB 1 A340). Similarly, SrgUrZ B13.
‘stand (up)’ (see also [190]; cf.
–mu+gub ‘make stand’ [60]):[391] (5 individuals) Amar-ki-me nig
als), they are the (people of) Amarki—they stand at the ready’
(Nik 14 i 6–7 [= FAOS 15/1, pp. 117–119]).
mu-bi ‘“Bau
stands at the prayers of Uruinimgina” is (this object’s) name’
(Uruinimgina 14s [= FAOS 5/1 Ukg. 53]).
[393] alan-na-ni mu-tud nam-
fashioned his (own) statue—“It stands in prayer”’ was the name
he gave to it on behalf (of Geshtinana)’ (Gudea St. M ii 7–iii 3).
[394] hur-sag za-gin
-na-gim mu-mu
hur-sag nu
‘He made (the temple) grow like a mountain
range of lapis lazuli—it stood to be admired like a mountain
range of white alabaster’ (Gudea Cyl. A xxiv 15–17).
[395] ga
‘I am truly one who stands to be
admired’ (In
[396] an ki
nam tar-tar-[ra] na-nam na-nam [dingir an-na
eŠ] ‘The gods of heaven, with their ready assent,
‘The young man
raised his hands heavenward to Utu’ (DzD 164).
[404] igi a-a ugu-na
‘(Sin) went
down on his knee before Enlil, the father who begot him’ (LSU
2-bi uruki
-bi dingir-bi-e-ne ki-bi-
(Asag’s) might the gods of those cities bow down to the ground’
(Lugal-e 40).
ha-ba-gur-re-en ‘I rose up like
earth trembled, the storm churned incessantly (lit. the storm
never slept). The heavens darkened (lit. became confused), a
shadow hung over it’ (LSU 81–82).
writing for
‘be awake’ (see Michalowski 1989: 79–80 ad
81). Cf. agentive
5.1.2 Motion Events
rst observed by Thureau-Dangin a century ago, verbs of motion
often have a separative nuance, denoting movement
away from a reference point, which, in the default scenario, when no
other location is speci
ed, is
, the location of the speaker. This sense
Dingir-a-mu nu-kiri
‘Uremush, the merchant of the “women’s house,” took away
1 mina of re
ned silver for the purchase of an
From this amount, he brought back 1
-worker, costing 14
shekels of silver—Urki, the gardener, took him away. (Uremush)
brought back 1 man, costing 1/3 mina of silver—Lugalda, the
shepherd of the wool-bearing sheep, took him away. (Uremush)
brought back 1
-worker, costing 15 shekels of silver—Din-
giramu, the gardener, took him away’ (Nik 293 [= FAOS 15/1,
pp. 521–522]).
[417] ud e
igi nu-ni-du
-ga e
‘Lubalasaga, at the temple of Nanna, swore that he had not seen
(the missing worker) since the palace took him away and that
if he does see him, he will bring him here’ (NG 190: 23–27).
[418] gi-NE e
60 men (to collect) reeds for the household. Then, the reeds
will be delivered (there)’ (LEM 77: 7–10).
2 u3 kug-babbar igi-ne-ne du
‘The slave woman and the silver
are considered equal—when (the debtor) brings the silver, he
can take away his slave woman’ (SLHF viii 11–15).
, as discussed in §4.1.2, this separative sense is a consequence
of this pre
x’s broader middle voice function.
, focuses
on the dynamic and ingressive change-of-state that characterizes the
from placing
(see p. 180 n. 10).
Observe in this connection that unlike a language such as English, in
which motion verbs are lexically speci
ed for direction or orientation,
, in Sumerian it is the pre
x that serves to
give direction or deictic content to basic motion verbs, such as
‘carry, take’, which are intrinsically neutral in this regard.
That the separative sense that
conveys derives from its broader
middle voice function—that, indeed, motion events are not fundamen-
tally different from the other middle situation types with which the
x correlates—goes some way to accounting for those situations
that do not necessarily suggest a separative meaning. There are many
contexts in which
(§4.1.2), simply emphasizes the
‘When evening came
dUtu nig2-si-sa2 inim gi-na ka-ta ba-da-an-kar ‘Utu took away
the pronouncement of justice and the
rm decree’ (LSU 62).
6-e mu
en-ra mu-ni-in-si
at Bird, and then
ed into the waters’ (BdFh 109).
[429] gal
y (away)’ (see also [401]); compare the Akkadian equivalent
y’, the lexical use of the N-stem, as with
ee, escape’, ‘disappear’, re
ecting the inherent middle semantics of
the event. The separative force of the pre
x is often reinforced with
the ablative
zag dib-ba ba-an-sag
-bi i
-gul-e (var. i
-gul) ‘(Gilgamesh)
ung a
haunch at Inana—she
ew off like a pigeon and it destroyed
the rampart’ (GgBH D50 [Me-Turan]).
ªug3º? ni2 ba-da-te simmušen
‘The people have
become frightened—they
y away like swallows’ (Cohen Er-
emma no. 32: 55). For
[440] me gal-gal-la-ni a-gim
‘How did Enlil make his
greatest divine powers
y off?’ (LNip 114).
[441] sim
‘Like a swallow that has
own from its nest (lit. house), (Ibbi-
[447] uru-mu
‘I arose over
my city like Utu, suspended in its midst’ (
[448] a-ba nam-til
(var. ba-an-e
) ‘What mortal has
ascended to heaven?’ (
[449] ur gal-gal 7-bi ba-e-u
‘You ride on seven
great beasts when you go forth from heaven’ (Inana C 105).
Emotion Events
[450] nu-kar
ur-mu in-bala-a-
‘Although I had given no insult,
the bodyguards of my king overturned my table—I became
ed, my
esh crept’ (RCU 1: 27–28). Similarly, see [439].
šag4-ba En-me-er-kar
‘In the midst (of the troops)
Enmerkar son of Utu became afraid, became troubled—the
clamor made him anxious!’ (LgB 2: 268–269). For
ious’, note: en al-BAD =
[452] lugal-me-en
Compare the anomalous and presumably erroneous construal with
-me-en sila-
dNin-urta-me-en ki-tuš mah-ga
(var. im-hu-luh-ha-en-na-g[im]) ‘Since you frightened me, the
lord Ninurta, on my great throne’ (Lugal-e 422).
5-ša4-zu ama gal
‘At your thun-
dering, great mother Ninlil cowers in fright before you’ (Cohen
emma no. 23: 14).
-mu ba-dar
‘I have grown scared, I am
heart-broken’ (Dial 3: 136). Similarly,
-mu ba-
‘I have grown afraid, I am heart-broken’ (Edub A 23).
[462] dur
ki-a nu-ub-za
(var. ba-an-tum
) ‘No one rose from his seat (for me),
no one bowed down (before me)—I grew anxious’ (RCU 1:
[463] iti
-zal i-zalag-ga-am
2-am3 inim 5-am3 u3-na-dug4 zi gi4-ba-an-ze
‘I have
told each one of you time and again, “Calm down!”’ (Dial 2:
nim guruš uru-na-še3 šag4-ga-ni an-hul
‘(Gilgamesh) was pleased with the advice of the young men of
[468] en-e ur-sag-ra
‘The lord rejoiced at the hero—father Enki rejoiced at
the hero Ninurta’ (NinTrtl B13–14).
al ba-an-dug4 al ba-an-dug4 ki-nu
craves it, she craves it, she craves the bed!’ (DzIn D1: 18).
ki ha-ba-an-ag
hul ha-ba-an-gig
ki ha-ba-an-ag
š am3-da-la2-e-en ‘If she loves her city, but hates me—why then does she
bind the city to me? If she hates her city, but loves me—why
(var. ki ha-ba-da-ag
ki la-ba-ra-ag
(var. ki la-ba-ra-da-
‘I love righteousness. I do not love wickedness—indeed, I hate
words uttered malevolently’ (
m-mi-a nig2 na-me ba-an-zu-a (vars. bi-zu-a, bi
sag ba-gid
-za ak-mu-un (var. ak-bi) ‘The
teacher, who had learned everything, became extremely angry
(saying,) “Do as you wish!”’ (Dial 3: 179–180). For
[645]–[653]; note that
, emphasizes the control
and volition of the potential agent.
Hu-wa-wa zu2 ba-an-da-zalag sag-ki
[476] ma
‘As for
the sorcerer, his face darkened and his mind became confused’
Self-Benefactive Events
Like verbs of emotion, self-benefactive verbs involve animate subjects,
but unlike the former, the subject, nominally at least, exercises volition
and control over the event. Further, the subject is generally not as thor-
oughly affected by the action as in the case of many spontaneous and
emotion events. Predicates corresponding to self-benefactive middles
denote actions in which the
of the verbal action (Table 3). As with other situation types,
and in accord with the broader character of the pre
with verbs for which self-benefactive semantics is an inherent part of
the lexical semantics (see §4.1.4). Unlike
add this re
exive reading when not lexically present, but redundantly
reinforces it when it is part of the verbal meaning. As expected,
further differs from
in viewing the event as an atomic occurrence,
as an uncomplicated
whole. Finally, with
often plays a more
role in the event as compared with
Self-benefactive verbs commonly occurring with
include, among
with the sense of ‘take a wife’, i.e., ‘marry’,
The verb
‘receive, take’, naturally, represents an inherently low-
transitive middle event. It also has
but the recipient does not necessarily do anything at all” (1991a: 331).
[484] Abzu Eridug
-ga me
‘She received the divine powers in the Abzu,
in Eridu. Her father, Enki, presented them to her’ (IdDgn A
Common in literary texts is the topos
to accept tears
, which is regularly
expressed with
2-še dMu-ul-lil2 a i-bi2-za Šu ba-e-Ši-in-ti-a ‘Even now, (speak
supplications to) Enlil, who will accept your tears from you!’
dUtu er2-na kadra-gim Šu ba-an-Ši-in-ti ‘Utu accepted his
tears as a gift’ (GgHw-A 34).
u ba-an-
accepted his tears and gave him life’ (LgB 1 A226).
with the meaning ‘marry’ (lit. ‘acquire [as a
spouse]’), occurs frequently with
, and less often with the neutral pre
(the latter,
, is common in Old Babylonian law codes and
related texts, see §3.3). As with other verbs of
, the event may
have separative overtones (i.e., in the sense of taking a woman away from
the home of her father),
but any such force is certainly secondary, and
the event is to be understood, primarily, as representing a self-benefac-
tive action ‘take s.o.
for o.s.
’. As with many of the events that correlate
, this is a dynamic and punctual action, one that terminates in
a resultant state, speci
cally in this case, that of
being married
3 inim A2-na-na ab-ba-ta-am3 Šes-kal-la-a Nin9-ab-b[a-n]a
a]-an-tuku ‘And, at the word of Anana, the father, Sheshkala
married Ninabbana’ (NG 16: 12–14).
INin-ezen dumu Lu
-su-ka A-kal-la dumu Ba-a-ke
‘Akala, the son of Ba, took Ninezen,
the daughter of Lu-Ningirsu, in marriage’ (NG 191: 5–8).
[490] tukum-bi . . . igi ad-da til
‘If . . . (the
dNin-kiri3-u3-tud dNin-a-zu ha-ba-an-tuku-tuku ‘Ninkiriutud shall marry Ninazu’ (EnkNh 275).
ku-li-mu-ne-me-en dam
‘In my
city, I am among my friends, and they all have already married.
I am among my companions, and they too have all already
married’ (MarMartu 28–29).
A certain Ur III attestation of
occurring in an eviction clause is provided
šag4 in-dab5-ba-na na-ba-a-du3 ‘He cannot keep for himself
eld that he has seized!’ (LEM 131: 7–8).
[496] A-kal-la u
-uru-sag-ra en-na igi-mu-
See Steinkeller 1989: 54 for the argument that a transitive construal of this idiom
developed secondarily as an Ur III innovation.
2-kur za-gin3-na gišgidru ha-ba-dab5-ba barag babbar-ra
‘In the lustrous Ekur, where
I took hold of the scepter, I lifted my head toward heaven on
a shining dais, a throne with a
rm foundation’ (
Note the body-action middle
sag ba+il
[500] lugal-e muhaldim-gal nu-me-a gir
(var. [b]a-
) ‘The king, though not a master chef, took a knife in
his hand’ (GgBH D49 [Me-Turan]).
eger-ra-ni sa-bi
‘The turtle was able to
take hold of (Ninurta’s) tendon from behind’ (NinTrtl B40).
Ak-ka3 dumu En-me-barag-ge
‘Aka, the son of Enmebaragesi, (and his army) laid
siege to Uruk (lit. held on to the side of Uruk)’ (GgAk 49).
Spontaneous Events
Spontaneous events are actions that are perceived as occurring autono-
mously and that result in the subject attaining some end state. Often
the event is conceptualized as unfolding independently of the interven-
tion of an external cause, either because an explicit or salient—that
is, human—agent cannot be identi
ed, as for example, in physical
processes such as
turn green
, or because, for pragmatic reasons,
the salient agent is completely de-emphasized, e.g.,
be born,
(Kemmer 1993a: 142–144; see Table 3). Spontaneous events are,
therefore, Agent-defocusing situations that share much in common with
passives, as they code only one participant, the affected Endpoint (see
§5.2). And, like passives, implicit to spontaneous events is the notion
that the subject has undergone a resultant change-of-state and that
this state represents the end of some dynamic telic action, cf.
He died
(spontaneous middle).
However, unlike passives,
which semantically possess an Agent (§2.9.1), which is merely ignored
in the surface coding, in spontaneous events the action is viewed as
emanating from the subject, who ful
lls the role of Actor-Initiator in
addition to that of Undergoer-Endpoint (Kemmer 1994: 211–212). As
observed by Merlan, in a description that aptly captures the middle
semantics of spontaneous events, “Such predicates have in common
that the subject is the locus of the action or of whatever is described
by the verb’s meaning; this is going on entirely within the subject with-
out external agency, and also is not directed outwards toward another
entity. These characteristics are related to lexicogrammatical features
that Indo-Europeanists have attempted to capture by the designation
‘middle’” (1985: 351).
Where spontaneous events differ from other middle-situation types,
but where they are again allied with passives, is in the characteristic lack
of control and volition exhibited by their subjects in most actions of this
type. The lone participant coded in the event, the subject, undergoes
a change-of-state in which it has little choice. These are often incho-
ative or inceptive events in which the subject is, essentially, a
Further, these object-like characteristics of the subject account for the
association with inanimacy, a property with which
larly close af
nity. Again, like the prototypical Object, which serves as
the subject of the passive clause, the subjects of spontaneous events
are frequently inanimate, e.g.,
turn green,
for Akkadian
‘die’, as well as in (LU
adjective expressing the resulant state of the event (cf.
U.BA.(AN.)TI =
[§5.1.4]). As in the case of
‘receive’, the combina-
tion is so frequently attested, so obvious to anyone with even a passing
familiarity with the language, that only a few examples are necessary
ba-su-a-ba ba-su-a-ba ud gišma
‘When it sank, when it
sank, when the Magan-boat sank, when the
-barge sank’
ªgišma2º Dilmun-na tumu nu-mir
‘A Dilmun-ship
may sink, even though the wind is not raging’ (SP CT 58,
Note the absence of the pre
x with both
‘die’ and
‘sink’ in the
following negative assertions, hypothetical statements that concern
generic subjects and actions that will not take place. Subject affected-
ness is obviously only marginally at issue in these events and so
omitted (conversely, for the absence of
and the neutralization of
, for these same reasons, see §3.3).
[513] gar
(var. e-ug
(var. bi
(var. nu-
u sum-mu-de
bar-re ‘He should release one wooden mortar to be given to
Lugalkugzu, (as a replacement for) the one that was lost (or:
the one [Lugalkugzu] lost)’ (LEM 224: 3–7).
u2-gu a-ba-de2-a ba-uš2 Ur-
inim nu-
‘If (the sold woman) disappears, and then dies, Ur-
Sin will not raise a claim against Ur-Shulpae (on her account)’
(FAOS 17, no. 94**: 8–11).
[519] ki
‘He declared, “The
‘In my city people are dying, and
hearts are full of distress. People are lost—that
lls me with
dismay’ (GgHw-A 23–24).
‘grow, sprout’, commonly referring to plants (see also [664]; cf.
‘grow s.th.’ [57], [223],
‘grow s.th.
for o.s.
)[526] u2numun
‘Alfalfa grass grew, alfalfa grass grew, mourning reeds grew!’
presumably represents
, following the
suggestion of Civil (apud Michalowski 1989: 97 ad 321).
2 gišma
gigir-ra ba-gar-ra-ba u
‘On (Akade’s) canal
banks, where boats are towed, grass grew long. On its roads
laid for wagons, lamentation grass sprouted’ (CAk 273–274).
2 a-ni]r ba-am3-mu
3-bi u2 gid2-da ba-am3-mu
‘Lamentation grass sprouted, alfalfa grass
sprouted—by the walls grass grew long’ (Nisaba B 8–9).
‘My city, before Dilmun
existed, palm trees grew there. Nippur, before Dilmun existed,
palm trees grew there’ (NSJN 34–35).
[530] hur
-mu nig
‘On my black mountain
white gypsum has appeared (lit. grown)’ (OmYg 29
oat, drift, sail’ (cf. the Akk. N-stem equivalent
). This
event intersects with the translational motion type and may additionally
have separative overtones, i.e., ‘
oat away’, deriving from the incho-
[532] hur-sag galam kad
constructed mountain range that
oats on water’ (EnkJN 74).
gišma2 kar gi-na li-bi2-ib-taka4-a-gim tumu-e
(var. ba-ab-dirig-ge-en) ‘Like a boat that is not moored to a
[539] dingir-m
god, the day shines bright over the land, but for me the day is
dark’ (ManGod 69).
‘fall’. With inanimate subjects, this predicate denotes a typical
spontaneous event in which volition is not in consideration [540]–[543]
(note the juxtaposition of the semantically-close spontaneous and
passive events in [542]–[543], as well as the instance of indirect cau-
sation represented by [540]; see also [610]). With animate subjects,
however, there is the possibility of control, in which case the event is
not truly spontaneous. This last scenario describes intransitive events
that approach the body-action class, as, for instance, is represented
by [546] (note the sequence of middle events; cf. the non-volitional
events involving an animate subject in [544]). In transitive clauses,
typically indicates a lack of volition and control (a nuance that
is captured by the translation
in [545]) and the combination may
be seen as standing in semantic opposition to volitional
–[540] u3-šub mu-dub
mold and the brick fell out into the daylight’ (Gudea Cyl. A
xix 3). Cf. non-agentive
[541] a-ra2 1-a-kam Tum-ma-al
‘For the
rst time, the Tum-
mal fell (into ruins)’ (Tummal 6).
nam-lugal-bi Bad
‘Eridu fell,
and kingship was carried off to Badtibira’ (SKL 8–10). Note
) in the parallel passages of
ll. 18, 24, and 30 (cf. Finkelstein 1963: 41–42, Jacobsen 1939:
2-e gi
-si-bi ba-gul ‘The great
(var. ba-
ub) sahar-ra
(vars. im-da-an-tu
) tug
contexts, which may be conceived as occurring without an external
ned to a
-subject. To be included
here are predicates that denote actions of the type ‘become or turn
ed state)’, ‘change’ (see Table 3 and [590]–[592]), which
are quite common with
and often occur in sequences of spontane-
ous events [551]–[555]. As discussed in connection with [654]–[659],
this use of the pre
nds a parallel in languages that middle-mark
verbs derived from adjectives of state. There are also events in which
There are many other events in which the subject is inanimate and
thus non-volitional, which could be included here; particularly common
among these are expressions involving
(var. ba-e-zal-la-ri) mu
(var. mu-un-da-dar) ‘Five years,
then ten years had passed, the tree had grown massive—but its
bark did not split’ (GgEN 127–28). The variant
views the event, presumably, as belonging to the domain of a
potential Agent, i.e., ‘(but) no one was able to split its bark’.
[558] ud Ki-en-gi-ra
kur-re he
kur-re he
-eb-zal ‘The storm that passed over Sumer,
passed also over the foreign lands. The storm that passed
over the country, passed also over the foreign lands’ (LSU
Finally, there are events, such as
discussed above
[547]–[550], in which the subject is animate and exercises volition, but
which share with conventional spontaneous situations the quality that the
action emanates from the subject without the notion of external agency;
cation, than the passive function of
. Some, such as Poebel,
Falkenstein, and Oberhuber, have claimed that this use of the pre
was a secondary development, dating to the Ur III period. Others, such
as Horsnell and Christian, have speci
cally denied its existence, while
many writing more recently have been careful to note that the only
about the pre
x is that it may be suitably translated by the
passive in English (e.g., Postgate 1974: 27). Thomsen expanded on this
sentiment when she summed up the consensus at the time and wrote
that “/ba-/ has been called a ‘passive pre
x’ because of its frequent
occurrence in one-participant forms . . . this use of /ba-/ depends on its
inanimate/non-agentive reference, and it has nothing to do with the
category ‘passive’” (1984: 183).
The problem, of course, revolves around how the passive is under-
stood. If one understands it in traditional terms, as serving solely a
“role remapping” function by which an Object is moved to the subject
position, and one further demands—unreasonably, given that polysemy
is an intrinsic characteristic of language—that there be an exclusive
previous action, an action that has been brought about by some—per-
haps unspeci
ed, perhaps unknown—Agent. Indeed, it is this quality
in passive function, which is on display in the Old Babylonian
grammatical texts. Although the vast majority of the
xed verbs
(136 of 151 attestations, nearly 90% according to Black’s calculation
[1991: 27]) are equated with Akkadian forms with the
x (Gt
fteen that remain are equated with the
N-stem, and also, revealingly, with
-stem statives. The latter highlight
the fact that some underlying Agent has brought about the state, a
what remains—what is relevant—is the object-like affectedness of the
subject, from which the event is perceived to emanate. Similarly, with
passive situations the Agent is defocused to the extent that it is typically
omitted from the clause. Again, what remains is the affected object,
the Undergoer, in some lasting resultant state predicated by the verb.
In both cases, the elaboration of events is by default zero, and in both
cases the perspective on the event is from its Endpoint. Notably,
greater af
nities to the Initiator or Actor role—more
broadly, representing a less extreme manifestation of Endpoint focus—is
less frequently called upon to express the passive (§4.2).
Well-known as exemplifying the passive function of
are the alter-
native construals of Ur III and later Old Babylonian year names. The
verbs involved in these clauses often represent prototypical transitive
events. When the clause is agentive, the pre
x is regularly
, when
the Agent is de-emphasized, that is, omitted,
is often the pre
x of
[563] mu
ul-gi lugal-e bad
‘The year: Shulgi, the
king, built the wall of the land’ ~ mu bad
‘The year: the wall of the land was built’ (
[564] mu
Suen lugal-e Ur-bi
‘The year: Amar-
Sin, the king, destroyed Urbilum’ ~ mu Ur-bi
‘The year: Urbilum was destroyed’ (Amar-Sin YN 2).
[565] mu
Suen lugal-e ma
‘The year: Shu-
Sin, the king, fashioned the magni
-barge’ ~ mu
‘The year: the magni
As we have observed, there is a reluctance in the language to couple
with predicates that lend themselves to the expression of canoni-
cally transitive events, unless the transitivity of the clause as a whole is
downgraded. Stripping the verb of its Agent, as in the agentless passives
above [563]–[565], is obviously one, particularly effective, strategy for
accomplishing this. The absence of an Agent in a clause that contains
a canonically agentive predicate represents a marked construction. And
in this way,
can be seen as signaling a disruption of the normal
peripheral members of the passive category, those akin to the impersonal
passive in which the Agent is de-emphasized but not omitted. In other
words, implicit to this de
nition is the understanding that the passive
is a graded category, a progressive continuum of Agent de-emphasis.
At its most extreme, it is bounded by the paradigmatic role-remapping
passive and, at its minimum, it shades over to the active voice. This
domain may also be described in terms of the broader continuum
of event types portrayed in Fig. 8, itself a graded scale.
with the gamut of middle and passive event-types, being weighted
most heavily to the extreme end—events in which there is no percep-
tible Agent, as in spontaneous events, or events in which the Agent is
semantically present, but clausally off-stage, as in the canonical passive.
More peripherally,
extends to events in which an Agent may be
present, both semantically and syntactically, but may not be the most
salient participant in the clause. These are clauses in which the Agent
chapter five
The passive, or, more accurately, the Agent-defocusing function of
the pre
x may be most clearly illustrated by the year names, but it is by
no means restricted to this genre. Attestations of the pre
x appearing
in contexts in which the Agent is omitted, but necessarily part of the
semantics of the clause, are ubiquitous. The early attestation of much
of this evidence (e.g., [568] ED IIIA; [569], [570] ED IIIB) underscores
that this is a primary function of
and refutes the suggestion that
it was a secondary development (see §1.3). In all instances, the pre
emphasizes the Endpoint of the event—it focuses upon the outcome of
the action and the affectedness of the Object, overlooking the initiation
of the action and the role of the Agent.
[568] 1 ab
‘1 cow was
supplied to Inana; 1 cow was supplied to Nanna’ (Alberti and
Pomponio 1986, no. 45 i 1–6).
[569] 1 ma
‘1 (dead) goat
was removed from the fallow land of Sagatura-
eld’ (Nik 179
i 1–3 [= FAOS 15/1, pp. 404–405]).
ª10(gur) šeº gur-
of barley on
(system) were bought with silver’ (Nik 77 i 1–2
[= FAOS 15/1, pp. 292–293]).
[571] mu Ku-li
› eger-a-ni u
dam dumu-ni dumu Ba-
‘Because Kuli has been killed,
his estate, and his wife and child are given over to the sons of
Babamu’ (NG 41: 5–8).
[572] ud 7-am
‘For seven days grain was not ground’
[573] gu
‘(Gudea’s) call was heard’
2 ud ul-li2-a-ta sig4 E2-babbar-ra
a i
‘From the time
when the brickwork of Ebabbar was (
rst) constructed’ (Sam-
–)Similar to the year names in [567] are Agent-defocusing/Object-focus-
ing clauses in which the Object assumes the leftmost position [575]-
[582]. In these clauses the addressee’s or reader’s attention, which
typically follows the word order and topical importance of the clausal
NPs, does not
ow in the expected unmarked direction, from Agent
to Object, but rather takes the marked, unexpected route, namely,
from Object to Agent (i.e., OSV rather than the standard SOV word
order). Again, the pre
x functions to indicate a disruption in the normal
[582] nig
3-g[e-ne] zu2-lum gišpeš3 ga-
tu-ra i
-ne kid-da
dŠul-gi-me-en ba-tu-de3-en-ta nita kalag-ga-me-en ‘I am
Shulgi, since I was born I have been a mighty man’ (
nam-tar gig-ga-ka
‘I was born on an ill-fated
day’ (SP 2.5).
dŠul-gi bala nam-he2 ba-tud-da ‘Shulgi, who was born for a
reign of prosperity’ (
2-mu kur Dilmun
kur Dilmun
‘My house,
before the land of Dilmun ever existed, was created from a
date palm. Isin, before the land of Dilmun ever existed, was
created from a date palm’ (Ninisina A 93–94).
Similarly, there is
, which may denote ‘become, turn (into a
ed state)’, ‘change’, a spontaneous event along the lines of [551]–
[555]. Joined with
the verb has passive overtones; the common
incorporation of an inanimate agent pronoun
, may suggest
an underlying, if indistinct, Agent (see also [663]–[665]).
ki-ga igi nigin-bi ba-kur
-a ba-ab-dug
‘In Eridu, (everything) was reduced to ruin, was wrought with
confusion’ (LEr A18).
[591] Ke
‘Kesh, built
all alone on the high steppe, was haunted’ (LSU 143).
[592] ud-ba nam-lugal kalam-ma-ka
those days, the kingship of the Land was de
led’ (LSU 99).
Even with canonically transitive actions (e.g.,
‘destroy, demolish’),
the construal of the event may be such that there is no sharp distinc-
[594] im
ba] ‘When the rain had
rained, when the walls had been demolished’ (
Of course, passivization is not limited to highly transitive predicates,
but may intersect with the full range of middle situation types. A verb
such as
‘steal’, for instance, commonly occurs with
, the pre
[596] nin-mu
ba ‘My lady, you selected me—
[597] dingir arhu
-mu ki-
mu-bi ‘“My god,
who is full of compassion, selected me from the horizon” (is)
the name (of this libation cup)’ (Gudea 69: 3
gišªtukulº-ba lugal-mu
mu-bi ‘The name of this
weapon is “My lord selected me”’ (Nammahani 16: 3
Backgrounding Functions
Note, in this connection, that in the
construction, which is essentially
a reduced relative clause, the Object assumes the topical, leftmost position, with the
form sharing much in common semantically with the passive.
Gragg further observes that “of the other 16, 7 involve a third person plural suf
(there are no occurrences of ba + Verb + e
+ a + ba) and 3 . . . involve the pre
their inherent Endpoint character, note the use of
in the frozen (nominalized) verbal
‘response, answer’ (see PSD B s.v.
Also note: e
ud-mu he
the shrine of Larsa, the city where I was created, may my days be long’ (Warad-Sin
22: 39–41); see also [586], [588]. Similarly, in the following passage, although syntacti-
cally active, the Object, as in the passive, is more topical than the Agent: di
2] ‘For the little ones to whom I have given
birth may rewards not be lacking’ (EnkNh 272).
2 ªUmmakiº-k[e4] ªeger
ba-hul-a-ta nam-tag dNin-gir2-su-da e-da-ak-ka-am
‘The leader of Umma, when he
destroyed Lagash, committed a sin against Ningirsu’ (Uruin-
For a different understanding of the semantics of this expression (namely, as a
causative construction) and the rationale behind this substitution, see Zólyomi 1999:
-ra mu-
3] ‘When Ningirsu had chosen
seized from Adaga’s place after being on the run for four years’
IGu-u2-gu arad2 Ur-
in-gar mu lugal ud a-ra
ga-hul bi
‘Gugu, the slave of Ur-Nungal, was able to escape, but
was caught. He appeared (before the judges) and swore by the
name of the king, “May I be mutilated the day that I
ee for a
second time!”’ (NRVN 1, 1: 1–8 [= Molina and Such-Gutiérrez
Subordinate clauses provide a particularly transparent context for
observing the pragmatic functions of the pre
xes. There is, however,
additional, albeit more subjective, evidence from the alternative portrayal
of events in narratives, which speaks to similar pragmatic functions, spe-
cally for
like me,
fty of them, act on my behalf (lit.: act at
my side
2 tuku e2-a-ni-še3 ama tuku ama-a-ni-še3 nita sag-dili e-ne-gim ak 50-am3 a2-ni-Še3 ba-an-ak-eŠ (var. [. . .]-
hu-mu-u[n-. . .])
‘Whoever had a household (went) to his household. Whoever
had a mother (went) to his mother. Bachelors, those like him,
there were
fty, acted on his behalf (lit.: acted at
his side
(GgHw-A 52–53). The Ur source that witnesses
n- . . .] confuses the Þ rst- and third-person versions of the account (cf.
in [607]); note that the other three exemplars in
which the pre
x is extant attest
, i.e.,
A similar distribution is in evidence in the famous passage from
The pragmatic grounding functions of the pre
xes are likely in evidence in
gišellag-mu kur-
-ma-mu Ganzir-
(var. im-ta-e
) ‘My
has fallen down
you order me around, I will not be ‘your big brother’”’ (Dial 3: 5–6). What appears
to be salient here is a difference in relative pragmatic emphasis—the
rst clause (
is an asserted action on the part of the speaker, the second (
on the part of the addressee.
nite. Also to be included here, with particular relevance for verbs
that are naturally trivalent and therefore regularly take a
as part of their lexical frames—e.g.,
‘address, speak, call’—are instances of
). These are clauses in which the
role is—against the typical usage of the verb in ques-
tion and so against expectations—not
lled. Such clauses correspond
to events of the type ‘give away’, ‘speak out’, as well as ‘give for doing
s.th.’, ‘call for s.th.’, which similarly lack a
All of these cases can be accounted for in terms of the Nominal
. In both Sumerian and Akkadian the primary
son value is unmarked vis-à-vis its use with second and third, as the
location of the
rst-person is intrinsically
. In Sumerian, when the
dative argument is a third-person
(i.e., of the
NP which falls in the middle of the Nominal Hierarchy, the speaker
may deem that person to be highly salient and so mark that person as
if he is
, that is, like the SAPs (
‘he gave to him’, cf.
‘he went to him’). Alternatively, in the absence of such empathy
or identity, the speaker or writer may leave the argument unmarked
). In the extreme case of the absence of a
argument, when one is typically expected (again,
‘address, speak, call’), a type of Endpoint at
nity—to use a term that captures the iconicity involved—the pre
nuance is apparent (§5.1.2): compare
’ (also ‘give for doing s.th’) with
’. Representing an
extension of the coding of this limiting case, from the other end of
the spectrum as it were, is the marking of inanimate recipients, NPs
Yoshikawa (1992a) considers the primary function of
to be signaling the absence
of a dative argument. In his view, this applies to all cases—the pre
x, he contends,
does not indicate a decrease in transitivity (the primary function in my view), but a
valency decrease with respect to the dative (
forms are, therefore, labeled
of the action, the subject himself may be perceived as the locus of the
action’s principal effects, e.g.,
And with the pre
x proximal to the verbal root, the likelihood that it
will interact directly with the verbal meaning, speci
cally, functioning as
a voice marker, is increased. The notion that animate, that is, sentient
are more likely to absorb the effects of the
action than inanimates parallels—indeed, may be seen as an extension
Observe, in this connection, that
may also (although more rarely than
occur in these same contexts, in which it appears to function with its regular middle-voice
meaning as is suggested by the following sequences of middle events: lugal
(vars. gu
-e, gu
(vars. gu
mu-un-na-ab-sum-mu) ‘Before
in the limiting scenario, in which a
nite objects to be
more marked
than less animate and de
This tendency, as Croft explains, is a consequence of the
This claim is supported by the fact that the third-person singular is the least
marked person cross-linguistically, frequently taking zero marking, for instance, note the
Akkadian predicative construction in which the 3
m. form is
Ø, but all other
persons and numbers are marked, i.e.,
of Ø
, proceeding to inanimates, collective and
nite animates, and concluding,
nally, with speci
. This category is represented prototypi-
cally by [612]–[613]. Also included here are
[614]–[615], which correspond to in
nitival constructions in English
(e.g., ‘give for doing s.th., in order to do s.th.’) as well as purpose con-
structions lacking an adressee [616]-[617] (e.g., ‘call for s.th.’):
[612] ki-bala hul-gig
‘May An cast off (lit.: give away) the rebel land, as it is hateful
to your Nanna’ (Inana B 93).
[613] ga
-e [
(var. [g]u
te (vars.
, ga-am
-da-te) ‘I will shout until the heavens
draw near to the earth!’ (GgBH B51).
[g]az-de3 ba-an-sum-ªmuº-uŠ ‘(The judges) delivered him
over to be executed’ (PBS 8/2, 173 rev. 13
[= NG 1, p. 132:
n. 5]). Similarly, SLHF ii 37–38 and RA 71 (1977): 6, 25.
pu-uh2-ru-um Nibruki-ka
‘The assembly of
Nippur delivered Ilshu-muballit over to swear an assertory oath
at the gate of Ninurta’ (WO 8 [1976]: 160 ll. 19–22). Similarly,
[616] ur-sag nig
‘Warrior, you called for what is
tting’ (Gudea Cyl. A viii 20).
[617] En-suh-gir
kin-e ‘Ensuhgirana asked for advice, searching for an answer’
, both specific and non-
Since I am here concerned with semantic roles and animacy, and not case mark-
ing, the morphology of this element (e.g., Edzard’s understanding of

term.] [2003a: 135]) is of secondary importance for my purposes.
, as I have noted, typically belong to the
class, and as such,
marks a departure from expectations, a deviation from the
2-e . . . lugal
provided oracular decisions for the temple’ (Gudea Cyl. B iv
3 70 še gur-am
he should give 70
of grain for the tithe of (the temple of)
Ningal’ (LEM 190: 7–9).
dŠul-gi lugal Urim
-ma-me-en u
nam-nar-ra (vars. nam-nar-e,
nam-nar-a) gu
(var. ha-ma-sum) ‘I, Shulgi,
king of Ur, have also devoted myself to music’ (
dInana-ke4 me3 šen-šen-na ki-bala-e ba-an-sum ‘Inana handed
over (victory in) battle and strife to a rebellious land’ (LSU
‘(Enlil) gave (Huwawa’s)
rst aura
to the
elds. He gave his second aura to the rivers’ (GgHw-A
še3 tumu duru
‘I will call up to heaven for a
certain verbs take prototypically inanimate
‘add’ in the sense of ‘add s.th.
to s.th.
’. As expected, the verb is regularly
with this meaning, e.g., nam-til
-mu nam-til
(Enlil) add life to my life’ (Lugalzagesi 1 iii 19–21); nig
‘These being the
bridal gifts for Bau for the new house, which Gudea, ruler of Lagash, the builder of
the temple, had increased (i.e., added gifts to gifts)’ (Gudea St. E vii 15–21); 30 gun
‘He should add 30 talents of dry
bitumen to the dry bitumen (already assigned) for the (boat-)mast’ (LEM 222: 1-3);
added to hand, and a man’s house is built up. Stomach (lit. food) added to stomach
and a man’s house is destroyed’ (SP 2.138).
workers in the
rst clause [
ed number
in the second [
]), as well as those that are generic
–[628] lugal-
‘My lord should
give (the
eld) to the work troop’ (FAOS 19, Gir 16 rev. 2
). For
the value sur
in (Pre-)Sargonic sources, see Steinkeller 1990 (note
)-ra in this text [FAOS 19, Gir 16: 5]).
[629] eren
e nu-tuku Ur-
e lugal-ta
‘The troops on
duty have no barley. Until Ur-Gatumdug says (otherwise), he
should give them 1 royal
of barley each’ (TCS 1, 141:
še-ba gur Ti-e2-mah-ta 3 gur A-lul 1(gur) 1(bariga) Nin
of barley-rations for
for Alul, and 1
for Ninani—he
should give these barley rations to them’ (TCS 1, 151: 3–7).
[631] 4 dumu-dab
-ba 1(bariga)
e lugal-ta
e gur lugal-am
-ba bala sumun-na-ke
‘He should give the four agricultural laborers 1
of barley each . . . and he should give 8 royal
barley to the agricultural laborers of the last term’ (TCS 1,
ªi3-du8º g[al] ªkurº-ra 7-bi nig2-ba ba-ab-sum-mu
‘He gave gifts
1991; see [676] for
with a perfect meaning). Speci
cally, the
x in these cases, generally legal and administrative texts, functions
independently of the following dative in
x and combines with the verbal
root, despite the morphological
of the two:
Ba-zi dumu
‘Lu-Huwawa was given to Bazi, son of Sheshshesh,
as his slave-girl’ (NG 126: 12–13).
dam dumu u
Ur-mes Lu
wife, child, and slave girl were con
rmed to Lugula’ (NG 42:
[637] 2-a-ne-ne nam-erim
‘Both (the owner
of the slave and the guarantor) were given over before (the
witnesses) to take an assertory oath’ (NG 51: 16–17).
And, occasionally, the pre
x is encountered in Old Babylonian literary
texts in what appears to be its regular middle-voice capacity, despite
the presence of an intervening dative in
2 a-ra-zu im-me-a-bi
‘May your heart
relent toward him who utters prayers (to you)’ (LUr 432).
[639] ga
‘At that time, I will,
myself, submit to him’ (EmkLA 291).
The correlation of
, as has long
been observed (§1.3), extends beyond the dative to the other case in
of the verb. As with the dative, the pre
x often re
ects the degree of
animacy of the peripheral argument as represented by the immediately
following in
x. Thus, when Enlil looks upon Ningirsu, a
whom he empathizes, the pre
x is
[640], but when Gudea looks
upon a clay pit, the pre
x is
[641]. Similar alternations with the
comitative are also common. And, as observed in §5.1.2,
occurs with the ablative with v
erbs of motion. H
, in many of these cases, is preferably understood as ful
lling a
middle-separative function and modifying the meaning of the verbal
root, rather than re
ecting animacy (see §5.1.2).
-e en
favorably upon the lord Ningirsu’ (Gudea Cyl. A i 3).
še3 igi zid ba-Ši-bar ‘(Gudea) looked favorably upon
the clay pit’ (Gudea Cyl. A xiii 18).
With the locative, the pre
x’s expression of object-focus may be seman-
tically similar to that exhibited in dative scenes. This is particularly
true when the chain
occurs with compound verbs in which an
to s.th.
(see Attinger 1993: 282), with the locative
often marking the instrument when applicable. In many cases, these
verbs, intrinsically or contextually, represent what may be referred to as
object-oriented events. An activity is performed on an object for what
may be perceived as the bene
Also to be included among events of this general type is
sound the horn’, which occurs regularly with
(attested already at Fara, e.g., SF 40
]—see Karahashi 2000a: 111; Krebernik 1998: 324), e.g., uru
‘In his city, as if to a single man, (Gilgamesh) sounded the horn’
-ru-um-ma-ta nimgir-e sila sila-a
accord with the order of the assembly, the herald has sounded the horn throughout all
and the action itself. Correspondingly, representing the opposing node of
the event,
embraces the Endpoint categories of inanimacy, patiency,
. And, with respect to the temporal qualities of the event, this
iconicity extends to telicity, perfectivity, and most notably—given its
cation in bilinguals with the Akkadian category known by this
the present result of a past event, e.g.,
John has broken the window
Comrie 1976: 12; Bybee, Perkins, and Pagliuca 1994: 54–55).
perfect in particular, then, shares much in common with the passive,
that other category with which
liated, for both are
concerned with the states that result from previous actions.
This distinction is summarized by Binnick: “Most scholars today see the perfect as
resultative or stative, referring to the enduring resultant state subsequent to a completed
action, while the perfective is viewed by most as, like the Greek aorist, representing an
action as a whole, without regard for the progress or development of the action—that
is, as ‘complexive’ or ‘totalizing’” (1991: 161).
and aspect, Dixon observes that the passive and antipassive often carry
aspectual meaning, pointing out that Tzutujil (Mayan) has a ‘comple-
tive passive’, which, as its name suggests, “emphasizes the result of
the activity on the patient as well as the termination of the activity”
(1994: 148 citing Dayley 1985a: 342; see also Dayley 1978: 31–33;
Dayley 1985b: 205–207). As the perfective can be seen as subsuming
the perfect, the latter adding a current relevance reading, there are
intent or other internal executive capacity of the actor (non-eventive), and
the actor as the dominant participant in the scene (active voice). When
one takes the viewpoint or orientation of the goal, one ‘sees’ primarily
the termination of the action or its relevance to the current state of the
goal (perfective or perfect aspectuality), the actual realization of the action
as it affects some entity (eventive modality), and the goal as the dominant
participant in the scene (passive voice). (1982: 206)
A perfect meaning for
, then, as the above discussion makes clear,
would be entirely consistent with the pre
x’s other Endpoint-oriented
functions and could be placed on a semantic map of its functions
adjacent to its use with passive situations and spontaneous events. Of
course, direct evidence for this meaning is relatively late, stemming from
the Old Babylonian period, as the bilinguals in which the Sumerian
and Akkadian forms are equated, and the Akkadian perfect itself, are
rst and foremost phenomena of this period. The question that arises
motion (§§4.1.2, 5.1.2) that this dynamic quality is most conspicuously
on display, for when coupled with
, the inceptive
I thank P. Steinkeller for the Polish parallel.
-ge (var. he
-ge) ad-da-mu
u sa
mu-e-en (var. mu-un-e
) ‘My vagina is too small, it does not know
pregnancy. My lips are too young, they do not know kissing.
If my mother learns of it, she will slap my hand! If my father
learns of it, he will lay hold of me!’ (EnlNl 30–33).
Similarly, note the following attestations of
)ing ‘recognize’ and ‘learn’ (lit.: ‘come to know’), and ‘become known’;
compare [652]–[653] with these two
rst-millennium passages: 7-bi
‘to learn (Sum.
he learned) about the activities of the Seven’ (CT 16, 45: 122–123);
an-]na-a-ti kul-lu-mi-im-ma ‘to reveal (Sum. he made known) these procedures’ (AAA 22 [1935]: 82
ll. 86–87 [Utukku Lemnutu]).
[646] kur-ra a
-in-gar igi he
har-ra-an kug An-na-ka-
will defeat the mountain, I will examine it, and I will learn its
length! I will go out on the holy road of An and I will learn
its depth!’ (InEb 85–86).
u nar-ra
u tam-tam-ma-gim gi
a ki-mu-u
mu-ni-ib-DU-[x-x] u
-ne-en (vars. u
‘Like a musician with perfect
hands, if they bring me a musical instrument, which has never
been heard before—when I play it, its true sound (lit. heart)
becomes known!’ (
and variants are under-
stood to represent
, i.e.,
base of
[648] zi gi
) ‘Calm down! Do not decide anything until
you have learned more!’ (EmkEsg A132).
dEn-ki inim m[u-un-dug
-ga-a Abzu-a
‘Father Enki in the Abzu was able to learn what had been said’
(NinTrtl B9).
2 mu-lu-ne-ka malga
‘You learn what is being deliberated in the
houses of men’ (Dial 5: 113).
[653] dumu lu
(vars. ba-e-
/bi-zu-zu) ‘Don’t rape a man’s daughter—the court-
yard will learn of it!’ (In
ur 62). Note the Akkadian version with
, i.e.,
ma-ar-ti a-me-li ina
‘Don’t rape a man’s daughter—the assembly
will learn of you’.
Relatedly, the pre
x often occurs with verbs that are derived from
what are basically adjectival roots. Although there are no clear mor-
phological means of distinguishing adjectival roots (see Black 2000)
from verbal, or, for that matter, nominal roots, lexemes such as
esh/heart of s.o./s.th.’, which
similarly makes use of the D-stem of
in factitive function (cf.
an-na zag-bi-
= AN-
ana pat-ti-

‘He (Sum.: They) pleased the heavens to their borders’ [
13: 7]). Notably, transitivity is minimized in those clauses in which the
semantic object is relegated to the locative or locative-terminative case
(i.e., lit. ‘be pleasing
s.o./s.th.’ [= Akk.
]; cf.
‘please s.o./s.th.’ [58]).
Consequently, as in [655] and
In this same vein are the Ur III personal names based on the typically stative
‘be beautiful’. The combination
denoting a process, while
designates ‘be beautiful’, signifying a state, e.g.,
‘The king made (him/her) good/beautiful’,
In [660]–[662] the context supports an inchoative reading for
the union of pre
x and verb signifying a state that results from an
autonomous, punctual process much along the lines of the spontane-
ous events:
hul ba-gal2-la gu2-zu la-ba-ši-šub ‘You don’t scorn bread
which has turned bad’ (SP 1.21).
[661] a-gar
-ta a um-ta-kud-a-ta ki-duru
the water had been cut off from the arable tracts, and the work
on the damp earth had begun’ (HoPl 80–81).
[662] uru
-mu du-lum gig
en an-na-gim
‘Because bitter misery appeared in my city, I beat my wings like
a bird of the heavens and
ew to my city’ (LUr 104–106).
With the verb
‘be suitable’, the combination often re
ects the
dynamic event of
becoming suitable
. Attestations with
a pre-radical
, which may suggest an underlying agency—generic,
unknown, or unimportant—that brings the state into being (see [590]–
[592]; cf.
)[663] ki-tu
residence of valor, which has been made suitable for dwelling’
2 gišma2-gur8-ra ba-ab-du7-a-zu šag4-ba u2teme
the middle of your watercourses, which were made suitable
for barges,
-plants grow’ (LUr 367). Note the spontaneous
event represented by
barag mah-ha sag il2-la-a ba-ab-du7-me-en3 ‘You are one who
has been entitled (lit. made suitable) to hold his head high on
As these examples clearly illustrate, the resultant change-of-state
serves to highlight readily lends itself to a perfective or perfect
reading. Naturally, the choice of one over the other for a given pas-
evidenced by many of the examples scattered throughout this chapter,
to translate verbs pre
xed with
as perfects when the resultant state
is seen as persisting and affecting conditions at the time that the mes-
sage was coded. The grammar may support either a past or a perfect
translation, but the latter is often chosen because the context suggests
a persistent state that is relevant to the reference time, e.g.,
[666] ma-da dim
‘The allegiance of the territory
has changed’ (RCU 17: 23).
2 mu-zu i
(vars. mu-da-gul-e/en)
-mu u
‘O city, your name exists, but you have been destroyed! O city,
your wall rises high, but your land has perished! O my city,
(you are) like a faithful ewe, your lamb has been separated from
you! O Ur, (you are) like a faithful goat, your kid has perished!’
(LUr 64–67). Note that the anomalous variants
compelled to equate the two in bilinguals. And in this connection, we
cannot overlook the highly suggestive, albeit late, evidence from the
Neo-Babylonian grammatical texts (a g
rammatical genre,
which has its origins in the Old Babylonian period), in which the pre
is described as
Black dismissed the identi
cation of
as a designation for the
perfect, already suggested by the editors of
4 (p. 186 s.v.
attractive” (1991: 98); as discussed below, Black believed that the perfect was excluded
from the Old Babylonian grammatical texts, preferring to see all
xed verbs as
t, and Nt stems.
The restorations are based on the unilingual Sumerian version STVC 3+4 iv
26–30 and SP 3.179 also cited by Lambert (BWL, p. 274).
copy cited here provides a syllabic Sumerian version with an Akkadian
š-še um-ni-im im-ma-an-gen
‘Now the troops have arrived’ (RCU 10: 18 [Susa]).
mar, particularly with respect to the Akkadian perfect, which by this
period had lost its current relevance value and had come to replace the
of legal phrases, the so-called Sumerian Laws Handbook (Roth 1997),
as a comparision of [677] with [678] suggests. Indeed, in many cases
nd direct counterparts in the warranties of Ur III con-
tracts, cf. [518], [547], [548].
ªbiº [ba]-ªuŠ2º [ba-an]-zah2-a ªu2º-gu ba-an-de2 u3 tu-ra ba-an-ku4 ‘If she has died, run off, vanished, or fallen
[t]ukum-b[i] lu2 sag-ga
ugu bi-an-de
-e ga
-la ba-an-dag
tu-ra ba-ab-ak
um-ma a-wi-lum ar-da i-gu-ur-ma
im-tu-ut i
-ta-liq it-ta-ba-ta
‘If a man hired a slave, but
(the slave) has (subsequently) died, run off, vanished, stopped
work, or fallen ill’ (Ai. VII iv
The events represented in the parallel clauses of [677] and [678]
while simultaneously stressing the culminating event in a manner that
to occur in this stem. As Black explained, “it seems that we are dealing
These perfects, then, take their place in the grammatical texts along-
side what we would traditionally distinguish as
-stem forms, e.g., [233],
[413]. The argument that the acceptance of perfects would ruin the
systematic arrangement of the paradigms (Black 1991: 28) can only
hold true if the Old Babylonian grammarians drew the same rigorous
Note that -
xed verbs—the only forms for which the perfect can be for-
mally distinguished from the
Grammatical voice is a linguistic means of altering the perspective from
which events may be viewed, providing speakers with a series of options
‘fear’, and stretching as far as to occur with middle situations that
involve volitional subjects, as is the case in many body-action and
self-benefactive events, e.g.,
‘lie (down)’,
‘receive’. When expressed with
, what these events share
with the more prototypical spontaneous events and passive situations
is a singular focus on subject affectedness. Complementing
, which
is preoccupied with the Initiator role,
focuses on the Endpoint, the
on events,
has no direct connection with
. If
a morphological doubling of
(see §1.3) one would, reasonably, expect
the pre
x to function as
does, perhaps even for the reduplication
to re
ect an iconic intensi
cation of the qualities intrinsic to
other words, one would expect the pre
x to occur in the expression of
prototypical transitive events, which it does not. If we are to seek to
derive the pre
x from the other members of this class, the anchor for
any such compound must be
. Given the ventive force the pre
with verbs of motion, at least in certain contexts, the most viable can-
didate, on the basis of function and meaning, remains the one implied
by the analysis of the ancients, namely,

Iconicity is integral to the functions attributed to each pre
x and to
the system as a whole. It is on display with the aforementioned distinc-
tory in
give to me
), nearly so in
give to you
option that is exercised in the case of
give to him
), when
the speaker empathizes with this third person and seeks to identify this
individual, who is uninvolved in the discourse, with the SAPs, the
one category to another, accumulating properties from each along the
way (Bybee, Pagliuca, and Perkins 1994: 17–18; Kemmer 1994: 220).
* * *
The theory presented in this book, that the primary conjugation pre
represent a system of grammatical voice in Sumerian, is based on the
observation that each pre
x tends to occur with certain verb-types and
more frequently in certain environments than in others. This is a basic
active-middle voice system in which middle marking is often used to
express the passive. To summarize, the markers in this system consist
, representing what I have described as the marked active voice,
and two middle markers of different
. Although there is near isomorphic overlap in
types of situations in which the two middle markers may occur—both
xes span the range of middle domain—
clusters much more
toward the low end of this spectrum.
gravitates in terms of proto-

situations. The function of this pre
x derives precisely from its neutral-
ity, allowing a speaker to de-emphasize or neutralize various qualities
of a proposition, including agency, subject-affectedness, empathy, and
information salience.
As these last comments suggest—and as I have stressed throughout
this book—the choice of pre
extended category in Sumerian as constituting a broad focus system
that includes aspectual distinctions and various types of locative focus
in addition to grammatical voice.
The modern scholar cannot help but see a certain chaos in the
behavior of the pre
xes precisely because there are so few semantic and
morphosyntactic environments in which a given pre
x can be expected.
Much of this, as we have seen, stems from the critical role played by
pragmatic factors. And it is exacerbated by the multiple functions of
the pre
xes and the polysemy that characterizes the system as a whole.
Undoubtedly, speakers must have relied heavily upon the broader con-
expresses them by more grammatical means, exploiting the
of an iconically structured system. And this speaks to what is most
profoundly interesting about the pre
xes—the glimpse they provide
into how Sumerian speakers conceptualized reality and experience as
ected in their language.
Aalderen, C. T. van. 1982. “Some Observations on Ergativity,”
Behrens, H. 1978.
Enlil und Ninlil: Ein sumerischer Mythos aus Nippur
(Studia Pohl: Series
Maior 8). Rome: Biblical Institute Press.
Benito, C. A. 1969. “Enki and Nimah and Enki and the World Order.” Ph.D. diss.,
University of Pennsylvania.
Benveniste, E. 1971.
Problems in General Linguistics
. Coral Gables, FL: University of
Miami Press.
Berlin, A. 1979.
Enmerkar and Ensu
danna: A Sumerian Narrative Poem
lications of the Babylonian Fund 2). Philadelphia: University Museum.
Bertin, G. 1886. “L’incorporation verbale en Accadien,”
Time and the Verb: A Guide to Tense and Aspect
. Oxford: Oxford Uni-
versity Press.
Black, J. 2000 [published 2005]. “Some Sumerian Adjectives,”
Sumerian Grammar in Babylonian Theory
, 2nd ed. (Studia Pohl: Series Maior
12). Rome: Editrice Ponti
cio Istituto Biblico.
——. 1986. Review of M.-L. Thomsen,
The Sumerian Language,
——. 1985. “A-
e-er Gi
-ta: A Balag of Inana,”
Chafe, W. L. 1976. “Givenness, Constrastiveness, De
niteness, Subjects, Topics, and Point
of View,” in
Subject and Topic
, ed. by C. N. Li, pp. 25–55. New York: Academic Press.
Charpin, D. 1992. “Les malheurs d’un scribe ou de l’inutilité du sumérien loin de
Nippur,” in
(CRRA 35), ed. by M. de Jong Ellis, pp. 7–27.
Philadelphia: University Museum.
Christian, V. 1957.
Beiträge zur sumerischen Grammatik
(Österreichische Akademie der
Wissenschaften. Philosophisch-historische Klasse, Sitzungsberichte, 231. Band, 2.
Abhandlung). Vienna: Rudolf M. Rohrer.
Chung, S. 1981. “Transitivity and Surface Filters in Chamorro,” in
Studies in Paci
Languages & Cultures
, ed. by J. Hollyman and A. Pawley, pp. 311–332. Auckland:
Transitivity and Discourse Continuity in Chamorro Narratives
(Empirical Approaches
to Language Typology 4). Berlin: Mouton de Gruyter.
——. 1982. “Topicality, Ergativity and Trasitivity in Narrative Discourse: Evidence
from Chamorro,”
Studies in Language
Cristofaro, S. 2003.
Oxford: Oxford University Press.
Croft, W. 2003.
Typology and Universals
, 2nd ed. Cambridge: Cambridge University
Radical Construction Grammar: Syntactic Theory in Typological Perspective
. Oxford:
Oxford University Press.
——. 1995. “Modern Syntactic Typology,” in
Approaches to Language Typology
, ed. by
M. Shibatani and T. Bynon, pp. 85–144. Oxford: Clarendon.
——. 1994. “Voice: Beyond Control and Affectedness,” in
Voice: Form and Function
(Typological Studies in Language 27), ed. by B. Fox and P. J. Hopper, pp. 89–117.
Amsterdam: John Benjamins.
Syntactic Categories and Grammatical Relations: The Cognitive Organization of
. Chicago: University of Chicago Press.
Typology and Universals
. Cambridge: Cambridge University Press.
Croft, W., and D. A. Cruse 2004.
Cognitive Linguistics
. Cambridge: Cambridge Univer-
sity Press.
Croft, W., K. Denning, and S. Kemmer, eds. 1990.
Studies in Typology and Diachrony
for Joseph H. Greenberg
(Typological Studies in Language 20). Amsterdam: John
Croft, W., H. Shyldkrot, and S. Kemmer. 1987. “Diachronic Semantic Processes in the
Middle Voice,” in
Papers from the 7th International Conference on Historical Linguistics
, ed.
by A. Giacalone-Ramat, O. Carruba, and G. Bernini, pp. 179–192. Amsterdam:
John Benjamins.
Crystal, D. 2003.
schrift für Rykle Borger zu seinem 65. Geburtstag am 24. Mai 1994: tikip santakki mala
(Cuneiform Monographs 10), ed. by S. M. Maul, pp. 9–38. Groningen: Styx
Fillmore, C. J. 2003.
Form and Meaning in Language.
Vol. 1:
Papers on Semantic Roles
for the Study of Language and Information [CSLI], Stanford, CA, Lecture Notes
121). Stanford: CSLI.
Lectures on Deixis
(Center for the Study of Language and Information
[CSLI], Stanford, CA, Lecture Notes 65). Stanford: CSLI.
Aspects of Inversion,” in
Voice and Inversion
(Typological Studies in Language 28),
ed. by T. Givón, pp. 3–44. Amsterdam: John Benjamins.
Syntax: A Functional-Typological Introduction
, vol. 2. Amsterdam: John Benjamins.
Syntax: A Functional-Typological Introduction
, vol. 1. Amsterdam: John
On Understanding Grammar.
New York: Academic Press.
——. 1976. “Topic, Pronoun, and Grammatical Agreement,” in
Subject and Topic
, ed.
by C. N. Li, pp. 149–188. New York: Academic Press.
Voice and Inversion
. Amsterdam: John Benjamins.
Gonda, J. 1960. “Re
ections on the Indo-European Medium I (and) II,”
Gordon, E. I. 1959.
Sumerian Proverbs: Glimpses of Everyday Life in Ancient Mesopotamia
Philadelphia: The University Museum.
Gragg, G. 1973a. “A Class of ‘When’ Clauses in Sumerian,”
Heimpel, W. 1994. “Meaning 5 of ba-al,”
Heine, B., and M. Reh. 1984.
Grammaticalization and Reanalysis in African Languages
Hamburg: Helmut Buske.
Hirsch, H. 2002.
Gilgamesch-Epos und Erra-Lied: Zu einem Aspekt des Verbalsystems
Beiheft 29). Vienna: Instituts für Orientalistik der Universität Wien.
Hobbs, J. R. 1985. “Granularity,” in
Proceedings of the Ninth International Joint Conference on
cial Intelligence
, vol. 1, ed. by A. Joshi, pp. 432–435. Los Altos, CA: International
Joint Conference on Arti
cial Intelligence, Inc.
Holisky, D. A. 1987. “The Case of the Intransitive Subject in Tsova-Tush (Batsbi),”
Hook, P. E. 1991. “The Emergence of Perfective Aspect in Indo-Aryan Languages,”
Approaches to Grammaticalization.
Vol. 2:
Focus on Types of Grammatical Markers
logical Studies in Language 19/2), ed. by E. C. Traugott and B. Heine, pp. 59–89.
Amsterdam: John Benjamins.
Hopper, P. J. 1991. “On Some Principles of Grammaticalization,” in
Approaches to
Volume 1:
Jackendoff, R. 1987. “On Beyond Zebra: The Relation of Linguistic and Visual Infor-
Jagersma, B. 2007. “A Concise Sumerian Grammar” (unpublished manuscript).
——. 2006. “The Final Person-Pre
xes and the Passive,”
——. 1993. Review of J. L. Hayes,
A Manual of Sumerian Grammar and Texts
Jelinek, E. 1993. “Ergative ‘Splits’ and Argument Type,”
MIT Working Papers in Lin-
Jespersen, O. 1924 [rev. ed. 1992].
The Philosophy of Grammar
. Chicago: University of
Chicago Press.
Jestin, R. 1976. “Quelques notes complémentaires sur le système pré
xal sumérien,”
Kramer Anniversary Volume: Cuneiform Studies in Honor of Samuel Noah Kramer
25), ed. B. E. Eichler, pp. 261–263. Kevelaer: Verlag Butzon & Bercker.
Le verbe sumérien: Pré
medagan: Originality and Dependency in Sumerian Royal
Hymnology,” in
Bar-Ilan Studies in Assyriology Dedicated to Pinhas Artzi
, ed. by J. Klein
and A. Skaist, pp. 65–136. Ramat-Gan: Bar-Ilan University Press.
——. 1981a.
ulgi Hymns: Sumerian Royal Hymns Glorifying King
ulgi of Ur
. Ramat-
Gan, Israel: Bar-Ilan University Press.
——. 1981b.
The Royal Hymns of Shulgi King of Ur: Man’s Quest for Immortal Fame
Lambert, M. 1972–1973. “Probleme essentiel dans l’Étude du Sumerien,”
Comptes rendus
du Groupe Linguistique d’études chamito-sémitiques
Lambert, W. G. 1960.
Babylonian Wisdom Literature
. Oxford: Oxford University Press.
Lambrecht, K. 1994.
Information Structure and Sentence Form: Topic, Focus and the Mental
Representation of Discourse Referents
Cambridge University Press.
Landsberger, B. 1923. “Der ‘Ventiv’ des Akkadischen,”
Langacker, R. W. 2002.
Concept, Image, and Symbol: The Cognitive Basis of Grammar
ed. (Cognitive Linguistics Research 1). Berlin: Mouton de Gruyter.
Foundations of Cognitive Grammar
. Vol. 2:
Descriptive Application
. Stanford:
Stanford University Press.
Foundations of Cognitive Grammar
. Vol. 1:
Theoretical Prerequisites
. Stanford:
Stanford University Press.
Non-Distinct Arguments in Uto-Aztecan
(University of California Publications
in Linguistics 82). Berkeley: University of California Press.
Langacker, R. W., and P. Munro. 1975. “Passives and their Meaning,”
Langdon, S. 1911.
A Sumerian Grammar and Chrestomathy: With a Vocabulary of the Principal
Roots in Sumerian and a List of the Most Important Syllabic and Vowel Transcriptions
. Paris:
Librairie Paul Geuthner.
Lazard, G. 1998.
(Empirical Approaches to Language Typology 19). Berlin:
Mouton de Gruyter.
Lenormant, F. 1873.
Papers from a Colloquium Sponsored by the King’s College Research Centre, Cambridge
, ed. by
E. L. Keenan, pp. 61–83. Cambridge: Cambridge University Press.
Introduction to Theoretical Linguistics.
Cambridge: Cambridge University
MacLaury, R. E. 1991. “Prototypes Revisited,”
Annual Review of Anthropology
MacWhinney, B. 1977. “Starting points,”
Marchesi, G. 2004. “Who Was Buried in the Royal Tombs of Ur? The Epigraphic
and Textual Data,”
Martin, J. B. 2000. “Creek Voice: Beyond Valency,” in
Changing Valency: Case Studies in
, ed. by R. M. W. Dixon and A. Y. Aikhenvald, pp. 375–403. Cambridge:
Cambridge University Press.
Matisoff, J. A. 1991. “Areal and Universal Dimensions of Grammaticalization in
Approaches to Grammaticalization.
Vol. 2:
Focus on Types of Grammatical Markers
(Typological Studies in Language 19/2), ed. by E. C. Traugott and B. Heine, pp.
383–453. Amsterdam: John Benjamins.
Matthews, P. H. 1991.
, 2nd ed. Cambridge: Cambridge University Press.
Merlan, F. 1985. “Split Intransitivity: Functional Oppositions in Intransitive In
Grammar Inside and Outside the Clause
, ed. by J. Nichols and A. C. Woodbury, pp.
324–362. Cambridge: Cambridge University Press.
Michalowski, P. 2004. “Sumerian,” in
The Cambridge Encylcopedia of the World’s Ancient
, ed. by R. D. Woodard, pp. 19–59. Cambridge: Cambridge University
Osgood, C. E., G. J. Suci, and P. H. Tannenbaum. 1957.
The Measurement of Meaning
Urbana, IL: University of Illinois Press.
Pallis, S. A. 1956.
The Antiquity of Iraq: A Handbook of Assyriology.
Palmer, F. R. 2001.
, 2nd ed. Cambridge: Cambridge University
Grammatical Roles and Relations
. Cambridge: Cambridge University Press.
. Cambridge: Cambridge University Press.
Payne, T. E. 1997.
Describing Morphosyntax: A Guide for Field Linguistics
. Cambridge:
Cambridge University Press.
Payne, D., M. Hamaya and P. Jacobs. 1994. “Active, Inverse and Passive in Maasai,”
Voice and Inversion
(Typological Studies in Language 28), ed. by T. Givón, pp.
283–315. Amsterdam: John Benjamins.
Perlmutter, D. M. 1978. “Impersonal Passives and the Unaccusative Hypothesis,”
——. 1977. “Human Categorization,” in
Studies in Cross-Cultural Psychology
, vol. 1, ed.
by N. Warren, pp. 1–49. London: Academic Press.
——. 1975. “Cognitive Representations of Semantic Categories,”
Journal of Experimental
Psychology: General
——. 1973. “On the Internal Structure of Perceptual and Semantic Categories,” in
Cognitive Development and the Acquisition of Language
, ed. by T. E. Moore, pp. 111–144.
New York: Academic Press.
Roth, M. T. 1997.
Law Collections from Mesopotamia and Asia Minor
——. 1985. “Passives and Related Constructions: A Prototype Analysis,”
Shibatani, M. and T. Bynon. 1995. “Approaches to Language Typology: A Conspec-
tus,” in
Approaches to Language Typology
, ed. by M. Shibatani and T. Bynon, pp. 1–25.
Oxford: Clarendon.
Shibatani, M., and S. A. Thompson, eds. 1996.
Grammatical Constructions: Their Form and
. Oxford: Oxford University Press.
Shopen, T., ed. 1985. “Introduction,” in
Language Typology and Syntactic Description
. Vol. 1:
Clause Structure
, pp. 1–2. Cambridge: Cambridge University Press.
Siewierska, A. 1984.
The Passive: A Comparative Linguistic Analysis
. London: Croom
Sigrist, M. and T. Gomi. 1991.
à la grammaire sumérienne
. Geneva: Librairie E. Droz [Reprint: Niederwalluf bei Wies-
baden: Martin Sändig oHG, 1971].
——. 1950. “Études de linguistique sumérienne,”
Cahiers Ferdinand de Saussure
Steible, H. 1982.
Die Altsumerischen Bau- und Weihinschriften
, 2 vols. (FAOS 5). Wiesbaden:
Franz Steiner.
Steiner, G. 1994. “Die sumerischen Verbalprä
xe mu= und e= im sprachtypologischen
——. 1986. “Der Grenzvertrag zwischen Laga
Steinkeller, P. 2004. “A Building Inscription of Sin-Iddinam and Other Inscribed Mate-
rials from Abu Duwari,” in
The Anatomy of a Mesopotamian City: Survey and Soundings
at Mashkan-shapir
, ed. by E. C. Stone and P. Zimansky, pp. 135–152. Winona Lake,
IN: Eisenbrauns.
——. 1990. “The Value sur
of ÉREN in Third Millennium Sources,”
Sale Documents of the Ur-III Period
(FAOS 17). Stuttgart: Franz Steiner.
Steinkeller, P. and J. N. Postgate. 1992.
Third-Millennium Legal and Administrative Texts in
the Iraq Museum, Baghdad
(MC 4). Winona Lake, IN: Eisenbrauns.
Streck, M. P. 2003.
Die akkadischen Verbalstämme mit
(AOAT 303). Münster:
——. 1999. “Das ‘Perfekt’
im Altbabylonischen der Hammurapi-Briefe,” in
Tempus und Aspekt in den semitischen Sprachen: Janaer Kolloquium zur semitischen Sprachwis-
(Janaer Beiträge zum Vorderen Orient 1), ed. by N. Nebes, pp. 101–126.
Wiesbaden: Harrassowitz.
——. 1998. “The Tense Systems in the Sumerian-Akkadian Linguistic Area,”
Steele, F. R. 1948. “The Code of Lipit-Ishtar,”
tions from the Biblical World
, ed. by W. W. Hallo and K. L. Younger Jr., pp. 575–588.
Leiden: E. J. Brill.
——. 1997b. “6. School Dialogues,” in
The Context of Scripture
. Vol. 1:
tions from the Biblical World
, ed. by W. W. Hallo and K. L. Younger Jr., pp. 588–593.
Leiden: E. J. Brill.
——. 1985. “On the Verbal Pre
x /i/ in Standard Sumerian,”
Veldhuis, N. 2004.
Religion, Literature, and Scholarship: The Sumerian Composition Nan
the Birds, with a Catalogue of Sumerian Bird Names
(Cuneiform Monographs 22). Leiden:
E. J. Brill/Styx.
——. 2000. “Sumerian Proverbs in their Curricular Context,”
——. 1993. “An Ur III Incantation against the Bite of a Snake, a Scorpion, or a Dog
Volk, K. 1995.
——. 1977b. “Some Remarks on the Sumerian Verbal In
xes -n-/-b- in the Preradical
Der Rechtsfall der En-
(AOAT 246). Munster:
Zólyomi, G. 2005. “Sumerisch,” in
Sprachen des Alten Orients
, ed. by M. P. Streck, pp.
11–43. Darmstadt: Wissenschaftliche Buchgesellschaft.
——. 1999. “Directive In
x and Oblique Object in Sumerian: An Account of the
History of Their Relationship,”
——. 1993. “Voice and Topicalization in Sumerian.” Ph.D. diss., Eötvös Loránd
University, Budapest.
p. 291
Alberti and Pomponio 1986
no. 45 i 1–6
p. 445: 6
72 rev. 17–18
Bird & Fish
p. 274: 10–18 (SP 3.179) 670
Charpin 1992
Code of Hammurapi
Copper & Silver
The following index of passages cited refers to the corresponding example numbers,
unless otherwise speci
ed (i.e., pages and footnotes are indicated by p. and n. respectively):
Counsels of Wisdom
44: 96–97 p. 293
45: 122–123 p. 291
Curse of Akade
p. 222 n. 2
Death of Gilgame
H11 [Me-Turan] p. 284 n. 21
343 ii 2–iv 2 154
392 i 4–ii 2 237
416 ii 1–iii 3 355
470 iii 1–5 p. 214 n. 26
614 iv 1–v 2 153
Dumuzi & Ge
Dumuzi-Inana B
Dumuzi-Inana C
Dumuzi-Inana D1
Dumuzi-Inana G
Dumuzi-Inana O
p. 293 n. 24
Dumuzi’s Dream
1 ii 24–26 288
1 rev. i 16–19 177
1 rev. i 31–32 80
5 iv 12–19 37
6 iv 16–v 4 37
Edubba A
Edubba C
Elegy on the Death of Nannaya
Enanatum I
2 x 6–xi 2 49
Enki & Ninhursag
p. 269 n. 12
Enki & The World Order
Enki’s Journey to Nippur
625, p. 277 n. 16
p. 277 n. 16
104–109 p. 170 n. 4
1001 ii 11–12 657
Enmerkar & Ensuhgirana
Enmerkar & the Lord of Aratta
text index
537–538 483
A ii 21–22 480
A iii 14–15 205
A iv 25–v 3 p. 174 n. 6
A v 22–vi 4 p. 174 n. 6
A vii 22–23 349
A viii 13–14 148
A ix 18–19 241
A xii 16–17 128
A xiii 12–13 357
A xvii 26–28 679
A xviii 6–9 246
A xviii 8–9 208
A xviii 21–22 p. 214 n. 26
A xx 17–18 31
A xx 21–23 190
A xxii 9–10 57
A xxiii 13–18 348
A xxiii 22–23 604
A xxiv 13–14 444
A xxiv 15–17 394
A xxiv 21–22 536
A xxv 22–23 81
B ii 23–iii 1 210
B xviii 6–7 155
B xviii 12–13 238
B xviii 17–18 112
Gudea Cylfrgm.
Gudea Statues
A iii 5–iv 2 56
p. 225 n. 3
p. 214 n. 26
B vi 77–vii 4 136
B vii 21–48 328
B vii 38–41 267
B vii 42–43 155
B vii 49–54 576
D ii 13–iii 2 644
E vii 15–21 p. 281 n. 20
M ii 7–iii 3 393
Hoe & Plow
p. 198 n. 20
Home of the Fish
p. 273 n. 14
p. 273 n. 14
Iddin-Dagan A
p. 198 n. 20
p. 190 n. 15
Lament for Uruk
p. 222 n. 2
Laws about Rented Oxen
Laws of Lipit-I
Laws of Ur-Namma
Nanna-Suen’s Journey to Nippur
p. 211 n. 25
153 i 6–ii 4 386
161 i 1–v 1 235
Ninurta & the Turtle
Römer 2001
197: B31–34 185
Royal Correspondence of Ur
10: 18 [Susa] 673
19 ii 1–iv 3 151
19 v 1–vii 5 152
25–32 (Akk.) 680
55–57 (Akk.) 574
12–17 (Akk.) 141
Sargon & Ur-Zababa
Sheep & Grain
p. 211 n. 24
141 i 15–18 36
Temple Hymns
TuM 1/2
Tummal Inscription
659, p. 222 n. 2
WO 8 (1976)
160: 19–22 615
Year Names
Amar-Sin 2 564, pp. 10–11
Apil-Sin 1 567
Apil-Sin 2 566
Sabium 1 567
Sabium 10 566
Sin-muballit 1 567
Sin-muballit 7 566
Sumulael 13 567
Sumulael (unidenti
ed) 566
p. 293
The following index of pre
x and verb combinations refers to the corresponding page
a—ru ‘dedicate’
‘wash, bathe’
‘order, instruct’
‘raise the arm’
—sud ‘swing the arms, run’
ad—gar ‘twitter, resound’
ak ‘do, act, make’
‘desire, crave’
—bala ‘curse’
ba-al ‘excavate, grub, recover’
BAD ‘be anxious, worried’
bala ‘cross, transgress’
bu(r), bu
‘release, loosen’
‘seize, hold’
205, 206, 214, 215
246, 247
dah ‘add, increase’
dar ‘break, split’
‘bring, take, carry’
128, 146, 148, 156,
157, 158, 159, 231,
232, 310
181, 206, 231, 232,
233, 234, 267, 310
‘go, pass’
221 n. 1
‘fashion, create’
oat, drift’
dirig ‘be excessive’
‘build, erect, work’
gaz ‘kill, crush’
—ak ‘pay attention’
rm’, ‘
x, establish’
ha-lam ‘destroy’
haš ‘break’
hul ‘destroy’
hul—gig ‘hate’
-la-gim—gar ‘be happy, joyful’
194, 195, 299, 300
hun ‘hire’
igi—bar ‘look at’
igi—gar ‘look at’
igi—lib ‘be awake’
‘raise, carry’
isiš—gar ‘wail, moan’
‘grow, sprout’
mud ‘be scared, anxious’
mul ‘glitter, shine’
‘cease, stop’
‘separate, remove’
naga—su-ub ‘rub with soap, wash,
—kud ‘swear an oath’
—dub ‘relax’
—te(g) ‘fear, be afraid’, ‘tremble’
—te(n) ‘cool off, relax’
ll, load’
si(g) ‘place, cover, wear’
—ra ‘sound the horn’
si—mul ‘make the horns radiate, shine’
‘make straight, put in order’,
‘perform properly’
(g) ‘hurl, cast down’
‘be green, verdant, beautiful’
‘sink, submerge’
su-ub ‘rub, sweep (away)’
‘have goose bumps’, ‘be
sum ‘give’
sumun ‘be old, dilapidated’
‘be humble’
sur ‘press, plant’
‘be worried’
—dar ‘be heartbroken’
—hun ‘be calm’, ‘soothe, relent’
icted, distressed’
‘hold, seize, capture’
‘make straight, put in order’,
šu—tag ‘seize, take hold’
‘decorate, adorn’
‘send, deliver’
šu—ti ‘receive, take’
šub ‘fall, drop, throw’
atten, collapse’
‘abandon, leave behind’
te(g) ‘approach’
ee, escape’, ‘disappear’
zal ‘pass,
ow’ (see also ud/mu—zal)
zalag, zalag
‘shine, brighten’
zi—ir ‘be troubled, distressed’
zi-ir ‘remove, erase’
‘take refuge’
zu ‘know’, ‘learn’
—ra-ah ‘chew, munch, nibble’

Приложенные файлы

  • pdf 15819581
    Размер файла: 3 MB Загрузок: 0

Добавить комментарий