Lingua Aegyptia

Journal of Egyptian Language Studies

25
2017

Widmaier Verlag ∙ Hamburg 2018

 LINGUA AEGYPTIA – Journal of Egyptian Language Studies (LingAeg)
founded by Friedrich Junge, Frank Kammerzell & Antonio Loprieno

Editors
Heike Behlmer Frank Kammerzell Antonio Loprieno Gerald Moers
(Göttingen) (Berlin) (Basel) (Wien)

M a n a g i n g E d i t o r R e v i e w E d i t o r s
Kai Widmaier Eliese-Sophia Lincke Daniel A. Werning
(Hamburg) (Berlin) (Berlin)

i n c o ll a b o r at i o n w i t h
Tilmann Kunze
(Berlin)

A d v i s o ry B o a r d
James P. Allen, Providence Elsa Oréal, Paris Wolfgang Schenkel, Tübingen
Joris F. Borghouts, Leiden Richard B. Parkinson, Oxford Thomas Schneider, Vancouver
Christopher J. Eyre, Liverpool Stéphane Polis, Liège Ariel Shisha-Halevy, Jerusalem
Eitan Grossman, Jerusalem Sebastian Richter, Berlin Deborah Sweeney, Tel Aviv
Roman Gundacker, Wien Kim Ryholt, Copenhagen Pascal Vernus, Paris
Janet H. Johnson, Chicago Helmut Satzinger, Wien Daniel Werning, Berlin
Matthias Müller, Basel Jean Winand, Liège

Lingua Aegyptia (recommended abbreviation: LingAeg) publishes articles and book reviews on all aspects
of Egyptian and Coptic language and literature in the narrower sense:
(a) grammar, in­cluding graphemics, phonology, morphology, syntax, semantics, pragmatics, lexico­graphy; (b)
Egyptian language history, including norms, diachrony, dialectology, typology; (c) com­para­tive linguistics, in-
cluding Afroasiatic contacts, loanwords; (d) theory and history of Egyptian liter­ature and literary discourse; (e)
history of Egyptological linguistics. We also wel­come contri­butions on other aspects of Egyptology and neigh-
bouring disciplines, in so far as they relate to the journal’s scope.
Short articles on grammar and lexicon will be published in the section “Miscella­nies”. Authors of articles or
reviews will receive electronic off-prints. Periodically, we would also like to put the journal at the colleagues’
disposal for a forum in which an important or neglected topic of Egyptian linguistics is treated at some length:
in this case, a scholar who is active in this particular area will be invited to write a conceptual paper, and others
will be asked to comment on it.
Authors should submit papers electronically to the managing editor ([email protected]). Please send
contributions in both doc/docx and pdf format. Further information (incl. guidelines and a template) is available
from www.widmaier-verlag.de. The decision whether to publish a manu­script is taken by the editors in agree-
ment with the advisory board.

Addresses
Departement Altertumswissenschaften: Ägyptologie, Universität Basel
Petersgraben 51, 4051 Basel, Switzerland
Institut für Archäologie: LB Archäologie und Kulturgeschichte Nordostafrikas, Humboldt-Universität zu Berlin
Unter den Linden 6, 10099 Berlin, Germany
Seminar für Ägyptologie und Koptologie, Georg-August-Universität Göttingen
Kulturwissenschaftliches Zentrum, Heinrich-Düker-Weg 14, 37073 Göttingen, Germany
Institut für Ägyptologie, Universität Wien
Franz-Klein-Gasse 1, 1190 Wien, Austria
The annual subscription rates are 49 € for individual and 69 € for institutional subscribers while single issues are
available for 99 € (incl. German VAT, excl. shipping). Orders should be sent to the publisher:
Widmaier Verlag, Kai Widmaier, Witthof 23F, 22305 Hamburg, Germany ([email protected]).

www.widmaier-verlag.de
ISSN 0942-5659 | ISBN 978-3-943955-65-1
 Proceedings of the Fifth International
Conference on Egyptian-Coptic Linguistics
(Crossroads V)
Berlin, February 17–20, 2016

edited by
Daniel A. Werning

Widmaier Verlag ∙ Hamburg

2018
 Titelaufnahme:
Daniel A. Werning (ed.)
Proceedings of the Fifth International Conference on Egyptian-Coptic Linguistics
(Crossroads V)
Berlin, February 17–20, 2016
Hamburg: Widmaier Verlag, 2018
(Lingua Aegyptia – Journal of Egyptian Language Studies; 25)
ISSN 0942-5659
ISBN 978-3-943955-65-1

© Widmaier Verlag, Kai Widmaier, Hamburg 2018

Das Werk, einschließlich aller seiner Teile, ist urheberrechtlich geschützt.
Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung
des Verlages unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen,
Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen.
Gedruckt auf säurefreiem, archivierfähigem Papier.
Druck und Verarbeitung: Hubert & Co., Göttingen
Printed in Germany

www.widmaier-verlag.de
 Contents

PREFACE......................................................................................................... vii–x

Articles

Marc Brose
Das ägyptische Verb und der ägyptisch-semitische Sprachvergleich.......... 1–40

Gaëlle Chantrain & Camilla Di Biase-Dyson

Making a Case for Multidimensionality in Ramesside Figurative
Language...................................................................................................... 41–66

Roberto A. Díaz Hernández

Von wn.|n und wn zu ⲛⲉ. Ein diachroner Fall der
Hintergrundinformation............................................................................... 67–81

Barbara Egedi
Two Kinds of Definiteness in Coptic........................................................... 83–99

Roman Gundacker
Where to Place ‘Ältere Komposita’? Traces of Dialectal Diversity
Among Early Toponyms and Theonyms...................................................... 101–176

Elsa Oréal
Nominalizations as a Source for Verbal Morphology. Grammaticalization
Paths of Modality and Information Structure in Earlier Egyptian............... 177–209

Helmut Satzinger
Second Tenses in Egyptian-Coptic and Some Other African Languages.... 211–229

Wolfgang Schenkel
|n-/= „sagt“ < |(.|) |n- „sagt(e), nämlich“................................................... 231–279

Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

The Question of Sumerian “Determinatives”. Inventory, Classifier
Analysis, and Comparison to Egyptian Classifiers from the Linguistic
Perspective of Noun Classification.............................................................. 281–344

Nathalie Sojic
Identifying Late Egyptian Virtual Relative Clauses..................................... 345–372
vi Contents

Elisabeth Steinbach-Eicke
Experiencing is Tasting. Perception Metaphors of Taste in Ancient
Egyptian....................................................................................................... 373–390

Sami Uljas
Where to Stick an Adverbial in Earlier Egyptian......................................... 391–414

ADDRESSES OF THE AUTHORS............................................................... 415–416

ADVERTISEMENTS
LingAeg Studia Monographica: New Publication
LingAeg 25 (2017), 281–344

The Question of Sumerian “Determinatives”

Inventory, Classifier Analysis, and Comparison to Egyptian Classifiers
from the Linguistic Perspective of Noun Classification

Gebhard J. Selz, Colette Grinevald & Orly Goldwasser1

Abstract
The two most ancient writing systems, Sumerian cuneiform and Egyptian hieroglyphs share
one feature: the use of so-called “determinatives” thought to be of purely graphic nature and
unpronounced. After considering the state of the art discussion of these cuneiform determina-
tives, the first contribution of this paper is to present a consolidated list of the alleged cuneiform
determinatives, including a short discussion of the various entries, related to their semantics,
estimated origin, frequency and chronological distribution. The second import of this paper is
to further demonstrate that the Sumerian determinatives constitute a “noun classifier” system
strikingly similar to better-studied classifier systems. This demonstration starts by establishing
the particular categorization domains and functions of the Sumerian system, in order to then
compare it with two classifier systems: one a noun classifier system in a contemporary Mayan
language (Jakaltek), the other that of the Ancient Egyptian script, a much more complex system
in its inventory, use and function but including a similar classification function. It is suggested
in conclusion that a future path of research should discuss in detail how the Sumerian classi-
fier system emerged and in what ways it forms the basis for the later evolution of classifiers
in the cuneiform world, taking up both the issue of noun formation and noun classification in
that script, to eventually establish Sumerian as the earliest attested language with true noun
classification.

Keywords
cuneiform determinatives; graphemic classifiers; lists of cuneiform classifiers; early writing
systems; Sumerian classifiers; Egyptian classifiers, Jakaltek noun classifiers; domains of noun
classification.

1 Gebhard J. Selz, University of Vienna (gebhard.selz[at]univie.ac.at); Colette Grinevald, University

of Lyon (Colette.Grinevald[at]univ-lyon2.fr); Orly Goldwasser, Hebrew University, Jerusalem and
Göttingen University (orly.goldwasser[at]mail.huji.ac.il). The co-authors met through the framework
of the European COST A31 project: Stability and adaptation of classification systems in a cross-
cultural perspective, directed by Thekla Wiebusch in 2005–2010.
We are grateful to Halely Harel for preparing the manuscript for publication.
282 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

0 Introduction
Cuneiforms and hieroglyphs are the two earliest writing systems known, both born before
the beginning of the 3rd millennium BCE. Despite its origins as a pictorial system, the
Sumerian script, developed along the shores of the Euphrates, evolved rapidly into the
abstract script known today as “cuneiform”, while the hieroglyphic script, on the banks
of the Nile in Egypt, kept its highly pictorial nature. These two complex script systems,
cuneiform and hieroglyphic, both comprise hundreds of signs, but while they differ from
each other in many respects, they do have in common a unique written phenomenon which
has been traditionally called “determinatives”. These determinatives fulfill one of the three
different semiotic functions that the signs may assume in both systems, the others being
logograms and phonograms. Determinatives have always posed a challenge for analysis
because, while they are almost always present, they are allegedly unpronounced.2
Both Assyriologists and Egyptologists have traditionally found these determinatives
to be of little interest, since they consider them to be an extra-linguistic phenomenon
that merely provides paralinguistic or metalinguistic information. The overall aim of this
paper, however, is to show how momentous and elaborate this neglected phenomenon of
determinatives is and how it can be demonstrated to resemble systems of categorization
known in linguistic typology as “classifier systems”. As is the case with all other classi-
fier systems of the world, a semantic analysis of the determinatives should also provide
modern scholarship with another source of information about knowledge organization in
the different cultures that were using the cuneiform script.3
The paper reveals the importance of the application of state-of-the-art methods from
another academic discipline, in this case contemporary typological and cognitive linguis-
tics, to the traditional fields of cuneiform studies. A similar analysis of determinatives
as classifiers has recently been applied in the field of hieroglyphic studies, most notably
by Orly Goldwasser, Frank Kammerzell and Eliese-Sophia Lincke.4 A trans-disciplin-
ary research model has been developing over the last decade by two of the co-authors,
­Goldwasser and Grinevald, resulting in a pilot study of grammatical and cultural aspects
of the classification system of the Egyptian hieroglyphic script (Goldwasser & Grinevald
2012). This model is being applied here for the first time to the analysis of the cuneiform
script, with the new collaboration of the Sumerologist Gebhard Selz. Among other things,
this innovative interdisciplinary approach aims at building a new bridge between the fields
of Assyriology and Egyptology.

2 See below 6.1 for a preliminary discussion whether Sumerian determinatives were ever pronounced
or not, however all scholars agree that Egyptian determinatives are unpronounced.
3 For classifiers system as mirror of knowledge systems in various cultures; see, inter alia, Craig
(1986b), Denny (1976), Lakoff (1986), Senft (2000), Kilarski (2014).
4 For the theoretical framework, see Goldwasser (1999, 2002, 2005, 2006a), Lincke (2011); Lincke
& Kammerzell (2012); Kammerzell (2015); Lincke (2015a). Werning (2011) presents the only
up to date corpus analysis of classifiers, with many innovative discussions and insights as well as
statistics.
 The Question of Sumerian “Determinatives” 283

This collaborative research between a Sumerologist, an Egyptologist, and a linguist

– specialist in classifier systems – focuses on establishing that, from a linguistic typo-
logical perspective, the Sumerian determinatives constitute a classifier system. This article
therefore presents for the first time a description of the Sumerian “determinatives” as clas-
sifiers, and does so by showing how they are comparable in many ways to other known
classifier systems of the world. Two parallels are drawn to support this new analysis of
cuneiform classifiers. One is a parallel with a well-studied classifier system of a contem-
porary Mayan language of Mesoamerica living as an oral tradition language, and the other
with the classifier system of the Egyptian hieroglyphic script, for which a parallel demon-
stration has already been established.
The article will start (section 1) with a state-of-the-art description of determinatives
in cuneiform script. In a quick review, it will recall that the existence of determinatives is
widely acknowledged, and point to the fact that, to this date, no attempt has been made to
place them within a wider typological linguistic context of parallel phenomena in other
languages or scripts. It will then proceed (section 2) to establish a consolidated inventory
of cuneiform determinatives, preceded by a presentation of the secondary sources used
and the organization of information contained in the entries, then followed by an initial
analysis of this list.
The next two sections, will turn to demonstrate how Sumerian determinatives show
characteristics of classifier systems. The thematic organization of the list (section 3) will
reveal how far the semantic domains covered by this system correspond to those of known
classifier systems, in an expected mix of universal categories and categories specific to
the local culture. The demonstration of how the Sumerian determinatives can be put in
perspective with other classifier systems will first consider (section 4) the classifier system
of Jakaltek, a contemporary Mayan language of Mesoamerica, and then the one of the
Ancient Egyptian hieroglyphic script system (section 5) more recently described as such.
The concluding section (section 6) will consider some of the implications of this study and
point to possible future domains of research. Finally (section 7), it is underlined that ap-
plying the proposed term “classifier” identifies the Sumerian determinatives as indicative
of a linguistic system.

1 State of the art of determinatives in cuneiform script

In cuneiform studies, little attention has been given so far to the determinatives in the
cuneiform script. In this section, we intend to summarize some prevailing opinions which
present the current state of the art research on these “determinatives”.

1.1 Determinatives in recent grammars of cuneiform languages

None of the recent grammatical studies – nor the respective sign lists – provide new in-
sights in the phenomenon of “determinatives”. Two prominent scholars supply the follow-
ing description, with the following labels and definitions.
284 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

Edzard (2003: 8–10) declares that determinatives are one class of “cuneograms” –
alongside (1) logograms, (2) syllabograms, (3) “phonetic indicators” (a sub-class of [2]),
and (4) “signs for numbers or the combined notation of measuring”. He then specifies that
“they are signs which precede or follow words or names in order to specify them as be-
longing to semantic groups”, adding that “determinatives can be proven not to have been
pronounced (although doubt may exist in specific instances)”; cp. fn. 36.
Jagersma (2010: 16, 18) labels determinatives as “auxiliary logograms” and estab-
lishes their function as one that identifies the following or preceding word as belonging
to a specific semantic class. He further describes determinatives as “in origin word signs
stripped of a pronunciation”. Jagersma’s remark points to a process which gave rise to the
cuneiform determinatives which will be considered below.
So far, neither the different origins nor the various functions of these determinatives
have been researched, and in general, no further significance is attributed to them. As a
starting point, we will first provide an overview of the phenomenon known as “Sumerian
determinatives” and discuss some of their distinctive features.

1.2 No single list of cuneiform determinatives

Whereas in Egyptology Alan Gardiner’s list of 1957 of “generic determinatives”5 (which
is by no means comprehensive or conclusive) serves as a basic reference point for the
identification of such determinatives, there is no single list for the cuneiform world, al-
though, as will be seen, a number of partial and differing lists do exist.
The reason for that rather surprising state-of-the-art may be that the development
and evolution of such classifiers in Mesopotamia (and adjacent regions) occurred in a
­multi-lingual environment, right from the beginning. Already at the time when writing
was invented, in the late 4th millennium Mesopotamia, there coexisted several linguisti-
cally diverse groups, although their various contributions to the evolution of the writing
system is presently still obscure. Therefore, one has at least to consider the possibility that
this linguistic environment influenced the development of the concept of determinatives,
even if one contends that they are exclusively a written phenomenon. It can be further
demonstrated that the replacement of the previously dominant Sumerian by Akkadian (and
other Semitic languages) also affected the implementation of the classifiers in the script.
Most important here is the fact, that in the cuneiform writing system the so-called Sumero-
grams played a salient role, right down into the 1st millennium BCE. Thus, the consulted
authors’ differing judgment of the origin, evolution and adaptation of the cuneiform deter-
minatives partially explains the variations in the extant lists.

5 The list that was compiled for teaching purposes (Gardiner 1957: 31–33) refers mainly to Middle-
Egyptian with some later additions. It suffers from many shortcomings. The first and foremost is
that the most common classifiers, such as A1 or B1 get the same status on the list as other,
rare, or very rare classifiers. Some of the classifiers are hosted by many words, while others forge
very small categories that harbor 3 or 4 words only. All-encompassing determinatives’ lists in the
grammars published since then are based on this list.
 The Question of Sumerian “Determinatives” 285

Looking at the graphemic features of the cuneiform classifiers, the early loss of iconicity
of the ideograms, in contrast to the more stable and detailed iconicity of hieroglyphic
script, might have furthermore influenced the system, imposing highly standardized sign
forms on the cuneiform determinatives.

1.3 A diversity of common determinative lists used in cuneiform studies

There is no complete and generally accepted inventory of this script phenomenon. In
­cuneiform, especially for Sumerian, the most commonly used lists – selected here rather
arbitrarily – are quite diverse, depending on the author’s individual notion of the phe-
nomenon and the specific purpose of teaching the script (and language) of a given pe-
riod. It may well be that the newly compiled list presented here will need further elabora-
tion, ­especially for what concerns the proposed frequency and chronology of attestations.
While existing data-bases may help, such statistics are presently difficult, if not outright
impossible, to obtain.
The number of determinatives varies according to the aims of the various studies but
remains in the range of few dozens, from 15 to 46. Borger (1978) has 46, Labat (19886) 36,
Ellermeier (1979) 31, Huehnergard (1996) 25, Falkenstein (1964 and 1949) 20. The lists
are reduced to 22 and 14, in Foxvog (2016) and Edzard (2003) respectively, both authors
exclude those items that are allegedly attested only in later stages of the system. Borger,
Labat and Ellermeier are the standard cuneiform sign lists. Whereas Borger and Labat
cover both Sumerian and Akkadian texts, Ellermeier concentrates on Sumerian.
Cuneiform determinatives are attested in all sorts of texts – historical, literary and
administrative – but also, especially in the late periods, in all sorts of scholarly or “scien-
tific” texts. Indeed, the latter texts make heavy use of Sumerian logograms, and there are
indications that the Sumerian of these texts was indeed a pronounced technical language.
The determinatives clearly appear at the time when writing was invented (late 4th mil-
lennium) and there can be little doubt that this took place in close contact with Sumerian,
probably being the dominant language at the time.
It is well-known that the concept of cuneiform determinatives spread also to the fringes
of cuneiform writing in the west and in the east and even entered the writing of an entirely
different language, the Indo-European Hittite.6 To deal with these issues is beyond the
scope of this paper; however, we contend that even the newly introduced determinatives
of the 2nd and 1st millennium reflect the Sumerian system – as described below – even when
used in Akkadian contexts. As it seems and as will be elaborated in the following, the type
of noun classification that the determinatives reflect is a salient feature of the Sumerian
language, no matter whether these classifiers were silent or pronounced.

1.4 The problem of transliteration

To this day, the transliteration and identification of determinatives in cuneiform texts is
almost exclusively based on one or the other of the above mentioned standard sets. Their

6 Cohen (2010); for Luwian hieroglyphs, see Yakubovich (2008: 9–36) and Payne (2014).
286 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

number amounts, as we have seen above, generally to less than twenty and never exceeds
thirty to forty in any given period. However, there is no common agreement among
scholars on the corpus of determinatives and even the scholarly mode of transliteration is
not always consistent. Most often, what individual authors see as (silent) determinatives is
– in transcription – rendered in upper case: thus, the buru4 bird (“a crow or a vulture”), in
cuneiform (nu11.bur.mušen) is transliterated as buru4(=NU11.BUR)mušen
that is buru4 . Occasionally the identification of such (silent) classifiers is inconsistent.
bird

One finds, for instance, for both transliterations ǧešgušur and ǧešúr “beam”. The
Akkadian loan gušūru is almost certainly derived from ǧeš-úr meaning something like
“wooden base, basis” and demonstrates that in this case ǧeš- wood has the function of a
pronounced(!) classifier, i.e. ‘wood’.

1.5 Recognized characteristics of “determinatives”

Previous studies have already established a number of semantic and characteristic features
of the determinatives:

A Semantic classification
The existence of semantic classification through determinatives in cuneiform script is
widely accepted.

B Variation in position
Variation in position (pre- and post-noun) is acknowledged by all scholars, but little
researched. In the cuneiform system, most determinatives appear before the word, as
in “nest”, where GI reed is the determinative and the logograms
plants+earth+put convey the semantics of a nest, or in “tent”,
where the determinative KUŠ skin/leather is followed by AB shrine/
dwelling. However, in some cases, the determinative follows the word as is the case for
determinatives for place and fish or bird (see list below).
This is in sharp contrast to Egyptian, where such “silent” hieroglyphs are always found
at the end of the word, as in: sSy, “nest” logogram with house determinative,
and imw, “tent” – phonograms im-(m)w and same determinative house.7

C Variation in the choice of “determinatives”

Occasionally attested is the possible variation in the choice of a determinative, as
in the case of the alternation of the reed and wood classifiers. For instance, the noun
“cedar” can be written simply with the logogram EREN = “cedar tree”, but also
with either the pre-classifiers (pre-determinatives) for ǦEŠ tree/wood, or perhaps
ŠIM aromatics, occasionally also with both. = aromatics+tree/woodcedar.

7 See here the very rare exception in CT IV 269 in a single version, in pre-positon, coffin T2Be, b –
. We are grateful to Wolfgang Schenkel for this example.
 The Question of Sumerian “Determinatives” 287

This possibility of multiple classifiers is very rare or even doubtful in cuneiform, but is a
common feature of the Egyptian system (see below, 5).

2 A consolidated inventory of cuneiform “determinatives”

As there exists, so far, no in-depth study of cuneiform determinatives, the consolidated list
to be presented in this section may be useful for Sumerian studies in general. It was actually
conceived, at the same time, as a major step towards building the ensuing argumentation
that these “determinatives” are to be considered “classifiers”.

2.1 Information contained in the consolidated list of determinatives

The inventory to be proposed next in 2.2 offers a novel synthesis of much information of
different nature, gathered from different sources and analyzed from different perspectives.

A The sources
The inventory is based on a selection of works, sign lists and grammars, widely used in
teaching cuneiform. The selection includes the following: Borger (1978: 48); Falkenstein
(1964: 21 and 1949: 34–358); Labat 19886 (19481): 20–22, Ellermeier (1979: XXXIV–
XXXV); Huehnergard (1996: 537); Foxvog (2016), and Edzard (2003).
From the collection presented below it can be easily seen that the choice of which
cuneogram is considered as determinative depends first on an author’s focus on a particular
period, and also on the major language (Sumerian/Akkadian) considered. The lists also
reflect the teaching purpose for which they were established, quite evidently sign lists
(Borger, Labat, Ellermeier) differ from grammars (Falkenstein, Huehnergard, Foxvog, and
Edzard). The most extensive list is by Borger and, for convenience sake, the reference to
his sign numbers will be given first.9
In each entry of the proposed new list, the references to the works quoted will be
presented in the following order, with the following abbreviations:
Borger = Borger (1978: 48);
F = Falkenstein (1964: 21);
F-2 = Falkenstein (1949: 34–35);
L = Labat (19886 [19481]: 20–22);
E = Ellermeier (1979: XXXIV–XXXV);
H = Huehnergard (1996: 537);
Fo = Foxvog (2016);
Ed = Edzard (2003).

8 Falkenstein’s short list can be found in Falkenstein (1964: 20–21); I also include references to
Falkenstein (1949: 34–35 [F-2]); there Falkenstein collects the instances in the inscriptions of Gudea
where writing with and without determinatives (“eine etwas grössere Freizügigkeit”) are attested.
9 Numbers in brackets ([…]) are provided by the authors and follow the sequence of enumeration in
a given source.
288 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

The number following these authors’ name abbreviations (except for Borger, taken as the
initial source) will correspond to the position of an item in the respective list of origins,
thus L [3] refers to item 3 in Labat’s list, or H [8] to item 8 in Huehnergard’s list.

B The form of the entry itself

Each entry of the list includes these features:
– A number, as established in the standard ordering according to the later sign forms
used e.g. by Borger (arranged according to their elements – horizontals, winkelhaken,
and verticals).
– A hyphen (–) in front or after the determinative to indicate its use as “pre–” or “–post”
classifiers, they also abbreviated as:
PR for Pre-position classifier
PO for Post-position classifier
– The (Old Babylonian) standardized written cuneiform with reference to earlier forms
(at least when they have a clear iconic referent),10 showing the graphic type of a sign.
– The standardized transcription indicating the supposed Sumerian word from which
the meaning is derived. These transcriptions refer to an alleged (standardized) reading
of a sign and its meaning(s); minor variations are attested.
– An occasional asterisk indicates a suggestion or correction. For instance, “*didli” is
the proposed transliteration and meaning of this sign, for which some scholars use the
(less informative) sign name “HAL” instead.
– The references to the source lists, starting with Borger’s list (and his numbering) and
mentioning others by an initial. The serial number provided there refers simply to
the sequence in which the determinatives are listed (the authors did not provide such
numbering).

C Additional information contained in the entries

Following the informative presentation of the entry itself just outlined, each entry provides
the following additional information about each determinative:
a (Literal) Meaning and its classifying Use
b Position and Lexical Origin, occasionally with a brief discussion

10 The archaic sign forms (from Uruk) are based on Green & Nissen (1987), see also Falkenstein
(1936); most of the forms represented here are bsed on digitally adapted (and more distictive)
forms by the Cuneiform Digital Library Initiative (CDLI) at http://cdli.ucla.edu/tools/SignLists/
protocuneiform/archsigns.html (accessed 09-07-2017).
 The Question of Sumerian “Determinatives” 289

c Earliest (estimated)11 Period in which it appears and its estimated Frequency of use12,
categorized as follows:

Period Frequency
AR = Archaic (before 2800 BCE) FR = Frequent
EA = Early (before 2000, mid or late 3rd mill. RF = Relatively Frequent
CL = Classical R = Rare
PC = Post-Classical (after 1750 BCE) D = Doubtful
LATE = 2nd or 1st millennium

D Sample entries showing all the information provided

Two entries have been chosen to illustrate the diversity of the information provided in
this list. The first line shows the entry itself, and is followed by the additional information
from the various fields just mentioned (and fully spelled out here for the demonstration
but simplified a-c in the list):
19. ǧiš–/ǧeš– (Borger 296; F [4]; L [11]; E [7]; H [5]; Fo [7]; Ed [3]) .
(Meaning; Use) a. “wood, lumber, tree; wooden objects”.
classifies tree-names and wooden things.
(Position; Origin) b. PR; originally (in Sumerian) part of compound lexemes.
(Period; Frequency) c. AR/EA; frequent.
50. –ku6, (Borger 597; F [17]; L [20]; E 17]; Fo [19]; Ed [6]) .
(Meaning; Use) a. “fish”; also: “amphibians, crustaceans”.
classifies all kinds of fish and extended to aquatic animals.
(Position; Origin) b. PO; originally part of compound lexemes.
(Period; Frequency) c. (AR?)/EA (from mid-3rd mill.); frequent.

2.2 The revised consolidated list of cuneiform determinatives

Below is the inventory of determinatives in its present stage of development, organized to
provide the different types of information mentioned in the previous section, as much as
it has been collected so far. It contains in total 50 entries, some entries offering more ex-
tensive information than others, and some being still tentative analyses. That list of deter-
minatives corresponds to the items that will be analysed later as “classifiers” in section 3.

11 This may occasionally be difficult, especially because the change from an alleged lexeme status
of a classifier in noun compounding to a purely graphemic classifier is rarely traceable, cp. below
fn. 35–36 and also Krebernik (2013: 188).
12 The preliminary status of these remarks is obvious, especially because precise statistics are pres-
ently not available.
290 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

1. –*didli (HAL) (Borger 2; F [0]; L [27]; E [0]; H [0]; Fo [0]; Ed [0]).

a. lit. “one (and) one”: individual things; later marking pluralis paucitatis “a few” or
“several, various” (mensural).
b. PO; originally a Sumerian adjective meaning “single, countable”.
c. Cl/LATE; rare.

2. *kuš (SU)- (Borger 7; F [0]; L [16]; E [18]; H [13]; Fo [9]; Ed [14])

.
a. “skin, hide; leather”; classifies all sorts of leather products.
b. PR (and PO?); originally part of compound lexemes; count word and class noun?
c. AR(?)/EA (early 3rd mill.); relatively frequent.

3. *diǧir (AN)– (Borger 13; F [1]; L [4]; E [2]; H [2]; Fo [4]; Ed [1]).
a. “god (sky; heaven)”; iconically the depiction of a star, classifies (astral) deities.
b. PR; original lexeme status.13
c. EA (early 3rd mill.14); frequent.

4. uru/eri– (Borger 38; F [12]; F-2 [7]15; L [7]; E [32?16]; H [23]; Fo [15] Ed [0])
.
a. “city”, “town”; classifies names of cities.
b. PR; originally lexeme status.
c. EA (late 3rd mill.); relatively frequent.
comment: Sometimes used together with the post-classifier –ki “place” which re-
sults in a kind of double classification (eri-CITYNAME-ki = “city name place”),
e.g. uruǧír-suki designating the city Ǧirsu; at least in such cases uru might have
been pronounced, that is “the city Ǧirsu”. In the writing ki-CITYNAME-ki, e.g.
ki lagaški is to be interpreted as “the region Lagaš place”; here the initial ki was
certainly pronounced.

5. iti/itu(d)– ,17 (Borger 52; F [0]; L [5]; E [1418]; H [8]; Fo [0]; Ed [0]).
a. “month”; classifies month names; functions as a (unpronounced) classifier probably
due to Akkadian influence.
b. PR; originally part of month names in compounds.
c. EA/CL (from the late 3rd mill. onwards); relatively frequent.

13 An attempt to assemble arguments for an original pronunciation of the divinity classifier is made in
Selz (2016).
14 Cp. Selz (2008: 22) and fn. 33 and cp. also Selz (2016).
15 Falkenstein’s example (1949: 35 [F-2]) may rather attest speech variation, e.g. “city Ǧirsu” vs. “Ǧirsu”.
16 A question mark after E refers to Ellermeier (1979: XXXIVf.) and his remark: „sum. belegt?”
17 This sign form provides interesting evidence for early sign formation; the sign combines the logo-
gram for day UD and the number 30. 30 days was the (administrative) length of the month since the
Late Uruk period. In contrast to Egypt, Mesopotamia followed a calendar based on moon cycles.
The system of incorporatiing numbers to or into UD “day” was very elaborate already in the Uruk
periods; cp. Green & Nissen (1987: sign 569), and the more specific account in the CDLI list.
18 Ellermeier (1979: XXXIV) “vor Monatsnamen wohl nur akkadisch“.
 The Question of Sumerian “Determinatives” 291

6. –*mušen (HU) (Borger 78; F [18]; L [21]; E [23]; H [17]; Fo [20]; Ed [7]) .19
a. “bird”; also “insects” and generally winged animals” ; later graphemic determina-
20

tive for birds and insects.

b. PO; originally part of compound lexemes.
c. (AR[?])/EA (mid-3rd mill.); frequent.

7. gi– (Borger 85: F [3]; L [12]; E [6]; H [4]; Fo [6]; Ed [11]) .

a. “reed”; classifying reeds and reed products or objects.
b. PR; originally part of compound nouns; sometimes alternating with (19) ǧiš –.
c. (AR(?)/EA (after the mid-3rd mill.); relatively frequent.

8. gada – (Borger 90; F [0]; L [0]; [5?]; H [0]; Fo [0]; Ed [0]) .

a. “flax, linen”; only material; classifying linen garments (never used for the plant!) is
more specific than (40) the general garment classifier and contrasts (44), the wool clas-
sifier.
b. PR; originally part of compound lexemes.
c. EA (from the late 3rd mill. onwards); relatively frequent.

9. sa– (Borger 104; F [0]; L [19c21]; E [0]; H [0]; Fo [0]; Ed [0]) .

a. “bundle”, “net”; count noun and measure word (mensural); cp. (11) in the meaning
“piece”.
b. PR; originally part of compound lexemes, later marking braided or bundled items,
like reed, fish etc.
c. CL (late 3rd mill.); relatively frequent.

10. ganá/aša5–, also –iku22 (Borger 105; F [0] L [0]; E [1123]; H [7]; Fo [0]; Ed [0]) .
a. “field(-measure)”; classifying field names.
b. PR; PO if used as numeral for area measures (iku) (mensural); originally part of
compound lexemes.
c. Early/CL; relatively frequent.

19 Iconically most probably the depiction of a water-bird.

20 Cp. Foxvog (2016: 13).
21 According to Labat this determinative as well as síg (siki), má, sa, anše and udu) “peuvent être
considérées come faisant partie de l’idéogramme” and are therefore not considered by Labat as
“true” determinatives (Labat 19886: 21).
22 Following numbers, the sign specifies area measures; see Ellermeier (1979: XXXIV); Huehnergard
(1996: 537).
23 Following Ellermeier (1979: XXXIV) considered as determinative only “nach Flächenmassen”
with the reading iku; the origin of GANÁ remains somewhat unclear; cp. GANÁ A.ŠÀ probably
to be interpreted as GANÁa-šà FIELD (LOGOGRAM)field (reading).
292 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

11. saǧ– (Borger 115; F [0]; L 0; E [0]; H [0]; Fo [0]; Ed [0]) .

a. “head”; as graphic marker in front of slaves’ names, originally designating slave (as
counted items, also used as a count word: “piece”, mensural(?); cp. (9).
b. PR; originally lexeme status.
c. Early/Cl.; relatively frequent.

[11a. má– (Borger 0; F [0]; L [19b]; E [0]; H [0]; Fo [0]; Ed [0]); probably not a clas-
sifier; very doubtful; meaning: “ship”. ]
12. mul– (Borger 129a; F [0]; L [6]; E [2124]; H [16]; Fo [10]; Ed [0]) ; cp. further
(29) múl and (35) mul4 25.
a. “star ; planets ; constellations”; classifying names of stars/constellations.
b. PR ; originally part of compound lexemes.
c. CL/PC (early 2nd mill.); rare.

13. –urud(u/a)(–) (Borger 132; F [13]; L [16]; E [3326]; H [24]; Fo [16]; Ed [4])
.
a. “copper” used before (and after?) “mediocre” metal (copper, bronze) and things
made thereof; cp. (30) –zabar “bronze”.
b. PR and PO(?); originally part of compound lexemes?
c. EA/CL (late 3rd mill); relatively frequent.

14. uzu– (Borger 171; F [14]; L [18]; E [34]; H [26]; Fo [17]; Ed [0]).
a. “flesh”; by extension also “body; entrails; omen” classifies occasionally some body
parts like liver, stomach, heart, lung; also meat cuts (in recipes for soups!).27
b. PR; originally part of compound lexemes?
c. PC/LATE; rare.

15. anše– 28
(Borger 208; F [0]; L [19d]; E [1]; H [0]; Fo [0]; Ed [0]) .
a. “donkey; onager (wild donkey)”;  classifies equids (extended later to the newly
introduced horses, and still later also camels [which are not attested in Mesopotamia
before ca. 1500 BCE]).
b. PR; originally lexeme status.
c. EA/CL (from late 3rd mill.); relatively frequent.

24 Ellermeier (1979: XXXIV) “nicht eigentlich sum”.

25 UL = mul4; based probably not only on phonetic similarity; the early meaning of /ul/ “flower; bud”
may have motivated the use of this sign as a logographical classifier for “star(s)”.
26 The only determinative which seems attested in pre- and postposition: “vor und nach Gegenständen,
die ganz od. teilweise aus Kupfer of. Bronze gefertigt wurden” (see section 3.2.1). Ellermeier
(1979: XXXV).
27 According to Steinkeller (1982: 358–359) an Akkadian invention; cp. Foxvog (2016: 13).
28 There are number of animal head signs, although not easy to distinguish; basic research on these
signs was done by Mittermayer (2005).
 The Question of Sumerian “Determinatives” 293

16. kaš – (Borger 214; F [0]; L [0]; E [15]; H [0]; Fo [0]; Ed [0])
.29
a. “beer”; may classify different varieties of beer.
b. PR; part of compounds (as often, difficult to distinguish between part of a compound
vs. unspoken determinative).
c. EA (from late 3rd mill. onwards [?]); relatively frequent.

17. šim– (Borger 215; F [10]; L [0]; E [23]; H [0]; Fo [12]; Ed [0]) .
a. “aromatics; perfume”30; used before aromatic plants etc.
b. PR; originally lexeme status.
c. EA (late 3rd mill.); relatively frequent.

18. na4 – (/ ià4 /zá/) (Borger 229; F [9]; L [1331]; E [24]; H [18]; Fo [11]; Ed [12])32
(earlier form is different: NUNUZ = ZATU 423 33
).
a. “stone; stone objects; jewels”; used in front of stone names or items, graphically
differentiated from na “stone” . (This sign is also most often used to write the
syllable /na/); a writing like na4na stone34, is a case of a “repeater”; classifies
stones and stone objects, also jewels.
b. PR; originally lexeme status35.
c. AR(?)/EA (mid 3rd mill); relatively frequent.

19. ǧiš– / ǧeš– (Borger 296; F [4]; L [11]3; E [7]; H [5]; Fo [7]; Ed [3]) .
a. “wood, lumber, tree; wooden objects”; classifies tree names and wooden things.
b. PR; originally part of compound lexemes.36
c. AR/ EA; frequent.

29 The major difference in the archaic forms are depictions jars with and without a spout.
30 Cp. also interesting writings of /šimgig/ = Akk. kanaktu “an aromatic tree” as: šim-gig, ĝeššim-gig;
ĝeš
ŠIM and further ŠIM-ga “a plant” wr. úŠIM-ga.
31 Note Labat’s rendering of the sign as zá. Pronunciation and possible differentiation of readings
remain unclear.
32 Cp. further the interesting writings for “clay; pebble”: a) im-na4na CLAY+stoneSTONE
and b) 4im-na
na stone
CLAY+STONE.
33 Depiction of a string of beads.
34 This example looks like a clear case of a “repeater” classifier as it was already defined by Allan
(1977: 295), may be compared to the Egyptian “repeaters”, see below 5.6.2 (also 4.1.2). Ellermeier
(1979: XXXV); Green & Nissen (1987: 261 [ZATU 423]) speculated about an original consonantal
cluster (or morpheme [?]) /nza/ or /nza/of the conventionally differentiated reading /na/ “stone” and
/za/ “precious stone, jewel”; similarly, Civil (2008: 51–52), also mentioning the reading ia4.
35 Edzard’s (2003: 9) example for na4-nunuz = erimmatu “egg-stone, a bead” demonstrates the con-
textual dependency for the question of “reading” the determinatives; In a given context the pronun-
ciation /nunuz/ “egg” may have clearly referred to this type of jewelry, in another not; cp. “cherry”,
German “Kirsche” (fruit), also used for Kirschholz (cherry wood) and Kirschbaum (cherry tree).
36 Edzard (2003: 9) states: “Akkadian loanwords clearly show that ĜEŠ cannot be part of the word”;
he then, nevertheless, continues “note in contrast ǧeš-ùr with the Akkadian loanword gušūru
showing that ǧeš is part of the word”.
294 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

20. gu4– / gud– (Borger 297; F [0]; L [0]; E [8]; H [0]; Fo [0]; Ed [0]) .
a. “bull; oxen”; classifies all sorts of bovine animals.
b. PR; originally lexeme status.
c. late 3rd mill.; relatively frequent.

21. –ta-àm (Borger 139; F [0]; L [30]; E. [0]; H [0]; Fo [0]; Ed [0]).
a. “each”; marking distributive numbers.
b. PO; originates as grammatical (Sumerian) morpheme.
c. CL/PC; rare.

22. dug– also written (BI = dugx) (Borger 309; F [2]; L [17]; E
[3]; H [3]; Fo [5]; Ed [5]).
a. “jar, jug, pitcher; vessels”; count noun, better measure word (mensural); classifies
all types and names of vessels (for liquids).
b. PR; originally part of compound lexemes?37
c. AR(?)/EA from early 3rd mill.; relatively frequent.

23. ú– (Borger 318; F [11]; L [10]; E [30]; H [21]; Fo [14]; Ed [10]) .

a. “plants; grass; herbs, by extension also food and fodder; bread, loaf”; classifies all
sorts and names of plants.
b. PR; originally part of compound lexemes?
c. (AR(?)/EA (from early 3rd mill.); relatively frequent.

24. é– (Borger 324; F [0]; L [0]; E [4?]; H [0]; Fo [0]; Ed [0]) , and cp. the sign
ĜÁ 38
.
a. “house” > “container ” ; may classify names of buildings.
b. PR; originally lexeme status “house” (original depicting the (front of a) house, later
extended designating “container”).39
c. PC (2nd mill. onwards); rare/doubtful.

37 Well attest as “frame sign” in the archaic Uruk texts (ZATU 88–124), where it seems to func-
tion – as I suggest to term it – as a “quasi-determinative” designating “containers for liquids” =
­liquid; see also Green & Nissen (1987: 189). A similar frame sign is GÁ/PISAN (ZATU 162–183)
referring to other containers box; cp. also EZEM (ZATU 150–157), MAH (?) (ZATU 340–351)
and several other modified simple signs in the ZATU list. Compare here a similar case for liquid
(mensural?) classifier in the Egyptian system, see below, Table 6.
38 First group represents probably the front of a house (with sopraporte). The second group is the sign
ĜÁ (ǧá is also Emesal - the alleged women’s language - for house!).
39 The opposite evolution can be observed with sign ĜÁ “container” which can later also designate a
built structure. The sign forms, however, are sharply distinguihed; both archaic froms are provided
here for comparions only.
 The Question of Sumerian “Determinatives” 295

25. lú– (Borger 330; F [7]; L [2]; E [1940]; H [14]; F [7]; F-2 [5]41; Fo [2]; Ed [8])
a. “human”; “man”; classifies all sorts of professions and peoples (personal names,
inhabitant of cities and countries)42.
b. PR; originally lexeme status (also count noun).
c. EA (late 3rd mill.); relatively frequent, especially in PC period.

26. –sar or –nisig (Borger 331e; F [19]; L [23]; E [2543]; H [0]; Fo [21]; Ed [9])
.
a. “garden plot” (sar) or better “vegetables” (nisig); used to mark names of herbs,
vegetables (lettuce, cucumbers, onions). Note the differing classification of plants;
cp. (23) ú–, (7) gi–, (19) ǧiš– / ǧeš–, and (17) šim– (by nature or smell).
b. PO; originally part of compound lexemes.
c. EA; relatively frequent.

27. kur– (Borger 366; F [0]; L [8]; [E [0]; H [12]; Fo [0]; Ed [0])
.
a. “mountain (region); foreign land”; designates countries and mountain names.
b. PR; originates very early as lexeme in compounds.
c. EA/Cl (likely to be reduced to pure graphic status only after early 2nd mill.);
relatively frequent.

28. še– (Borger 367; F [0]; L [0]; E [27]; H [0]; Fo [0]; Ed [15a]) .
a. “barley”; extends to a graphic marker for all sorts of cereals (barley, wheat, emmer …).
b. PR; originally lexeme status.
c. CL/PC; relatively frequent.

29. múl (TE)– (Borger 376; F [0]; L [0]; E [0]; H [0]; Fo [0]; Ed [0]).
a. “star” used like (12) mul, related to (35), perhaps based on sign similarity to the
ancient form of TE ; classifies names of constellation of stars in astronomical
texts.
b. PR; origin uncertain.
c. PC/Late, problematic; rare.

40 Ellermeier (1979: XXXIV) comments: “vor Bezeichnungen für Angehörige bestimmter Personen-
gruppen wie Berufen etc.”; cp. Huehnergard (1996: 537).
41 Because of various spellings (without lú) in other inscriptions of Gudea, Falkenstein (1949: 35
[F-2]) argues for lú as unpronounced determinative; it seems, however, that this is rather a case of
speech variation, e.g. “the impure” vs. “the impure man”.
42 It seems likely that we can observe here how lú developed into a silent classifier. It may have
started with simple genitive compounds like lú-alan “man (responsible for) the statue”. Whether
compound nouns like lú-ú-bíl is “man (of the) firewood” (genitive compound) or “man burning
fire material” remains uncertain. However, as it seems that in expressions like “scribe” < “(the one)
writing a tablet” (dub-sar) – an exocentric compound - it was felt that the implicit classification
“human” should be made explicit, at least in writing, e.g. lúdub-sar “scribe”.
43 Ellermeier (1979) suggests the reading /nisig/ or /nissag/ for SAR meaning “greenery or the like”;
in this case /nisig/ designates a superordinate class, like fish, bird, and place.
296 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

30. –zabar ,44 older form: 45

(Borger [0]46; F [0]; L [0]; E [0]; H [0]; Fo [22]; Ed
[0]).
a. “bronze”; bronze objects; classifies bronze (metal) objects.
b. PO, often combined with PR urudu- “copper”; lexeme status.
c. EA; rare.

31. –hi-a (Borger 396+404; F [0]; L [26]; E [0]; H [0]; Fo [0]; Ed [0]); also, incorrectly
transliterated as –há47.
a. lit.: “various/mixed” (countable) items; originally a Sumerian adjective meaning
“different (classes of) items”.
b. PO; later marking pluralis paucitatis “a few”; then also general plurals.
c. Cl/Late; rare.

32. IM/tu15– (Borger 399a; F [0]; L [0]; E [12]; H [0]; Fo []0]; Ed [0]) 48
.
a. “wind” cp. no. (33); classifies graphically the names of winds, the reading of this
sign is debated.
b. PR; originally part of compound lexemes.
c. EA/CL (late 3rd mill.); relatively frequent.

33. im– (Borger 399b; F49; L [0]; E [13]; H [0]; Fo [0]; Ed [0]) .
a. “clay”; classifies things made of clay.
b. PR; clay originates as a rebus writing for im– “rain”, cp. (32)50.
c. EA/Cl; rare.

34. –kam, -kám , (Borger 406; F [0]; L [28+29]; E [0]; H [9]; Fo [0]; Ed [0]).
a. “of N (number) is it”; marks ordinal numbers; later writing also –kàm (Borger
143, L [28]); designates ordinalia.
b. PO; originates as a (Sumerian) grammatical morpheme: /(a)k/ + 3.ps. sg. of encl.
copula /am/.
c. CL/PC; relatively frequent.

44 The sign -zabar is also attested in combination with the pre-determinative urudu/a-; cp. Foxvog
(2016: 13).
45 Many former “compound signs”, here KAxUD, are, in later periods, written analytically, here
KA.UD; bar being in both cases a phonetic (disambiguating) element.
46 Not considered as determinative by Borger; cp. his no. 381 and 28 (and 29).
47 Note also HI.A.SAR (see ePSD), perhaps hi-a nisig „various (things), vegetables”; but
cp. hi-issar and HIsar = Akk. hisu, a vegetable; see further Selz (1993: 155–156,
503–504).
48 The iconic origin of this sign is uncertain. The sign may well depict a “sail”, see already Deimel
LAK 376 „Segel” and LAK 377 „Segel+Taue”. This would provide an interesting parallel to the
Egyptian situation, see below 5.5 with fn. 149; but for a different opinion see Lincke & Kammerzell
(2012: 71–75).
49 Problematic, but cp. Falkenstein (1949: 35).
50 Rebus writings play a salient role in the evolution of Sumerian glottographic writing, which relates
to the fact that the Sumerian language has (allegedly) numerous homonyms.
 The Question of Sumerian “Determinatives” 297

[34a KAM- sc. útul (?) only in F [6]; very doubtful51.]

35. **ul– (Borger 441; F [0]; L [0]; E [0]; H [0]; Fo [0]; Ed [0]); doubtful.
a–b. “bud; flower”; PR; due to phonetic (?) similarity and mythological speculations
used for marking names of stars; sometimes read as mul4, corresponding to (12)
mul and (29) múl.
c. PC(?) doubtful and problematic; rare/doubtful.

36. gig– (Borger 446; F [0]; L [0]; E [0]); H [0]; Fo [0]; Ed [0]) .
a. “sick; illness” meaning of sign is unclear; classifies names of illnesses.
b. PR; originally part of compound lexemes(?), a late invention?
c. PC/Late; rare.

37. –ki (Borger 461; F [16]; L [22]; E [16]; H [10]; Fo [18]; Ed [2]) .
a. “place; cities and other geographic entities”; classifies names of countries and places.
b. PO; lexeme status.
c. EA (early 3rd mill.); frequent.

38. *diš– (Borger 480; F [15]; L [1]; E [0]; H [1]; Fo [1]; Ed [0]).
a. “one”; marks in the early 3rd mill. lines and entries; = one (item), later restricted to
marking (usually male) personal names; mensural.
b. PR; originates as numeral.
c. EA/CL; frequent.

39. –meš (Borger 533; F [0]; L [24]; also abbreviated as –me : L. [25]; E [0]; H [0];
Fo [0]; Ed [0]).
a. “they are”; later marking noun plurals, esp. in non-Sumerian contexts; mensural.
b. PO; originally pl. of encl. copula.
c. EA; cp. also (1) and the remarks below; relatively frequent.

40. túg– (Borger 536a; F [0]; L [15]; E [29]; H [19]; Fo [13]; Ed [13])
.
a. “cloth; garment; textiles”; classifies garments in general (different from more
specific (8) linen (44) wool).
b. PR ; originally part of compound lexemes, meaning (piece of) cloth.
c. AR(?)/EA; frequent.

51 Mentioned only in Falkenstein (1964: 21) “Gericht vor Speisen”; reading /kam/ here is obsolete;
perhaps simply útul- “bowl” or similar; used as a count word like saǧ(-) (11)?
298 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

41. zíd– Borger 536b; F [0]; L [0]; E [35]; H [0]; Fo [0]; Ed [15b]) (?)52.
a. “flour”; classifies all sorts of flour.
b. PR; originally part of compound.
c. EA (from late 3rd mill.); relatively frequent.

42. tukul– (Borger 0; F-253: L [0]; E [0]; H [0]; Fo [0]; Ed [0]).

a. “weapon”; iconic origin? If Falkenstein is correct it may compare to “unique clas-
sifiers” in Jakaltek and in Egyptian.
b. PR; part of compound nouns.
c. EA; rare/doubtful.

43. udu– (Borger 537; F [0]; L [19e?]; E [31]; H [22]; Fo [0]; Ed [0]) .
a. “sheep”; classifies sheep and other ovine animals (cp. (15) donkey and (20) bull).
b. PR; originally part of compound lexemes.
c. EA (from late 3rd mill.); relatively frequent.

44. sík(i)– (Borger 539; F [0]; L 19a]; E [26?]; H [19]; Fo [0]; Ed [0])

a. “wool”; classifies all sorts of (woolen) fabrics, with extension from wool to all sorts
of other fabrics.
b. PR; originally part of compound lexemes, “made of wool”.
c. EA (mid 3rd mill.); relatively frequent.

45. mí/munus– (Borger 554; F [8]; L [3]; E [22]; H [15]; Fo [3]; Ed [0]) .
a. “female; woman”; pure graphemic use for marking female personal names, profes-
sions, and female animals.
b. PR; lexeme status.
c. EA (late 3rd mill.54); relatively frequent.

46. –min (II) (Borger 570; f [0]; L [27]; E [0]; H [0]; Fo [0]; Ed [0]).
a. lit. “two”; marks dualis (esp. in Sumerogram in Akkadian contexts).
b. PO; numeral.
c. EA/Cl.; relatively frequent.

52 The iconic origin of the sign read zíd remains unknown; it is perhaps an abbreviated depiction of a
“sieve” as suggested by O. Goldwasser.
53 A. Falkenstein’s description of tukul- as determinative is exclusively based on the variants tukulmit-
tum (RTC 198 rev 13) vs simple mi-(ì-)tum in in RTC 197 rev., Gud. Cyl. B 7: 24 et passim; see
Falkenstein (1949: 35).
54 In later Akkadian context, the sign is used to mark grammatically feminine of Akkadian nouns, e.g.
MUNUS+HUL = FEMALE+EVIL = lemuttu “evilness”; cp. Borger (1978: 192 no. 554). A rare
parallel for this process may be found in Egyptian. The Semitic loanword brkt –“blessing” ‫ הכרב‬is
a feminine noun in Biblical Hebrew. It gets the feminine classifier in a 20th Dynasty example, see
Hoch (1994: 103–114).
 The Question of Sumerian “Determinatives” 299

47. –àm (Borger 579a; F [0]; L [30]; E [0], H [0];99 Fo [0]; Ed [0]).
a. “(X) is it”; originally 3.ps. sg. of the encl. copula /am/ (from the verb /me/ “to be”).
b. PO; following Sumerian logograms; not a true determinative; mensural.
c. CL/PC; relatively frequent.

[47a –aya (Borger 579b; F [0]; L [30]; E [0], H [0]; Fo [0]; Ed [0]); doubtful!
a. “times”; marking multiplication number numbers.
b. PO; origin unclear; technical term? Only in Assyrian? Mensural?
c. PC; rare/doubtful.]

48. íd/i7– 55
(Borger 579c; F [5]; L [5]; E [1456]; H [6]; Fo [8]; Ed [0]).
a. “river; watercourse, canal”; classifies names of rivers and canals.
b. PR; originally part of compound lexemes.
c. (AR[?])/EA (mid 3rd mill.); relatively frequent.

49. ninda/índa–57 (Borger 589; F [0]; L [0]; E [0]; H [0]; Fo [0]; Ed [0])
.
a. “bread”; classifies breads and pastries, also count noun.
b. PR; originally part of compound lexemes.
c. EA (from late 3rd mill.?)/CL/PC; relatively frequent.

50. –ku6, (Borger 597; F [17]; L [20]; E [17]; H [11]; Fo [19]; Ed [6])
Cp. also archaic GIR
a. “fish”; also: “amphibians, crustaceans”; classifies all kinds of fish58 and – by exten-
sion – other aquatic animals.
b. PO; originally part of compound lexemes.
c. (AR?)/EA (from mid-3rd mill.); frequent.

2.3 Comments on the proposed consolidated inventory

This section, in a kind of bird’s eye perspective, offers a preliminary analysis and synthesis
of the information given in the entries of the list just presented. The partial neglect of

55 In ED texts the composite sign ÍD = A+ENGUR = WATER+GROUND WATER DEPTH

was also written A for which an (artificial) reading id5 is proposed; likewise, the simple sign
ENGUR alone may be transcribed as id3.
56 Ellermeier (1979: XXXIV) “als Det. vor Monatsnamen wohl nur akkad”.
57 The modern transcription inda3 replaces traditional ninda (cp. Hh XXIII v 19); the connection
between inda3 and inda2 was already proposed in Selz (1999). The loss of an initial /n/, well
attested for many other Sumerian words, seems the plausible explanation.
58 Together with (6) bird one of the most frequently used determinatives, pointing to the importance
of birds and fish in early Mesopotamian diet.
Note, that in a few cases were ku6 fish is attested preceding the noun (see ku6suhur vs. suhurku6
in Falkenstein 1949: 35 n.2) this may be indicative of a reading ku6-suhur as is suggested by the
phonetic (Emesal) writing ku6-da-s(š)uhur(ku6); cp. Falkenstein (1952: 62).
300 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

chronological (and linguistically based) varieties is justified by the fact that, on one hand,
no sufficiently verifiable and calibrated data are yet available; on the other hand, our main
intention is to discuss the overall system of cuneiform determinatives in order to reanalyze
them as “classifiers”.

2.3.1 The consolidated list: Information analysis

After the various comments made on the variation between the source-lists used for
establishing the proposed inventory, the following section will consider the additional
information contained inside the entries, in the order it is given in the entries.
a. Meaning: First we give here the (general) lexical semantics of an entry; cp. also the
discussion of the thematic organization of the inventory in section 3 below.
Use and function: This is an abbreviated description of the use and the most important
function of a determinative; see further the extensive discussion in Section 3 of the
thematic organization of the system.
b. Position: PR or PO, or both. It is a striking phenomenon of the system of Sumerian
classifiers that they can be found both in pre- (PR) and post- (PO) position. Generally,
the assignment to a position before or after the noun is strict, the only possible exception
being (13) –urud(u/a)(–) for which some scholars also accept a post-noun position.59 A
possible post-position for this determinative (13) might be motivated by the fact that the
comparable determinative (30) –zabar “bronze” is exclusively attested in post-noun
position. The position marked as PR (i.e. pre-posed to the noun) is the most common,
while the PO position (postposed to the noun), which is rather rare (type), is attested on
the other hand with some widely-used classifiers (token): For instance, PO is used for
(6) bird, (13) copper, (26) vegetable (garden plots), (30) bronze, (37) place name,
and (50) fish.60
Most puzzling in terms of position is the interesting thematic field of animals, with
both PR and PO, with PR for mammals, quadrupeds, terrestrial animals, but PO for non-
terrestrial ones such as (6) birds and (50) fish, which can be counted by the hundreds.
So far, the striking and clear differentiation between pre- and post-classifiers has found
no explanation. Of course, this PO position could be pointing to their possible origin
as lexemes in final position of a compound. Therefore, we suggest that a connection
may exist to the fact that in Sumerian two kinds of noun+noun compounding are
attested: the common left-headed and the rather rare right-headed nominal compounds
(See Jagersma 2010: 117–120).61 The PO system does not seem to be productive in
the historical periods which coincides nicely with the common hypothesis that the

59 Ellermeier (1979: XXXV): “vor und nach Gegenständen, die ganz od. teilweise aus Kupfer od.
Bronze gefertigt wurden”.
60 The generally post-position “déterminatifs grammaticaux” are not considered here; see below
2.3.2, subsection C and 3.2.3.
61 If this assumption is accepted, we may connect it with our theory of a spoken origin of the Sumerian
classifier system, to which a separate article is in preparation.
 The Question of Sumerian “Determinatives” 301

right-headed noun+noun compounds allegedly belongs to a protohistoric stratum of the

Sumerian language. This would mean that such right headed compounds are considered
to be not any longer productive in the historical phases of the Sumerian language – so
the parallel to the PO system is obvious.
Origin: Most important for any further research are the great majority of determinatives
which, in our list, are designated as “originally part of compound lexemes” or having
“lexeme status”, attested both as pre- and post-classifiers. This notion is not entirely
new, but it is neither generally accepted nor discussed in detail by anyone. Labat (19886:
21) remarked about a selected number of such determinatives: “Il existe quelques
autres déterminatifs qui, en réalité, peuvent être considérés comme faisant partie de
l’idéogramme”. 62 As already mentioned, Edzard’s statement (2003: 9) is more rigorous:
“Determinatives can be proven not to have been pronounced (although doubt may exist
in specific instances)”. In our opinion the original lexeme status of the overwhelming
majority of the classifiers is a prominent feature linking language and script.
Thus, almost all of the entries in the list are attested as independent lexemes, so they
have lexeme status. A possible exception is (48) na4–/ ià4– /zá–. This seems to be the
only case where the system uses a specific grapheme, however its origin and evolution
remain presently uncertain. Therefore, the use of all determinatives may mirror the
alleged Sumerian system of noun-compounding, especially the left-headed and right-
headed noun+noun compounds. The fact that this may point to their having been
originally pronounced, as already mentioned, is a suggestion that merits a profound
discussion which is at this point beyond the scope of the present article, but will be
treated in a separate article in preparation.63
Another interesting group in our list are those determinatives which sometimes are
(allegedly) later inventions, sometimes originating in a specific Akkadian language
environment: (1), (5), (10), (12), (14), (21), (24), (29), (30[?]), (31), (33), (36), (39). They
prove that the system of classifying nouns was productive long after spoken Sumerian
had died out. However, even these younger inventions must be considered part of the
Sumerian system, in much the same way as the ‘newly introduced’ Medieval Latin is
a form of Latin. The well-known fact that the use of determinatives as Sumerograms
spread from the Mesopotamian core to neighboring languages (as far as they are written
in cuneiform) corroborates this notion, notwithstanding the fact that West Semitic or
Hittite languages induced alterations.

62 Cp. also Borger (1978: 48), who commented on this list in the following way: “Es ist mehrmals
nicht mit Sicherheit zu entscheiden, ob ein Determinativ oder ein Logogramm bzw. ein Teil eines
Logogramms vorliegt”.
63 Selz suggested already in (2008: 15, fn. 6) a connection between the well-known phenomenon
of lexical lists (thematic lists) and the “determinative” phenomenon. Recent elaborations of the
implicit semantic classifications in logographically written scripts  (Hyman 2006, Hilgert 2009,
Zand 2009, Civil 2010 and Johnson 2012) indicate that an in-depth research of determinatives
versus lexical lists may not only improve our understanding of the emic mental mapping of the
Mesopotamian world but also support the hypothesis that the Sumerian classifiers originated as
elements of the language; see also below 6.1.
302 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

c. Periods of use: Rather complicated – and questionable – is our preliminary attempt to

indicate the (earliest) period(s) in which the determinatives are attested. This depends
very much on whether one accepts the identification of these determinatives as (always)
unpronounced graphic elements or whether they should be considered as (initially) pro-
nounced parts of the language. Therefore, our indication AR (Archaic) is speculative.64
It is only with those determinatives marked as Cl(assical) or L(ate) that the assump-
tion of pure graphic character (as reading help) is virtually certain. For the time being,
we would simply refer to the fact that the thematic organization of the most important
Sumerian determinatives corresponds not only to the organization of the earliest “the-
matic lists65” but also shows a surprising parallelism with other classifier systems, as
will be shown later.
The outcome of this tentative periodization is, however clear enough: The system
overall is productive, even as late as the 1st millennium BCE and seems to keep a – more
or less – stable number over time.
Frequency: Almost all entries are marked as having some frequency, whether they are
“relatively frequent” – as 31 of them are, or very “frequent” for 7 of them. Only 9 of
the entries are marked as “rare”, and 3, resp. [5], as “rare/doubtful”. For lack of reliable
statistics, the proposed and estimated frequency is not very informative, and depends,
of course, in addition, very much on the genre of texts in which they are attested, and,
last but not least, on our familiarity with the texts. However, judged to be rather rarely
attested, and marked as rare, are (1), (12), (14), (21), (29), (30), (31), (33), (36), and
rare and doubtful are: (24), (35), (42), [(11a)], [(47a)]. Setting aside the mensural ones
(discussed below in 3.2.3) perhaps the most interesting cases of rare attestation are (14)
“flesh”, (29) “star”, and (36) “illness”, which, while being quite late inventions, actu-
ally prove the productivity of the system.

2.3.2 Different authors and different lists

The list was drawn up, somewhat arbitrarily, by choosing widely used lists as given in sign
lists and grammars by various authors. Little is known of the reasons that led some authors
to include a particular determinative not found in the other scholars’ lists. In general, the
motivation for such inclusion or exclusion is rarely made explicit.
Indeed, the variation in the numbers and the types of determinatives chosen by the
individual authors is astonishing, as can easily be seen in the following table, in which
determinatives shared by all lists are in bold and the ones shared by all but one are in bold
and italics:

64 From a methodological point of view, we restricted our research the phenomenon of the allegedly
silent determinatives.
65 These “thematic lists” (Englund 1998 and Civil 2008, also Civil 2010: 58–147) classify the surround-
ing physical and societal world (Veldhuis 2004), and already their earliest specimens (3000 BCE)
group their material into different classes: (numbers), grain and grain products, fish, birds, domestic
animals, wood and wood products, dairy products, containers, textiles, metals, persons, place names
and time indications (Veldhuis 2006: 188; slightly different labels are used in Englund 1998: 88).
 The Question of Sumerian “Determinatives” 303

Table 1 | Comparison of different lists of Sumerian “determinatives”

Selz Borger F(alkenstein) L(abat) E(llermeier) H(uehnergard) Fo(xvog) Ed(zard)

a few 1 2 - 27 - - - -
hide 2 7 - 16 18 13 9 14
god 3 13 1 4 2 2 4 1
city 4 38 12 7 32(?) 23 15 -
month 5 52 - 5 14 8 - -
bird 6 78 18 21 23 17 20 7
reed 7 85 3 12 6 4 6 11
linen 8 90 - - 5(?) - - -
net 9 104 - 19c - - - -
field 10 105 - - 11 7 - -
head 11 115 - - - - - -
ship 11a - - 19b - - - -
star 12 129a - 6 21 16 10 -
copper 13 132 13 16 33 24 16 4
flesh 14 171 14 18 34 26 17 -
donkey 15 208 - 19d 1 - - 15a
beer 16 214 - - 15 - - -
aromatics 17 215 10 - 23 - 12 -
stone 18 229 9 13 24 18 11 12
wood 19 296 4 11 7 5 7 3
bull 20 297 - - 8 - - -
each 21 139 - 30 - - - -
jar 22 309 2 17 3 3 5 5
plant 23 318 11 10 30 21 14 10
house 24 324 - - 4(?) - - -
human 25 330 7 2 19 14 2 8
vegetable 26 331E 19 23 25 - 21 9
mountain 27 366 - 8 - 12 - -
barley 28 367 - - 27 - - -
star 29 376 - 10 - - - -
bronze 30 - - - - - 22 -
various 31 396+404 - 26 - - - -
wind 32 399a - 12 - - - -
clay 33 399b ? - 13 - - -
(ordinalia) 34 406 - 28+29 - 9 - -
? 34a ? 6 - - - - -
bud(?) 35 441 - - - - - -
illness 36 446 - - - - - -
place 37 461 16 22 16 10 18 2
one 38 480 15 1 - 1 1 -
(plural) 39 533 - 24 (25) - - - -
textile 40 536a - 15 29 19 13 13
flour 41 536b - - 35 - - 15b
weapon 42 - F-2 - - - - -
sheep 43 537 - 19E(?) 31 22 - -
304 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

Selz Borger F(alkenstein) L(abat) E(llermeier) H(uehnergard) Fo(xvog) Ed(zard)

wool 44 539 - 19A 26(?) 19 - -
woman 45 554 8 3 22 15 3 -
two 46 570 - 27 - - - -
„is it“ 47 579a - 30 - - - -
(times) 47a 579b - - - - - -
river 48 579c 5 5 14 6 8 -
bread 49 589 - - - - - -
fish 50 597 17 20 17 11 19 6

Remarks to Table 1:

A. 18 core items
Of the 50 determinatives listed, there is a core of 18 items widely shared across the
different lists (Table 2 below), including 11 being mentioned by all the authors (in bold)
and 7 by all but one (in bold italics). Those 18 items will be identified again in bold and
bold italic in the general table showing the thematic organization of the whole system
(Table 3 in subsection 3.2.1) in order to capture then what seems to constitute maybe the
core components of this classification system.
Table 2 | Core determinatives mentioned by all authors or all but one

Determinatives mentioned by Numbers in the new list Total

All 3, 6, 7, 13, 18, 19, 22, 23, 25, 37, 50 = 11
All but one 2, 4, 14, 26, 40, 45, 48 =7
Core determinatives = 18

It is very likely that the authors of our lists grasped the most salient determinatives of the
Sumerian system, those that we may assume to offer a glimpse – when taken all together
– of the major items of this classification system:
All: (3) deity, (6) bird, (7) reed, (13) copper, (18) stone, (19) wood,
(22) jar, (23) plant, (25) human, man, (37) place, (50) fish.
All but one: (2) skin/leather, (4) city, (14) flesh, (26) vegetables, (40) cloth,
(45) female, (48) river.
Despite the fact that – following our authors – we take a bird’s-eye perspective and ignore
any chronological features, this list apparently provides a good insight into the most salient
classes of the cuneiform classification system that persisted from the beginning of writing
in the 4th millennium down to the late 1st millennium BCE.

B. The most restrictive list – Edzard

Edzard enumerates only 14 determinatives, lacking 4 of the items (4) uru-, (14) uzu-, (45)
munus-, and (48) íd, considered determinatives in the opinion of all other authors. This
fact may imply that from Edzard’s point of view these items are either later inventions or
 The Question of Sumerian “Determinatives” 305

have been pronounced (as part of composite nouns) and therefore should not be included
in a list of determinatives. This leaves us with a minimum core of 14 determinatives.

C. “Déterminatifs grammaticaux” – excluded by most authors

“Déterminatifs grammaticaux” are the determinatives labelled so by Labat (19886: 22)
and adding up to a handful: (1), (21), (34), (38), (39), (45), (47a[?]). These determinatives
actually possess a different classifying function and correspond in classifier systems to the
so-called “mensural classifiers”, as opposed to the larger other set corresponding to more
familiar “sortal classifiers”, an issue considered in 3.2.3 below. While the co-occurrence in
the Sumerian inventory of determinatives of both sortal and mensural types of classifiers
is in itself a striking characteristic of the system, the emphasis of this initial study will be
on the sortal ones.

D. Purely graphic – or echoing noun+noun compounding?

It is worth underlining here again that, according to the opinion of the authors used as
sources, all these determinatives are considered as unpronounced, purely graphic ele-
ments. However, as will be argued below (section 3), there is no doubt that they constitute
an interesting classification system, akin to the classifier systems discussed in the linguis-
tic literature. And it is also worth noting how this classification function might be mirrored
in a feature of the Sumerian language, specifically its noun+noun compounding forms, a
fact explicitly mentioned by Labat (19886: 20, and alluded to also by Edzard 2003: 9).

2.4 In conclusion
The inventory of signs presented above is a revised list of Sumerian determinatives, which
is offered as reference for future work and re-consideration. As mentioned, it is based on
a compilation of established lists that all differ one from the other, and it constitutes an
altogether more complete inventory. This list is innovative in pointing to the kind of infor-
mation that would be desirable or even necessary to gather systematically. It provides new
(preliminary) information by estimating the frequency of use and by dating the periods of
(earliest) use of these determinatives. At this point, this list represents just the initial step
of a larger research project aiming at completing the information provided in this inven-
tory through a more systematic study of the texts available. It is also an essential initial
step to provide evidence that the determinatives here collected attest to the existence of a
cuneiform classifier system, as will be shown in the following section.

3 Thematic organization of Sumerian “determinatives”: a “classifier

system”
The inventory given above provided some general information about various basic proper-
ties of the items, such as the forms of the determinatives, their origin and their use, all with
the explicit aim of completing and rearranging some of the information found in already
306 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

established lists. This section aims now at taking a different approach to the proposed new
list, by considering how this inventory of the determinatives of the Sumerian script (and
language) projects an interesting thematic organization of the world of the Sumerians, of
the kind which is very characteristic of classifier systems around the world.

3.1 The classification of noun entities mirrors a “noun classifier” system

To readers unfamiliar with the field of classifier studies, the thematic organization of the
inventory of the (allegedly unpronounced) classifiers to be proposed here may on first
sight appear somewhat arbitrarily constructed. It will however be shown later how it actu-
ally corresponds in convincing ways to ontologies commonly revealed by other systems
of classifiers from around the world. In fact, this discussion of the thematic organization of
the Sumerian system will be later compared with that of two other systems: one the (pro-
nounced) noun classifier system of a contemporary Mayan language and the other the gra-
phemic (unpronounced) classifier system of the Ancient Egyptian hieroglyphic writing.
An overall characteristic of the thematic organization of the Sumerian system is that it
corresponds in an interesting way to specifically one of the subtypes of classifiers identi-
fied in the literature on classifier systems: the particular one known in the literature on the
typology of classifier systems as a “noun classifier” system.66 Systems of “noun classifi-
ers” have been found so far mainly in native oral tradition languages of the American and
Australian continents.
One of the defining traits of “noun classifier” systems is that their categorization sche-
ma relies principally on the essence of the entities, therefore distinguishing classes for
humans, animals, plants, etc.67 This is said in contrast to so-called “numeral classifier”
systems which are better known and widespread in languages of Asia. Those are used
specifically when counting entities (hence their name “numeral”) and have the character-
istic of relying for their categorization schemas principally on the shape of those entities,
distinguishing minimally between 1D long objects, 2D flat objects and 3D round objects.68
What identifies the Sumerian system as a “noun classifier system” is its combination of
classes for animates, humans and animals, with an array of classes for inanimate entities
of the world (section 3.2).69 These inanimate classes include a number of different classes
for the vegetal and the mineral worlds, as well as for the natural environment.

66 Identified early in Jakaltek in Craig (1986a) and argued as one subtype of classifier systems,
in Grinevald (2000, 2004), and refined in Grinevald (2015). For an extensive source of data on
classifier systems in a great variety of languages, see Aikhenvald (2000). On noun classifiers, verb
classifiers sortals and mensurals, see Bisang (forthcoming).
67 E.g. Senft (2000).
68 So, while in a noun classifier categorizing by material/essence, one could find for instance:
a ‘man-hunter,’ a ‘wood-canoe’; in a numeral classifier doing it by shape, one could find
instead ‘two-biped hunter,’ ‘one-long rigid canoe.’
69 The basic distinction between animate and inanimate entities is also reflected in the morphology of
the Sumerian grammar in a gender system marked in noun and verbal phrases, in which the mor-
pheme /n/ refers to the animate / personal class / human class of a given noun and the morpheme
/b/ to a class combining inanimate / impersonal / non-human entities. For a recent critique of the
 The Question of Sumerian “Determinatives” 307

In addition, as is known of classifier systems in general, the Sumerian system also

includes classes that reveal the importance of certain cultural elements linked to this part
of the world in ancient times. Finally, it will be seen that while the Sumerian classifier
system is principally based on criteria of the essence of the nominal entity, it also relies in
places in its categorization on the notion of the functionality of the items. This is also found
in other noun classifier systems, where it is taken as a way of underlining the importance
of particular interactions of humans with certain elements of their environment. This will
be discussed in particular with the case of the categorization of animals.

3.2 The thematic organization of the Sumerian “classifier” inventory

The first step in analyzing the Sumerian determinatives as classifiers is therefore to consider
their overall thematic organization. This will be done first by presenting an overview of all
the items given in the proposed consolidated list by arranging them in a table according to
the semantic domains they cover, independent of the periods in which they appeared and
of their formal characteristics (Table 3 below).70 The general presentation of this thematic
table of the items of the new list will then be followed by a consideration of the semantic
principles of categorization evidenced by the semantic domain cuts observed, and then by
two case studies that of the classification of illnesses and of animals – in anticipation of
illustrative comparisons with other systems discussed in sections 4 and 5 below.

3.2.1 Overall thematic organization of the determinatives list as evidence for a Sumerian
classifier system
The main point of Table 3 below is to offer a first overview of all the semantic domains
covered by the 50 items of the newly proposed list, while keeping most of the information
already presented on the items. For instance, all the items of the list are identified by their
given number and their position, indicated again by a hyphen (–) placed either before or
after a classifier, depending on whether it is found before or after a word. The information
about which belong to the core determinatives of Table 2 is also repeated, with the items
mentioned by all sources used given in bold and those in all sources but one given in bold
italics.

extant terminologies for such a marking system, see Jagersma (2010: 102) who proposes “to call
the two gender classes human and non-human”. On the complex relations of gender systems and
classifier systems see Kilarski (2014) with bibliography.
70 Ignored here is the question of whether “new” classifiers were introduced later by Akkadian
speaking scribes, in arguing that all the cuneiform classifiers attested are Sumerian logograms that
attest to the existence of a Sumerian system, no matter whether Sumerian was still the vernacular
or not (see discussion above).
308 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

Table 3 | Thematic organization of the Sumerian classifier system

SORTAL MENSURAL
ANIMATE INANIMATE Unitizers/plural
Human and deity Vegetal Mineral world
–deity (3) –wood/lumber/ (and man-made –one, unit (38)
–human, male (25) tree (19) objects) individual things–
–reed (7) –stone (18) (1)
–human, female
(45) plant/grass– (23) –clay (33) –head (11)
–Human (38) –barley, cereal (28) (–)copper– (13)
–illness (36) vegetables/garden bronze– (30) dualis– (46)
plot– (26) –weapon (42) plural– (31), (39)
[bud– (used only Mensuratives
Animals for stars) (35); cp. Arrangements and
Quadrupeds (12), (29)] aromatic, Natural elements containers
–bovine (20) perfume– (17) –wind (32) –bundle (9)
–sheep (43) Manufactured –star 1 (older) (12), –jar, jug, pitcher
(35)
–donkey (15) From vegetal (22)
–star 2 (younger) –bowl (34a) (?)
More animals –flax linen (8)
(29); cp. (35)
fish– (50) Never used for the
plant! Spatial
bird (and insect)–
(6) –cloth/textiles place– (37)
(40) –mountain/land (27)
Body Parts of
humans and –beer (16) –river (48) 1
animals –flour (41) Man-made spaces
–flesh (14) –bread (49) –field/piece of land
–skin/leather (2) –ship (11a[?]) (10)
From animal –house (/container)
–animal hides (2) (24)
–wool (44) –city (4)
–cloth/textiles –garden plot/
(40) may include vegetables (26)
wool items Time
See also mineral –month (5)
world

3.2.2 Domains and subdomains

As is commonly found in noun classifier systems, the organization of the world in the
sortal Sumerian classifiers falls first into two major domains: one of animates and another
of inanimates, and a number of subdomains within each.
Several remarks need to be made about the domain of animates in the Sumerian system:
first that the subdomain of humans and deities is not very developed; second that it was
 The Question of Sumerian “Determinatives” 309

decided to place the class of illness in this domain on the basis of how this item is treated
in other systems, as will be mentioned in (3.3.2), and third that the animal domain will be
the object of more attention below (3.3.3), in order to point in particular to the absence of
generic animal classifier.71
As for the organization of the inanimate world, it is subdivided into vegetal, mineral
and natural world semantic regroupings. As expected, the vegetal domain has more classes
than the mineral one, reflecting the universality of the more intense and diversified interac-
tions of humans with this domain, as evidenced by the several classes of products derived
from vegetal entities. For the domain of natural elements, it is interesting to note that it
includes neither sun nor moon, which are commonly found in other classifier systems.
A domain of time has been included although it is an intangible entity that could have
been considered a mensural, a measure of time. In any case, it includes only one class,
that of month.72 One of the most interesting domains of the Sumerian system is the well-
developed set of classes that have been regrouped here into a domain of spatial entities.
This domain includes not only natural spaces (indeterminate place, mountain/foreign land/
netherworld, river) but clearly what could be considered as man-made spaces (field, city,
house, garden plot[?]) a categorization that may reflect their importance for this great civi-
lization, which is known for its sophisticated urban centers and advanced administrative
control of land and agricultural plots.

3.2.3 Setting aside the Sumerian determinatives considered as mensural classifiers

It is worth noting that the first level organization of all the items of the list into semantic
domains and subdomains is a primary distinction between the two semantic types of
classifiers, known as “sortal” and “mensural”. While sortals classify according to “sort”,
categorizing the entities according to salient characteristics – such as essence (like most
of the Sumerian classifiers) or shape (in the case of numeral classifiers), mensurals are
concerned with counting either individual units and called “unitizers”, or types of measures
or of arrangements, called “mensuratives”.73
The coexistence of both sortals and mensurals within the Sumerian system of determi-
natives (being reanalyzed here as noun classifiers) is a very special feature of this classifier

71 There are, however, Sumerian lexical terms which attempt to establish more generic animal classes;
see below 3.3.3.
72 Note, however, that Englund (1988: 164–168) identified passages with time indications in the
earliest lexical lists from Uruk; see also Veldhuis (2014: 99).
73 Mensurals are known from numeral classifier systems, where they co-exist with sortals. Sortals
are those classifiers that categorize by some semantic feature of the entity, in the case of numeral
classifiers, by the shape of the entity as in “2- quadruped cats” or “3-round oranges” while
“mensural” classifiers provide some meronymic information on the nature of the units to be counted,
either as unitizers like “1-head of cattle”, “2-pair of glasses” or as mensuratives indicating a kind
of measure, like types of containers, as in “3-bags of oranges” or “2-bottles of wine”, see Craig
(1992: 281) and Grinevald (2000: 1020). An attempt to relate this terminology to the varieties of
“number notation systems” attested in the archaic Uruk tablets, seems to be a highly promising
endeavour; cp. Green & Nissen (1987: 114–144).
310 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

system, since it combines in one system sortals – typical of noun classifier systems and a
few mensurals usually associated with numeral classifier systems.
The determinatives relabelled here as “mensural classifiers”, together with those for
ordinal (33)74 and distributive (21) numbers, are, with the exclusion of DIŠ (38), attested
in post-position only. Most scholars do not include (all of) them in their lists of determina-
tives (section 2.3.2) and in classical Sumerian texts they preserve their literal meaning.
However, later logographical (“Sumerian”) written texts make extensive use of them as
graphemic classifiers at which point they were used in a stereotypical manner. Most proba-
bly their wide-spread use was motivated by linguistically mixed texts in which the earliest
examples often use Sumerian for the formulaic textual parts.75 Therefore, original code-
switching froze, and these “déterminatifs grammaticaux” became a merely graphical ele-
ment of the script and in this case their position after the nouns is easily understandable.76
Typologically, these “déterminatifs grammaticaux” treated here as mensural classifiers
include both unitizers and mensuratives (see fn. 73). The Sumerian unitizers include the
written number (diš, [38]) “one” that identifies individual units (animate and inanimate)
and is amply used already in the earliest texts marking specific entries and it is the only
“grammatical determinative” written in front of the noun (PR). The repetition of the num-
ber “one” in the form aš77 is didli, lit. meaning “one by one”, designating a countable
plurality of items, else pluralis paucitatis. Related classifiers are min (46) (repeated DIŠ),
referring to the dualis, and the plural markers hi-a (31) and meš (39). Hi-a is a past perfect
particle of the verb /hi/ “to mix” attributing a (given) set of countable items to a superordi-
nate class, e.g. udu-hi-a sheep + mixed > “various (types of) sheep”.78 The plural marker
-me(š) has the form of the enclitic copula in the 3.ps.pl., in the literal meaning “(sheep)
they are”. The singular form -àm of the copula (47), with the meaning “(X) is it/he/she”,
singularizes an item much in the same ways as the number diš (38).
The mensurals of Sumerian include also classifiers that originate as count words, such as
head (piece) (11), bundle (9), jar/jug (22), and perhaps also bowl (34a), and eventually
even field measure (10). They may correspond to unitizers and mensuratives expressing

74 The formation of ordinal numbers in Sumerian is interesting. The literal interpretation is “of
number is it/he/she”. Grammatically, these are headless genitives+encl. Cop. (3. ps). The ordinal
2, “the second”, in Sumerian */min-ak-am/, designates that something /someone is belonging to
the quantity domain “two”.
75 On this phenomenon see Selz (in press a). This sort of multilingualism is a phenomenon to be
studied from a contact linguistic point of view and should be kept apart from the well attested
forms of bilingualism which, in Assyriology, designates the phenomenon of text transmitted in two
languages, e.g. Sumerian and Akkadian (or other languages).
76 A related phenomenon in the hieroglyphic script are perhaps what Werning has called “Grammato-
klassifikatoren” (Werning 2011: 101–113) such as the 1st sg., Dualis and Plural classifiers. In
Sumerian, however, the “déterminatifs grammaticaux” originated as frozen (and standardized)
language elements.
77 For the difference between AŠ (ZATU 37) and DIŠ (ZATU 81) in the archaic periods see Green &
Nissen (1987: 177 and 187).
78 In so far as it refers often implicitly to brand, shape, color or age of the sheep, hi-a is not a true
mensural, but a sortal classifier.
 The Question of Sumerian “Determinatives” 311

the units and measurement used.79 They probably developed to unpronounced classifiers
simply by adaption to an already existing system of unpronounced determinatives.
To sum up, the main point here is that the mix within a single system of noun classifiers,
of two sorts of classifiers – the mensural (typical of numeral classifier systems) and sortal
(common to all subtypes of classifier systems) – is actually a very interesting typological
feature of our system.

3.3 Categorizing principles of the Sumerian determinatives analysed as sortal

classifiers
In Table 3 above the attested items of the proposed list of cuneiform classifiers are struc-
tured not only along the grid of sortal and mensural classification but more specifically ac-
cording to their thematic function. In what follows, all the attention will be given to the sor-
tal kind, in order to prepare the comparison with other classifier systems in sections 4 and 5.

3.3.1 Level of categorization, size of the classes and class extension

The Sumerian system does not include a high generic classifier, meaning just “a thing”,
although comparable nouns are, of course, attested (compare the Egyptian system discussed
below). However, as in all classifier systems, the classes defined by the different classifiers
vary in size, some being large classes with dozens or more of items and others much more
restricted classes with a more specific selection. An example of this difference in size in
Sumerian is, for instance, the contrast between the large class of cloth/textile, and
the more limited linen class for just linen items. There does not seem to be, however, a
clear case of “unique classifiers” identified so far in the Sumerian system, i.e. classifiers
for just a single item, as there are in the Jakaltek and the Egyptian systems (Craig 1986a,
Goldwasser & Grinevald 2012); the only possible exception in Sumerian might be the
class of weapon (42).
The larger classes in the Sumerian system may either be large by the sheer number of
items of the same nature (like the hundreds of birds and fish identified in the Sumerian
system), or by various processes of class extension, in which case the classifier loses
partially its original semantics. Several such cases of extension will be considered below.
In the animal domain of quadrupeds, the process of extension of a class by absorption
of new items that have entered the culture is illustrated by the case of the classifier donkey
(15) (Equus asinus asinus)- developed from the more iconic 80

extended beyond its prototypical value of the specific species of domestic animal to head

79 It is often unclear whether the sign GANÁ has to be interpreted as part of a field name or
determinative or whether even a reading iku, a square measure (of ca. 3.500 m2), is correct. We
note here, that in the cuneiform systems of number notations the form of a number sign often
implies a specific measure. The number 5 might be written differently whether 5 (talents), 5 (mina),
or 5 (shekels) are intended.
80 It is remarkable, that for animals, especially those of larger size, the cuneiform script depicts just
the heads of these animals, not their shape. The complex and diverse situation for the various
animal heads were studied by Mittermayer (2005); for the ANŠE sign cp. 28–35 and esp. 30.
312 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

a class that includes wild donkeys (probably the Asiatic onager, Equus hemionus, also
Equus hemionus hemippus) and perhaps even deer. It finally comes to head a class that
has absorbed various foreign or wild animals, such as the horse (first attested as anšesi2-
si2 [donkeysisi, /sisi/ being a wanderwort]), then called the “donkey from the mountain”
ANŠE.KUR.RA = read /sisi/ ), and later occasionally even the camel as
ANŠE.A.AB.BA “donkey of the sea (lands)”.81
Another example of class extension is the case of the bull/cattle (20) classifier used
for all sorts of bovines, including Bos primigenius and probably also Bos gaurus/frontalis.
The SHEEP (43) classifier seems occasionally also to be used for other small livestock,82
which might be motivated by the lexically well attested use of “sheep” as an intermediate
taxon meaning “small livestock”, specifically “sheep” and “goat”.83
In the plant domain, it is the case of the extension of the barley (28) classifier to
classify other cereals. Meanwhile, in the manufactured, man-made objects domain, one
finds, for instance, the wood (19) classifier being used first for all sorts of trees, but
also for objects “made of wood” (Egyptian shows a similar classification, see Goldwasser
2002: 51). Similarly, the copper (13) and bronze (30) classifiers, as well as the classifier
for animal hide skin/leather (2), regroup together all sorts of manufactured objects
“made of bronze/copper” or “made of leather”.

3.3.2 The case of illness

It is especially interesting that illness (36) has a classifier of its own in the (late) Sumerian
system, to the extent that, as will be mentioned later, the concept of illness is often marked
in classifier systems, although in different ways. In Sumerian, illness forms a class of
its own; it is perceived as a result of demonic forces. In Jakaltek (see below, section 4)
it is singled out by being included in the male human entities. In the Egyptian Middle
Kingdom, “be ill” may be classified by the seth classifier – the male god of confusion and

81 We note that on the language level a similar extension is attested. The most common syllabic
writings for the camel are am-si-har-ra-an and am-si-kur-ra, both with the Akkadian translation
ibilu. The first expression means literally “the elephant (am-si = pīru “elephant”) of the caravan
route”, the second “the elephant of the mountains”.
82 Cp. udu dur sheep ‘young male donkey’: ELA/Ur III/Susa 1(diš) udu dùr ga-lu-hu-ul-ga MDP
23, 162 293 3 and perhaps udu durah sheep ‘wild goat’: ELA/Ur III/Drehem: 1sila4 4 udu dara4!?(
DURAH) UCP 9-2-2, 036 1; /alu/ “ram” in Akk. = ālu, seems also written udu a-lu sheep ‘ram’
(AUCT 2, 358 4; even /šegbar/, in Akk. šapparu, “a deer or mountain goat”, may occasionally also
take the sheep classifier (TL 263 8(?). Compare here Goldwasser (2017).
83 A similar extension of lexical use is attested for maš designating the he-goat and commonly also
used to designate goats in general. However, occasionally even u(z)d5 “she-goat” is used as broader
term (Selz 1989: 422 = Nik I 193).
We note in passing that the maš sign “goat” has no obvious iconic referent and is generally
thought to belong to the group of abstract signs. However, the suggestion of Selz (2000: 195) of a
phonetic relation to maš “interest; share; ‘half’” depicted (iconically by two crossed wedges may
be corroborated by the iconically similar hieroglyph (also uses as classifier) break, divide (two
crossed sticks) , see Gardiner (1957: 538, Z9–10).
 The Question of Sumerian “Determinatives” 313

disturbance, who is most often personified.84 The point to be made here is that the concept
of illness seems to have some universal importance to humans that is revealed in classifier
systems of different times and continents, and is signaled in the Sumerian system as well.

3.3.3 The case of the classification of animals: a functional categorization

In the Sumerian knowledge organization as evidenced by the classifier system of the script,
there is first the noticeable absence of any generic category of “animal”.85 The animal
world is actually divided into five classes, which seem to correspond in an interesting way
to some utilitarian perception of the animals in addition to physical similarities.86 This
clear utilitarian classification distinguishes between classes of bird, fish, bovine, sheep,
and donkey. This statement is corroborated by the observation that for all these groups of
animals specific lexical lists are attested, right from the beginnings of writing onwards.87
The role of the classes of bird and fish is clear, as they are important element of the
nutrition system and have a clear economic value, as shown by numerous economic docu-
ments from the earliest phases of writing.
The same utilitarian function could be applied also to livestock classes of bovine,
sheep and donkey, which additionally relate to the late 3rd–2nd millennium lexical generic
terms: /niĝki/ lit. “things (creatures, moving) on earth”, in Akk. nammaštû “the ones that
move around” and /niĝurlimmu/ lit. “four-legged things (creatures)” > “quadrupeds”, in
Akkadian būlu “herd” (intermediate taxon) for livestock88; also /niĝziĝal2/ “living beings” >
“creatures”, lit. “breathing things” (= Akkadian šiknat napišti, also namaššû), or /zišagĝal/
“the breathing ones”.89 The livestock classes may also point to the functional importance
of these animals, including, among other things, their use as working animals. The bovine
class is set aside for being the most important working animals, also important for their
dairy production, and to a lesser extent for meat consumption and leather production.
The sheep class is set aside as smaller livestock also used for dairy products and meat
consumption, but with the additional, and especially important, industrial value of their
wool. The functional value of the class headed by the donkey, with its later inclusion of

84 On the Seth classifier, see Te Velde (1967: 22–23), Goldwasser (1995: 102–103, 2005: 108–110);
McDonald (2007) and Allon (2007). The common classifier for the lexeme mr “be ill” since the
Middle Kingdom is the negative classifier, see DZA 24.127.330–350 (for the “bad bird”
classifier, see fn. 137 below). The Seth classifier is an alternative classification for this lexeme.
Names of illnesses may take the gland classifier, see Gardiner (1957: 539, Aa2).
85 The Sumerian language, at least at the turn from the 3rd to the 2nd millennium, knows more generic
terms, see next paragraph, but this has to be kept apart from the “classifier system”.
86 See Denny (1976) for an early discussion in classifier studies of the three principles of categorization:
by essence, by form or by function. In the case of birds, form must have also played a central role
as some birds were not regular part of food supply (we are grateful to an anonymous reader for this
remark).On the “utilitarian classification” which is activated in many cultures, see Hunn (1982).
87 It was Niek Veldhuis who actually stressed the environmental foundation of the earliest lists, noting
that any reference to metaphysical aspects are missing in these texts; see Veldhuis (2006: 187) and
cp. Selz (2011: 57–58 with fn. 37).
88 Compare here intermediate taxa in the Egyptian script, see Goldwasser (2002: 69–78).
89 An extensive discussion of the taxonomy of the animal kingdom is given in Selz (in press b).
314 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

the horse and the camel, is principally that of a transport class, to the extent that none of
these animals is normally used for meat consumption in the Ancient Near East.90
The other argument for a functional classification of the animal domain resides in the
fact that the animals that remain unclassified, from large carnivores such as lions and pan-
thers to smaller quadrupeds such as dog, cat, monkey or mouse, are all of limited utilitar-
ian value. This extends to reptiles, such as snake or worm, which remain unclassified too.
To sum up, animal classification in the Sumerian script system shows the following
characteristics:
1. Animals are classified mainly according to utilitarian and functional considerations.
2. Animals which have no utilitarian function – such as big carnivores (e.g. lion, panther)
or other animals (e.g. dog, mouse, cat) remain unclassified.
3. The classification clearly differs between domesticated and non-domesticated animals.
4. Birds and fish are classified by their own specific classifier.
5. Snakes and worms are not classified.91

4 The Sumerian classifier system in typological and areal perspective

The analysis of the Sumerian determinatives as consisting of a noun classifier system
relies critically on two previous studies of other classifier systems to be introduced in this
section. The first one is of a classifier system from a Mayan language of Mesoamerica
which was itself the original source of comparison to establish the analysis of Egyptian
determinatives as classifiers.92 These two systems will be considered in turn, from both a
typological and areal perspectives, in order to point to interesting similarities and expected
differences between them and the Sumerian (cuneiform) system.93

4.1 Comparison with the Jakaltek (Mayan) noun classifier system

The particular classifier system of Jakaltek94 to be considered first was at the origin of
typological discussions of the variety of classifier systems, and was instrumental in the
identification of a particular subtype of classifiers of nominal entities, now established in
the literature under the label of “noun classifiers”. A quick overview of the contribution

90 On the donkey, see, Way (2011: 107) and Greenfield et al. (2012).
91 Compare here the Egyptian classifier system which has a single class for snakes and warms, see
Goldwasser (2002: 68–69), and discussion below.
92 See the analysis in Goldwasser & Grinevald (2012) of the determinatives of Egyptian scripts as a
classifier system, on the basis of a comparison with the classifier system of Jakaltek.
93 In fact, from a historical point of view, the ancient Mayan logosyllabic writing system shows
astonishing similarities with the cuneiform system. G.J. Selz thanks Alfonso Lacadena with whom
he extensively discussed the different features of the two script systems during a workshop of the
COST A 31 (Stability and adaptation of classification systems in a cross-cultural perspective).
94 A Mayan language of the Q’anjob’alan branch spoken in Guatemala. Originally spelled Jacaltec
(when Grinevald was still called Craig), later written Jakaltek following the establishment of
official alphabets for Mayan languages of Guatemala, and renamed later yet Popti.’ Today usually
called Jakaltek Popti’ but here simply referred to as Jakaltek.
 The Question of Sumerian “Determinatives” 315

of Jakaltek to the study of classifier systems in the context of its comparison with the
Egyptian system can be found in the Appendix of Goldwasser & Grinevald (2012).95

4.1.1 About “noun classifier” systems

As argued early in Craig (1986a, 1986c, 1992) and then further in Grinevald (2000, 2015),
principally and originally on the basis of the Jakaltek system of classifiers, there exists a
“noun classifier” subtype of classifiers, which is to be distinguished from the much better
known “numeral classifier” subtype found, for instance, in East Asian languages such as
Chinese or Thai (see discussion above 3.1).96
The argumentation to make such a distinction relies on a combination of morpho-syn-
tactic97, lexical and semantic characteristics, most of which can be shown to be shared
(even if through another medium) by the cuneiform and Egyptian “determinative” sys-
tems.98 From a structural point of view, noun classifiers stand in a position right next to the
nouns they classify and form with them a morpho-syntactic unit99 (while numeral classifiers
form units with quantifiers and numbers instead, and possessive classifiers with possessive
forms100). From a lexical point of view, they originate in nominal lexical items which come
to be grammaticalized as classifiers, with expected accompanying phonological reductions
and semantic changes. From a semantic point of view, they categorize nouns by their char-
acteristic of essence or nature (for instance human vs. animal vs. vegetal vs. mineral etc.)
rather than by shape – the way numeral classifiers do (as in 1D-long vs 2D-flat vs 3D-
round, etc.).101

4.1.2 Jakaltek as a noun classifier system

Jakaltek classifiers consistently stand directly to the left of the noun they classify and form
a construction unit: classifier+noun, i.e. all classifiers are in pre-position (PR). It is easy to
argue that the classifiers have originated in nouns, and that the classification is done on the
basis of the essence of the items and the material they are made of. Part of the evidence lies

95 The Appendix by Grinevald is entitled “Basics on Classifier Systems” (pp. 46–51­).

96 For overall views of classifier systems see Aikhenvald (2000) and Senft (2000). For numeral clas-
sifiers, see Bisang (1999). The small group that we termed “mensural classifier” in the cuneiform
system was discussed above.
97 “Morphosyntactic » to the extent that the forms belong to some established paradigms and appear
in specific positions as part of certain constructions of the language, independant of the medium of
their expression.
98 Classifiers may surface in different media. For a general discussion of classifiers in Sign Languages
see K. Emmorey (2002), and for specific comparison of classifiers in Sign Language and Egyptian
script system, see Lincke & Kutscher (2012).
99 The (unpronounced) classifiers of the cuneiform and Egyptian scripts appear in similar syntactic
positions.
100 For a discussion of the different morphosyntactic constructional characteristics of the different
types of classifiers see Craig (1992), Grinevald (2000).
101 Talking of the sortal classifiers which categorize the entities themselves, and not the mensural ones
which specify either units for counting or arrangements (see table 3 in 3.2.1.).
316 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

in the existence of what are known as repeater classifiers, classifiers that are a (somewhat
modified) replica of a lexical noun source.102 For all but one of the Jakaltek classifiers the
source nouns have remained identifiable as illustrated below (with classifiers indicated
in caps): (1) with no reduction, but others exhibiting different degrees of phonological
reduction (2) with a case of truncation and (3) cases of fusion in the speech chain103:
(1) a. ix ix b. ixim ixim c. ch’en
ch’en
woman woman CORN corn rock rock

(2) a. naj winaj b. no’ noq’

man man animal animal

(3) a. tx’otx’ tx’otx’ > [tx’otx’otx] b. te’ te’ > [tete’] c. ha’ ha’ > [haha’]
SOIL soil PLANT tree WATER water

Although rare in the Sumerian system, a similar instance of repeater may be the case of
na
4
na = ‘stonestone’ (18).
Jakaltek classifiers come from source nouns that define three levels of categorization:
generic, specific, unique.104 This terminology refers to the size of the category discussed
and not to the relationship between the classifier and host word. An example of generic
classifiers is the one for plants and plant products, to be contrasted to the specific one
for corn and products derived from it, and to the unique one just for dog, while there is
a generic one for all the other animals. This distribution between generic, specific and
unique classifiers is very clear and central to the categorization system of the Jakaltek noun
classifier system (Craig 1986a, Grinevald Appendix in Goldwasser & Grinevald 2012).

4.1.3 The thematic inventory of the Jakaltek system

The analysis of the thematic inventory of the Jakaltek classifier system originally appeared
in Craig (1986a). Table 4 below summarizes the organization of the inventory and shows
first how the two dozen classifiers of the system can be subdivided into two even subsys-
tems. The first one counts a dozen classifiers of social interaction, labelled this way to
specify that they not only organize humans (plus deities and spirits) according to gender,
kinship and prestige parameters, but can also express discursively marked choices. They
allow speakers some freedom of choice to express either positive or negative feelings to-
wards the person classified, indicating special closeness for a non-kin by using a kin clas-
sifier, for instance, or distance from a kin by using a non-kin classifier (Craig 1979: 43–44;
1986a: 269–270). The other subsystem of so-called physical and functional interaction
is basically organized around physical characteristics (form) of entities (whether of the
animal or vegetal domain for instance) with no speaker variation possible, while pinpoint-

102 For the term “repeater” see Allan (1977: 292); Senft (2000: 22); Grinevald (2003: 98); Goldwasser
& Grinevald (2012: 48–49). For the list of lexical sources of all Jakaltek classifiers see Craig
(1979: 45).
103 Examples are from Craig (1979: 45); Grinevald (2002: 1026); Craig (1986a: 256).
104 The Jakaltek system does not have a general or neutral classifier, one devoid of semantic content.
 The Question of Sumerian “Determinatives” 317

Table 4 | Jakaltek classifiers – thematic organizationa

Social interaction Physical and Functional interaction

1. deity male Generic Specific Unique
kumam
2. deity female Animal 13. animal 14. dog
kumi’ b no’ metx’
3. respected human
ya’
Vegetal 15. plant/wood 16. corn
Non-kin te’ ixim
4. male
naj 17. thread
5. female tx’al
ix
6. young male 18. twine
naj ni’an tx’an
7. young female
ix ni7an 19. cloth
k’ap
Kin
8. male
ho’ Mineral 20. rock 22. salt
9. female ch’en atz’am
xo’
10. young male 21. dirt/earth
ho’ ni7an tx’otx’
11. young female
xo’ ni7an
12. infant Natural 23. water 24. fire
unin elements ha’ k’a’

a Table 4 gives a compounded presentation of information found in Craig (1986a :266–267 and 278),
Goldwasser & Grinevald (2012, Appendix 50–51).
b The Jakaltek classifiers are written here in the new official alphabet: hence different transcriptions
from the early source (Craig 1986a) ; so c (as in 1. cumam and 2. cumi7) was changed to k (as in 1.
kumam and 2. kumi’) and all glottal stops written originally with 7 (as in 2. cumi7) are now written
with apostrophe ’ (as in kumi’).

ing the items with culturally important functions either through special class extensions or
through identification of specific and unique classes.105
There is a similarity in the way the classifier systems of both Jakaltek and Sumerian
divide up the world that humans live in and interact with. Admittedly, there is a major

105 Craig (1986a: 263–293) offers cultural explanations for these cases: the specific ones of corn and
weaving for instance as being pan-Mayan and locally essential cultural items, respectively, and
the unique ones including the dog as emblematic of male maturity, salt as essential currency in
commerce in the region and fire as not having any derived product.
318 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

difference between the two inventories in the fact that half of the Jakaltek system is
concerned with the categorization of human entities, with detailed distinctions of status,
sex, kinship and age, which means much more detail than what the Sumerian system
grants to this domain.106 But, on the other hand, the other half of the system, the one
which categorizes the non-human world, is quite similar in both systems, relying on
the categorizing feature of the material essence and consistency of the entities, and
occasionally also on that of their function for humans. So, in both systems, the non-human
classifiers cover similar general domains of the surroundings of the speakers, which for
the sake of comparison of inventories will be divided into animal, vegetal, mineral and
natural domains.107 One noticeable difference in the list of domains is how Sumerian has
developed an interesting set of classifiers of spatial entities like houses, fields, (cultivated
plots), and cities, while Jakaltek – a language spoken in a predominantly rural and
agricultural environment – does not include in its classification such a domain linked
to a notion of functional or administrative locations. Therefore, in Jakaltek, a house as
a building is in the wood/plant class and there is no classification and no lexical item
equivalent of house as a functional place, while villages, market places and fields, while
being named, are not classified at all.108
In terms of size of the classes, both systems are comparable in the contrast between
large classes headed by a generic and smaller, more specific, classes that categorize items
with marked cultural value. For instance, both systems set apart some culturally signifi-
cant food plant from the general class of plants but, interestingly, it is corn in the Jakaltek
Mayan system and barley in the Sumerian one. And while unique classifiers, i.e. classi-
fiers heading a class of just one item, are present in the Jakaltek system, they do not seem
so far to have been detected reliably in the Sumerian system, although some are found in
Ancient Egyptian.109

4.1.4 About animals and illnesses in the Jakaltek system

The categorization of animals is an interesting case for comparison between the three
classifier systems considered. Jakaltek has one very large animal class, headed by a
lexeme that stands for the generic term “animal” in the lexicon. This class includes all
the different kinds of animals, independent of shape or size, domestication or habitat
(quadrupeds: cats, pigs or horses; bipeds: birds, chickens; but also, fish, snakes and bugs).
The class is in fact very large, since it has been further extended to include all animal parts
and products (egg, milk, meat, horns, wool), as well as all manufactured objects, made

106 There are, however, indications that in the early Uruk texts similar differentiations are attested
lexically.
107 Our etic English terminology reflects typical dissection of the world by western societies and
is widely used for discussion in scientific literature, including classifier studies, cognitive
linguistics and ethno-biology (E.g. Brown 1999; Berlin 1992; Hunn 1982). The emic information
is represented, ipso-facto, in the classifier systems discussed in the article, and it naturally fits only
partially the modern nomenclature.
108 The system interestingly only classifies tangible items of identifiable material, leaving nouns
like sun/moon/stars or beer/cocacola and other drinks of unidentified origin, as well as locations
unclassified, see Craig (1986a: 273–274).
109 See Goldwasser & Grinevald (2012: 20).
 The Question of Sumerian “Determinatives” 319

with animal material (hides, sandals, woollen blankets). However, in a very idiosyncratic
way, dogs are excluded from this animal class and are assigned a class of their own, in a
culturally significant manner (considered in 4.1.2-4.1.3 above).
Meanwhile, the Sumerian classifier system shows (see 3.3.1 and 3.3.3) neither a generic
classifier for animal or any similar class,110 but has several specific classes with functional
value. This fact stands in contrast to the Ancient Egyptian classifier system which has
developed a large superordinate category classified by of hide & tail =“(having) hide &
tail” classifier which was extended from a class created around the prototype of (made
of) hide/leather (see below 5.6), ending up classifying the live animal world of Ancient
Egypt (see below).
Another interesting point of comparison between the three classifier systems considered
here is the special treatment they consistently give to illnesses within their classificatory
scheme, although each in its own way. So, while illnesses are not given a specific
class in Jakaltek, they are indeed singled out in that system too, and given importance
in the categorization schema by being assimilated into the class of male/human non-
kin, actually an act of personification through the classifier system which resembles the
Egyptian personification of illness by the classifier of the god Seth (see below111).

4.1.5 Variation across systems

Beyond finding some interesting parallelisms of a typological nature in the way any two
noun classifier systems categorize their worlds, such as the one pointed out between the
spoken Jakaltek system from native America today, and the written Sumerian one from
the Old World in Ancient times, one should not expect any two systems of classifiers to
ever have the same inventory or classification scheme. This is so to the extent that classi-
fier systems develop very locally and are lexically based and reflect local culture. There
will never be two fully matching classifier systems. This can be illustrated by the docu-
mented differences found even between very close languages, such as the sister languages
of Jakaltek, all languages of the same Q’anjob’alan branch of the Mayan family, and all
spoken in the same isolated area of the Cuchumatanes Mountains. Table 5 below shows
a sample of the classifiers of four close sister languages in order to illustrate the kinds
of variation that are found across the otherwise very similar systems. In this sample, the
same conventions used for the Sumerian system earlier apply: classifiers shared by all the
languages are in bold, those found in all but one language in bold italics the rest show dif-
ferent patterns of variation, including cases of a classifier appearing in only one language.

110 To prevent possible misunderstandings: The Sumerian language increasingly developed inter-
mediate and superordinate taxa to name various categories of animals (see above 3.3.3), yet this
phenomenon has to be sharply distinguished from the established set of classifiers.
111 On the cognitive analysis of this process within the Egyptian script as an intrusion of Lakoff’s
“Myth and Belief Principle” into the classifier system, see Goldwasser (2005: 107–109).
320 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

Table 5 | Sample Variations in noun classifiers inventories in the Q’anjob’alan languages of the
Guatemalan Cuchumatanes (after Hopkins 2012)a

Generic Specific Unique Q‘anjob‘al Jakaltek Akatek Chuj

Illness +
as male/
human
illness Ø Ø Ø +
animal + + + +
dog Ø + Ø Ø
tree/ + + + +
wood
maize, + + + +
grain
plant, + Ø + +
herb
cord, vine + + + +
cloth Ø + Ø +
thread Ø + Ø ?
plastic Ø Ø Ø +
earth/ + + + +
clay

a For the actual forms of all these classifiers, see Grinevald (2016: 288, Table 8).

The table shows a selection of core classifiers shared by all the languages (animal, tree/
wood, maize/grain, cord/vine, earth/clay), and how Jakaltek distinguishes itself from
the other languages, by either being the only one having invented a certain classifier dog
or the only one not to have a classifier plant/herb.

4.1.6 Conclusion
The point of this section was to underline the similarities that can be drawn by a
comparison between the Jakaltek and Sumerian systems of classifiers, in particular in the
thematic organization of their respective worlds. Of course, a major difference between
the two classifier systems is that the one in Jakaltek and sister Q’anjob’alan languages
consist of pronounced items, clearly identified in the last decades in new studies of these
contemporary oral tradition languages of the Mayan family of languages spoken in the
Mesoamerican region of the so-called New World. By contrast, the pronunciation of the
Sumerian determinatives analyzed here as comparable classifiers is generally denied, as
they are mostly taken to be a feature of script only (but compare 6.1). As shown, the spoken
classifiers in Jakaltek can easily be linked to lexical nouns still used in the language, even
if they sometimes show reduction when used as classifiers (see above). A similar link to
 The Question of Sumerian “Determinatives” 321

lexical nouns is argued by Selz for the Sumerian determinatives considered as classifiers,
a link repeatedly noted as part of the new information in the consolidated list presented
in section 2 above. Formally, all classifiers in Jakaltek show pre-position like most of the
classifiers in the cuneiform system, and unlike the Egyptian classifier system (see below).
The other parallel drawn between the two systems relates to the process of class
extension, which was shown to be more open and more extensive in Jakaltek than in
Sumerian. In Jakaltek, extensive class extensions of the three basic generic classifiers for
the animal, vegetal and mineral worlds, have created very large classes for all the objects
or products derived from those basic materials. So, for instance the wood/plant class
includes all objects “made of wood”, such as house, pieces of furniture (bench, table, bed)
or kitchenware (spoon, bowls), but also drinks “made of plants” like coffee or alcohol,
demonstrating that the classification is really by essence or natural source of the entities. A
similar process of class extension was detected in the Sumerian system, although at a more
specific categorization level, where classifiers of different materials (wood, reed, copper,
bronze, and stone) mark not only the material itself, but also other objects or products
made thereof. The same phenomenon is well known in the Egyptian system.112 As noted,
the Sumerian system appeals more often to a notion of functionality in its classificatory
schema, as seen with the specifics of the animal classes and the spatial classifier pointing
to man used and built entities (cp. above 3.2.2).

5 A short comparative perspective on the Egyptian classifier system

The Egyptian hieroglyphic script is the other great script system that was born in the
Ancient Near East and timewise was probably parallel to the Sumerian system. Both sys-
tems present a long, documented, diachronic development, from their early stages in the
second part of the 4th millennium to the first centuries AD in Mesopotamia and Egypt. The
Egyptian script has already been shown to boast a richly documented, long-lived system
of classifiers which provides a rare source of diachronic data for the typological study of
classifier systems.113
Unlike the cuneiform script, the Egyptian script recorded throughout its long history
a single language: Ancient Egyptian – an Afro-Asiatic language. A number of other
fundamental differences between the Ancient Egyptian and the Sumerian systems are
noted below.

5.1 Formal characteristics: presence, position, complexity

Although the diachronic development of the Egyptian classifier system can clearly be
traced, it was actually little noticed and hardly studied until the last two decades.114 The
system is not as rigid as the Jakaltek system just described, and in this respect closer to the

112 See Goldwasser (2002).

113 E.g. Goldwasser (2002, 2006a); Goldwasser & Grinevald (2012); Lincke & Kammerzell (2012);
Kammerzell (2015); Lincke (2015a).
114 For history of research and its problematics, see Goldwasser (2006a) and Kammerzell (2015).
322 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

cuneiform one. As a matter of fact, the presence of Egyptian classifiers was not absolutely
obligatory, as it could depend on various factors, such as period of use and materiality,
technical considerations of space, context or writing-form – hieroglyphs, cursive hiero-
glyphs, hieratic – or different textual genres.115
Once the identification of Egyptian “determinatives” as classifiers is acknowledged,
one can turn to their basic structural characteristics. The first one is that all of them are
post-position (PO) classifiers,116 and that unlike other classifier systems – the Sumerian
one included – the Egyptian system exhibits frequent complex constructions of multi-
classifiers. These classifier constructions combine up to 4 or even 5 classifiers that obey
strict ordering principles, as discussed in Goldwasser & Grinevald (2012) and Lincke &
Kammerzell (2012).

5.2 Kind of classifier system

While the cuneiform classifier system is strictly of the “noun classifier” subtype of classi-
fiers, and as such comparable to other noun classifier systems found in world (such as the
Jakaltek presented above in section 4, or others discussed in Aikhenvald 2000), the Egyp-
tian system is of a somewhat different type, until recently not identified in the typology of
classifier systems. It constitutes, interestingly, a rare type of system that classifies at once
nouns, verbs, and sometimes even adverbs or prepositions.117 So while Ancient Egyptian
has many classifiers clearly categorizing nominal entities similar to the Sumerian ones,
such as: human/male , plant , house , etc., there exists in parallel a number of
“event oriented” classifiers that classify lexemes clearly of a verbal nature.
The reasons for the tendency of the Egyptian system to classify both nouns and verbs
may well be related to Ancient Egyptian being a language of the Afro-Asiatic branch, well
known for a particular typological feature of the structure of its lexicon that sets those
languages apart from other languages, such as Indo-European languages: The central and
salient notion of a highly productive consonantal lexical-semantic “root skeleton”.118 The
vast majority of the languages of the world are characterized instead by the fundamental
contrast between the linguistic categories of “noun” and “verb”, a contrast less imminent
in Egyptian.119

115 Here see Lincke (2011) on classifiers in the Pyramid Texts, Shalomi-Hen (2000, 2006) on classifiers
in the Coffin Texts, Werning (2011) for classifiers in New Kingdom religious and lapidary
historical texts and Kammerzell (2015) on tendencies in New Kingdom hieratic classifiers. Also
Sumerian displays a number of variations with and without classifiers (for a collection of examples
see Falkenstein 1949: 34–35). These variations often serve as proof for the claim that Sumerian
classifiers are unpronounced, a pure phenomenon of script.
116 For a very rare example from the Middle Kingdom Coffin Texts, see fn. 7 above.
117 Verb classifiers (to be carefully differentiated from verbal classifiers), see Goldwasser & Grinevald
(2012: 47), such as those of Australian languages, were not identified until relatively recently, see
Schultze-Berndt (2000), Schultze-Berndt & Sagna (2010), for one of the first descriptions of such
systems.
118 For a discussion of the Semitic root from the view point of Historical Semitics see Kienast (2001:
59–68 [with extensive bibliography]).
119 E.g. Goldwasser (2006b).
 The Question of Sumerian “Determinatives” 323

A question that comes to mind immediately is why the Mesopotamian system that
later was also used for recording various Semitic languages and dialects never turned
to the classification of roots (hence of verbs too). One could argue that the reason for
this may well be because the Mesopotamian classifier system discussed above was in
fact devised for Sumerian, a language which is clearly a non-Semitic language, since
no cognates of Sumerian have been identified so far (Edzard 2003: 2–3). Therefore, the
cuneiform classifier system developed in a non-Semitic language devoid of a key notion of
consonantal “root”. As argued above, the cuneiform classifying system maintained itself
as a basically Sumerian system in its essence, and even remained productive until the
1st millennium BCE. However, this argument is challenged by recent studies on Luwian
hieroglyphs which developed independently in Anatolia mainly from the 13th century to
the 9th and 8th centuries BCE. The hieroglyphic Luwian system also shows a mixed system
of noun cum verb graphemic classification.120 As Luwian hieroglyphs record mainly
Luwian, an Indo-European language, the explanation of “root classification” may be not
entirely satisfactory. Moreover, the Chinese writing system also shows a mixed graphemic
classifier system that classifies nouns cum verbs, as has been recently shown by Chen.121

5.3 Number and inventory of classifiers in the Egyptian script

Even without a thorough study, it is very clear that the Egyptian classifier system counts
many more classes than the Sumerian system. An inventory of the overall Egyptian clas-
sifier system was never established in a methodical way in Egyptology.122 The basic “list
of classifiers” available (but not even numbered!) was compiled more than half a century
ago by Alan Gardiner. This list, which he presented for teaching purposes in his grammar
book of Middle Egyptian, is however the fruit of his many years of profound experience
with Egyptian texts and remains, to this day, the standard list in Egyptology.123 The list
contains what he considered to be about 90 “generic determinatives”, although the number
of classes represented in it should probably be considered somewhat smaller, partly for
reasons explained below. Meanwhile, a number of other classifiers should be added to
it, since many logograms could be activated as classifiers under specific circumstances.
Almost all classifiers possess inherently a lexeme (and logogram) status as argued above
for Sumerian and Jakaltek (lexeme only).124 Some classifiers in Gardiner’s list seem to be
variations of a single classifier meaning. These variations are in part due to the high iconic
nature of the hieroglyphic script that enforces in many cases the choice of a single picto-

120 See Payne (in press).

121 See Chen (2016). On verb classification in numeral classifier system, see fn. 131 below.
122 Inventories are in the process of being created for single texts and some corpora, see Werning (in
preparation).
123 Gardiner (1957: 31–33). Borghouts (2010: 49–50) presents a “lightweight” list of less than 70
classifiers, yet the essence of the list as its order is not changed; other shortened similar lists appear
in different grammars. Werning, in a new approach, gives a new combined list of classifiers used
in a few texts studies by him, see Werning (2015: 15).
124 For a recent semiotic analysis of the role Egyptian hieroglyphs may play in the script system, see
Polis & Rosmorduc (2015) with bibliography.
324 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

rial prototype for the representation of a complex generic concept.125 Such a complication
is not found in the cuneiform system, since it lost its iconicity very early.126 Moreover, in
different periods of the Egyptian classifier system, different prototypes may compete for
the representation of the same generic category. Such is the case of the concept god for
which the earliest pictorial classifier is a falcon god (falcon on a standard), later chal-
lenged, from the 5 dynasty onwards, by another prototypical pictorial representation –
th

that of a human manifestation of the divine – which was probably originally that of the
god Osiris.127 The human prototype manifestation implies a clear development in the
conceptualization of the divine that is now presented as a personified god capable of lov-
ing, thinking, being angry and finally having mercy and showing care – all characteristics
of humans, but hardly of a falcon. A third contestant for the representation of the divine is
the banner which is a metonymic, non-specific pictorial representation of the concept
god through its function as “temple banner”. However, the three variations of the god or
divine classifier continue to exist side by side throughout history.128 Next to this example,
there are other classifiers in Gardiner’s list that are either rarely used or head very small
classes with very few members.
Table 6 below is meant to open up a “dialogue” with the new consolidated list of the
Sumerian classifiers presented in section 2.2 and 3.2.1 with Table 3. It displays those
Egyptian classes (as defined by their classifiers) that refer to categories somewhat similar
of those found in the cuneiform script. As is to be expected of classifier systems, there is
no complete parallelism between the two systems, in which categories differ greatly in
range, number of members, centrality, semantic range etc. The suggestion of mensurals
in Egyptian is tentative.129 However, it is hoped that the table provides the reader with an
overview of how the repertoire of the Egyptian and the cuneiform systems compare.

5.4 Prominent Egyptian classes with no parallel in the cuneiform system

There exist two types of classes in the classification system of Ancient Egyptian that are
not found in Sumerian: one classifying events generally realized as verbs, and a few very
general and inclusive “abstract” classes.

125 See Goldwasser (2002).

126 Right from the beginning, the Sumerian writing system embraces all sorts of “abstract” signs; in
other cases, the iconicity of a sign might have been grounded in cultural conventions not traceable
for us. The loss of iconicity was certainly promoted by the most common writing material – clay.
127 See Shalomi-Hen (2000: 98–101; 2006). For the anthropomorphic divine, see Goldwasser
(2002: 113–114). See the “classification game” of the word nTr in the Old Kingdom tomb of
Nyankhnefertem, in Myśliwiec, Kuraszkiewicz & Kowalska (2010: pls. cxiii, cxxxviii).
128 The banner hieroglyph gains popularity in the Middle Kingdom in the Coffin Texts. Its use
declines in the New Kingdom, yet it is well known in royal texts, even sporadically during the
early days of Akhenaten. It becomes popular again in the Late Period, see Shalomi-Hen (2000:
91–92); Goldwasser (2002: 115) and Goldwasser (2006c: 269 [fig.1], and 273–274) on possible
reasons for its use in certain contexts.
129 Here compare the discussion of Werning on ‘grammato-classifiers’ in the Egyptian system, in
Werning (2011: 102–113 and 154).
 The Question of Sumerian “Determinatives” 325

Table 6 | Egyptian classes that correspond to cuneiform classesa

Classes in bold are large classes, with many members. Classes with asterisk* are small-scale classes.

a This list is by no mean inclusive from the point of view of Egyptian classifiers repertoire. Addi-
tionally, all English “names” of categories are tentative. In many cases, it is difficult to reach a
defined translation, a correct “name” for a class, even if the classifier is iconically transparent.
When activated as classifiers, the hieroglyphs distance themselves from their iconic meaning. A
clear example is the hieroglyph sun mentioned in the table. It embraces a class of nouns such as
‘sun’, ‘day’, ‘light’ – with extensions of ‘spend the day’, ‘yesterday’ and verbs like ‘shine’, ‘rise’.
Later it clearly extended to time, in lexemes such as ‘hour’, ‘period’, ‘eternity’ or ‘moment’, e.g.
Gardiner (1957: 485, [N5]), and Faulkner (1962: 1). For discussions on the “name” of categories
see Goldwasser (2002: 13–14) and Lincke & Kammerzell (2012: 67–75).
326 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

5.4.1 Principal event classes130

Considering that actual events cannot easily be captured directly in the pictorial, they are
classified by one of two types of entities: either the body part involved in an action or
the tool used by the agent performing an action. Below are some examples of the most
common classifiers for such categories. The “names” of the categories are of course highly
tentative (categories that embrace a large member of words are in bold):

a. actions – by body part131

movement (walking legs)

action of force (man holding a stick – also in tools below)

‘eat, drink, talk, feel, think’ (man putting something into his mouth)132
see (eye)
smell (nose)
hear (ear)
action of teeth and mouth (tooth)
copulate (phallus)
embrace (embracing hands)

b. actions – by tool
cultivate (hoe)
bind (rope)
write (scribe’s tools)
break, divide (crossed sticks)
cut (knife)
action of force (man holding a stick – see above body parts)
These ‘event classifiers’ that classify roots in verbal function may also classify semantically
related nouns. In such cases the tools or body-parts are necessary central semantic
components in the nominal entity classified. This is the case of the phallus classifier
linked to the action of copulating but at the same time classifying the nouns ‘male’, ‘bull’,
‘husband’ and others. On the other hand, the very widely used event classifier movement

130 For the seminal works on verb-event classifiers in the Egyptian script, see Kammerzell (2015) and
Lincke (2011 and 2015a). On event classes in classifier languages, see Bisang (forthcoming).
131 Chinese numeral classifiers (in spoken language) has been shown lately to classify also verbs. In
this case we see a similar cognitive procedure of classification of an action by a body part involved,
see the examples of fist classifier for the verb ‘to punch’ in Bisang & Wu (2017: 258–259).
132 This classifier represents the deep-structure conceptual metaphor [the body is a container].
In this case it unites material and abstract entities that ‘go in and out’ of the body or reside in the
body (stomach) from the emic Egyptian point of view, such as food, speech or feelings, as well
siA translated as ‘perception’ ‘knowledge’ (DZA 28.913.940 and Goldwasser 2006b: 479–480 for
explanation and discussion). For conceptual metaphors in the Egyptian script see Goldwasser
(2005) and in Egyptological studies in general Nyord (2015). For explanation and definition of
this specific conceptual metaphor in different cultures, and the concept in general, see Lakoff &
Johnson (1980).
 The Question of Sumerian “Determinatives” 327

includes very few “entity nouns” that are not created from verbal roots. One wonders
if it has to do with the fact that this classifier actually seems to refer iconically to the “event
of moving” and not just to a static limb necessary for the action.
Almost all classes show diachronic extensions, e.g. action of force is extended in
the New Kingdom into administration.133 smell is extended to “things you could see on
the ‘nose-face’”, e.g. joy, contempt etc. writing, represented by the scribe’s palette, is
extended to include a word for ‘red’, probably the specific color used on the scribal palette
to be differentiated from other shades of red.134

5.4.2 Overriding generic classes

Already by the end of the Old Kingdom a few large abstract generic classes had become
conspicuous in the Egyptian script system. The most prominent ones are:
document “abstract” later residual135 (default) – (written, sealed papyrus roll)
foreign136 (weapon, typical of enemies – throw-stick)
small/negative (sparrow)137

5.5 Similarities and differences in thematic classes of the Egyptian and

cuneiform systems: some remarks
An important domain where the Egyptian classifier system shows more classes than the
cuneiform is in the animate human classification, where, next to the large superordinate
classes of human/male , human/female , one finds a number of classifiers for more
specific classes such as exalted man , important man , as well as the pejorative
classifier138 dead man/enemy . This large number of classifiers is closer to the detailed
human classification shown in the Jakaltek case above.
Body parts of humans and animals also play a specific role in both systems. In
Egyptian, the flesh classifier embraces a large taxonomic category of flesh and limbs
as well as different sorts of meat, and a similar classifier is also attested in Sumerian.
The classifier has an iconic origin in a sign of hide/leather, a class known also in
Sumerian. In Egyptian it gets by the end of the 3rd millennium an intriguing extension
to be considered next (in 5.6. below). Illness and sufferings related lexemes are usually
classified by gland, negative, or both.139 Yet during the Middle Kingdom certain words

133 Administration was obviously related in the mind of the writers to coercion and use of power.
134 Goldwasser (1995: 70).
135 For “residual” categories in the Egyptian system, see Kammerzell (2015: 1403).
136 For the diachronic development of this classifier, see Allon (2010).
137 See David (2000) on the early development and use of the classifier during the Old King-
dom. For the later dramatic extension of the category into negative, see Kammerzell (2015:
1407).
138 See Goldwasser & Grinevald (2012: 25–26). Considered as “referent classifier” by Lincke &
Kammerzell (2012: 48–50) and Lincke (2015a).
139 See Gardiner (1957: 539–540). This classifier is of high frequency in medical texts.
328 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

pertaining to illnesses take an alternative classification, that of the god seth (see above
3.3.2).140
Vegetal classification is represented in the Egyptian script by the classifiers wood
, tree and the combined classifier 141
, plant and later, rather rarely, wine
. There is a general classifier for all sorts of grain – a measuring container
with some seeds escaping from it.142 However, there is no classification for gardens, and
vegetables get the generic classifier plant , their importance as food products being
unmarked in the Egyptian script system. Incense is not a highly-marked category even
though an incense  classifier is known.
For human productions from plant material, Egyptian has cloth/textile but
wool has no recorded importance in the Egyptian sources and is indeed absent from the
classifier system – pointing to a remarkable difference to the Mesopotamian (clothing)
customs. In edible substances, the bread/cake classifier enumerates dozens of varia-
tions of cakes and bread indicating how the Egyptians were very fond of cakes, as can
be seen in the Onomasticon of Amenemope for example.143 Several kinds of drinkable
substances are also classified: beer is classified by a beer jug 144, later extended to create
a category of jug or generic (drinkable) liquid, beer being the prototypical drink of the
Egyptian society in a way quite similar to the Sumerian situation (cf. no. 16 in the Sume-
rian list above). In comparison, urine which is a non-drinkable liquid takes the water
classifier.145 wine and milk are also marked during earlier periods by a specific jug-
classifier of their own, but are later classified in many texts by the generic (drinkable)
liquid classifier – the generic beer-jug.146 There are also other different containers that
play the role of classifier, such as the coffin/box .
The mineral world is represented with stone, copper, and sand, also for sand-like
derived materials. Yet clay which in the cuneiform system is used for classification of
all sorts of wet earth, mud, etc.147 is unmarked as a specific classifier in the Egyptian
system.148

140 See Goldwasser (2005: 108).

141 For a detailed discussion on the relation of the concepts “wood” and “tree”, see Goldwasser (2002:
39–55).
142 A container that was used for many sorts of measurable grains stands here metonymically for the
grains themselves, see Gardiner (1957: 516–517, U9; U10).
143 Gardiner (1947: 228*–233*).
144 The high iconicity of the Egyptian script made it possible to differentiate various liquids according to
their container form. Oil, beer, wine and milk and other liquids were stored in different containers.
145 DZA 22.577.750.
146 See DZA 21.157.060-090. For the “generic” jug see Gardiner (1957: 530, W22).
147 Note that clay possessed a salient utilitarian value in Sumerian society for both, building and
manufacturing containers – and tablets!
148 The word qAH that carries the meaning, ‘clay’, ‘mud’ also takes the stone classifier. Here a procedure
of chaining, through the shape of the stone, to ‘brick’ and then to ‘clay’ may be reconstructed. See
the discussion in Lincke & Kammerzell (2012: 23–25), and DZA 30.316.620. Another word for
clay is sin, that shows different “lump form” classifiers, see DZA 28.932.220–240.
 The Question of Sumerian “Determinatives” 329

The Egyptian script includes the natural elements wind (represented by a concrete
item, sail)149 and star , as well as sky , a classifier sometimes extended to the abstract
notion above. However, rather surprisingly, a classifier for sun is not attested in the
cuneiform system, while it is a very central member in the Egyptian classification system
and culture. The sun held the most important religious significance in Egypt. It is a high,
superordinate concept that puts together all words pertaining to sun, actions of sun, light
etc. It is mainly sun (and not the moon like in the Sumerian world) that classifies the
category time in Egyptian, as the Egyptian calendar is a sun-calendar and not a moon-
calendar. However, the moon hieroglyph is used as a logogram or classifier in the
words iaH ‘moon’ and Abd ‘month’.150
water as a natural element is a conspicuous classifier in the Egyptian system. It
classifies “water”, “urine”, as well as water bodies such as “sea”, while another classifier
stands for the class of water-way . In the New Kingdom, these two classifiers are
usually combined in a double–classifier construct. The Sumerian system shows a single
classifier for “water-ways, rivers and canals” (see (48) in the list above).
The spatial domain is represented in Egyptian in a number of classifiers which echo
those of the spatial domain of the cuneiform system shown in Table 3, such as the ones
for mountain (27), flat land , field/piece of land (10). The Egyptian
script shows in addition a classifier for road and related words which does not exist
in the cuneiform corpus.151 Man-made spaces are also represented first by the house
classifier – a large class that includes all building, institutions and extends further also
to the notion of habitat to optionally include stable, bird’s nest and lion’s den as well as
tent. The town classifier152 is an early and very productive classifier. As shown earlier,
town is an important classifier also within the cuneiform system. The prominence of
town classifiers in both systems seems to coincide with the fact that these two cultures
were the most advanced, largest urban societies of the Ancient Near East.153

149 The same in the Sumerian system; cp. (32). For a different analysis of this classifier and its relation
to wind in the Egyptian system, see Lincke & Kammerzell (2012:18–19).
150 For the sun classifier during the Amarna period, see Goldwasser (2010). For the moon, see Gardiner
(1957: 486, N11,12). Note, however, that in Sumerian the sign for the (younger) classifier month
itu(d) is a composite sign consisting of UD (“day”) and the number 30, thus explicitly referring
to the administrative calendar developed sometime in the 4th millennium. In this (normalized)
administrative system a year had 12 months of 30 days each. Altogether, we can say the moon
calendar was very important for the Mesopotamians structuring of time.
151 The classifier road classifies regularirly also the adverb “here”.
152 Regulski (2010, 162–163).
153 The definition of “town” in Egypt and Mesopotamia is a topic that reaches out of the scope of
this article. For the problematics of town definition in archaeology in general, see recently Smith
(2016) with bibliography. In Egyptology, see Bietak (1979), Loprieno (2003: 242–246), and
Lincke (2015b) – a view from the classifier system.
330 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

5.6 Animal classification in the Egyptian script – a brief comparative overview

The animal world, on the other hand, is a domain where the classifier system of the Egyptian
script reflects a different cultural attitude when compared to the cuneiform system.

5.6.1 classifier hide & tail – stage 1

From the Old Kingdom onwards, the classifier which carried the iconic meaning hide
or hide (& tail) was used as a classifier for different types of hide. Somewhat later, it was
extended to classify artefacts “made of hide”. All members in this class (stage 1) were
always inanimate, and almost all members that were not a “kind of hide/leather” could be
defined as manufactured (“made of”). 154
Central members155 in this class are the nouns mskA “leather” known since the Old
Kingdom, and dHr (since the Middle Kingdom) both referring to types of treated hides.
During the Old and Middle Kingdom, the word inm used to denote human skin, could also
get the hide & tail classifier.156
The noun xnr “reins” probably a loanword, is linked to the appearance of
the horse in Egypt in the New Kingdom. It is an example (there are many others) of the
continuation of the extension process of the “made of hide/leather” – “manufactured”
relations, the reins being analysed as made of leather and classified accordingly.157

5.6.2 The classifier hide & tail – stage 2

During the Old Kingdom, animals are classified in most cases by their own icon (e.g.
‘dog’ = phonograms – Tsm + dog classifier).158 In these cases, the classifying icon
repeats through iconic depiction the same information given by the previous hieroglyphs
which are themselves used as phonograms. These cases correspond to what is known
in the literature on classifier systems as “repeaters”159, originally classifiers that do not
represent a concept ranking higher in the taxonomic hierarchy and seems to “repeat” the
lexical information, although they generally later come to head themselves classes of
items at a lower taxonomic rank.160

154 As we have seen similar extensions are well-attested in both, the Sumerian and the Jakaltek system
of classification. For an extensive discussion of this classifier see Goldwasser (2002: 57–89) and
discussions and statistics for different animals in Müller (2002).
155 I consider a word a central member in the category if most of its occurrences, or almost all, take a
certain classifier.
156 For mskA, see DZA 24.393.860–24.394.010; dHr, see DZA 31.447.960–DZA 31.447.980. inm
“skin”, “human skin”. (DZA. 20.910.280–300) is a less central member in the category. When
referring to human skin the lexeme inm shows- especially in the medical papyri – the hair
classifier, as part of the concept of “what covers the human being”?
157 Goldwasser (2002: 57–63, 79).
158 See Müller (2002: Appendix II, 25).
159 See above in 4.1.2 with examples in Jakaltek.
160 Already Allan (1977: 292, 295); see also Senft (2000: 22). Similar cases are attested in the cuneiform
system. Occasionally, a syllabically written word is followed by the corresponding logogram, e.g.
ĝeš-túg
PI that is the syllabically written word for “ear; reason, wisdom”/ĝeštug/ is repeated by
 The Question of Sumerian “Determinatives” 331

By the end of the Old Kingdom, the classifier was suddenly extended to the clas-
sification of living animals that “have (?) hide & tail”. In stage 1, as seen above, the hi-
eroglyph of the hide& tail was representative of the concept of hide, the tail being mostly
ignored. Only when came to classify live quadrupeds, the tail did become a meaningful
part of the icon.
The first animal known to get the classifier was the lion, a big carnivore with no
obvious utilitarian qualities. Another example appears on clay tablets recently published
and found in the provincial town of Balat, dating to the end of the Old Kingdom. The
hide & tail is found as classifier in a place-name mw-mA “the water of the
antelope”.161 The antelope species may have played here the lexical role of prototype for
all desert animals that come to quench their thirst in the pools of the oasis. However, the
appearance of hide & tail here, cannot be the due to a reluctance to present the full animal
hieroglyph. Although written in hieratic on clay, the tablets from Balat never refrain from
presenting the names of the various animals with rather iconic repeaters, showing in detail
the differences between various quadrupeds, the difference being especially marked by
their different horns. Yet, a repeater classifier would therefore represent only one sort
amongst the many sorts of quadrupeds that existed in the area and were also attracted to
the pond.
Was it the high iconicity of the script (very different in that respect from the Sumerian
one) that limited generalizations, what pushed the writers in some cases into the use of the
more generic classifiers?
In the Middle Kingdom, the extension into the “animate” animals became a clear
tendency in the script, with dog, cat, mouse, as well hippo and pig, getting the , and
being all recorded also in lapidary hieroglyphic inscriptions as well as in cursive script
and hieratic.162
The schematic diagram below, represents the development and extension of the
category from the Middle Kingdom period to the New Kingdom. The category is extended
by the New Kingdom to include many members that have no hide or no tail, or both. Clear
examples are the scorpion or the turtle. The turtle is a clear fringe member in the category
and its classification oscillates between fish and hide & tail.163

the logogram PI , which standing alone has also the reading ĝeštug) and is iconically the depiction
of an ear. However, in many related instances the syllabically written pronunciation of a logogram is
only partially rendered. Therefore, in cuneiform studies these syllables are perceived as reading help
(matres lectionis) for the logogram and are never considered in terms of classification.
161 Pantalacci & Lesur-Gebremariam (2009: 247). Compare here “animaux sauvages”, Meeks (2012:
525, fn. 74).
162 Here see Müller (2002: 23*–32*). Interestingly, the panther gets the classifier only in the New
Kingdom and mostly in hieratic texts, see (DZA 20.042.590).
163 See for such examples in the lexeme STw ‘turtle’ in three versions in the coffin texts of the very
same sentence – CT V 30f (I am grateful to Niv Allon for this example), and see also Goldwasser
(2002: 68).
332 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

In the Middle Kingdom and in the New Kingdom imaginary animals make part (even
if as fringe members) of the hide & tail category as well.164

Figure 1  | A tentative schematic representation of the mental organization of the Egyptian animal
universe: Middle Kingdom vs. New Kingdom (first presented at « Langage et Cognition »
Issu de l’ACI COGNITIQUE, Paris 2008, based on Müller 2002.).

At this stage, a quick comparison of animal classification in the Egyptian script and the
cuneiform script points to following:
1. The Egyptian hide & tail classifier is not sensitive to the differentiation between do-
mesticated and non-domesticated, or between carnivores and herbivores. Lions, ante-
lopes, mice, donkeys, dogs and cats belong to the same category.
2. Unlike the cuneiform classification, the Egyptian classification of animals in the script is
basically non-utilitarian. It is a sortal classification based on observable characteristics

164 Parallel versions of the Coffin Texts show an alternating classification of the ttSS ‘griffin’
with hide & tail and in the parallel text with divine classifier, see CT V 91b; e. For this
griffin in the Middle Kingdom, see Gerke (2014: 139,16). For the classification of the
ammyt ‘dead devourer’ in the Book of the Dead with the hide & tail classifier, see DZA 21.721.330.
 The Question of Sumerian “Determinatives” 333

of the animals, disregarding their function and or specific relations to man. Members
of the category are those “having hide and tail”. In its essence, this classification
is similar to the Aristotelian category quadruped which observes the characteristic of
“having four legs”. 165 The most dominant members in this category of the Egyptian
script are mammalia (“having mammae”) by the Linnaus classification.166
3. By extension processes, almost all quadrupeds may get classified in the script through
by the New Kingdom.167 Newcomers, such as the horse, first appearing in the
Egyptian texts in the middle of 16th century BCE, are immediately analyzed as and
put into the script with the correct classifier. The only animal that has a kind of “hide
168

& tail” but keeps being classified almost always with a repeater is the crocodile.169
However, even some clear fringe-members, such as the turtle – in the Middle-Kingdom
– and flea and scorpion in the New Kingdom are occasionally classified in this class.170
4. bird and fish which have no hide & tail are kept in separate categories. As
we have seen above, fish and bird also appear as separate categories in cuneiform,
which, in both Egyptian and Sumerian systems, have dozens of members. By the end
of the New Kingdom it seems that the generic category hide & tail starts to be further
extended, to occasionally include also birds – thus moving towards the higher generic
concept animal.
Unlike the Sumerian system, snakes and worms do get classified by the Egyptian
system. The classifier is hosted by words describing all kind and sizes of snakes
and worms – termed sworm in an earlier work by Goldwasser.171 By the end of the New
Kingdom the category hide &tail starts, in rare cases, to include sworm members –
thus confirming the extension of the hide & tail category into a higher superordinate
concept that includes now also reptiles.
However, the Egyptian lexicon lags behind. It is worth noting that the Egyptian
lexicon, unlike the Sumerian one, lacked a lexeme that would refer to all these sorts of
creatures at once. When the Bible is translated into Coptic, the Egyptians resort to using
a Greek loanword – “zὡoν”.172

165 This specific animal categorization is likewise known in the Sumerian lexicon (see above 3.3.3),
but it never entered the classifier system.
166 The word “mammal” is modern and comes from the scientific name Mammalia coined by Carl
Linnaeus in 1758, derived from the Latin mamma (“teat, pap”). All female mammals nurse their
young with milk, which is secreted from special glands, the “mammary glands”.
167 The Egyptian script system always keeps the parallel option to classify an animal by repeater or
unique, or both by repeater and the generic hide &tail. See the tables by Müller (2002: Appendix
II).
168 See Goldwasser (2017). Also all loanwords referring to quadrupeds get the , e.g. Goldwasser
(2002: 67).
169 This may be due to the special place of the crocodile within the Egyptian culture. It may be parallel
to the case of the dog in Jakaltek, Craig (1986a: 281).
170 See examples in (Müller 2002: 42*–44*) and discussion in Goldwasser (2002: 68). Fringe
members are defined as such if they only rarely show the hide & tail classifier.
171 Goldwasser (2002: 57, 68).
172 Crum (1939: 904). Vycichl (1984: 191). I am grateful to Ariel Shisha-Halevy for calling my
attention to this fact.
334 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

5.6.3 The special case of “things moving” (on the earth) – a comparative note
In the Sumerian classifier system, as we have seen in the consolidated list above, bovine
sheep and donkey are three different intermediate categories that organize the animals
according to their utilitarian qualities. In the meanwhile, the Sumerian lexicon developed
a few generic superordinate lexemes for animals such as “things (creatures, moving)
on earth” (see 3.3.3 above), however these organizing principles do not surface in the
classifier system.
The Egyptian lexicon shows a parallel development to the one noted in the Sumerian
lexicon. Egyptian shows, from the Middle kingdom on, a new lexeme mnmnt
“herds” literally “the roaming ones” used in many administrative texts to describe privately
owned herds of different kinds of large quadrupeds.173 These herds were a sort of “roaming
property” and this may be the reasoning behind the semantics of this term in Egyptian.

Figure 2 | “Dismantling” the collective noun “roaming ones” by five(!) classifiers (after Möller 1910: 34).

A playful writing of the word mnmnt with 5 different classifiers174 for different animals
(bull, gazelle, oryx, wild boar (?)175 and ram) seems to point to the long-rooted habit
of the Egyptians of mixing what is regarded in the Western tradition as “desert/steppe
animals” with livestock such as bull, donkey etc, a situation which the five classifiers of
the above lexeme seem to manifest clearly. Recent studies of tomb decoration since the
Old Kingdom have shown very clearly that the Egyptian treated in a similar way desert
quadruped (including hyena) and domesticated quadruped.176
However, in most occurrences the word mnmnt takes the bovine classifier alone, the
prototype that stands for all big quadrupeds of this sort.177 It is noticeable that bovine
is somewhat differentiated from the other members of the category. The fact that
bovine is rather a late comer in the hide &tail group is probably due to its high utilitarian
importance in the Egyptian culture, that kept it apart from other animals.178

173 DZA 24.080.670–700. For a discussion of mnmnt, see Goldwasser (2002: 74–78), and Meeks
(2012: 528).
174 For a lexicographical discussion of different lexemes referring to groups of animals, see Meeks
(2012). He compares the Egyptian lexical categories to the Aristotelian classification of ‘animal’,
not referring to any modern discussions on animal categorization in anthropology, cognitive
linguistics, etc.
175 To be differentiated from Gardiner’s E12, wild boar image after Borghouts (2012: 43, sign E12).
For the pig and wild boar in Egypt, see Vernus & Yoyotte (2005: 556–560).
176 Herb & Foerster (2009); Fitzenreiter (2009).
177 Müller (2002: Appendix II, 21).
178 See detailed discussions in Goldwasser (2002: 58–89). The major intermediate taxon for livestock
and desert animals, awt/iAwt, which is known since the Old Kingdom and its various animal
 The Question of Sumerian “Determinatives” 335

Worth noting is the fact that the horse does not join the classifiers groups of mnmnt,
thereby signifying its different prestigious status in the Egyptian society. As a matter of
fact, unlike the other “herd animals”, horses were not used in pharaonic Egypt as source
for meat or hide products, as far as is known.
However, when birds are concerned, the cognitive extension that includes bird in
higher generic order can be detected in script and lexicon at once. As birds start to push
into the hide & tail category in the classifier system, the same tendency can be detected
for the semantic extension of the lexeme mnmnt. Already during the 19th Dynasty, we find
a rare example of the lexeme mnmnt classified by the bird classifier (!).179 Birds (such
as geese) were probably able to join the fringes of this collective noun category as they
were also tended in “herds” and could be conceived as “moving property” used for man’s
needs.180 In this case the function has overridden the perceptual consideration of wings
and two legs. Here we clearly have a utilitarian analysis that interferes with the perceptual
analysis of form that usually stands in the base of Egyptian fauna classification. However,
all these “herds” were naturally domesticated or partially domesticated.

6 Concluding overview
The main aim of this paper was to review the little-discussed and poorly understood phe-
nomenon of the cuneiform determinatives, the so-called “Sumerian determinatives.” The
main motivation for this endeavour came from outside Assyriology, most directly from the
many insights into the system of Egyptian classifiers, which has received much attention
in recent decades. A comparison with the cuneiform system seemed promising, especially
because in both fields the determinatives had been considered a pure graphemic feature
said to have been established by scribes for the supposed purpose of disambiguating lin-
guistic information. Granted that in both scripts the determinatives indeed fulfil this func-
tion, we have argued here that in addition they fulfil another clear “classifying” function:
that is, they provide additional semantic, pragmatic, and cultural information about the
host words. Using a linguistic approach borrowed from the now well-established field of
classifier studies of oral and signed languages, an earlier study of Egyptian determinatives
had already proposed to reanalyse determinatives as a new type of graphemic classifiers.
This path of research on Egyptian determinatives having already yielded highly interest-
ing results, it had become evident for some time that taking the same approach to analyse
the determinatives of the cuneiform system seemed highly promising.

classifiers (including pig!) is discussed in Goldwasser (2002: 60–72), and also by detailed tables
and statistics by Müller (2002: 13*–18*). See also Meeks (2012: 525–528).
179 See Goldwasser (2002: 75, Nauri decree).
180 In English, the word ‘bird’ carries a strong semantic component of “flying”. The Biblical Hebrew
compound ‫“ בעלי כנף‬the ones that have wings” i.e. “the winged ones” may be more fitting to the
description of the category in Egyptian.
336 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

6.1 Sumerian classifiers in a comparative perspective

In order to understand the cuneiform classifier system, we first reviewed what are consid-
ered as determinatives by various scholars in the field of Assyriology. In a second step, we
built a new inventory of these determinatives, consolidating the widely-diverging lists of
determinatives. After observing that the differences between the various authors were par-
tially explainable by their specific aims, we proceeded to tentatively discuss the individual
determinatives with respect to a number of parameters: their (literal) meaning, position,
lexical origin and use and, in addition, estimates of their (relative) frequency and of the
periods of their earliest attestation. The first result of our observations is that the cuneiform
system of determinatives can correctly be labelled a “Sumerian system” of determina-
tives. The salient reason being the fact, that the overwhelming majority of the cuneiform
determinatives possess a clear origin in Sumerian lexemes. This also holds for newly
introduced determinatives like flesh (14) or star / constellation (29). This conclusion
might be surprising, as Sumerian vanished as a vernacular at the turn of the 3rd to the 2nd
millennium BCE. However, some varieties of Sumerian survived, much like Latin in me-
dieval times, not only in temples and schools, but also as a professional jargon. This use
as a technical language is amply attested in numerous ancient reference works, especially
from the 1st millennium BCE, and has to be studied as a late form of Sumerian, which has
rarely been done until now. In this perspective, the fact that a number of “grammatical
determinatives” originate from frozen Sumerian phrases, e.g., Sumerian numerical ex-
pressions, seemed interesting and prompted us to include them in our discussion about
mensural classifiers. Although the Sumerian system continued to live on in a dominantly
Semitic (Akkadian and Aramaic) speaking region, there are no traces of any influence of
these languages upon the Sumerian classifier system.
Next, we compared the different lists we used in order to get comparative data of what
the individual authors considered as determinatives. The picture is complicated, as the
system of remained productive over time: some determinatives disappear but others are
newly invented. The new determinatives follow the use established at the beginning of the
3rd millennium. However, for the Early and Classical periods (ca. 2600–1750), we were
able to identify 18 core items (Table 2). One of the most striking and well known facts
of the Sumerian system is that its classifiers occur before or after the respective nouns.
This situation partially corresponds to pre-position classifiers in most classifier languages,
including Jakaltek as discussed above. However, it stands in sharp contrast to the post-po-
sition of Egyptian classifiers. The suggested explanation is that the Sumerian phenomenon
is related to a specific feature of Sumerian noun compounding, where both left-headed and
right-headed noun+noun types of compounding are allegedly equally attested.
Typologically, we took note of the coexistence in Sumerian both mensural and sortal
classifiers, which makes it a rare noun classifier “mixed system,” since mensural clas-
sifiers have only been known so far from numeral classifier systems. The domains of
sortal classifiers that classify either by shape (numeral classifiers) or by essence (noun
classifiers) are particularly interesting, in that they are especially valuable for the study
of lexical semantics. In agreement with a central feature of Sumerian grammar, which
 The Question of Sumerian “Determinatives” 337

morphologically marks two genders, human and non-human, also described as animate vs,
inanimate,181 the sortal domain of Sumerian classifiers can be subdivided into animate and
inanimate worlds. We also observed how the size of classes and the level of classification
vary, and how occasional superordinate classes such as cloth/textile (40) are created.
However, generic classifiers (e.g. quadruped, animal) as well as unique classifiers (clas-
ses that include a single member) are not attested in the cuneiform script,182 a major diffe-
rence from the other two systems considered, the Jakaltek and the Egyptian.
The extension of classes is rather well attested and seems to reflect cultural considera-
tions, where, e.g. bull/oxen (20) or sheep (43) classifiers extend semantically to become
intermediate taxa “cattle” or “small livestock.” Highly interesting is also the case of the
donkey (15) classifier which later incorporated horses and even camels. The Sumerian
classification of animals into five classes is remarkable, reflecting their utilitarian and
functional significance – as in the thematic lexical lists, attested already in the earliest pe-
riod of cuneiform writing. Another kind of extension, also well known in the Egyptian and
Jakaltek systems, is the type of extension “made of,” observed with the classifiers wood
(19), reed (13), copper (26), bronze (30), and stone (18). They come to designate not
only the material itself, but objects or products made thereof.

6.2 Comparisons in the present and the past

Not only for morpho-syntactic reasons but especially for their semantic implications the
Sumerian system and the system of Jakaltek, a living Mayan language, show parallels and
cover basically the same entities of the world. Like the Jakaltek and most other contem-
porary oral classifier systems, Sumerian classifies exclusively nominal entities (a “noun
classifier system”). Although showing general resemblance of the semantic repertoire,
­Jakaltek unlike Sumerian is more detailed in categorizing human entities, with distinction
of status, sex, kinship and age playing an important role. On the other hand, the classi-
fication of spatial entities, well attested in the Sumerian classifier system is absent in the
Mayan language. A strong parallel exists in the two systems between the status of corn
classifier in Mayan Jakaltek and the functionally equivalent Sumerian barley (28) classi-
fier. In addition, Jakaltek, like Egyptian but unlike Sumerian, knows three unique classi-
fiers dog, salt and fire. Jakaltek also attests a level of superordinate, generic, classifiers
as in the very extensive animal class. A similar generic animal classifier is unknown in
the Sumerian classifier system.
In order to demonstrate that one should not expect a perfect matching of the classifier
systems, a short comparison to the similar but different noun classifier systems of the

181 On earlier discussions on the various terminologies see Selz (in press b) where it is suggested that
the morphological grid /b/ vs. /n/ actually refers to the general vs. an individualizing notion of the
referents.
182 Of course, this does not mean that generic classification was unknown; the language has elements
to form such abstract nouns; however, they did not enter the domain of (silent) classifiers, the
so-called determinatives, on which we focus here; see various discussion above, and also Selz
(forthcoming).
338 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

Q’anjob’alan languages and the Guatemalan Cuchumatanes was included. One way to
understand such variation between systems of even very closely-related languages is their
dependence on local cultural issues, yet the overall picture of the thematic organization
of those types of classifier systems coincides with what we have observed for both the
Sumerian and the Jakaltek systems.
The Egyptian graphemic system of classifiers was further shown to have several spe-
cifics. That these classifiers are unpronounced is generally accepted. But when compared
with the Mesopotamian multilingual situation, a major difference emerges. The Egyptian
classifier system was used for three millennia in a script which recorded a single, albeit
ever-changing, language. Another specific trait is that, as a rule, classifiers are post-posi-
tioned in the Egyptian script. Moreover, Egyptian classifiers, like the classifiers in the Chi-
nese script and hieroglyphic Luvian (but unlike Sumerian and Jakaltek), operate not only
on nouns but also on verbs. At least two dozen “event classifiers” could be safely identified
in Egyptian. The number of classifiers attested in the Egyptian system is certainly much
higher than in Jakaltek or Sumerian. A precise figure183 during different periods and in
different genres is difficult to establish, due to numerous considerations, including graphic
variations connected to the high iconicity of Egyptian hieroglyphs. Classification of a host
word by several co-occurring classifiers is common in Egyptian but rare in cuneiform and
unknown in Jakaltek. Changes and alternations in classifiers are well known, and reflect
changes or developments in Egyptian concepts, as well as specific contextual and/or prag-
matic considerations. All in all, the Egyptian system is more multi-layered in its inventory
as well as in its highly productive nature.

7 A final word – why classifiers?

At this juncture, it is worth clarifying yet another point, which is that calling Sumerian
determinatives classifiers is not just a way of simply renaming them. The traditional term
“determinative” is neither informative about the role that these signs play in the script, nor
is it productive in relating the Sumerian system to any phenomenon in linguistic typology.
Moreover, the connection to linguistic typology is, in the case of the Sumerian classifiers
system, also corroborated by several points of contact between the script and the Sumerian
language.
The recognition that determinatives are indeed classifiers brings along with it the un-
derstanding of what such systems can “be” and what they can “do”, as shown by the multi-
ple typological studies of such systems in the languages of the world. So, this is not a case
of simple relabelling, but one of identifying a linguistic system, which entails expectations
about finding a number of characteristics already outlined from the study of other such
systems in other languages.  This includes questions about the formal characteristics of
the systems, their origin and path of evolution, their categorization principles – including
the different possible levels of categorization and the possibilities of class extensions – and
their “raisons d’être”. Particularly interesting is how such systems are expected to outline

183 Here compare Werning (2011: vol. I, 326); Werning (1998), for first attempts.
 The Question of Sumerian “Determinatives” 339

the particular classification of the world of the speakers, in a mixture of universal catego-
ries and culture-bound ones. The new analysis of Sumerian “determinatives” as classifiers
brings Sumerian cuneiform classifiers and the language(s) they encode into the domain of
linguistic typology, which studies how the languages of the world function, also through
their various classifier systems.

8 References
Aikhenvald, Alexandra. 2000. Classifiers: A Typology of Noun Categorization Devices, Oxford:
Oxford University Press.
Allan, Keith. 1977. Classifiers, in: Language 53, 285–311.
Allon, Niv. 2007. Seth is Baal – Evidence from the Egyptian Script, in: Ägypten und Levante 17, 15–22.
––– 2010. At the Outskirts of a System. Classifiers and Word Dividers in Foreign Phrases and Texts, in:
Lingua Aegyptia 18, 1–17.
Berlin, Brent. 1992. Ethnobiological Classification: Principles of Categorization of Plants and Animals
in Traditional Societies, Princeton (NJ): Princeton University Press.
Brown, Cecil H. 1999. Lexical Acculturation in Native American Languages. New York: Oxford
University Press.
Bietak, Manfred. 1979. Urban Archaeology and the ‘Town Problem’ in Ancient Egypt, in: Kent Weeks
(ed.), Egyptology and the Social Sciences, Five Studies, Cairo: American University in Cairo Press,
97–147.
Bisang, Walter. 1999. Classifiers in East and Southeast Asian Languages. Counting and Beyond, in:
Jadranka Gvozdanovi (ed.), Numeral Types and Changes Worldwide, Berlin & New York: de
Gruyter, 113–185.
––– forthcoming. Nominal and Verbal Classification – A Comparative Perspective”, in: William B.
McGregor & Søren Wichman (eds.), The Diachrony of Classification Systems, Current Issues in
Linguistic Theory, Amsterdam: John Benjamins.
Bisang, Walter & Yicheng Wu. 2017. Numeral Classifiers in East Asia, in: Linguistics 55(2), 257–264.
Borger, Rykle. 1978. Assyrisch-Babylonische Zeichenliste, Kevelaer, Neukirchen & Vluyn: Butzon &
Bercker/Neukirchner Verlag.
Borghouts, Joris F. 2010. Egyptian: An Introduction to the Writing and Language of the Middle
Kingdom. Volume II: Sign Lists, Exercises and Reading, Leuven & Leiden: Peeters/NINO.
Chantrain, Gaëlle. 2014. The Use of Classifiers in the New Kingdom. A Global Reorganization of the
Classifiers System?, in: Lingua Aegyptia 22, 39–59.
Chen, Yongcheng. 2016. The Prototypical Determinatives in Egyptian and Chinese Writing, in: Scripta
8, 101–126.
Civil, Miguel. 2008. The Early Dynastic Practical Vocabulary A (Archaic HAR-ra A), Archivi Reali di
Ebla Studi IV, Roma: Missione Archeologica Italiana in Siria.
––– 2010. The Lexical Texts in the Schoeyen Collection, Bethesda (MA): CDL.
Cohen, Yoram. 2010. The Glosses of the Emar Lexical Lists: Akkadian or West Semitic?, in: Leonid
Kogan (ed.), The Proceedings of the 53rd Rencontre Assyriologique Internationale, Moscow and
Saint Petersburg, Winona Lake: Eisenbrauns, 813–839.
Craig, Colette. 1979. Jacaltec: Field Work in Guatemala, in: Timothy Shopen (ed.), Languages and
Their Speakers, Cambridge (MA): Winthrop Publishers, 3–57.
––– 1986a. Jacaltec Noun Classifiers, in: Colette. Craig (ed.), Noun Classes and Categorization. Pro-
ceedings of a Symposium on Categorization and Noun Classification, Eugene, Oregon, October
1983, Amsterdam & Philadelphia: John Benjamins, 263–293.
340 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

––– (ed.). 1986b. Noun Classes and Categorization. Proceedings of a Symposium on Categorization
and Noun Classification, Eugene, Oregon, October 1983. Amsterdam & Philadelphia: John
Benjamins.
––– 1986c. Jacaltec Noun Classifiers: A Study in Grammaticalization, in: Lingua 70, 241–284.
––– 1992. Classifiers in a Functional Perspective, in: Michael Fortescue, Peter Harder & Lars
Kristoffersen (eds.), Layered Structure and Reference in a Functional Perspective. Papers from
the Functional Grammar Conference in Copenhagen 1990. Amsterdam & Philadelphia: John
Benjamins, 277–303.
CT= de Buck, Adriaan. 1951. The Egyptian Coffin Texts, Vol. IV,. Chicago: The University of Chicago
Press.
Crum, Walter E. 1939. A Coptic Dictionary. Oxford: Clarendon Press.
David, Arlette. 2000. De l’infériorité à la perturbation: l’oiseau du « mal » et la catégorisation en
Égypte ancienne. Wiesbaden: Harrassowitz.
––– 2007. Ancient Egyptian Forensic Metaphors and Categories, in: Zeitschrift für ägyptische Sprache
und Altertumskunde 134, 1–14.
Deimel, Anton. 1922. Die Inschriften von Fara. I. Liste der archaischen Keilschriftzeichen, Wissen-
schaftliche Veröffentlichungen der Deutschen Orient-Gesellschaft 40, Leipzig.
Denny, John P. 1976. What are Noun Classifiers Good For?, in: Sanford B. Steever et al. (eds.),
Proceedings of the 12th Regional Meeting of the Chicago Linguistic Society, Chicago: Chicago
Linguistic Society, 453–471.
DZA= Digitized Slip Archive, http://aaew.bbaw.de/tla/servlet/DzaBrowser (accessed 10.2016).
Edzard, Dietz O. 2003. Sumerian Grammar, Leiden: Brill.
Ellermeier, Friedrich. 1979. Sumerisches Glossar. Band 1, Teil 1. Nürten-Hardenberg bei Göttingen:
Selbstverlag.
Emmorey, Karen (ed.). 2002. Perspectives on Classifier Constructions in Sign Languages, Mahwah,
New Jersey: Lawrence Erlbaum Associates.
Englund, Robert K. 1988. Administrative Timekeeping in Ancient Mesopotamia, in: Journal of the
Economic and Social History of the Orient 31(2), 121–185.
––– 1998. Texts from the Late Uruk Period, in: Josef Bauer, Robert K. Englund & Manfred Krebernik
(eds.), Mesopotamien: Späturuk-Zeit und Frühdynastische Zeit, Fribourg & Göttingen: University
Press Fribourg/Vandenhoeck & Ruprecht, 13–233.
ePSD = The electronic Pennsylvania Sumerian Dictionary: http://psd.museum.upenn.edu/epsd/index.
html (accessed 10.2016).
Falkenstein, Adam. 1936. Archaische Texte aus Uruk, Berlin: Harrassowitz & Leipzig.
––– 1949. Grammatik der Sprache Gudeas von Lagaš I. Schrift- und Formenlehre, Roma: Pontificium
Institutum Biblicum.
––– 1952. Zu Einem Syllabisch Geschriebenen Emesal-Text, in: Archiv Für Orientforschung 16, 60–
64.
––– 1964. Das Sumerische. Handbuch der Orientalistik. Der Nahe und der Mittlere Osten. Zweiter
Band: Keilschriftforschung und Alte Geschichte Vorderasiens, Leiden: Brill.
Faulkner, Raymond O. 1962. A Concise Dictionary of Middle Egyptian, Oxford: Griffith Institute.
Fitzenreiter, Martin. 2009. On the Yonder Side of Bread and Beer: The Conceptualisation of Animal
Based Food in Funerary Chapels of the Old Kingdom, in: Heiko Riemer et al. (eds.), Desert
Animals in the Eastern Sahara: Status, Economic Significance, and Cultural Reflection in
Antiquity. Proceedings of an Interdisciplinary ACACIA Workshop held at the University of Cologne
December 14-15, 2007, Köln: Heinrich-Barth-Institut, 309–340.
Foxvog, Daniel A. 2016. Introduction to Sumerian Grammar. Internet Publication. (Cuneiform Digital
Library Preprints http://cdli.ucla.edu/pubs/cdlp/cdlp0002_20160104.pdf [accessed 10.2016]).
Gardiner, Alan H. 1947. Ancient Egyptian Onomastica, Oxford: Oxford University Press.
––– 1957. Egyptian Grammar, Being an Introduction to the Study of Hieroglyphs, 3rd rev. ed. Oxford:
Griffith Institute.
 The Question of Sumerian “Determinatives” 341

Gerke, Sonja. 2014. Der altägyptische Greif: von der Vielfalt eines “Fabeltiers”, Hamburg: Helmut
Buske.
Goldwasser, Orly. 1995. From Icon to Metaphor, Fribourg & Göttingen: University Press Fribourg/
Vandenhoeck & Ruprecht.
––– 2002. Prophets, Lovers and Giraffes. Wor(l)d Classification in Ancient Egypt, Wiesbaden: Har-
ras  so­witz.
––– 2005. Where is Metaphor? Conceptual Metaphor and Alternative Classification in the Hieroglyphic
Script, in: Metaphor and Symbol 20, 95–113.
––– 2006a. A Comparison between Classifier Language and Classifier Script: The Case of Ancient
Egyptian, in: Gideon Goldenberg (ed.), A Festschrift for Hans Jakob Polotsky, Jerusalem: Magnes
Press, 16–39.
––– 2006b. On the New Definition of Classifier Languages and Scripts, in: Lingua Aegyptia 14, 473–
484.
––– 2006c. The Essence of Amarna Monotheism, in: Gerald Moers, Heike Behlmer, Katja Demuß &
Kai Widmaier (eds.), jn.t Dr.w: Festschrift für Friedrich Junge, Göttingen: Seminar für Ägyptologie
und Koptologie, 267–279.
––– 2010. The Aten is the ‘Energy of Light’: New Evidence from the Script, in: Journal of the American
Research Center in Egypt 46, 159–165.
––– 2017. What Is a horse? – Lexical Acculturation and Classification in Egyptian, Sumerian, and
Nahuatl, in: Tanja Pommerening & Walter Bisang (eds.), Classification through Times. Classification
from Antiquity to Modern Times – Sources, Methods, and Theories from an Interdisciplinary
Perspective, Berlin: de Gruyter, 45–65.
Goldwasser, Orly & Colette Grinevald 2012. What Are Determinatives Good for?, in: Eitan Grossman,
Stéphane Polis & Jean Winand (eds.), Lexical Semantics in Ancient Egyptian, Lingua Aegyptia
Studia Monographica 9, Hamburg: Widmaier, 17–53.
Green, Margret W. & Hans Nissen. 1987. Zeichenliste der archaischen Texte aus Uruk. Archaische
Texte aus Uruk Band 2, Berlin: Gebr. Mann.
Greenfield, Haskel. J., Itzhaq Shai & Aren M. Maeir. 2012. Being an “ass”: An Early Bronze Age Burial
of a Donkey from Tell es-Safi/Gath, Israel, in: Bioarchaeology of the Ancient Near East 6, 21–52.
Grinevald, Colette. 2000. A Morphosyntactic Typology of Classifiers, in: Gunter Senft (ed.), Systems of
Nominal Classification, Cambridge: Cambridge University Press, 50–92.
––– 2004. 97. Classifiers,  in:  Geert Booij et al. (eds.), Morphology, An International Handbook on
Inflection and Word-Formation. Volume 2, Berlin & New York: de Gruyter, 1016–1032.
––– 2015. Classifiers, Linguistics of, in: J. D. Wright (ed.), International Encyclopaedia of the Social
& Behavioral Sciences, Amsterdam: Elsevier, 1973–1978.
––– 2016. The Jakaltek Popti’ Noun Classifier System: Changes due to Spanish Contact, in: Andrea
L. Berez-Kroeker, Diane M. Hintz & Carmen Jany (eds.) Language Contact and Change in the
Americas. Studies in honor of Marianne Mithun, Amsterdam; Philadelphia: John Benjamins, 273–
295.
Herb, Michael & Frank Förster. 2009. From Desert to Town: The Economic Role of Desert Game in the
Pyramid Ages of Ancient Egypt as Inferred from Historical Sources (c. 2600–1800BC). An Outline
of the Workshop’s Inspiration and Objectives, in: Heiko Riemer et al. (eds.), Desert Animals in the
Eastern Sahara: Status, Economic Significance, and Cultural Reflection in Antiquity. Proceedings
of an Interdisciplinary ACACIA Workshop held at the University of Cologne December 14-15,
2007, Köln: Heinrich-Barth-Institut, 17–46.
Hilgert, Markus. 2009. Von Listenwissenschaft und Epistemischen Dingen. Konzeptuelle Annäherungen
an altorientalische Wissenspraktiken, in: Journal for General Philosophy of Science 40/2, 277–309.
Huehnergard, John. 1996 (21998). A Grammar of Akkadian, Atlanta: Scholars Press.
Hunn, Eugene. 1982. The Utilitarian Factor in Folk Biological Classification, in: American Anthro-
pologist 84, 830–847.
Hyman, Malcolm D. 2006. Of Glyphs and Glottography, in: Language & Communication 26, 231–249.
342 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

Jagersma, Abraham H.. 2010. A Descriptive Grammar of Sumerian. Leiden: Universiteit Leiden,
https://openaccess.leidenuniv.nl/handle/1887/16107 (accessed 10.2016).
Johnson, J.Cale. 2012. Indexical Iconicity in Sumerian belles lettres, in: Language & Communication
30(1), 26–49.
Kammerzell, Frank. 2015. Egyptian Verb Classifiers, in: Panagiotis Kousoulis & Nikolaos Lazaridis
(eds.), Proceedings of the Tenth International Congress of Egyptologists. Rhodes, 22–29 May
2008, Leuven: Peeters, 1395–1416.
Kienast, Burkhart. 2001. Historische semitische Sprachwissenschaft, Wiesbaden: Harrassowitz.
Kilarski, Marcin. 2014. The Place of Classifiers in the History of Linguistics, in: Historiographia
Linguistica 41,1, 33–78.
Krebernik, Manfred. 2013. Die frühe Keilschrift und ihr Verhältnis zur Sprache, in: Nicola Crüsemann
et al. (eds.), Uruk – 5000 Jahre Megacity, Petersberg: Imhof, 187–193.
Labat, René. 19886 (1948). Manuel d’epigraphie Akkadienne, Paris: P. Geuthner.
LAK see Deimel 1922.
Lakoff, George. 1986. Classifiers as a Reflection of Mind, in: Colette Craig (ed.), Noun Classes and
Categorization. Proceedings of a Symposium on Categorization and Noun Classification, Eugene,
Oregon, October 1983, Amsterdam & Philadelphia: John Benjamins, 13–51.
Lakoff, George & Mark Johnson. 1980. Metaphors We Live by, Chicago: University of Chicago Press.
Lincke, Eliese-Sophia. 2011: Die Prinzipien der Klassifizierung im Altägyptischen, Göttinger Orient-
forschungen IV/38, Classification and Categorization in Ancient Egypt 6, Harrassowitz: Wies-
baden.
––– 2015a. The Determinative is Prescribed and Yet Chosen. A Systematic View on Egyptian Classifiers,
in: Panagiotis Kousoulis & Nikolaos Lazaridis (eds.), Proceedings of the Tenth International
Congress of Egyptologists. Rhodes, 22–29 May 2008, Leuven: Peeters, 1425–1434.
––– 2015b. Raumwissen in Hieroglyphen. Die sprachliche Kategorisierung des ägyptischen Ortes
Sile (Tjaru) aus prototypentheoretischer Sicht, in: Kerstin P. Hofmann & Stefan Schreiber
(eds.),  Raumwissen und Wissensräume. Beiträge des interdisziplinären Theorie-Workshops
für Nachwuchswissenschaftler/innen. eTopoi, Journal for Ancient Studies 5, 127–155.
Lincke, Eliese-Sophia & Frank Kammerzell 2012. Egyptian Classifiers at the interface of Lexical-
Semantics and Pragmatics, in: Eitan Grossman, Stéphane Polis & Jean Winand (eds.), Lexical
Semantics in Ancient Egypt, Lingua Aegyptia Studia Monographica 9, Hamburg: Widmaier, 55–
112.
Lincke, Eliese-Sophia & Silvia Kutscher. 2012. Motivated Sign Formation in Hieroglyphic Egyptian
and German Sign Language, in: Eitan Grossman, Stéphane Polis & Jean Winand (eds.), Lexical
Semantics in Ancient Egyptian, Lingua Aegyptia Studia Monographica 9, Hamburg: Widmaier,
113–140.
Linné, Carl von. 1758. Systema Naturæ per Regna Tria Naturæ, Secundum Classes, Ordines, Genera,
Species, cum Characteribus, Differentiis, Synonymis, Locis, Stockholm: Laurentius Salvius.
Loprieno, Antonio. 2003. Is the Egyptian Hieroglyphic Determinative Chosen or Prescribed?, in: Lucia
Morra & Carla Bazzanella (eds.), Philosophers and Hieroglyphics, Turin: Rosenberg & Sellier,
237–250.
McDonald, A. 2007. A Metaphor for Troubled Times: The Seth Deity Determinative in the First
Intermediate Period, in: Zeitschrift für ägyptische Sprache und Altertumskunde 134, 26–39.
Meeks, Dimitri. 2012. La hiérarchie des êtres vivants selon la conception égyptienne, in: Annie Gasse,
Frédéric Servajean & Cristophe Thiers (eds.), Et in Aegypto et ad Aegyptum. Recueil d’études
dédiées à Jean-Claude Grenier, Université Paul-Valéry Montpellier III: Cahiers de l’Egypte
Nilotique et Méditerranéenne (CENiM), 517–546.
Möller, Georg. 1910. Hieratische Lesestücke für den akademischen Gebrauch. 2. Literarische Texte
des neuen Reiches, Leipzig.
Mittermayer, Catherine. 2005. Die Entwicklung der Tierkopfzeichen, Münster.
Müller, Matthias. 2002. Appendix, in: Goldwasser (2002), *iii–*46.
 The Question of Sumerian “Determinatives” 343

Myśliwiec, Karol & Kamil O. Kuraszkiewicz. 2010.  Saqqara IV: The Funerary Complex of
Nyankhnefertem, Varsovie.
Nyord, Rune. 2015. Cognitive Linguistics, in: Julie Stauder-Porchet, Andreas Stauder & Willeke
Wendrich (eds.), UCLA Encyclopedia of Egyptology, Los Angeles. http://digital2.library.ucla.edu/
viewItem.do?ark=21198/zz002k44p6 (accessed 10.2016).
Pantalacci, Laure & Joséphine Lesur-Gebremariam. 2009. Wild Animals Downtown: Evidence from
Balat, Dakhla Oasis (End of the 3rd millennium BC), in: Heiko Riemer et al. (eds.), Desert Animals
in the Eastern Sahara: Status, Economic Significance, and Cultural Reflection in Antiquity.
Proceedings of an Interdisciplinary ACACIA Workshop held at the University of Cologne December
14-15, 2007, Köln: Heinrich-Barth-Institut, 245–259.
Payne, Annick. 2014. Hieroglyphic Luwian: An Introduction with Original Texts, 3rd edition, Wiesbaden:
Harrassowitz.
––– in press. Compound Determinatives, in: J. Cale Johnson, Eva Cancik-Kirschbaum & Jörg Klinger
(eds.), Proceedings of ‘Die Sprache des Bewusstseins und das Bewusstsein von Sprache im Alten
Orient’, Berlin 17/11/2014.
Stéphane Polis & Serge Rosmorduc. 2015. The Hieroglyphic Sign Functions. Suggestions for a
Revised Taxonomy, in: Hans Amstutz, Andreas Dorn, Matthias Müller, Miriam Ronsdorf & Sami
Uljas (eds.), Fuzzy Boundaries. Festschrift für Antonio Loprieno. Volume I, Hamburg: Widmaier,
149–174.
Schultze-Berndt, Eva. 2000. Simple and Complex Verbs in Jaminjung. A Study of Event Categorization
in an Australian Language, Wageningen: University of Nijmegen.
Schultze-Berndt, Eva & Serge Sagna. 2010. Towards a Typology of Overt Verb Classification. Paper
Presented at the Conference Categorizing Human Experience: Classification in Languages
and Knowledge Systems, Final Conference of COST Action A31 Stability and Adaptation of
Classification Systems in a Cross-Cultural Perspective, May 14th–16th, 2010.
Selz, Gebhard J. 1989. Die Altsumerische Wirtschaftsurkunden aus der Eremitage zu Leningrad,
Freiburger Altorientalische Studien 15,1, Stuttgart: Franz Steiner.
––– 1993. Die Altsumerische Wirtschaftsurkunden aus amerikanischen Sammlungen. Freiburger
Altorientalische Studien 15,2/I–II, Stuttgart: Franz Steiner.
––– 1999. Vom ‘vergangenen Geschehen’ zur ‘Zukunftsbewältigung’. Überlegungen zur Rolle der
Schrift in Ökonomie und Geschichte, in: Barbara Bock, Eva Cancik-Kirschbaum & Thomas
Richter (eds.), Munuscula Mesopotamica. Festschrift für Johannes Renger, Alter Orient und Altes
Testament 267, Münster: Ugarit-Verlag, 465–512.
––– 2000. Schrifterfindung als reflexives Zeichensystem. Wiener Zeitschrift für die Kunde des Morgen-
landes 90, 169–200 (review article on Glassner 2000).
––– 2008. The Divine Prototypes, in: Nicole Brisch (ed.), Religion and Power, Divine Kingship in
the Ancient World and Beyond, Oriental Institute Seminars 4, Chicago: Chicago University Press,
13–31.
––– 2011. Remarks on the Empirical Foundation and Scholastic Traditions of Early Mesopotamian
Acquisition of Knowledge, in: Gebhard J. Selz & Klaus Wagensonner (eds.), The Empirical
Dimension of Ancient Near Eastern Studies/Die empirische Dimension altorientalischer
Forschungen, Wiener Offene Orientalistik 10, Wien: LIT, 48–70.
––– 2016. Who is a God? A Note on the Evolution of the Divine Classifiers, in: Paola Corò, E. Devecchi,
N. De Zorzi & M. Maiocchi (eds.), Libiamo ne’ lieti calici, Ancient Near Eastern Studies Presented
to Lucio Milano on the Occasion of his 65th Birthday, Alter Orient und Altes Testament 436,
Münster: Ugarit-Verlag, 605–614.
––– in press a. Scriptura Franca? Zur Rolle einer ideographisch basierten Schrift in einer mehr­sprachigen
Gesellschaft, in: S. Fink, M. Lang & M. Schretter (eds.), Sprachsituation und Sprachpolitik in
Mesopotamien, Alter Orient und Altes Testament, Münster: Ugarit-Verlag.
––– in press b. Reflections on the Pivotal Role of Animals in Early Mesopotamia.
––– forthcoming. A Mesopotamian Path to Abstraction (Melammu Conference Volume).
344 Gebhard J. Selz, Colette Grinevald & Orly Goldwasser

Senft, Gunter. 2000. What Do We Really Know about Nominal Classification Systems?, in: Gunter
Senft (ed.), Systems of Nominal Classification, Cambridge: Cambridge University Press, 11–49.
Shalomi-Hen, Racheli. 2000. Classifying the Divine: Determinatives and Categorisation in CT 335
and BD 17, Wiesbaden: Harrassowitz.
––– 2006. The Writing of Gods. The Evolution of Divine Classifiers in the Old Kingdom, Göttinger
Orientforschungen IV/38, Classification and Categorization in Ancient Egypt 2, Wiesbaden:
Harrassowitz.
––– 2008. The Bearded Woman and the Queen. The Formation and Transformation of Female Divine
Classifiers, in: Carolyn Graves-Brown (ed.), Sex and Gender in Ancient Egypt: ‘Don your Wig for
a Joyful Hour’, Swansea: Classical Press of Wales, 179–189.
Smith, Michael E. 2016. How Can Archaeologists Identify Early Cities: Definitions, Types, and
Attributes, in: Manuel Fernández-Götz & Dirk Krausse (eds.), Eurasia at the Dawn of History:
Urbanization and Social Change, New York: Cambridge University Press, 153–168.
Smoczynski, Wawrzyniec. 1999. Seeking Structure in the Lexicon – On Some Cognitive-Functional
Aspects of Determinative Assignment, in: Lingua Aegyptia 6, 153–162.
Steinkeller, Piotr. 1982. Two Sargonic Sale Documents Concerning Women, in: Orientalia Nova Series
51, 355–368.
Velde, Henk (te). 1967. Seth, God of Confusion: A Study of His Role in Egyptian Mythology and
Religion, Leiden.
Veldhuis, Niek. 2004. Religion, Literature, and Scholarship: The Sumerian Composition Nanše and
the Birds. With a Catalogue of Sumerian Bird Names, Cuneiform Monographs 22, Leiden: Brill.
––– 2006. How did they Learn Cuneiform? “Tribute/Word List C” as an Elementary Exercise, in: Piotr
Michalowski & Niek Veldhuis (eds.), Approaches to Sumerian Literature in Honour of Stip (H.L.J.
Vanstiphout), Leiden: Brill, 181–200.
––– 2014. History of the Cuneiform Lexical Tradition, Guides to the Mesopotamian Textual Record 6,
Münster: Ugarit-Verlag.
Vernus, Pascal & Jean Yoyotte. 2005. Le bestiaire des pharaons, Paris: Agnès Viénot Ed.
Vycichl, Werner. 1984. Dictionnaire étymologique de la langue copte, Leuven: Peeters.
Way, Kenneth C. 2011. Donkeys in the Biblical World, Ceremony and Symbol, Eisenbrauns: Winona
Lake.
Werning. Daniel A. 1998. Frequencies of Hieroglyphic Signs,
Handout: http://hdl.handle.net/21.11101/0000-0000-9DDF-5 (accessed 5.2017).
––– 2011. Das Höhlenbuch. Textkritische Edition und Textgrammatik, 2 Vols. Göttinger Orient­
forschungen IV/48, Wiesbaden: Harrassowitz.
––– 2015. Einführung in die hieroglyphisch-ägyptische Schrift und Sprache. Propädeutikum mit
Zeichen- und Vokabellektionen, Übungen und Übungshinweisen. Humboldt Universität zu Berlin.
http://dx.doi.org/10.18452/13650 (accessed 5.2017).
Yakubovich, Ilya. 2008. Hittite-Luvian Bilingualism and the Origin of Anatolian Hieroglyphs, in: Acta
Linguistica Petropolitana 4(1), 9–36.
Zand, Kamran V. 2009. Die UD.GAL.NUN-Texte Ein allographisches Corpus sumerischer Mythen aus
dem Frühdynastikum, PhD dissertation, Friedrich-Schiller-Universität Jena.
ZATU see Green & Nissen 1987.