Journal of Semitic Studies LVI/1 Spring 2011 doi: 10.

1093/jss/fgq056
© The author. Published by Oxford University Press on behalf of the University of Manchester.
All rights reserved.

SEMITIC ROOT INCOMPATIBILITIES AND

HISTORICAL LINGUISTICS

EULÀLIA VERNET
UNIVERSITY OF BARCELONA

Abstract
This paper focuses on root incompatibilities in Proto-Semitic and
examines the importance of these laws with regard to historical root
reconstruction. As is well known, these rules can only be applied to
verbal roots, not to derivative forms and affixed forms. The impor-
tance of these structural incompatibilities consists, then, in the fact
that they reduce the possible number of combinations of the tricon-
sonantal bases. Excluding onomatopoeic roots and loan words, these
laws of incompatibility are fully regular in the verbal roots (but not
in the nominal ones) and, therefore, do not have exceptions, as in all
phonological laws.
The structure of the Semitic verbal roots is, then, absolutely con-
ditioned by these restrictions of incompatibility. These rules are uni-
versal in character and apply also to the different families of the Afro-
Asiatic and Indo-European languages. The restrictions of
incompatibility are a tool of great importance in the historical recon-
struction of the roots (especially, of the verbal roots in Semitic).

Introduction

Among the characteristic features of Semitic languages, the laws for

the formation of its verbal roots have a special importance. This paper
focuses on root incompatibilities in Proto-Semitic and examines the
importance of these laws with regard to historical root reconstruction.
As is well known, these rules can only be applied to verbal roots, not
to derivative forms.1
1
As we will see further on, in his well-known study on the verbal root incompat-
ibilities in Arabic, Greenberg (1950) determined that a contiguous sequence of two
similar (homorganic) or identical consonants is only possible if one of them is a non-
root morpheme. That is, two homorganic or identical consonants can be found in
positions 1–2 and 2–3, whenever they belong to different morphemes (for example,
root and derivation or inflection affixes); the affixes, therefore, allow the presence of
a homorganic consonant in an adjacent position (see Frajzyngier 1979: 3–4). This
rule has an important role in the historical criteria of verbal root reconstruction.

94051_JOSS-2011-1_CS4ME_01.indd 1 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

To introduce the subject of root incompatibilities, I mention the

most important studies written so far, above all since the 1950s, when
the subject began to acquire more importance. I compare the incom-
patibility restrictions of the Semitic roots and the laws for the forma-
tion of Proto-Indo-European (PIE) verbal roots: as we will see, these
laws are a necessary instrument for identifying the roots etymologi-
cally and formally and for solving possible questions in historical
linguistics (for example, how to distinguish between roots and loan-
words or derivative forms).2

1. Incompatibility Rules and Semitic Verbal Root Structure

An analysis of the consonant structure of the Semitic verbal roots

shows that not all theoretical combinations of consonants are possi-
ble. The verbal root structure can be defined through a series of pre-
cise laws.
The Semitic verbal root, therefore, is a radical morpheme whose
constituent elements cannot be combined arbitrarily but are subject
to specific combinatory restrictions.3 These restrictions in the poten-
tial system of the formation of verbal roots are known, in this special-
ized field, as ‘incompatibility laws’ (see Petrácek 1964b: 133 and

2
For a concise summary of papers on this subject written during the first half
of the twentieth century, see Koskinen (1964: 17). The subject of the root incom-
patibilities has generated a large bibliography. For Afro-Asiatic, see Bender (1978);
Loprieno (1984); Petrácek (1974) and Voigt (1999). For Semitic, see Cantineau
(1950 and 1951–2); D. Cohen (1978); Diakonoff (1970 and 1975b); Edzard
(1991); Eilers (1978); Frajzyngier (1979); Goldenberg (1994); Greenberg (1950);
Herdan (1962); Lipinski (1991); Petrácek (1964b, 1985 and 1987a); Schramm
(1991); Stempel (1999: 73); Ullendorff (1958: 74); E. Vernet (2008); Voigt
(1981); von Soden (1973); Zaborski (1994). As far as Hebrew root incompatibilities
is concerned, see Aescoly (1937–40); Koskinen (1964); Tobin (1990) and Weitz-
man (1987). For classical Arabic, see Colin (1945–8); Rousseau (1984); Schramm
(1962); Weitzman (1987) and Zaborski (1996), among others. A statistical study
on root incompatibilities in Ge¡ez is found in Ambros (1991). For root structure in
Egyptian in relation to Semitic, see G. Conti (1980); Hintze (1947); Petrácek
(1988b); Rocquet (1973); Rössler (1971); Vergote (1975) and Voigt (1999: 366).
As we will see further on, the Indo-European languages also show a clear root
incompatibility; this fact makes them typologically comparable to Semitic languages.
In this regard, cf. Kury¥owicz (1935: 121) and Ammer (1950–2: 211–12).
3
The root must to be understood as a complete phonetic entity, not motivated
by any productive or living derivation process, because the root is the morphologi-
cal and abstract basis of the derivation, that is, the part of a word from which all
the historical forms of the Semitic languages can be derived and explained through
the established internal laws.

94051_JOSS-2011-1_CS4ME_01.indd 2 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

1982: 393). The verbal root structure in Semitic, then, is the feasible
result of the possible combinations between its root morphemes.4
The Semitic languages have specific structural incompatibilities
which reduce the possible number of combinations of a triconsonantal
verbal root. Loans and onomatopoeic roots excluded, these incompat-
ibility laws are fully regular for verbal roots (though not for nominal
ones) and, therefore, like phonological laws (see Zaborski 1994a: 1),
have no exceptions.
Greenberg’s research (1950), based on Arabic data, was decisive in
establishing the set of laws of consonant root incompatibilities,
although this special feature of root formation had been observed by
some Semitists before his paper.5
These laws of incompatibility could be summarized as follows.
Among the Semitic languages, no verbal root with two identical conso-
nants, or with two homorganic consonants in first or second position is
found (ppl roots in Semitic are actually nominal roots or denominative
verbs that do not have any other correspondence outside a specific lan-
guage; exceptions are found only in primary substantives with two con-
sonants, such as Akkadian babum ‘door’; see E. Vernet 2008: §4.8).6
Nor are they frequently found in first or third position (for exam-
ple Akk. ÌasaÌum ‘to need’, karakum ‘to gather’; a large part of these
plp roots, however, must be analysed historically as a result of a previ-
4
Each Semitic language also has specific laws. In Akkadian, for example, two
emphatics cannot be found belonging to the same root, because of the dissimilation
law (cf. Hbr. qa†on < qa†un-, but Akk. qatnum; Ar. ∂aba†a, but Akk. Òabatum).
5
Some of these laws were already observed by the Arab and Hebrew grammar-
ians of the Middle Ages. This tradition was used by nineteenth and twentieth cen-
tury Orientalists, and in fact root incompatibilities have become a question of refer-
ence in Semitic historical linguistics (see footnote 1).
6
With regard to ppl roots, Arabic has the noun dadan ‘toy’, probably a child’s
term, but this is only a nominal root. Ethiopic also has a few examples with an
identical first and second consonant root: these cases are derived from tetraradical
forms of a biconsonantal origin, for example Eth. ssl ‘to leave’ < slsl < *sl. The few
examples of Ge¡ez do not have correspondence in other Semitic languages. Hebrew
has only one example, ddh, a pi¡el form that means ‘to go’. This kind of verb is also
very limited in Akkadian: I have found only three examples of primae geminatae
verbal roots: ÌaÌû, ‘spitting, to vomit’; lullu, ‘to equip with abundance’ and papatu,
‘breaking a small piece’. In the first example, ÌaÌû seems clearly an onomatopoeic
formation (it is a comparatively isolated form). Akk. lullu (<lly) is a denominative
verb derived from lalû ‘abundance’. Finally, the Akkadian verb papatu seems to be
a reduction, with possible previous assimilation, of a plpl root, as the form ptpt ‘to
shatter’ attested by Arabic and Syriac. The almost total absence, then, of ppl verbal
roots in Semitic languages underlines the inflexibility with which the root laws
operate. Similarly, ppl roots in PIE. are very rare: in this regard, see Benveniste
(19623: 171).

94051_JOSS-2011-1_CS4ME_01.indd 3 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

ous reduplication of a biconsonantal root of the type *pl > plpl > plp,
see E. Vernet 2008: §4.8).7
Two identical consonants (redoubled) in second and third position
are frequently found (they are called verba mediae geminatae), but two
homorganic consonants are not.8 Consequently, geminated verbs are the
best known exception to these incompatibility laws of Semitic roots.9
Liquids l and r cannot be found in first and second root position.
Nevertheless, they are found sporadically in first and third position
(cf. Ar. rasila ‘to wave’).

7
In connection with plp roots, Semitic languages have very few examples of
verbs of this kind. As a matter of fact, Proto-Semitic has only one certain plp verbal
root: *ntn ‘to give’ (cf. Phoen. ytn), a possible case of lengthening of a biradical root
through the prefixation of a determinative already fossilized in Proto-Semitic. The
plp roots are usually of a nominal origin: curiously, Proto-Semitic has a large
number of nominal roots with identical first and third radical: *sms ‘sun’, *bwb
‘door’, *†l† ‘three’, *nwn ‘fish’; *tÌt ‘inferior’, *srs ‘root’. With regard to the verbal
roots, some of them have a denominative origin (for example Akk. kanaku < kun-
nukku ‘stamp’; salasu < salas ‘three’) and the others are isolated examples of an
individual language and therefore do not belong to common Semitic (Akkadian has
secondary formations with a plp structure as a result of an assimilation at a distance
between the first and third radicals: sps, < sps, sapasu, sapaÒu, sabaÒu: ‘wrapping’ and
sns, < sns, s/sanasu, also sanasu?: ‘to put’.
8
Apart from w and y, there are four sections of homorganic consonants (see
Greenberg 1950): the first section is formed by the former consonants (glottal,
pharyngeal, post velar and velar series: ¿, h, Ì, ¡, x, g, k, g, q). The second one, by
liquid consonants (r, l, n); the third one, by leading consonants (sibilant, dental and
interdental series: ∂, s, s, s, z, Ò, t, †, d, †, ∂, tÛ ) and finally the fourth one, by the
labial ones (p, b, m). The consonants of each section can be combined freely with
those of the other sections when forming the triradical verbal morphemes.
9
The mediae geminatae roots are actually the only ones in which the same con-
sonant occurs in second and third position. A usual way to lengthen a biconsonan-
tal root into a triconsonantal structure is, then, the gemination of the second radical
of the root: this is the case of the verba mediae geminatae, which in many cases have
expressive or iterative connotations, as for example *¡zz ‘to be strong’ (> Ar. ¡azza
‘to be strong’, Eth. ¡azzaza ‘to be strong’, Akk. ezzu ‘to be angered’, a denominative
verbal root possibly: cf. ¡izzu ‘strength’, Hbr. ¡oz ‘strength’); *sbb ‘to go round’ (>
Hbr. såbab, Eth. sabba, sababa, Ugar. sbb ‘to turn round’). In the case of verba
secundae geminatae, the biconsonantal root is clear for two reasons: due to the exist-
ence of irregular forms, on the one hand, and because the last radical can alternate
with other consonants (known as complements or Wurzeldeterminativa, see Dia-
konoff 1988: 49 and Stempel 1999: 73), for the other one (cf. for example the
following root variations: Akk. *dbb ‘to talk’ and Hbr. *dbr, or Akk. *sll ‘to capture’
and Ar. slb, etc; there is no reason to see an assimilation or a dissimilation of the
third radical in these cases). On this subject, see Eilers (1987: 514), Macdonald
(1963–5: 78–9), Moscati (1969: §16.122–16.127), Stempel (1999: 73), Voigt
(1981: 149 and 155) and del Olmo (2008: 2), among others. For pll roots in Egyp-
tian, see Greenberg (1950: 181).

94051_JOSS-2011-1_CS4ME_01.indd 4 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

The Semitic verbal root structure, then, is absolutely conditioned

by these incompatibility restrictions which, due to their universal
character, have also been applied to different Afro-Asiatic families.
These restrictions have been treated from the point of view of infor-
mation theory and from the point of view of the Obligatory Contour
Principle (OCP).10 These approaches have been useful in the recon-
struction of historical roots.
The restrictions in the combination of phonemes are universal in
character and can be useful as the basis for a specific statistical typol-
ogy. These incompatibility laws are, thus, detectable in a great
number of languages, such as the Indo-European ones,11 and must
be used in a historical reconstruction of a protolanguage, as is the
case, for example, for the Proto-Indo-European12 and Proto-Semitic
root reconstruction. In spite of the universal character of these laws,
root incompatibilities can be an effective criterion to establish differ-
ences and analogies between languages.

2. On Root Formation Laws and the History of Research

The analysis of incompatibility laws that can be found in the verbal

roots shows the importance of these laws for determining the etymol-
ogy and reconstruction of the Semitic roots — see Goldenberg (1994:
37) and Vergote (1975: 198); furthermore, the universal character of
these laws is very useful in the historical description of the Proto-
Semitic verbal root (see Petrácek 1964b and Goldenberg 1994: 54).
With regard to the origin of the incompatibility rules, some schol-
ars claim that Landsberger was the first to identify them, in his paper
‘Die Gestalt der semitischen Wurzel’ (1935). Nevertheless, many of
these rules had already been observed by the Arab and Jewish gram-

10
According to this principle, two contiguous elements of a melody cannot be
identical. This rule, then, has a great number of applications — especially regarding
the restrictions of the two first radical consonants, as we will see further on — and
is useful to explain the structure of the root, its diachronic development and its
morphological realizations.
11
For example, in the reconstruction of Indo-Iranian — see Kury¥owicz (1964:
13), Meier-Brügger (20028: 138, §L 347 2) — where Bartholomae’s law was very
important when making the reconstruction of the roots.
12
See Bender (1978: 47) and Jucquois (1970–2). For a study of the phono-
logical structure of the Indo-European root, see Petrácek (1982). As for the subject
of the root incompatibilities in PIE., see Ammer (1950–2: 211–12). With regard
to the incompatibilities in the Wurzelerweiterungen of PIE, see Ammer (1950–2:
202 and 205).

94051_JOSS-2011-1_CS4ME_01.indd 5 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

marians of the Middle Ages (see Greenberg 1950: 163 and Rousseau
1984).
In the twentieth century, and especially from the 1950s onwards,
many Semitists have been involved in this field.13 The most impor-
tant was Cantineau (1946), who, in his ‘Esquisse d’une Phonologie
de l’Arabe Classique’, gave a solid theoretical explanation of the pho-
netic and phonological incompatibility of the roots.
Three years later, Greenberg (1950) made a systematic and rigor-
ous study of the phonological root incompatibilities in Semitic lan-
guages. In his paper ‘The Patterning of Root Morphemes in Semitic’,
he worked with Arabic data (consulting two dictionaries: Lane 1863
and Dozy 1881), but also with other Semitic languages. In this pio-
neering work, Greenberg performed a statistical study, from some
3775 triconsonantal verbal roots, of the possible combinatory fre-
quency in Arabic based on each of the possible consonant pairs in
each of the positions that they can occupy in the root, that is, 1–2,
1–3 and 2–3 (see Greenberg 1950: 163).14
An alternative statistical study for the positions 1–2 only was pro-
posed by Kury¥owicz (1973), who, unlike Greenberg, excluded the
roots with semivowels or geminated consonants and admitted denom-
inative verbs of the basic stem. Koskinen (1964) applied the same
statistics to Hebrew: he included nominal roots and excluded mediae
infirmae roots.
Greenberg’s paper (1950) was actually one of the first studies to
investigate this question scientifically. The conclusions that are most
relevant to Proto-Semitic reconstruction are:

I. Consonants of each section (back consonants, liquids, front consonants

and labials) can be combined freely with those of any other section in the
formation of triconsonantal verb morphemes.
II. Different consonants of the same order tend not to appear in the same
triconsonantal verb morpheme, except that: a. In the section for back
consonants, the velars (k, g, q) occur freely both with the pharyngeals (Ì, ¡)
and the laryngeals (¿, Ì); b. In the front section, sibilants occur fairly freely
with the dental stops t, d and †. In 1–2 position, the sibilant always pre-
cedes the dental (rule of metathesis).

13
See e.g. Aescoly (1937–40) with regard to Hebrew and Aramaic.
14
‘The key position in the present study is accorded to Arabic because of the
abundance of lexicographical information and the relative archaism of its phonological
structure. The composition of 3775 verb roots was investigated based on the lexicons
of Lane and Dozy. The results of this study are set forth in the accompanying tables
1, 2 and 3 describing the patterning in positions I–II, and I–III, and II–III respectively’.

94051_JOSS-2011-1_CS4ME_01.indd 6 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

III. The above statements can only be applied to the verbal root mor-
phemes. Substantival morphemes frequently violate them. For example, of
the numerals from one to ten, no less than four transgress the usual rules
applicable to verb morphemes (¿Ìd ‘one’, †l† ‘three’, sd† ‘six’, ts¡ ‘nine’).
IV. There are no Proto-Semitic roots with identical consonants in the
first and second positions15 and probably none with identical consonants
in the first and third positions. On the other hand, identical second and
third consonants are very common.
V. Two similar or identical consecutive consonants occur only if one of
them is a morpheme not belonging to the root. That is, two homorganic
consonants are possible in positions 1–2 and 2–3 whenever they belong
to different morphemes, as for example, root morphemes and affixes of
derivation or inflection morphemes; the affixes, therefore, allow the pres-
ence of a homorganic consonant in an adjacent position, separated by a
vowel or not (see Frajzyngier 1979: 3–4).
VI. There are no Proto-Semitic verbal roots with liquids in the second
and third positions. In positions 1 and 3, roots with initial n and final l
can occur in Semitic languages and, therefore, a large number of these
roots can be reconstructed for Proto-Semitic.
VII. The sibilants and interdentals are not common in Proto-Semitic:
there are no Proto-Semitic verbal roots with ∂ and sibilant or with ∂ and
interdental. Dental plosive d combined with sibilants is not common.
VIII. A definite number of roots can be reconstructed with one sibilant
followed by a dental in positions 1+2, but not with a dental followed by
a sibilant. Nor can a root with an interdental consonant z followed by a
dental in root positions 1 and 2 be reconstructed for Proto-Semitic.

Two years after the publication of Greenberg’s paper, von Soden

dedicated a section to the incompatibility rules in Akkadian verbal
roots in his grammar Grundriss der akkadischen Grammatik (1952;
see GAG: §51).
In the 1960s, Herdan wrote his paper ‘The Patterning of Semitic Ver-
bal Roots Subjected to Combinatory Analysis’ (1962) and subjected the
possibilities of combination of the Semitic verbal roots to a mathematical
analysis. His conclusion (1962: 266) is clear: of the 6332 hypothetical
triconsonantal combinatory morphemes (following the rules of triradical
verbal roots formation in Arabic), only 3775 (that is, 60% of all the
triradical combinations) are used in the texical formation of the root.16

15
This is also the case of PIE. See Benveniste (19623: 171).
16
‘By applying combinatory mathematics in connection with the rules of forma-
tion of Arabic tri-literal verbal roots, we thus arrive at 6332 such morphemes. This

94051_JOSS-2011-1_CS4ME_01.indd 7 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

Macdonald applied Greenberg’s study to the practical part of his

research. Through the comparison of those triconsonantal verbal
roots of Arabic with a geminated radical of the root, he reached the
conclusion that all the Semitic roots have a biconsonantal origin
(1963–5: 84).17
Petrácek (1964) related Greenberg’s root incompatibilities with
information theory (Informationstheorie) previously applied by
R. Jakobson and M. Halle (1956) in their binary phonological anal-
ysis of morphemes in contact (which acts by opposition), which is
universal for all languages. Petrácek’s aim in this paper is to clarify
the phenomena of root incompatibilities linguistically (not only to
mention the laws concerning root incompatibility, as Greenberg had
done in his 1950 study).
As a conclusion of his work, Petrácek (1964: 138) notes the impor-
tance of Informationstheorie in the reconstruction of the Semitic root.
Like Greenberg (1950: 178, §3), he points out that the incompatibility
laws apply in contact position 1–2 or 2–3, but they cannot be applied
between the first and the third radical (position 1–3). Hence, the con-
tact position between the constituents of the root is of high importance,
because each element is conditioned by the one that precedes it.
In the early 1970s, Kury¥owicz (1973) developed Cantineau’s pho-
nological incompatibility principles and used Greenberg’s results,
refining them with still more rigorous principles. He excluded mediae
infirmae roots and denominative verbs, but he did not exclude the
onomatopoeic roots.
Bender (1978) directed his research to the incompatibilities of the
verbal roots in Afro-Asiatic; using the theories of Greenberg (1950)
means that even with Greenberg’s exemptions from free combination applied strictly
— which in reality they are not — there would be about 6332/3775 = 1.68 times
as many roots as we actually know. Apart from the rather unlikely possibility of
verbal morphemes having been discarded with time to such an extent as to account
for the difference 6332 – 3775 = 2557, or that the rules of consonant combinations
as given by Greenberg are not exhaustive, the conclusion may be that for positive
reasons only 3775/6332 = 60% of all admissible tri-literal combinations are used in
word formation’.
17
‘Whatever one may decide — the evidence from our Tables, the obvious
process involved in the development of the quadriliteral, the evidence of the
derived verb — all these suggest with some force that there was once a biliteral
verb morpheme, possibly many of them, which in the course of development in
many regions over millennia brought about a proliferation of triliteral verb mor-
phemes, and some triliterals in turn created quadriliterals in exactly the same way,
by extension, and it is to be noted that the quadriliterals and the few quinquelit-
erals are explained, as D is, mainly by gemination, or addition by insertion of a
new element’.

94051_JOSS-2011-1_CS4ME_01.indd 8 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

he found that isoglosses of incompatibilities also occur in the Afro-

Asiatic languages, where only the Chadic and Omotic families repre-
sent the weak points. He notes that Greenberg’s Semitic incompati-
bility laws can also be found in the Indo-European languages (1978:
10),18 although they are treated from a different point of view and
using a distinct terminology.19
Since the 1990s, the phenomenon of consonantal incompatibility
in Semitic has been studied from the point of view of the Obligatory
Contour Principle (OCP), formulated by linguists in the 1960s (see
del Olmo Lete 2003: 75), and already observed by Petrácek in his
Semitic linguistics research (1964b). According to this principle, two
contiguous elements of a melody cannot be identical.20 This principle
has a great number of applications, especially for those restrictions of
the two first radical consonants (see Edzard 1991: 403, n. 7),21 which

18
‘Regarding the first form of negative scoring, Grover Hudson (personal com-
munication) remarked that the true isomorph may be that of having triliteral roots
at all. The pattern for all world languages seems to be mono- or bi-consonantal
lexical roots: Semitic coped with the rising information load of advancing civiliza-
tion by adding a consonant, while other languages added a vowel. Regarding the
second criterion, the problem of how to interpret consonant clusters arises as a seri-
ous problem, especially in such languages as Proto-Indo-European, where only 23
out of 302 verb roots investigated were free of clusters. If Indo-European and Afro-
asiatic are indeed genetically related, how does one account for the consonant clus-
ters in nearly all PIE verb roots? Is a root of form CCVC or CVCC in PIE to be assumed
to be related to one of form CVCVC in Afro-asiatic, or is the development of clusters
a purely secondary phenomenon? The fact that so many of the clusters are homor-
ganic nasal or sibilant compounds seems to indicate the latter”.
19
See on this sense Petrácek (1982: 396–7): ‘Dans ce sens les langues indoeu-
ropéennes et sémitiques se dressent les unes à côté des autres et le trait sémitique
des incompatibilités spécifiques dans la racine ne peut pas détacher ces langues des
autres, le type des incompatibilités sémitiques dans la définition de J. H. Greenberg
(position dans les sections de phonèmes destinée par la place d’articulation) étant
détectable même dans les restrictions de la racine dans les langues indoeuropéennes
traitées des différents points de départ et sous une autre terminologie, sans parler
des restrictions directement dans le cadre de la même définition’.
20
With regard to Arabic, see Goldenberg (1994: 54), ‘In Arabic, OCP is employed
in arguing that gemination is dictated by the “prosodic template” (McCarthy Pro-
sodic Theory 383–389); also for explaining some restrictions in root-structure, as
quoted from J. Greenberg, viz. the impermissibility in Arabic of roots with 1st and
2nd radicals identical (like *ssm) and of roots with adjacent homorganic consonants
unless they are identical. If Arabic roots are subject to the OCP, and all autosegmen-
tal spreading is rightward, then, as McCarthy suggests, mediae geminatae can be
described as outspread biradicals and the incompatibility rules simplified’.
21
‘As for Semitic, the OCP has been successfully applied to co-occurrence restric-
tions within the root (C1 ≠ C2; tendency to avoid two adjacent homorganic radicals)
as well as to phenomena of ‘antigemination’ in Tiberian Hebrew (cf. McCarthy

94051_JOSS-2011-1_CS4ME_01.indd 9 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

are useful to explain the root structure, diachronic development and

morphological realizations (see Goldenberg 1994: 54).22
Bohas (1990b, 1991 and 1993c and 1993d) analysed the effect of
the OCP on the structure and changes in the Semitic roots and lexi-
con, especially in Syriac. His internal methodology of reconstruction
and conclusive results are useful for applying a comparative method
of historical reconstruction in Proto-Semitic.
In the 1990s, Zaborski’s research into radical incompatibilities in
Proto-Semitic is of great interest. The aim of his paper ‘Exceptionless
incompatibility rules and verbal root structure in Semitic’ (1994) is
to point out that, loans and onomatopoeic roots excluded, the rules
of incompatibility are fully regular and, therefore, have no exceptions
(1994: 1).23 As for the ‘exceptions’ mentioned, they must be ono-
matopoeic roots, or loan words or dialectal and late developed roots.
The author highlights that etymological research is necessary to find
the exceptions mentioned.
Zaborski uses the laws of root incompatibility to find possible solu-
tions related to questions of historical reconstruction (Zaborski 1994:
13); for example, in the reconstruction of the Proto-Semitic phono-
logical system, Zaborski proposes that [g] and [g] must be simple
allophonic variants of the same phoneme in Arabic and in Proto-
Semitic as a result of the incompatibility of /g/ with /k/ (1994: 2).
With regard to the glottal or pharyngeal realization of emphatic
consonants in Proto-Semitic, Zaborski also proposes an allophonic
solution to explain the different dialects of Proto-Semitic (Zaborski
1994: 11), because of the original incompatibility of the emphatic
consonants with a following /¿/ on the one hand, and the incompat-

1986 for details). An example for the exceptional case in Tigrinya is the transition
from the root *smsm to the root ssm. Examples of the language game in Amharic
include the expansions bet → bayt¢t ‘house’, b¢dda → bayd¢d ‘to make love’ etc.”.
22
‘On the other hand, the conclusions McCarthy has drawn from the “Obliga-
tory Contour Principle” are probably connected with the “Principle” itself as
adapted for morphophonological analysis. “The Obligatory Contour Principle”
(OCP) requires that multiple occurrences of a consonant in the stem be represented
by a single element of the root melody, so the root underlying, say, reconstructed
bazzaz (…) is /bz/. The second consonant of this root is spread to fill available slots
of the CV skeleton” (McCarthy Chaha 212). The same is assumed with regard to
the vowel melody’.
23
‘All the writers on the subject, including Cantineau and Kury¥owicz, mention
a number of exceptions to the incompatibility rules and although it has been rec-
ognized that onomatopoeic and borrowed roots may violate these rules, nobody has
asked the question whether apart from these exceptions the rules are exceptionless
or not’.

10

94051_JOSS-2011-1_CS4ME_01.indd 10 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

ibility of the other consonants (with the exception of (h/, /Ì/, /¡/, /g/
and /q/ with a following /¿/) on the other. All this suggests that
Proto-Semitic originally had a kind of glottalization as a non-distinc-
tive feature, although this does not imply that it was the only dif-
ferentiating feature of the emphatic consonants. Zaborski believes
that there must have been some kind of pharyngealization as well.
The most important feature is that there is no phonemic opposition
between glottalization and velarization (or, stricto sensu, pharyngeali-
zation) in the Semitic languages, and, therefore, this opposition can-
not be reconstructed for Proto-Semitic.
Zaborski concludes that the velarization and pharyngealization of
emphatic consonants in Proto-Semitic is a question that, from the
phonemic point of view, cannot be answered: some dialects of Proto-
Semitic must have some kind of glottalization as a non-distinctive
feature, while other dialects of this protolanguage did not have this
feature (Zaborski 1994: 11).
In his paper, then, Zaborski (1994) reviews the results reported by
Kury¥owicz (1973), and focuses especially on those cases involving
incompatibility exceptions of R1 and R2 in Arabic. The author con-
cludes that the exceptions that break the rules of the root incompat-
ibilities are due to the following three factors (1996: 654):

a. Late variants (for example, forms of an eighth original conjugation

with an infixed -t-)
b. Loan words from other languages, or from other Arabic dialects, where
the laws of root incompatibility are different (for example, some of the
roots described by the Arab lexicographers as ‘Yemenite’)
c. Onomatopoeic words, as well as new roots, created by poets for expres-
sive purposes

None of the cases mentioned above can strictly be exceptions, because

they are derivative morphemes (especially in the first and second
cases) and not root morphemes.
Kienast (2001: §51.1) has applied Greenberg’s classical laws of
incompatibility to the verbal roots and to the primary adjectives of
Akkadian: this case provides more pragmatic evidence of the univer-
sal character of these laws.
We have seen that the subject of root incompatibilities in Semitic
has been studied in depth since the second half of the twentieth cen-
tury. As we have observed, the restrictions in the combination of
phonemes (linguistically speaking, these restrictions are universal) are
very useful because they make it possible to reconstruct the roots and
11

94051_JOSS-2011-1_CS4ME_01.indd 11 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

solve questions in historical linguistics, such as, for example, the

reconstruction of original roots and the identification of derived
forms (as we have already seen, two homorganic consonants occur in
positions 1–2 and 2–3 only if they belong to different morphemes)
and the reconstruction of the Semitic phonological system (such as
the problem arising from the glottal or pharyngeal realization of
emphatics in Proto-Semitic).24

3. On Root Incompatibilities in Proto-Semitic and in

Proto-Indo-European

As in Semitic, Proto-Indo-European roots were restrained by a set of

dissimilations and assimilations. Not all the consonants of the Proto-
Indo-European phonological system can be found in the root with
the same frequency. According to the glottal theory,25 the restrictions
are due to three factors:

a. two identical consonants cannot occur in positions 1 and 2

b. dissimilation of a sequence of two consecutive glottals
c. assimilation of a sonority and no sonority feature
24
For example, when primary roots should be reconstructed and derivative
forms should be identified. In this connection, see Frajzyngier (1979: 3–4): ‘There
is independent evidence to support this claim in any of the Semitic languages; an
examination of affixes, either inflectional or derivational shows that its occurrence
is not inhibited by the presence of a homorganic consonant in an adjacent position,
whether separated by a vowel or not. A random check in Arabic shows the follow-
ing: t-∂ in 1–2 position: no occurrences within one morpheme (Greenberg, 1950:
164), but ta-∂arabu; m-f in 1–2 position: within one morpheme, no examples
(Greenberg, 1950: 164), but miftaÌ “key” (prefix mi-); d-t in 2–3 position: within
one morpheme, no examples (Greenberg, 1950: 166), but lidat ←walada (suffix
–at). It is therefore possible to have two consecutive homorganic consonants when
they belong to different morphemes, and so on phonological grounds it has been
shown that the geminated forms are probably derived. But this argument can be
rejected if one shows on independent grounds that in phonological rules two iden-
tical consonants do not behave as if they were two similar consonants. For the
particular constraint in Semitic, one would have to show that for some reason two
identical consonants behave as if they were not homorganic. The reason I mention
independent grounds is because one can use the facts discussed in this paper as
evidence for the non-operation of the constraint’.
25
The glottalic theory holds that the voiced aspirated stops reconstructed in PIE
were actually simple stops; the voiced stops were voiceless glottalized and the simple
voiceless were also simple voiceless. With the traditional phonetic reconstruction of
the three series, this set of restrictions does not make sense. However, the new
definition that propitiates the glottalic theory explains the restrictions perfectly.

12

94051_JOSS-2011-1_CS4ME_01.indd 12 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

With regard to the first point, it can be observed that this restriction
is applicable to Proto-Semitic — in Egyptian something similar
occurs, see Petrácek (1988b) and Voigt (1999: 366). Regarding the
other laws, these restrictions affect only stop consonants, even though
they are very strict and affect the mode of articulation (we do not
know why there is no restriction in the other Proto-Indo-European
consonants, or why no emphasis has been placed on their point of
articulation, as we see in Proto-Semitic).
Regarding the second point, the restriction of two consecutive glot-
tal stops must be explained as a process of dissimilation (this feature
often occurs between real languages). Interestingly, in Arabic, following
table no. 1 in Greenberg (1950), in position I-II there is no example
of † or ∂ followed by an emphatic consonant, although eleven examples
of q + emphatic are found (it should be investigated whether these
examples can be reconstructed for Proto-Semitic or not).
Regarding the third point, the restriction voiced / non-voiced and
non-voiced / voiced consonant (assimilation of sonority/no sonority),
also occurs in Arabic. In this language the following groups are not
found: pf,26 bf, td, dt, gt, tg, gk, kg. That is, there is a strong restriction
in the combination voiced / non-voiced and non-voiced / voiced, a fact
that contrasts with the high frequency of the stop consonants.27
Actually, the cases above, with the exception of gt and tg, can be
explained by means of a restriction already postulated by Greenberg
at the beginning of his paper, according to which no homorganic
consonant occurs in position I–II. We have seen, then, that two dif-
ferent families share a set of restrictions which, curiously, can only be
applied to verbal roots and primary adjectives and which work as laws
of dissimilation.

Conclusions

This paper has focused on the laws of incompatibilities of the ver-

bal roots in Semitic. I have stressed the importance of these laws for
the historical reconstruction of the root. These restrictions are applied
only in the verbal root and therefore do not occur in historical deriv-
ative and affixed forms.

26
In Ar. ps. *p > f.
27
In the other cases there are few examples. In the case of bt, 5 cases; bk, 5; dp,
7; dk, 2; gf, 6; fd, 8; fg, 7; tb, 5; kb, 11; kd, 9. All these examples should be com-
pared to other Semitic languages in order to exclude possible loans and innovations
of Arabic.

13

94051_JOSS-2011-1_CS4ME_01.indd 13 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

We have seen, therefore, that the Semitic verbal root is a radical

morpheme; its constituent elements cannot be combined in an arbi-
trary way, but are subject to very specific combinatory restrictions.
The importance of these structural incompatibilities consists, then, in
the fact that they reduce the possible number of combinations of the
triconsonantal bases. We have seen that, excluding onomatopoeic
roots and loan words, these laws of incompatibility are fully regular
in the verbal roots (but not in the nominal ones) and, therefore, do
not have exceptions, as is the case in all phonological laws.
The structure of the Semitic verbal roots is, then, absolutely con-
ditioned by these restrictions of incompatibility. These rules are uni-
versal in character and apply also to the different families of the Afro-
Asiatic and Indo-European languages. The restrictions of
incompatibility are a tool of great importance in the historical recon-
struction of the roots (especially, of the verbal roots in Semitic).
The last part of this paper considers the incompatibilities of the
root with regard to historical linguistics. We have seen that, in the
field of Indo-European linguistics, the laws of formation of the verbal
roots have been described in a different way from the Semitic verbal
roots, but that in fact they are fully identifiable. In both families of
languages the restrictions are useful to recognize their etymological
roots, and they can also help to solve problems on questions of his-
torical reconstruction, such as the reconstruction of loan roots and
the identification of derived morphemes, since, as we noted above,
two homorganic consonants are possible in position 1–2 and 2–3
only when they belong to different morphemes.

Address for Correspondence: eulaliavernet@ub.edu

Abbreviations

Akk. Akkadian
Ar. Arabic
Eth. Ethiopic
GAG Wolfram von Soden. 19953 (19521). Grundriss der Akkadischen Gramatik.
(Rome).
Hbr. Hebrew
Phoen. Phoenician
PIE. Proto-Indo-European
ps. Proto-Semitic
Ugar. Ugaritic

14

94051_JOSS-2011-1_CS4ME_01.indd 14 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

REFERENCES

Aescoly, Aaron Z. 1937–40. ‘Incompatibilités de phonèmes en hébreu et en ara-

méen’, Comptes Rendus du Groupe linguistique d’Études chamito-sémitiques 3,
54–6
Ambros, Arne A. 1991. ‘A Computer-Assisted Statistical Survey of Ethiopic Verb
Patterns Based on Wolf Leslau’s Concise Dictionary of ge¡ez’, in A.S. Kaye (ed.),
Semitic Studies in Honor of Wolf Leslau on the Occasion of his Eighty-fifth Birthday
(1st. vol. Wiesbaden). 56–71
Ammer, Karl. 1950–2. ‘Studien zur indogermanischen Wurzelstruktur’, Die Sprache
2, 193–214
Bender, Marvin Lionel. 1978. ‘Consonant co-occurrence restrictions in afroasiatic
verb roots’ in Pelio Fronzaroli (ed), Atti del secondo congresso internationale di
lingüistica camito-semitica. Firenze, 16–19 aprile 1974 (Quaderni di Semitistica
5, Florence). 9–19
—— 2007. ‘The Afrasian Lexikon Reconsidered’ in C.L. Miller (ed.), Studies in
Semitic and Afroasiatic Linguistics Presented to Gene B. Gragg (Chicago). 1–26
Benveniste, Émile. 19623 (19351). Origines de la formation des noms en indo-européen.
(Paris)
Bohas, Georges. 1985. ‘L’explication en phonologie arabe’, ZAL 15, 45–52
—— 1990a. The Arabic Linguistic Tradition. (London)
—— 1990b. ‘A Diachronic Effect of OCP’, Linguistic Inquiry 21/2, 298–301
—— 1991. ‘Le OCP, la composition des racines et les conventions d’association’.
Bulletin d’études orientales de la grammaire de l’arabe aux grammaires des arabes
43, 119–37
—— 1993a. Développements récents en linguistique arabe et sémitique: séminaire tenu
au Collège de France, Chaire de Langue et Littérature arabes classiques. (Paris)
—— 1993b. ‘La réalisation des racines bilitères en arabe’, in R. Contini, F.A. Pen-
nacchietti and M. Tosco (eds), Semitica. Serta philologica Constantino Tsereteli
dedicata (Turin). 11–13
—— 1993c. ‘OCP et la persistance des représentations sous-jacentes’, Langues orien-
tales anciennes. Philologie et linguistique (LOAPL) 4, 45–40
—— 1993d. ‘Le PCO et la structure des racines’, in G. Bohas (ed.), Développements
récents en linguistique arabe et sémitique (Paris). 9–44
—— 1997. Matrices, Étymons, Racines. Éléments d’une théorie lexicologique du vocab-
ulaire arabe. ORBIS / Supplementa 8. (Leuven)
—— 2000. Matrices et étymons. Développements de la théorie. (Lausanne)
Cantineau, Jean. 1950. ‘La notion de schème et son altération dans diverses langues
sémitiques’, Semitica 3, 73–83
—— 1951–2. ‘Le consonantisme du sémitique’, Semitica 4, 79–94
Colin, Georges S. 1945–8. ‘Les racines trilitères à première et troisième radicale
identiques en arabe classique’, Comptes rendus du groupe linguistique d’études
Chamito-Sémitiques 4, 82–3
Cohen, David. 1978. ‘À propos d’un dictionnaire des racines sémitiques’, in Atti del
secondo congresso internationale di lingüistica camito-semitica. Firenze, 16–19 aprile
1974. (Quaderni di Semitistica 5), 87–100
Conti, Giovanni. 1980. Studi sul bilitterismo in semitico e in egiziano. Il tema verbale
N1212. (Florence)
del Olmo Lete, Gregorio. 2003. Questions de linguistique sémitique. Racine et lexème.
Histoire de la recherche (1940-2000). (Paris)

15

94051_JOSS-2011-1_CS4ME_01.indd 15 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

—— 2005. ‘An Etymological and Comparative Semitic Dictionary. Phonology ver-

sus Semantics: Questions of Method’, Aula Orientalis 23, 185–90
—— 2008. ‘Phonetic distribution in Semitic Binary Articulation Bases’ (in
press)
Diakonoff, Igor M. 1970. ‘Problems of Root Structure in Proto-Semitic’, Archiv
Orientalní 38, 453–80
—— 1975a. ‘Opening Address’, in James and Theodora Bynon (eds), Hamito-
Semitica. Janua Linguarum. Studia memoriae Nicolai van Wijk dedicata (Series
Practica 200). 25–39
—— 1975b. ‘On root structure in Proto-Semitic’. in James and Theodora Bynon
(eds), Hamito-Semitica. Janua Linguarum. Studia memoriae Nicolai van Wijk
dedicata (Series Practica 200). 134–53
—— 1988. Afrasian Languages. (Moscow)
—— 1991–2. Proto-Afrasian and Old Akkadian. A Study in historical phonetics.
(Journal of Afroasiatic Languages, vol. 4, Princeton). n. 1–2
Dozy, Reinhart. 18811. Supplément aux dictionnaires arabes. 2 vols. (Leiden)
Edzard, Lutz E. 1991. ‘Semitic Phonology and Preference Laws for Syllable Struc-
ture’, in A.S. Kaye (ed.), Semitic Studies in Honor of Wolf Leslau on the Occasion
of his Eighty-fifth Birthday. (Wiesbaden). 1, 397–410
Eilers, Wilhelm. 1978. ‘Semitische Wurzeltheorie’, Atti del secondo congresso inter-
nationale di lingüistica camito-semitica. Firenze, 16–19 aprile 1974. Quaderni di
Semitistica 5: 125–31
Frajzyngier, Zygmut. 1979. ‘Notes on the R1R2R2 Stems in Semitic’, JSS 24, 1–12
Goldenberg, Gideon. 1994. ‘Principles of Semitic Word-structure’, in G. Golden-
berg and Sh. Raz (eds), Semitic and Cuschitic Studies (Wiesbaden). 29–64
Greenberg, Joseph H. 1950. ‘The Patterning of Root Morphemes in Semitic’, Word
6, 162–81
Herdan, Gustav. 1962. ‘The Patterning of Semitic Verbal Roots Subjected to Com-
binatory Analysis’, Word 18, 262–8
Hintze, Fritz. 1947. ‘Zur Struktur des Wortes im Aegyptischen’, Zeitschrift für Pho-
netik 1, 18–24
Jakobson, Roman and Morris Halle. 1956. Fundamentals of Language. (The
Hague)
Jucquois, Guy. 1970–2. ‘La théorie de la racine en indo-européen’, La Linguistique
6/2, 69–102; 7/1, 73–91; 8/1, 73–103
Kienast, Burkhart. 2001. Historische Semitische Sprachwissenschaft. (Wiesbaden)
Koskinen, Kalevi. 1964. ‘Kompatibilität in den dreikonsonantigen hebräischen
Wurzeln’, Zeitschrift der Deutschen Morgenländischen Gesellschaft 114, 16–58
Kury¥owicz, Jerzy. 1956. L’Apophonie en Indo-européen. (Wroc¥aw)
—— 1961. L’apophonie en sémitique. (Wroc¥aw, Warsaw, Krakow)
—— 1964. The Inflectional Categories of Indo-European. (Heidelberg)
—— 1972. Studies in Semitic Grammar and Metrics. (Wroc¥aw, Warsaw)
—— 1973. ‘Verbal Aspect in Semitic’, Orientalia 42, 114–20
Landsberger, Benno. 1935. ‘Die Gestalt der semitischen Wurzel’, International Con-
gress of Orientalists: Proceedings 19, 450–2
Lane, William Edward. 1863. Arabic-English Lexicon. (London-Edinburgh)
Lipinski, Edward. 1991. ‘Monosyllabic Nominal and Verbal Roots in Semitic Lan-
guages’, in A.S. Kaye (ed.), Semitic Studies in Honor of Wolf Leslau on the Occa-
sion of his Eighty-fifth Birthday. (Wiesbaden). 2, 927–30.
—— 1997. Semitic Languages. Outline of a Comparative Grammar. (Leuven)

16

94051_JOSS-2011-1_CS4ME_01.indd 16 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

Loprieno, Antonio. 1984. ‘L’indagine delle strutture verbali delle lingue camito-
semitiche’, in F.A. Pennacchietti and A. Roccati (eds), A Atti della terza giornata
di Studi Camito-Semitici e Indoeuropei (Rome). 65–74.
Macdonald, John. 1963–5. ‘New Thoughts on a Biliteral Origin for the Semitic
Verb’, The Annual of Leeds University Oriental Society 5, 63–85
McCarthy, John J. 1981. ‘A Prosodic Theory of Nonconcatenative Morphology’,
Linguistic Inquiry 12/3, 373–418
Meier-Brügger, Michael. 20028. Indogermanische Sprachwissenschaft. (Berlin, New
York)
Moscati, Sabatino (ed.). 1969. An Introduction to the Comparative Grammar of the
Semitic Languages. (Wiesbaden)
Petrácek, Karel. 1964a. ‘Die innere Flexion in den semitischen Sprachen V: Entste-
hung und Entwicklung des Systems’, Archiv Orientální 32, 185–222
—— 1964b. ‘Die Inkompatibilität in der semitischen Wurzel in Sicht der Informa-
tionstheorie’, Rocznik Orientalistyzny 27, 133–9
—— 1974. ‘La structure de la racine et la classification des langues hamitosémi-
tiques’, Acta Universitatis Carolinae, Philologica 4, 115–21
—— 1982. ‘La racine en indoeuropéen et en chamito-sémitique et leurs perspectives
comparatives’, Annali dell’Istituto Orientale di Napoli 42, 381–402
—— 1985. ‘La racine en sémitique d’après quelques travaux récents en russe’,
Archiv Orientální 53, 171–3
—— 1987a. ‘Sur le rôle des modalités sonantiques dans l’élaboration de la racine
en sémitique’, Arabica 34, 106–10
—— 1987b. ‘Les laryngales en chamito-sémitique. Essai de synthèse (résumé)’, in
H. Jungraithmayr and W.W. Müller (eds), Proceedings of the Fourth International
Hamito-Semitic Congress (Marburg). 65–71
—— 1988a. Altägyptisch, Hamitosemitisch und ihre Beziehungen zu einigen Sprach-
familien in Afrika und Asien. (Prague)
—— 1988b. ‘Nochmals über die Struktur der Wurzeln mit den Pharyngalen im
Ältagyptischen und semitischen und ihre Inkompatibilität’, in Y.L. Arbeitman
(ed.), Fucus. A Semitic/Afrasian Gathering in Remembrance of Albert Ehrman
(Amsterdam). 371–7
Rocquet, Gérard. 1973. ‘Incompatibilités dans la racine en ancien égyptien’,
Göttinger Miszellen 6, 108–17
Rosseau, Jean. 1984. ‘La racine arabe et son traitement par les grammairiens
européens (1505–1831)’, Bulletin de la Société de Linguistique de Paris 79, 285–
321
Rössler, Otto. 1971. ‘Das Ägyptische als semitische Sprache’, in F. Altheim and
R. Stiehl (eds), Christentum am Roten Meer (Berlin, New York). 1, 263–326
Sanmartín, Joaquín. 2005. ‘The Semantic Potential of Bases (‘roots’) and Themes
(‘patterns’): A cognitive approach’, Aula Orientalis 23, 65–81
Schramm, Gene M. 1962. ‘An outline of Classical Arabic Verb Structure’, Language
38, 360–75
—— 1991. ‘Semitic Morpheme Structure Typology’, in A.S. Kaye (ed.), Semitic
Studies in Honor of Wolf Leslau on the Occasion of his Eighty-fifth Birthday. (Wies-
baden). 2, 1402–8
Stempel, Reinhard. 1999. Abriß einer historischen Grammatik der semitischen
Sprachen. (Frankfurt a. M., Berlin, Bern, New York, Paris, Vienna)
Tobin, Yishai. 1990. ‘A Combinatory Phonology of the Hebrew Triconsonantal
(CCC) Root System’, La Linguistique 26, 99–114

17

94051_JOSS-2011-1_CS4ME_01.indd 17 14/01/11 15:22

 SEMITIC ROOT INCOMPATIBILITIES AND HISTORICAL LINGUISTICS

Ullendorff, Edward. 1958. ‘What is a Semitic Language? A Problem of Linguistic

Identification’, Orientalia 27, 69–75
Vergote, Joseph. 1974. ‘Le rapport de l’égyptien avec les langues sémitiques.
Quelques aspects du problème’. Actes du premier congrès international de linguis-
tique sémitique et chamito-sémitique. Paris 16–19 juillet 1969 (The Hague, Paris).
49–54
—— 1975. ‘La position intermédiaire de l’ancien égyptien entre l’hébreu et l’arabe’,
in James and Theodora Bynon (eds), Hamito-Semitica. Janua Linguarum. Studia
memoriae Nicolai van Wijk dedicata (Series Practica 200, The Hague, Paris).
193–9
Vernet i Pons, Eulàlia. 2008. ‘Formació i estructura de les arrels verbals en semític:
comentari etimològic dels determinatius radicals presents en els verba tertiae
infirmae de l’hebreu bíblic masorètic’ [on line]. Ph.D. Dissertation. Universitat
de Barcelona. Departament de Filologia Semítica http://www.tdx.cat/TDX-
0212109-111510. Barcelona.
Villar, Francisco. 1996. Los indoeuropeos y los orígenes de Europa. (Madrid)
Voigt, Rainer M. 1981. ‘Inkompatibilitäten und Diskrepanzen in der Sprache und
das erste phonologische Inkompatibilitätsgesetz des Semitischen’, Die Welt des
Orients 12, 136–72
—— 1999. ‘Ägyptosemitischer Sprachvergleich’, in S. Grunert and I. Hafemann
(eds), Textcorpus und Wörterbuch. Aspekte zur ägyptischen Lexikographie (Leiden,
Boston, Koln). 345–66
von Soden, Wolfram. 1973. ‘Ein semitisches Wurzelwörterbuch: Probleme und
Möglichkeiten’, Orientalia 42, 142–8
—— 19953 (19521). Grundriss der Akkadischen Gramatik. (Rome)
Weitzman, Michael. 1987. ‘Statistical Patterns in Hebrew and Arabic Roots’, Jour-
nal of the Royal Asiatic Society 15–22
Zaborski, Andrzej. 1994a. ‘Exceptioneless Incompatibility Rules and Verbal Root
Structure in Semitic’, in G. Goldenberg and Sh. Raz (eds), Semitic and Cushitic
Studies (Wiesbaden). 1–18
—— 1994b. ‘Archaic Semitic in the Light of Hamito-Semitic’, Zeitschrift für Althe-
braistik 7/2, 234–44
—— 1996. ‘Some Alleged Exceptions to Incompatibility Rules in Arabic Verbal
Roots’, in P. Zemanek (ed), Studies in Near Eastern Languages and Literatures.
Memorial Volume of Karel Petrácek (Prague). 631–58

18

94051_JOSS-2011-1_CS4ME_01.indd 18 14/01/11 15:22