The Origin of the Third Person Markers on the Suffix Conjugation in Semitic

Rebecca Hasselbach-Andee

University of Chicago

1 Introduction1

The origin of the third person affixes marking person and number on the Semitic

stative/suffix conjugation – that is the East Semitic stative (paris) and the West Semitic perfect

(qatala) – is still a matter of debate.2 Unlike the person markers of the first and second persons,

which are clearly derived from forms of the corresponding independent pronouns (see section

2 below), the third person markers do not exhibit any formal similarities with attested third

person pronouns. This discrepancy between second and first versus third person markers has

led to various approaches attempting to explain the origin of the third person markers. The

two main reconstructions that have been proposed are, first, that the third person suffixes are

not person markers originally but reflect nominal endings corresponding to nominal gender

1
I am honored to dedicate this paper to my esteemed colleague Dennis Pardee, who has inspired my work for
many years. Abbreviations used in this article include: ACC = accusative; ADJ = adjective; c = comunis; f = feminine;
FEM = feminine; GEN = genitive; m = masculine; MASC = masculine; NOM = nominative; OBL = oblique; p = plural;
PC = prefix conjugation; PL = plural; PRED = predicate; PRO = pronoun; PTC = participle; s = singular; SC = suffix
conjugation; SING = singular; SUBJ = subject; TOP = topic.
2
In this article, the form in question will be designated as stative/suffix conjugation (SC) in order to indicate its
functional differences in East and West Semitic. The term “stative” has been chosen over “predicative adjective”
or “predicative construction” because it is the most common term used to describe the East Semitic form. In
addition, it designates the whole construction, that is, the nominal base and person marker, while “predicative
adjective” can also simply mean the nominal element (paris-). “Stative” as used in this article thus indicates the
combination of nominal base plus person marker, while “predicative adjective” simply indicates an adjective in
predicative use. In this sense, the East Semitic “stative” is, as Buccellati suggests, viewed as a syntactic rather than
morphological unit (Buccellati 1968:6).
and number markers, and, second, that the suffixes are, despite the lack of a clear

correspondence to attested pronouns, nevertheless derived from pronouns.

In this article, it will be argued that it is more likely that the third person suffixes are of

nominal origin since they can directly be derived from gender, number, and case markers of

the nominal paradigm.3

The following discussion will first briefly review the reconstruction of the suffix

conjugation and its affixes in general, then discuss previous scholarship on the derivation of

the third person markers, and subsequently argue in favor of a nominal derivation of these

morphemes.

2 The suffix conjugation

The Semitic stative/suffix conjugation (SC) is characterized by the fact that person markers

are suffixed to the base, as in the following Akkadian and Classical Arabic paradigms:

Akkadian stative

1cs marṣ-āku 1cp marṣ-ānu

2ms marṣ-āta 2mp marṣ-ātunu

2fs marṣ-āti 2fp marṣ-ātina

3ms maruṣ-Ø 3mp marṣ-ū

3
This statement seemingly contradicts what was claimed in Hasselbach (2007:135), where it was proposed that the
external plural markers –ū and –āt can be derived from the verbal system, more specifically from the predicative
verbal adjective. As will hopefully be shown in the following discussion, this statement does not truly contradict
the reconstruction proposed in this article, since the base for the derivation, the predicative verbal adjective, is
the same, but is analyzed as part of the verbal system in Hasselbach (2007), and, more correctly, as part of the
nominal system in this article.

2
 3fs marṣ-at 3fs marṣ-ā

Classical Arabic perfect

1cs katab-tu 1cp katab-nā

2ms katab-ta 2mp katab-tum

2fs katab-ti 2fp katab-tunna

3ms katab-a 3mp katab-ū

3fs katab-at 3fp katab-na

In West Semitic languages, the consonant of the first and second person markers is either

uniformly /t/, as in the Classical Arabic paradigm provided above, as well as in Canaanite,

Aramaic, and Ugaritic, or /k/, as in Ethiopian Semitic, Old South Arabian, and Modern South

Arabian (e.g. Geˁez 1cs gabar-ku, 2ms gabar-ka, 2fs gabar-ki, etc.). It is generally agreed that the

original distribution of the suffixal consonants was that exhibited by Akkadian, with /k/ in the

1cs and /t/ in the second person forms.4 The West Semitic languages leveled one of these

consonants, resulting in the uniform paradigms found in these languages. The first and second

person forms are thus commonly reconstructed as:5

1cs *qatVl-ku 1cp *qatVl-nu

4
For the reconstruction of the first and second persons with different suffix consonants based on the principle of
archaic heterogeneity see Hetzron (1976:92-95).
5
For this reconstruction, see e.g. Huehnergard (2004:150). For the assumption that the final vowels were short in
these suffixes see Hasselbach (2004). It is debated whether or not the stative/SC had a linking vowel between the
nominal base and the pronominal element. For reconstructions that assume a linking vowel in certain
environments, see Tropper (1999) and Voigt (2002/2003).

3
 2ms *qatVl-ta 2mp *qatVl-tum(±ū)

2fs *qatVl-ti 2fp *qatVl-tin(±na/ā)

For the third person forms, the most commonly found reconstruction is as follows:6

3ms *qatVl-a 3mp *qatVl-ū

3fs *qatVl-at 3fp *qatVl-ā

The –ā of the 3fp is attested in Akkadian and Geˁez. Since the marker –ā is attested in two

chronologically and geographically distinct languages, it is commonly assumed that it is

original to the 3fp. The Classical Arabic FEM PL marker –na can be explained by leveling. The

same FEM PL marker –na is attested on the prefix conjugation yaktubna ‘they (fp) write’ and

provides the base for leveling –na to the SC, resulting in the form katabna given above. Hebrew

has lost the distinction of MASC and FEM in the SC and uses the original MASC PL marker –ū as

third person plural form, as in kāṯəḇû ‘they wrote (MASC and FEM)’. The fact that 3fp forms

that have a different marker than –ā can be explained by leveling or loss strengthens the

reconstruction of the 3fp as -ā for Proto Semitic.7

From a functional point of view, the West Semitic SC is a finite verbal form expressing

past tense and perfective aspect, as in Classical Arabic katabtu ‘I wrote’, Hebrew kāṯaḇtî ‘I

wrote’, Geˁez gabarka ‘you (ms) did’.

6
See, e.g., Huehnergard (2004:150).
7
For this reconstruction see, e.g., Nöldeke (1884:421); Brockelmann (1908:571); Diem (1997:61); Huehnergard
(2004:150).

4
 In East Semitic, the stative expresses the state of the underlying verbal form, as in

marṣāku ‘I am sick’ from the verbal root marāṣum ‘to be sick’. The East Semitic form has no

tense value in itself but describes a state that coincides with the point in time referenced by

the main verb or context.8 The functions of the West Semitic SC and East Semitic stative are

thus fundamentally different. It is commonly assumed that the stative represents the more

archaic stage of the two reflexes of the form and that it indicates its derivation from an

original verbless clause. This verbless clause originally consisted of a nominal base – in the

case of the stative/SC a verbal adjective of the form qatVl – that reflects the predicate of the

verbless clause, and a pronominal element in the form of an independent pronoun indicating

the subject.9 The derivation of a verbal form from an original verbless clause is a cross-

linguistically common phenomenon. In an influential article, Givón argues that verb

agreement paradigms in fact always arise from anaphoric pronoun paradigms.10 This means

that the person markers of conjugated finite verbs are always the product of a process in

which an originally independent pronoun was grammaticalized into an enclitic person marker.

8
For a more detailed description of the function of the Akkadian stative see Kouwenberg (2010:163-165).
9
Buccellati (1968:3); Huehnergard (1986:228); Huehnergard (2004:152); Kouwenberg (2000:21, 23); Tropper
(1995:492).
10
Givón (1976:180). Givón suggests that the origin of for this type of verbalization process is in topicalized
constructions of the type ‘the man (TOP), he (PRO) came’, in which the topicalized construction was reanalyzed as
the neutral construction through overuse. The reanalysis into a neutral construction results in ‘the man (SUBJ),
he (Agreement)-came’ (Givón 1976:155). Another derivation has been proposed by Cohen (1975), who assumes
that certain marked predicative constructions, such as Classical Arabic zaydun huwa l-muˁallimu ‘Zaid is the
teacher’, in which a pronoun is obligatory between SUBJ and PRED in order to distinguish predicative from
attributive constructions, can lead to the grammaticalization of new verbal forms (Cohen 1975:88). He proposes
three steps for the verbalization process: the first step is a nominal predicate that is often marked for being
predicative as in the Arabic example just given. The second step is that this marked predicate can be a sentence in
itself. Lastly, the form is integrated into the tense-aspect system (Cohen 1975:89).

5
 As mentioned above, the East Semitic stative has no tense value. Other characteristics

in which the stative does not agree with the behavior of finite verbal forms in Akkadian are

that morphemes such as the subordinate marker or ventive cannot be suffixed to the stative

except in the third persons.11 These features show that the Akkadian stative is still in the

process of being grammaticalized into a verb.12 The grammaticalization process is completed in

West Semitic, where the SC is a finite verbal form and is fully integrated into the tense/aspect

system.

The derivation of the stative/SC from an original verbless clause has long been

recognized and is generally accepted.13

11
Buccellati argues that the ventive cannot be attached to a second or first person form because the second
element of the stative of these forms is pronominal and the ventive cannot be attached to a pronoun. This also
means that the pronominal elements still retain their pronominal nature and are no mere agreement markers. In
the third persons, there is no pronominal element and the ventive can consequently be attached to the base
(Buccellati 1968:3). Kouwenberg disagrees with Buccelati’s analysis and states that the ventive or subjunctive
cannot be attached to first and second person statives due to morphosyntactic reasons since they can only be
attached to the stem plus number/gender marker (Kouwenberg 2000:24 note 5). Although this sounds similar to
Buccellati’s argument, Kouwenberg stresses that this morphosyntactic feature does not indicate that the stative is
still nominal. He further claims that statives in general do not take ventives for semantic reasons since the
ventive is a directional morpheme that is primarily used for verbs of movement (Kouwenberg 2000:25 note 5). The
latter argument, however, does not explain the difference between first and second versus third person forms in
terms of being able to take the ventive or subjunctive markers.
12
The degree to which the Akkadian stative is verbalized is highly debated. Buccellati, followed by Huehnergard,
considers the stative to be a nominal clause (Buccellati 1968:2, 1988:164; Huehnergard 1986:221 note 11, 1987:215),
while Kouwenberg considers it to be a finite verbal form (Kouwenberg 2000:21).
13
See, e.g., Cohen (1975:89); Hodge (1975:69-73); Tropper (1995:492); Kouwenberg (2000:21). Hodge claims that all
verbs in Semitic and Egyptian are derived from original nominal sentences since the general pattern of
replacement in these two language families that can be observed over the centuries is generally nominal (Hodge
1975:73).

6
 The pronominal origin of the first and second person forms is, as mentioned in the

introduction, still noticeable:14

1cs *ˀan-āku 1cp *niḥ-nu

2ms *ˀan-ta 2mp *ˀan-tum(±ū)

2fs *ˀan-ti 2fp *ˀan-tin(±na/ā)

However, no direct correspondence can be noticed in third person forms, which are commonly

reconstructed as follows:15

3ms *suˀa 3mp *sum(±ū)

3fs *siˀa 3fp *sin(±na/ā)

Although the 3ms SC marker –a could still theoretically be explained by the 3ms independent

pronoun ending in –(ˀ)a, no such correlation is possible for the other third person forms.16 The

problem that there is no direct correspondence between the third person markers of the

14
For this reconstruction see Huehnergard (2004:150).
15
Huehnergard (2004:150).
16
The longer forms that are indicated for the second and third person plural forms contain the same elements as
corresponding forms of the stative/SC suffixes. However, these elements seem to be secondary additions in the
third person pronouns since they do not constitute obligatory elements and should probably not be reconstructed
for Proto Semitic (Huehnergard 2004:150). They are provided here for the sake of completion. Even if one
assumed that the longer forms are original, the 3fs form of the stative/SC still remains unexplained. Furthermore,
the elements marking the longer forms are present in both the second and third person forms and thus do not
constitute person markers but simple plural markers. Consequently, they differ functionally from the pronominal
elements found as person markers in the first and second persons on the stative/SC.

7
stative/SC and the third person independent pronouns that are attested in the descendant

languages – and based on the attestations reconstructed for Proto Semitic – has led to two

main hypotheses on how to derive the third person markers, which will be discussed in the

following section.

3 Previous reconstructions

One of the earliest reconstructions for the third person suffixes of the stative/SC was proposed

by Nöldeke (1884), who reconstructs *-a for the 3ms, *–at for the 3fs, *–ū for the 3mp, and *-ā

for the 3fp.17 Concerning the origin the person markers, Nöldeke assumes that the first and

second person suffixes are related to the corresponding independent pronouns, while no such

relationship can be noticed in the third person forms. Although he does not propose an origin

for the third person markers, he notes that it is unlikely that the third person forms contain a

pronominal element.18 Despite hesitating to suggest a derivation for the third person markers,

Nöldeke nevertheless points out parallels for some of them. The –t of the 3fp form *qatalat,

according to Nöldeke, is also found in the 3fs prefix of the prefix conjugation (*ta-qtul). The

MASC PL marker –ū is most likely simply a plural marker, based on plural forms such as

Classical Arabic yaqtul-ū-na (3mp PC) and external masculine plurals on nouns such as Classical

Arabic qātil-ū-na. Concerning the 3fp, Nöldeke states: “Völlig unklar bleibt â als Endung des

weiblichen Plurals.”19

17
Nöldeke (1884:421).
18
Nöldeke (1884:421).
19
Nöldeke (1884:421).

8
Nöldeke thus refrains from suggesting an origin for the third person markers but nevertheless

distinguishes them from the second and first person forms that, as he claims, are related to

corresponding independent pronouns.

Nöldeke’s reconstruction of the forms of the third person markers has been very

influential and has been followed by a majority of scholars.20 A first proposal regarding the

etymological origin of the third person markers that had significant impact on the

understanding of the forms was given by Zimmern (1898). Zimmern, who follows the same

morphological reconstruction as given by Nöldeke, proposes that the third person endings are

nothing else than inflectional endings of nouns:

“Die dritten Personen des Singulars und Plurals stellen im Grunde wohl nichts anderes

dar, als flektierte Nomina (bzw. Adjektiva, Partizipia), sodaß das Afformativ der 3. m.

Sg., soweit ein solches überhaupt anzunehmen ist, den maskulinen singularischen

nominalischen Auslaut (-a, -u), das Afformativ der 3. f. Sg. den entsprechenden

femininen (-at) enthält…”.21

Zimmern is commonly regarded as one of the first scholars to propose the nominal origin of

the third person markers.22 His reconstruction became the predominant theory for the

derivation of the third person markers of the stative/SC and has been followed by numerous

20
See, e.g., Brockelmann (1908:571, 574); Bauer and Leander (1922:308); Bergsträsser (1928:13); Diem (1997:61);
Lipiński (2001:368-372); Huehnergard (2004:150).
21
Zimmern (1898:99).
22
See Diem (1997:42-46) for a detailed overview of the reception of Nöldeke’s and Zimmern’s reconstructions.

9
scholars, including Bergsträsser (1928:13), Bauer and Leander (1922:308), and, in a slightly

varied form, more recent scholars such as Tropper (1999:188) and Lipiński (2001:368).

Despite its predominance, the nominal origin theory faces several problems. First of all,

it has been argued that the ending of the 3ms -a is not attested in predicative use in the

normative nominal system as attested in Semitic languages, nor is the form of 3fp –ā. Nouns in

the MASC SING either end with a case ending or have zero-marking in Semitic, as in Classical

Arabic kitāb-u-n ‘a book (NOM)’, Geˁez nəguś-Ø ‘king (NOM/GEN), and Hebrew mεlεk-Ø ‘king’.

The FEM PL on nouns has the form –āt across Semitic and consequently does not correspond to

any of the FEM PL endings that are attested on the stative/SC.23

Another issue that has been cited as an argument against the nominal origin of the

third person markers is that the nominal base, that is the verbal adjective with first and second

person pronominal elements, does not agree with its subject in gender and number but always

reflects the same indeclinable base *qatVl. If the third person forms indeed reflect the nominal

endings of a presumed predicative state, then we face the question why the base remains

uninflected.24

23
Because of the discrepancy between the attested third person markers of the stative/SC and the common
nominal inflection, some scholars have suggested that the nominal base of the stative/SC appears in a special
state, the so-called “predicative state”, whose form differs from the common declination of nouns. According to
scholars such as von Soden, this “predicative state” can be considered identical to the indeclinable absolute state
attested in Akkadian (von Soden 1995:96). What exactly this predicative state represents and how it relates to the
Akkadian “absolute state”, however, has not been clarified sufficiently. For the idea that the stative reflects a
predicative state of the nominal base see also Buccellati (1988:168) and Buccellati (1996:167). Buccellati further
assumes that this “predicative state” is a bound form, that is, it occurs with pronominal suffixes of the subject
case (Buccellati 1988:168).
24
See, e.g. Kraus (1984:37, 40). A common example for a case where the nominal base of a similar construction as
the stative/SC exhibits gender and number inflection is the predicative participial construction of Syriac. In
Syriac, the participial base to which enclitic pronouns are attached to mark a predicative construction declines

10
 Problems such as these have led scholars to propose alternative derivations for the

third person markers. The most common alternative hypothesis is that the third person

markers are not of nominal, but of pronominal origin and thus correspond to the derivation of

the suffixes of the first and second persons. This reconstruction has in particular been

advocated by Huehnergard (1987) and Diem (1997).

Huehnergard assumes that in early Proto Semitic, the identity of the endings of the

independent pronouns and the stative/SC was complete, that is, even the third person markers

of the stative/SC are derived from independent pronouns, although this reconstruction faces

the problem that the third person markers do not correspond to any known third person

pronouns. Huehnergard therefore argues that the attested third person pronouns are actually

demonstrative pronouns and not original to the pronominal paradigm. Instead, there existed

another, unspecified, set of third person independent pronouns that corresponded to the

endings of the stative/SC –a, -at, -ā, and –ū and from which these markers were derived. This

original set of third person pronouns was then lost and replaced by the attested set of

anaphoric pronouns, a process that presumably already took place in Proto Semitic.25

Huehnergard thus assumes that there originally existed a more consistent paradigm of

independent pronouns that was the base for all stative/SC person markers. This still leaves the

problem that the nominal base of the stative/SC does not show agreement. Huehnergard

argues that it is not the nominal base, that is, the verbal adjective, that is the predicative

element of the stative/SC, but the whole construction of verbal adjective plus enclitic subject

for gender, as in the 1fs qāṭl-ā-nā, compared to the 1ms qāṭel-nā, and number, as in the 2mp qāṭl-īt-tōn (< *qāṭl-īn-
tōn) (Diem 1997:47).
25
Huehnergard (1987:222).

11
pronoun constitutes the predication. Neither element is predicative in and of itself. The base

can, according to Huehnergard, occur in two constructions: an attributive construction that

consists of {base + gender/number markers + case endings}, and a predicative construction

that consists of {base + enclitic subject pronoun}.26 In the predicative construction, the base

thus lacks agreement while it exhibits agreement in the attributive constructions.27 The key to

explaining the lack of agreement in the base of the stative/SC is thus that the base or verbal

adjective is not the predicative element but constitutes an element of a larger, two-partite,

predicative construction.

Diem does not go as far as Huehnergard to assume a second set of independent third

person pronouns but tries to derive the third person markers on the stative/SC from the

attested set of anaphoric pronouns. Before proposing his own reconstruction, Diem presents

arguments against the nominal theory. He claims that if the nominal origin hypothesis were

correct, we would expect the 3ms to have a final –u, not –a, since nominal predicates usually

appear in the nominative, which is marked by –u in Semitic languages with productive case

systems.28 According to Diem, the marker –a that is attested in West Semitic would thus have to

be secondary. A secondary origin of this –a, however, is difficult to explain. Furthermore, he

restates the argument that if the third person forms are reflexes of nominal gender and

number markers, it is difficult to explain why the nominal base is not declined for gender and

number throughout the paradigm. He cites Syriac as an example of a language that underwent

a similar development as the stative/SC but that shows agreement in the nominal base. Syriac

26
Huehnergard (1987:222).
27
Huehnergard (1987:223). Indeclinable predicative forms are also found in other languages such as German and
Egyptian (ibid.).
28
Diem (1997:46).

12
developed a predicative construction based on the PTC to which enclitic pronouns derived

from independent pronouns are attached, as in qāṭel-nā (1ms), qāṭl-ā-nā (1fs), qāṭl-īt-tōn (2mp),

etc.29 According to Diem, the difference in the behavior of the Syriac participial construction

and the Akkadian stative in terms of agreement clearly shows that they underwent different

developments. Furthermore, although bare nominal forms can be used as third person forms

in Syriac, according to Diem, this is the endpoint of a long process of grammaticalization in

which PTCs plus third person pronouns were replaced by the bare stem. Since no such

presumed grammaticalization process can be traced for the stative/SC, we cannot explain how

pure nouns became verbal forms if the third person forms are indeed derived from nominal

inflection markers.30 Another problem that has been mentioned before, is that the 3fp –ā has

no correspondence in the nominal system.31 According to Diem, these issues make a nominal

origin of the third person markers unlikely.

Instead, he proposes a reconstruction of the attested third person pronouns that allows

for the derivation of the third person markers of the stative/SC:

Diem’s reconstruction32

3ms *h/šuˀ-a 3mp *h/šum-ū

3fs *h/šiˀ-at 3fp *h/šinn-ā

29
Diem (1997:47). See also note 23 above.
30
Diem (1997:47).
31
Diem (1997:48).
32
Diem (1997:69).

13
Based on this reconstruction, Diem concludes that all suffixes of the stative/SC have a

pronominal origin.33

There are some obvious problems with both Huehnergard’s and Diem’s

reconstructions. The original set of third person pronouns that Huehnergard assumes to

underlie the third person markers of the stative/SC has left no other traces in Semitic and is

thus purely hypothetical. We have no possibility to prove (or disprove) that such a set of

pronouns existed, which weakens Huehnergard’s reconstruction significantly. Diem’s

reconstruction is based on the stative/SC markers rather than on evidence from actual

pronouns. Although longer forms of the third person plural forms exist, as has been indicated

in section 2 above, even if these are original, there is no evidence in Semitic that the 3fs

independent pronoun ever ended in –at. The reconstruction of this form is solely based on the

forms of the FEM SING in the nominal system and has no basis in the pronominal system.

Some of Diem’s premises for discarding the nominal origin hypothesis are equally

problematic. His claim that the final vowel of the 3ms should be –u and not –a if the form were

nominal originally has been shown to be incorrect. A nominal MASC SING ending –a is clearly

attested in predicative use in Old Akkadian, Eblaite, and Amorite personal names, as has

already been pointed out by Gelb, Diakonoff, Krebernik, and Streck among others.34 Amorite

names, for example, have forms such as /ṣūr-a ˁamm-u/ ‘the divine uncle is a rock’, where the

subject is marked by –u and the predicate by –a, although the marker –a can also mark the

subject, vocative, and genitive besides marking nominal predicates in Amorite.35 A similar type

33
Diem (1997:71).
34
Gelb (1965:79); Diakonoff (1988:59); Krebernik (1991:138); Streck (2000:288-290).
35
Streck (2000:264-268); Hasselbach (2013:38).

14
of functional range of –a is attested in Old Akkadian names, although in Old Akkadian names,

the marking of the predicate by –a is the most frequently encountered reflex of this

morpheme.36 The idea that the ending –a reflects a marker of nominal predicates in these early

names is widely accepted.

A connection between this “predicative” –a and the 3ms –a of the stative/SC has, for

example, been proposed by Tropper. Tropper argues that since the stative/SC is derived from a

construction that has a predicative verbal adjective as its base, the –a of the 3ms SC should be

related to the predicative ending –a attested in early Semitic proper names,37 which is an

assessment with which I fully agree.38 Consequently, the 3ms marker of the SC –a has an

equivalent in the nominal inflection, namely in the early Semitic predicative marker –a. Diem’s

criticism that the 3ms –a cannot be original to the nominal system is thus invalid. The fact that

there exists a predicative form of the noun ending in –a also indicates that the verbal adjective

that forms the base of the stative/SC is indeed the predicative element of the construction,

contrary to Huehnergard’s analysis mentioned in this section above that claims that none of

the two basic elements of the stative/SC are predicative in and of themselves, but that the

whole construction consisting of {base + enclitic subject pronoun} constitutes the predicative

form.

The second point of criticism, which has been brought forth by Kraus, Huehnergard,

and Diem, is that the nominal base of the stative/SC lacks agreement.39 This lack of agreement

seems to contradict comparable grammaticalization instances that can be observed in Semitic

36
Hasselbach (2013:41-43).
37
Tropper (1999:181-182).
38
For a more detailed discussion of this “predicative” –a see Hasselbach (2013:48-50).
39
Kraus (1984:37); Huehnergard (1987:217, 222); Diem (1997:47).

15
languages such as the aforementioned participial construction in Syriac, in which the base

shows agreement. The issue is, of course, that we are dealing with grammaticalization, as

rightly observed by Diem.

From a grammaticalization point of view, however, the lack of agreement in the

nominal base of the stative/SC does not pose any problem. The lack of agreement, as has

already been pointed out by Kouwenberg, reflects a well-known grammaticalization process

called “simplification”.40 “Simplification”, a term first used by Heine and Reh, means “the

development of regularities for formerly irregular aspects of grammar”.41 Simplification can

thus be considered a type of analogical leveling where a form is used in contexts in which it

has not previously been used – although in the case of simplification this happens within

paradigms and thus constitutes a “paradigmatic analogy”.42 Since the linguistic replacement in

this case happens on the paradigmatic level, it can lead to the loss of relevant paradigmatic

distinctions. The concept of simplification as an explanation for the loss of inflection has

particularly been developed by Lehmann (1985), who states that grammaticalization from

periphrastic to morphological constructions can lead to increasingly small and homogeneous

paradigms.43 Simplification, and consequently the loss of paradigmatic distinctions, is a typical

side effect of grammaticalization processes.44

This means that the lack of agreement in the base of the stative/SC can be explained by

the fact that this form underwent grammaticalization. That this is true is also obvious from the

40
Kouwenberg (2000:58).
41
Heine and Reh (1984:41).
42
Heine and Reh (1984:41).
43
Lehmann (1985:307).
44
See also Croft (2003:264).

16
fact that we are not dealing with a periphrastic construction anymore but clearly with an

already grammaticalized synthetic form that has inseparable subject markers.45 This also

indicates, as Kouwenberg states, that even the stative of adjectives cannot simply be a

synchronic derivation or combination of adjective and pronoun in Akkadian. Instead, it is the

result of a historical process and has already undergone a certain degree of

grammaticalization.46

It is thus not necessary to assume a pronominal origin of the third person markers in

order to explain the lack of agreement in the base of the stative/SC, or to assume different

agreement behaviors of predicative and attributive constructions in early Semitic. The

nominal origin hypothesis can account for the lack of agreement in the nominal base when it

is viewed as the result of grammaticalization.

The points of criticism traditionally brought forth against the nominal origin

hypothesis of the third person markers of the stative/SC thus do not really pose any problems

for this derivation.

The nominal theory, however, still faces another problem. While the forms of the 3ms,

3fs, and 3mp can be derived from forms attested in the nominal system, the form of the 3fp –ā

still defies explanation. Many scholars following the nominal origin hypothesis of the third

person markers simply do not address the problem that the FEM PL on nouns is regularly

expressed by the morpheme –āt and not –ā in Semitic languages.47 Other scholars, such as von

45
This does not mean, however, that the stative in Akkadian has to be analyzed as a verb. It clearly still has
nominal characteristics. Morphologically, it has grammaticalized from a periphrastic into a synthetic
construction, but functionally, it has not yet undergone the full grammaticalization from verbless clause to finite
verbal form.
46
Kouwenberg (2000:58).
47
See, e.g., Zimmern (1898:98); Bauer and Leander (1922:308); Bergsträsser (1928:13); Lipiński (2001:372).

17
Soden, consider the third person markers of the stative/SC to be identical with the endings of

the Akkadian absolute state.48 The FEM PL of the absolute state is attested as both –ā and –āt,

although –ā seems to be more common.49 Since determining the origin of the absolute state has

been problematic, however, connecting the absolute state with the stative/SC does not solve

the issue of the 3fp ending –ā but simply shifts it to a different context.

One of the few scholars who specifically addresses the origin of the 3fp –ā is Tropper.

Tropper, who adheres to the nominal origin hypothesis, argues that –ā is actually derived from

–āt. Evidence for such a derivation presumably comes from the Egyptian pseudo-participle that

ends in –t(j).50 Tropper further assumes that the Akkadian FEM PL of the absolute state is

likewise derived from –āt. In general, he assumes like von Soden that the markers of the

predicative construction from which the stative/SC is derived are the same as those of the

absolute state, which he calls “absolutive case” (“Absolutivkasus”). Tropper consequently

reconstructs them as MASC SING –a, FEM SING -at, MASC PL -ū, and FEM PL –āt.51 The

reconstruction of the 3fp of the stative/SC as *-āt has obvious problems: this form is not

attested in any Semitic language as a marker on the stative/SC and thus seems rather unlikely.

Other scholars have doubted Tropper’s connection of the “absolutive case” with the endings of

the stative/SC. Weninger, for example, states:

48
Von Soden (1995:96).
49
Most of these attestations come from cardinal numbers, such as ešrā ‘20’, šalāšā ‘30’ etc., that appear in the
“absolute state” when they stand in apposition before the counted entity (von Soden 1995:96; Huehnergard
1998:236, 238).
50
Tropper (1999:179).
51
Tropper (1999:191); see also Tropper (1995:494).

18
 “Tropper regarded the ending of the 3SG.M –a as PS and identical with the ending of

the alleged nominal absolutive case. As the existence of the absolutive case itself is

problematic, this reconstruction is doubtful, although there is no plausible alternative

suggestion.”52

Consequently, the origin of the 3fp –ā of the stative/SC is still unclear.

4 The nominal derivation of the third person markers

As the preceding discussion has shown, neither the nominal nor pronominal origin hypothesis

have yet been proven with any degree of certainty since both face problems that still need to

be resolved. In the following, I hope to show that the nominal origin hypothesis regarding the

derivation of the third person markers of the stative/SC is the more viable and less

problematic reconstruction. For this reconstruction, special attention has to be given to the

form of the 3fp –ā.

As has been described in the previous section, the pronominal theories face the

problem that they have to assume reconstructed forms of third person pronouns that cannot

be justified by evidence from attested pronouns in the Semitic descendent languages, which

weakens this approach significantly. The nominal theory equally was thought to face several

problems, although as mentioned in the previous section, most of the main points of criticism

against a nominal origin of the third person markers of the stative/SC have been shown to be

less problematic than initially assumed. The fact that the nominal base of the stative/SC does

52
Weninger (2011:162).

19
not show agreement, for example, can be explained as the result of grammaticalization, more

specifically the result of simplification.

This still leaves one more problem for the nominal origin hypothesis, namely the form

of the 3fp –ā. In order to address the form of the 3fp, it is first necessary to briefly look at the

form of the 3fs –(a)t and the appearance of an inflectional ending –ā marking the dual.

As has been argued above, the MASC SING originally ended in –a. This –a reflects an

inflectional ending that marks predicative function.53 For the FEM SING, the ending –t was

added to the MASC SING base *qatala, resulting in *qatala-t. The 3fs marker of the stative/SC

was thus simply *–t, not *–at as often reconstructed.54 It has been noted by a variety of scholars

that the morpheme –t was not originally a FEM marker but marked abstracts and singulatives

before it was reanalyzed as a general FEM marker.55

A non-singular marker –ā is attested in both the verbal system (Classical Arabic yaqtul-

ā-ni) and on nouns, where it marks the dual. On nouns, the ending –ā marks the nominative

dual, as in Classical Arabic bint-ā-ni ‘two girls (NOM)’. The morpheme –ā is thus attested on

53
See section 3 above. As argued in Hasselbach (2013), the ending –a is in reality no mere “predicative” marker but
reflects the vestige of an older marked nominative system, in which the accusative –a was the unmarked form and
had multiple functions, including marking nominal predicates, marking of the genitive, the direct object, and the
neutral or citation form of the noun (Hasselbach 2013:325-327). The “predicative” –a is thus not a separate
inflectional form of the noun but simply the accusative marker –a that preserves functions of an older system.
The idea that “predicative” –a is related to the accusative ending –a has also been proposed by Streck (2000:288).
The “absolute state” attested in Akkadian, Eblaite, Amorite, Old South Arabian, and perhaps Ugaritic, is directly
related to this form. The “absolute state” covers the same functions as the ACC across Semitic, especially marking
of adverbial functions and the use as neutral/citation form of the noun, and equally represents a vestige of the
earlier Semitic marked-nominative system (Hasselbach 2013:318-322). In Akkadian, the “absolute state”, which is
no state but a case, is unmarked in the MASC SING because Akkadian lost final short vowels.
54
For the reconstruction of the 3fs marker as *-at see, e.g. Huehnergard (2004:150).
55
For a summary of scholarship on the function of –t see Hasselbach (2014:323-328).

20
nouns, but not as a FEM PL marker. There are, however, important arguments that suggest

that the dual marker –ā on nouns goes back to an original FEM PL marker and, consequently,

that the FEM PL was originally marked by –ā on all types of nouns. When the morpheme –t was

reanalyzed from a morpheme marking abstracts and singulatives to a general feminine

marker, it was subsequently added to all nominal feminine forms independent of number,

including the original FEM PL marker –ā.56 The reanalysis and extension of the new feminine

marker -t resulted in the FEM PL marker –āt attested throughout Semitic. The original FEM PL

markers –ā (and its OBL counterpart –ay) were subsequently reanalyzed as dual markers, that

is, they were attributed a secondary function, according to Kuryłowicz’s fourth law of

analogy.57 There are several reasons to assume that –āt is secondary. First, it behaves

differently from the corresponding MASC PL markers –ū and –ī. The latter two are

monomorphemic, that is, they cannot be split up into further morphemes despite the fact that

they are multifunctional and mark gender, number, and case. The FEM PL marker –āt, on the

other hand, only marks gender and number. Case is indicated by adding the case endings of

MASC SING nouns to the plural marker, although with adjusting them to the diptotic system of

the MASC PL, that is NOM –āt-un and OBL –āt-in (as in Classical Arabic muˁallim-āt-un ‘teachers

[NOM PL]’ and muˁallim-āt-in ‘teachers [OBL PL]’). This means that the FEM PL is only

secondarily marked for case, unlike the MASC PL markers, and unlike the “DUAL” markers -ā

and –ay, which are monomorphemic like the MASC PL markers. Furthermore, the “DUAL”

ending –ay is still attested as PL marker in various Semitic languages, as in the Old Akkadian

56
For a more detailed discussion regarding the underlying process that led to the reanalysis of -t see Hasselbach
(2014:337-338). That –āt is a combination of an original nominal and verbal plural marker –ā to which the SING
ending –t was added secondarily has already been suggested by Bauer and Leander (1922:514).
57
Hasselbach (2007:132).

21
and Assyrian OBL PL –ē (< *-ay) and the construct ending –ê in Hebrew and Aramaic (e.g.

Hebrew malkê ‘kings of’). It is thus likely that the DUAL markers attested in Semitic languages

go back to original PL markers that were relegated to a secondary function, probably through

the use of these endings for paired body parts, after the reanalysis of –t as general FEM

marker.58

Another point that might strengthen the analysis of –āt as secondary is that it can be

observed that agreement marking for both gender and number most likely started on nominal

targets, that is ADJs including PTCs, not controllers, that is substantives.59 This can be deduced

from the following observations. It is clear that Semitic at an early stage did not mark gender

by morphological affixes. For certain nouns denoting human beings and domesticated

livestock, gender was indicated by stem alternation, that is by different words, such as Hebrew

ˀāb ‘father’ and ˀēm ‘mother’.60 Gender marking then developed on adjectival targets. This is

suggested by evidence from so-called epicene nouns. Epicene nouns do not distinguish gender

morphologically, but agreement targets, that is adjectives and verbs, distinguish gender, as in

Hebrew: gǝmallîm mênîqôt ‘nursing (FEM Pl) camels’ and gǝmallîm bāˀîm ‘camels were

approaching (MASC PL)’, where the controller has a gender neutral form, but the target, the

ADJ in this case, is marked for gender.61 Agreement morphology in Semitic thus originated on

attributive and predicative ADJs (including PTCs), that is on nominal targets, and spread from

there to nominal controllers. That this was indeed the case is also confirmed by the fact that

strict gender and number marking is only found on ADJs and PTCs. Nominal controllers exhibit

58
Hasselbach (2007:132-133). For the rise of –t as FEM and agreement marker see also Hasselbach (2014).
59
For a more detailed discussion of the typological literature and terminology see Hasselbach (2014:35).
60
For a more detailed discussion see Hasselbach (2014:43-44).
61
Epicene nouns are relatively frequent in Semitic for nouns denoting animals (Hasselbach 2014:45, 319-321).

22
numerous exceptions to strict agreement marking.62 The fact that gender and number marking

is only “regular” on ADJs and PTCs in the nominal system, suggests that these nominal targets

developed agreement morphology first and that agreement never completely spread to

nominal controllers, or, to say it differently, it never reached the same degree of regularity on

controllers as on targets.63

If nominal agreement morphology indeed first developed on targets, the stative/SC

would preserve an older system than the system attested on nouns. The 3fp –ā attested in the

stative/SC should consequently be considered more original than the ending –āt attested on

nouns.64

The ending –ā thus reflects an original nominal FEM PL marker in Semitic that is still

preserved on the stative/SC. Given this analysis, it is unnecessary to refer to the pronominal

system to account for the form. All third person markers of the stative/SC can be explained by

original nominal endings, resulting in the following paradigm:

3ms (original ACC)65 *qatal-a 3mp *qatal-ū

3fs *qatala-t 3fp (original FEM PL marker) *qatal-ā

62
In Akkadian, for example, many nouns that are MASC in the SING take a FEM PL marker, such as dīnum ‘legal
decision’ with the PL dīnātum. In Hebrew, a number of FEM nouns denoting female animates take a MASC PL
marker, such as nāšîm ‘women’, and vice versa, as in ˀābôt ‘fathers’. For more examples see Hasselbach (2014:46-
49).
63
Hasselbach (2014:323).
64
Attributive ADJs obviously followed the extension of –ā to –āt. The stative/SC, or at that point predicative ADJ
did not, however, since it had a different function and most likely was already in the process of being
grammaticalized into a synthetic form, which prevented the change from –ā > -āt.
65
See note 53 above.

23
The assumption that the third person markers of the stative/SC are simply inflectional endings

implies that third persons were not marked for person in the original predicative construction.

This phenomenon has parallels in numerous languages. As Kouwenberg states based on

Benveniste, the absence of an explicit subject marker in the third person is motivated by the

“nature of the third person as the “zero person””.66 The “zero person” indicates the person

who is not present at the speech situation. The lack of explicit third person marking in

predicative constructions agrees with the fact that many languages, including Semitic, do not

have third person pronouns.67 As is well known, anaphoric pronouns have taken over the

function of third person pronouns in Semitic. Given this cross-linguistic tendency, it is

unnecessary to assume a set of original third person independent pronouns as suggested by

Huehnergard. Semitic most likely never had a separate set of third person independent

pronouns.

Based on the arguments provided above, it seems likely that the third person markers

of the stative/SC are of nominal origin and reflect original gender, number, and case markers,

not person markers.

5 Conclusion

In this paper, it has been argued that the third person markers of the stative/SC in Semitic are

of nominal, not pronominal origin as claimed by a number of scholars. The 3ms –a can be

explained as the original ACC case marker that represented the neutral form of the noun and

marked nominal predicates, in addition to direct objects, in an early Semitic alignment system

66
Kouwenberg (2000:57).
67
Kouwenberg (2000:58).

24
that was marked nominative. The predicative function of the form was preserved in various

environments, such as early Semitic personal names and the stative/SC, where it marks the

predicate status of the first element of the construction, consisting of a verbal adjective. The

forms of the 3fs and 3mp do not pose any difficulty since they can directly be derived from the

corresponding forms of the nominal system. The 3fp in -ā is the most problematic form

regarding its derivation from the nominal system since there is no direct correspondence of

such a morpheme marking FEM PLs. As has been argued above, the form can, however, be

reconstructed as an archaic FEM PL marker that was secondarily extended by the morpheme –t

after the latter was reinterpreted as a general FEM marker. The stative/SC did not undergo the

same extension since it was already in the process of being grammaticalized from a

periphrastic into a synthetic construction and functionally differed from other nominal

constructions in being an overtly marked predicative construction.

The fact that the nominal base of the stative/SC does not show agreement can be

explained by the cross-linguistic tendency to simplify paradigms from more heterogeneous to

more homogeneous paradigms as a result of grammaticalization. The lack of agreement in the

stative/SC shows the degree of grammaticalization the form had undergone already at an early

stage of Semitic. As has been shown above, it is equally common cross-linguistically to not

explicitly mark the third person. That the third person forms of the stative/SC have no overt

person marker but simply constitute inflectional endings is thus not unprecedented but rather

common. This also ties well into the fact that Semitic has no original third person independent

pronouns but uses a paradigm of anaphoric pronouns as third person pronouns instead. The

latter confirms that Semitic is a language in which the third person can be left unmarked.

25
 Given that the issues that have been brought forth as arguments against the nominal

origin of the third person markers of the stative/SC can be explained by diachronic

developments and, in particular, common grammaticalization processes, the nominal origin

hypothesis seems to be the more likely derivation for the third person markers of the

stative/SC, especially since the problems faced with a pronominal derivation, cannot be solved

without having to reconstruct forms that have no basis in the attested Semitic languages.

Bibliography

Bauer, H. and P. Leander. 1922. Historische Grammatik der hebräischen Sprache (reprint 1991).

Hildesheim: Georg Holms.

Bergsträsser, G. 1928. Introduction to the Semitic Languages: Text Specimen and Grammatical Sketches

(translated by P.T. Daniels, 1983). Winona Lake: Eisenbrauns.

Brockelmann, C. 1908. Grundriss der vergleichenden Grammatik der semitischen Sprachen (reprint

1961). Hildesheim: Georg Olms.

Buccellati, G. 1968. An Interpretation of the Akkadian Stative as Nominal Sentence. JNES 27:1-

12.

Buccelati, G. 1988.The State of the “Stative”. In Fucus: A Semitic/Afrasian Gathering in

Remembrance of Albert Ehrman, Y.L. Arbeitman (ed.), 153-189. Amsterdam: John

Benjamins.

Buccellati, G. 1996. A Structural Grammar of Babylonian. Wiesbaden: Harrassowitz.

Cohen, D. 1975. Phrase nominale et verbalization en sémitiques. In Mélanges linguistiques offerts

à Émile Benveniste. Louvain: Peeters, 87-98.

26
Croft, W. 2003. Typology and Universals (second edition). Cambridge: Cambridge University

Press.

Diakonoff, I.M. 1988. Afrasian Languages. Moscow: Nauka.

Diem, W. 1997. Suffixkonjugation und Subjektpronomina: Ein Beitrag zur Rekonstruktion des

Ursemitischen und zur Geschichte der Semitistik. ZDMG 147:10-76.

Gelb, I.J. 1965. The Origin of the West Semitic Qatala Morpheme. In Symbolae Linguisticae in

Honorem Georgii Kuryɫowicz, W. Taszycki (ed.), 72-80. Krakow: Polska Akademia.

Givón, T. 1976. Topic, Pronoun, and Grammatical Agreement. In Subject and Topic, C.N. Li (ed.),

149-188. New York: Academic.

Hasselbach, Rebecca. 2004. Final Vowels of Pronominal Suffixes and Independent Personal

Pronouns in Semitic. JSS 49: 1-20.

Hasselbach, Rebecca. 2007. External Plural Markers in Semitic: A New Assessment. In Studies in

Semitic and Afroasiatic Linguistics Presented to Gene B. Gragg, Cynthia L. Miller (ed.), 123-

138. Chicago: The University of Chicago.

Hasselbach, Rebecca. 2013. Case in Semitic: Roles, Relations, and Reconstruction. Oxford: Oxford

University Press.

Hasselbach, Rebecca. 2014. Agreement and the Development of Gender in Semitic (Part 1 and

2). ZDMG 164:33-64, 319-344.

Heine, B and M. Reh. 1984. Grammaticalization and Reanalysis in African Languages. Hamburg:

Helmut Buske.

Hetzron, R. 1976. Two Principles of Genetic Reconstruction. Lingua 38:89-108.

Hodge, C.T. 1975. The Nominal Sentence in Semitic. Afroasiatic Linguistics 2:69-75.

Huehnergard, J. 1986. On Verbless Clauses in Akkadian. ZA 76:218-249.

27
Huehnergard, J. 1987. “Stative”, Predicative Form, Pseudo-Verb. JNES 47:215-232.

Huehnergard, J. 2004. Afro-Asiatic. In The Cambridge Encyclopedia of the World’s Ancient Languages,

R.D. Woodard (ed.), 138-159. Cambridge: Cambridge University Press.

Kouwenberg, N.J.C. 2000. Nouns as Verbs: The Verbal Nature of the Akkadian Stative. Orientalia

69:21-71.

Kouwenberg, N.J.C. 2010. The Akkadian Verb and its Semitic Background. Winona Lake:

Eisenbrauns.

Kraus, F.R. 1984. Nominalsätze in altbabylonischen Briefen und der Stativ. Nieuwe Reeks, Deel 47

No.2. Amsterdam: Noord-Hollandsche Uitgevers Maatschappij.

Krebernik, M. 1991. Buchbesprechung: Gelb, I.J. und B. Kienast: Die altakkadischen

Königsinschriften des dritten Jahrtausends v. Chr. Zeitschrift für Assyriologie 81:133-143.

Lehmann, C. 1985. Grammaticalization: Synchronic Variation and Diachronic Change. Lingua e

Stile 20:303-318.

Lipiński, E. 2001. Semitic Languages: Outline of a Comparative Grammar (second edition). Leuven:

Peeters.

Nöldeke, T. 1884. Untersuchungen zur semitischen Grammatik. ZDMG 38:4-7-422.

Soden, W. von. 1995. Grundriss der akkadischen Grammatik (3rd edition). Rome: Pontificium

Institutum Biblicum.

Streck, M.P. 2000. Das amurritische Onomastikon der altbabylonischen Zeit (Band 1). Münster:

Ugarit.

Tropper, J. 1995. Die semitische “Suffixkonjugation” im Wandel – Von der Prädikativform zum

Perfekt. In Festschrift für Wolfram Freiherrn von Soden zum 85. Geburtstag am 19. Juni 1993, M

Dietrich and O. Loretz (eds.), 491-516. Neukirchen-Vluyn: Butzon and Bercker Kevelaer.

28
Tropper, J. 1999. Die Endungen der semitischen Suffixkonjugation und der Absolutivkasus. JSS

44:175-193.

Voigt, R. 2002/2003. Die beiden Suffixkonjugationen des Semitischen (und Ägyptischen).

Zeitschrift für Althebraistik 15/16:138-164.

Weninger, S. 2011. Reconstructive Morphology. In The Semitic Languages: An International

Handbook, S. Weninger (ed.), 151-178. Berlin: de Gruyter.

Zimmern, H. 1898. Vergleichende Grammatik der semitischen Sprachen: Elemente der Laut- und

Formenlehre. Berlin: von Reuther und Reichard.

29