(PDF) The Semitic “Perfect” and the Prob

(PDF) The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme
The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme
Volume 145 / Number 2 / April–June 2025
ISSN 0003-0279

Articles
Sri Sathvik rayala, Code-Switching of Regal Identity in the Early Modern Deccan . . . . . . . . . . . . .
yaSuko Suzuki, Development of r-clusters in Gandhari: Interaction of Consonant Cluster Reduction
and Mobile Rhotics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
lu kou, “As If Heaven Was Warning”: Omens and Interpretations in Shen Yue’s Song shu 宋書 . . . .
JianJun he, A Young Lady’s Passion for Cats: Sun Sunyi and Three Qing Compendia on Cats . . . . . . .
na‘ama Pat-el, The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme . . . . . . .
nili ShuPak, The Joseph Story (Genesis 37–50) from an Egyptological Perspective: Multilayered
Story or Exilic Novella? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
eli tadmor, Counting Lines in Erra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
mohammad amin manSouri, In the Name of Letters: Basmala as the Cosmic Design . . . . . . . . . . . .
Sara omar, Gendering Sex: Delineating the Licit from the Illicit . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

229
253
285
309
327
349
369
381
401

Brief Communication
Guy ShaPira, mAa xrw before the Fifth Dynasty? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419
Reviews . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423

JAOS

JAOS (ISSN 0003-0279) is published quarterly by the
AOS, Harlan Hatcher University Library, 913 S . University Ave, Ann Arbor, MI 48109-1190 . All information
about the Society and the Journal, including membership
dues and the Journal style sheet, can be found at www .
aos-site .org . The Society and the Editors assume no
responsibility for the views expressed by authors in the
Society’s publications .

EDITOR-IN-CHIEF
PERI BEARMAN
Harvard University
ASSOCIATE EDITORS
GARY BECKMAN
University of Michigan
STEPHANIE W. JAMISON
University of California, Los Angeles

None of this content may be reprinted, reproduced,
distributed, or disseminated without express permission of
the author. Using this content in any generative AI model
is also prohibited.

THOMAS MAZANEC
University of California, Santa Barbara
COMPOSITION BY LOCKWOOD PRESS, P.O. BOX 1080, COLUMBUS, GA 31902

PRINTED AT SHERIDAN PA, HANOVER, PA 17331

Reviews
Conan-Wu, Lure of the Supreme Joy: Pedagogy and
Environment in the Neo-Confucian Academies of Zhu Xi
(Chu Ming-kin) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
lei, Yuan Tongli nianpu changbian 袁同禮年譜長編
(Nicholas Morrow Williams) . . . . . . . . . . . . . . . . . . . . . . .
roWlandSon, BaGnall, and thomPSon, eds ., Slavery
and Dependence in Ancient Egypt: Sources in Translation
(Juan Carlos Moreno García) . . . . . . . . . . . . . . . . . . . . . . .
WellS, ed ., The Cambridge Companion to Law in the
Hebrew Bible (Dylan R . Johnson) . . . . . . . . . . . . . . . . . . . .
Budin, Gender in the Ancient Near East (Martti Nissinen) . .
tor and Beihammer, eds ., The Islamic–Byzantine Border in
History: From the Rise of Islam to the End of the Crusades
(Nicolas Drocourt) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

423
425

426
428
431

434

Fahmy, In Quest of Justice: Islamic Law and Forensic
Medicine in Modern Egypt (Nahyan Fancy) . . . . . . . . . . . .
meStyan, Modern Arab Kingship: Remaking the Ottoman
Political Order in the Interwar Middle East (Paul L . Heck)
Brann, Iberian Moorings: Al-Andalus, Sefarad, and the
Tropes of Exceptionalism (David J . Wasserstein) . . . . . . . .
diez, ed ., Al-Makīn Ğirğis Ibn al-ʿAmīd: Universal History.
The Vulgate Recension, Part 1, Section 1: From Adam to
the End of the Achaemenids (Alastair Hamilton) . . . . . . . .
TeSei, The Syriac Legend of Alexander’s Gate: Apocalypticism at the Crossroads of Byzantium and Iran
(Christopher Bonura) . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

438
440
443

445

447

The Semitic “Perfect” and the Problem
of the Third Person Zero Morpheme
na‘ama Pat-el
univerSity oF texaS, auStin
In many languages, third person forms in a verbal paradigm are unmarked, and
scholars have suggested that such cases are either a result of loss or nondevelopment . In this paper I will argue that in the perfect/stative paradigm in Semitic, the
third person morphology is a result of nondevelopment . I suggest that these forms
are constructed as predicative adjectives, without person markers, because Semitic
never developed third person nominative pronouns . I further discuss other innovative verbal formations in Semitic and show that when subject clitics are noncanonical, for example in Neo-Aramaic, third person forms are clearly marked .

1. introduCtion

Many linguists, both typologists and historical linguists, have pointed out that third person
subjects on verbs are frequently unmarked, or marked with zero, while non-third person
subjects are overt (Siewierska 2009; Bickel et al . 2015; Cristofaro 2021); they disagree,
however, on the diachronic cause for this phenomenon . The options are, broadly, nondevelopment of third person markers or loss of these markers, and explanations for this zero vary .
Siewierska (2009) distinguishes between absolute zero, where a language lacks any person form in a particular paradigmatic slot, and paradigmatic zero, where one slot in the
paradigm, mostly third person singular, may have zero, but not others . Third person markers
may be expressed as zero in adnominal possessive position, object, or subject . Siewierska
examines the frequency and cause for a zero morpheme in all three and suggests that crosslinguistically subjects tend to be of the paradigmatic type (p . 434) . She concludes that a loss
scenario is more likely to explain zero third person subjects (p . 436) .
Indeed, there are several widely quoted examples of such loss that resulted in zero .
Watkins (1962) shows how the third person marker in the Old Irish preterit ended up being
zero through phonological reduction and the reanalysis of the third person marker as a tense
marker. Such an example is the third person form of the verb “be” in several Indo-European
languages, where the dental, which was originally a person marker (e .g ., Middle Welsh oe-d),
was absorbed in the stem and resulted in a synchronic reinterpretation of the third person as
zero; cf . Middle Welsh 3s oed vs . 1s oed-wn; Persian 3s hast vs . 1s hast-am (Watkins 1962:
94–95); Vallader Romantsch 3s chantet ‘he sang’ vs. 1s chantet-an (Haiman 1977: 322), etc .
On the basis of these and other examples, Watkins formulated a law, stating “it is the form
of the 3Sg which will determine the rest of the paradigm in an analogical transformation”
(Watkins 1962: 198) . The position that third person zero is a result of a synchronic principle
is reflected, for example, in Bybee (1985: 56): “This phenomenon demonstrates that the
Author’s note: The research for the paper was funded by a EURIAS Fellowship in Uppsala, Sweden, and was largely
conducted during my stay there (2012–13). Versions of the paper were presented at the University of Vienna (2015),
Saarbrücken (2017), Princeton University (2018), and the University of Cambridge (2023). I thank audiences there
for their comments and engagement . I greatly benefited from feedback on earlier drafts from Phillip Stokes, and
especially John Huehnergard, as well as from two anonymous reviewers for JAOS . All bibliographic abbreviations
follow the Reallexikon der Assyriologie und Vorderasiatischen Archäologie .

JAOS 145 .2 (2025)

327

328

JAOS 145 .2 (2025)

occurrence of zeroes in paradigms is not just an accident of the diachronic developments in
the syntax, but is a viable synchronic principle that reflects the psychological structuring of
verbal paradigms.”
Koch (1995) expanded on the work of Watkins (1962), Bybee (1985), and others and
suggested that synchronically in a paradigm one value is unmarked and least specific . Since
singular is semantically the unmarked value of the number paradigm and third is semantically the unmarked value of the person paradigm, the third person singular is the unmarked
value in any paradigm that marks such distinctions, i .e ., subject, possessor, etc . (Koch 1995:
45) . Koch takes this generalization even further and suggests a diachronic principle, according to which “a word-form which expresses by means of a non-zero marker a property which
is typologically expected to be coded by zero is liable to be reanalyzed as containing a zero
marker” (Koch 1995: 65). According to Koch, therefore, morphological change consists of a
drive to semantically create more iconic coding . 1
As is often the case, these generalizations tend to be insensitive to language-specific
processes, which are frequently motivated by internal micro changes or restrictions . Indeed,
recently Cristofaro (2021) has shown that what motivates the lack of third person subject
marking is not just a single phenomenon . She argues that it is not at all clear that third person
zeros are a unified phenomenon, in which case we cannot explain them as result of general
typological principles . Cristofaro advocates for a diachronic, source-oriented approach to
the study of this phenomenon, rather than an analysis based on the synchronic expression of
the third person, which she calls result-oriented . She states that “explanations for individual
patterns should refer to the diachronic processes that give rise to the pattern, rather than the
resulting synchronic pattern in itself” (p. 3).
The Semitic verb provides a good ground to demonstrate this claim . In this paper I concentrate on the diachrony of a form known as stative in East Semitic and perfect, or suffix
conjugation, in West Semitic (see Table 1), both of which are reconstructed to a single protoSemitic paradigm. In the paradigm of this verbal—or verboid—form, the third person is
morphologically different from the first and second persons . In the following I will describe
the morphology of the form and explore the reasons for this difference . I will make two
claims . First, that the third person is historically a nominal predicative form, which is only
marked for gender-number, not for person . Second, that the reason for this state of affairs
has to do with a feature of the pronominal system and restrictions on the morphosyntax of
demonstratives in Semitic .
In West Semitic this form is part of the verbal system and marks the simple past . In both
paradigms, the third person is constructed differently than the other persons . It has long
been recognized that the third person suffixes are not pronominal but are likely related to
the nominal inflection (Nöldeke 1884; Brockelmann 1908–13; Diem 1997; Huehnergard
2019) . Semitists, however, disagree on the exact origin of the suffixes and their relationship
to nominal inflection .
Table 1 . Stative / Perfect Paradigm
Akkadian (ES)

Geez (WS)

1cs

marṣ-ā-ku

nabar-ku

2ms

marṣ-ā-ta

nabar-ka

1. This claim turns out to be highly problematic, as Bickel et al. (2015) recently showed that there is in fact no
universal trend toward third person zeros .

Pat-el: The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme 329
2fs

marṣ-ā-ti

nabar-ki

3ms

mariṣ

nabar-a

3fs

marṣ-at

nabar-at

1cp

marṣ-ā-nu

nabar-na

2mp

marṣ-ā-tunu

nabar-kəmu

2fp

marṣ-ā-tina

nabar-kən

3mp

marṣ-ū

nabar-u

3fp

marṣ-ā

nabar-a

I will examine the Semitic perfect (in West Semitic) / stative (in East Semitic) and will
show that unlike the first and second persons, which are marked for person-gender-number,
the third person is a historically nominal predicative form, which is only marked for gendernumber . Proto-Semitic predicative adjectives carried a different set of suffixes than attributive adjectives, and I will show that the perfect/stative reflects that distinction . I will further
argue that the East Semitic form, where the 3ms is zero, represents a more original stage than
the West Semitic form, where the 3ms is marked with *-a, and will offer a scenario by which
this suffix was acquired . Thus, the Semitic data reflects a lack of development rather than
loss . I argue that the reason for the lack of development is structural: Semitic does not have
a third person personal pronoun and uses the demonstrative in this function . These demonstratives do not inflect for case and cannot be cliticized as subjects and so the language lacks
a source for pronominal clitics on newly formed verbs . An example of a development in
Aramaic shows that this feature is still at play in more recent new verbal formations .
1.1. The Semitic Language Family and Its Verbal Formation
The reconstructed verbal system of Proto Semitic is minimally built upon 1) a present
tense form, which includes gemination of the second radical (base -qattvl-) and a set of
person prefixes and gender-number suffixes; 2) a past tense form, whose base lacks gemination (-qtvl-), which is also marked with the same set of affixes used to mark the present .
The past is also used as a jussive with the additional prefixation of the particle la- (Huehnergard 1983) . This set is best exemplified with the East Semitic verbal paradigm attested, for
example, in Akkadian (see ex . 1) . Akkadian had additionally a perfect, whose semantics in
early Akkadian roughly correspond to the English present perfect; this verbal form, however,
is probably an Akkadian innovation .
1 .

a . Present:
b . Preterit:
c . Perfect:

i-parras :
3-decide .pres
i-prus :
3-decide .pret
i-ptaras :
3-decide .pf

i-parras-ū
3-decide .pres-mp
i-prus-ū
3-decide .pret-mp
i-ptars-ū
3-decide .pf-mp

Akkadian had an additional conjugation, conventionally known as “stative.” 2 The stative
is typically constructed from a predicative verbal adjective qatVl with clitic pronominal sub2. Whether it is a part of the verbal system in East Semitic is hotly debated among Semitists (Buccellati 1968;
Huehnergard 1987a) . Kouwenberg (2000) makes a robust argument in favor of taking the stative as a verb, and
recently so does Kamil (2023) . While I take the position that the stative is not part of the verbal system, the question
has no relevance to the issues I address here .

330

JAOS 145 .2 (2025)

jects . 3 This stem of the stative is indeclinable . Between the predicate and the subject pronoun, there is a long vowel, ā, whose exact function is still unclear . This vowel is not found
when the subject is third person . The connection between the independent personal pronouns
and the clitic subject pronouns is transparent, as Table 2 of the Akkadian forms illustrates
(independent pronouns are in parenthesis) . As this table makes clear, the third person is not
derived from the corresponding pronoun . Moreover, the third masculine singular has no
markers of any kind .
The stem of the stative is also used as an attributive adjective with the typical morphology
for attributive adjectives, namely number-gender-case-state . In other words, the stem of the
stative is not restricted to predicative position with clitic subjects; its distribution is typical
to all adjectives .
The pronominal clitics listed in Table 2 are more often found with verbal adjectives, but
in East Semitic they can also be used on nouns or non-verbal adjectives (exx. 2–3), complete
with the long vowel, an indication that the pronouns are indeed clitic, and are not fused to
the base . The stative is, therefore, constructed of a base that has other functions, and subject
clitics that can be used with other bases .
Table 2. Akkadian ”Stative” Conjugation
Singular
1c

qatl-ā-ku (anāku)

Plural
qatl-ā-nu (nīnu)

qatl-ā-ta (atta)

qatl-ā-tunu (attunu)

qatl-ā-ti (atti)

qatl-ā-tina (attina)

qatil (šū)

qatl-ū (šunū)

qatl-at (šī)

qatl-ā (šinā)

rēš-ētunu
ūlū-ma
hubbul-ātunu
slave-2mp
or
indebted-2mp
‘You are slaves or indebted’ (Old Babylonian, ARM 2 94: 15)
lā
sinniš-ātini
šūt
zakar
neg
woman-2fp
3ms
male
‘Are you not women (and) is he not a man?’ (Old Assyrian, CCT 5, 8b: 25–26)

In West-Semitic, the verbal system has been rearranged . The preterit/jussive is used as a
non-indicative form, and vestigially as an indicative past . It also took over the present and
ousted the old geminated present *yiqattal . The most revolutionary innovation of this branch
is the verbalization of the stative into a new past tense: a verbal form whose gender-numberperson markers are suffixed and are indivisible from the base . The base has no independent
existence outside this paradigm and the subject markers are fused to the base, unlike the East
Semitic stative . West Semitic also exhibits loss of the long vowel between the base and the
pronoun, and a clear marking of the 3ms with a final short vowel .
3. The adjective in this function covers a number of different meanings, depending on the basic semantics of
the root (Huehnergard 1987a: 224): Active transitive verbs will produce a passive (imḫur ‘he received’ : maḫir ‘it is
received’), active intransitive a resultative verbal adjective (ušib ‘he sat down’ : wašib ‘he is seated’); statives will
produce adjectival sense (irpiš ‘it became wide’ : rapaš ‘it is wide’), etc. (Huehnergard 1987a: 225).

Pat-el: The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme 331
The older and newer West Semitic verbal systems are represented in Classical Ethiopic
(ex . 4) and Classical Arabic (ex . 5) .
4 . Classical Ethiopic
a . Present:
b . Jussive:
c . Perfect:

5 . Classical Arabic
a . Present:
b . Jussive:
c . Perfect:

yə-qattəl :
3-kill .pres
yə-qtəl :
3-kill .juss
qatal-a :
kill .pf-3ms

yə-qattəl-ū
3-kill .pres-mp
yə-qtəl-ū
3-kill .juss-mp
qatal-ū
kill .pf-3mp

ya-qtul-u :
3m-kill .pres-ind
ya-qtul :
3m-kill .pres
qatal-a :
kill .pf-3ms

ya-qtul-ū-na
3m-kill .pres-mp-ind
ya-qtul-ū
3m-kill .pres-mp
qatal-ū
kill .pf-3mp

Table 3 . Classical Arabic Perfect Conjugation
Singular

Plural

qatal-tu (ʾanā)

qatal-nā (naḥnu)

qatal-ta (ʾanta)

qatal-tum (ʾantum)

qatal-ti (ʾanti)

qatal-tunna (ʾantunna)

qatal-a (huwwa)

qatal-ū (huma)

qatal-at (hiyya)

qatal-na (hunna)

The West Semitic perfect paradigm shows a very similar inflection pattern with several
differences (see Classical Arabic conjugation in Table 3) . 4 Like the East Semitic stative,
Arabic has an asymmetric verbal paradigm, where the first and second person markers are
transparently derived from personal pronouns, while forms of the third person are not .
2. the ProBlem

The third person stative/perfect paradigm presents us with a number of problems . First,
why are the third person slots not structured in the same way as the first-second person?
Second, where do the third person markers come from? In the following I will review the
most recent literature on this topic . I will offer a detailed reconstruction of each of the morphemes that is internally consistent and also explains the attested variation . Finally, I will
point to more recent innovations that show the same restrictions that I argue affected the
proto-Semitic paradigm .
4. Note, for example, the Classical Arabic conjugation (Table 3), where the -t- typical of the second person
spread to the first person, but essentially maintains the same relationship between the suffixes and the independent
pronouns as the Akkadian paradigm (Table 2) . Some changes have taken place, the most conspicuous of which
is the innovative ending on the 3fp -na, instead of the expected *-ā . This is possibly a result of analogy with the
relationship between 3mp and 3fp in the present: yaktub-ū (pres .3mp) : yaktub-na (pres .3fp) :: katab-ū (pf.3mp) : X
= katab-na (pf .3fp) (Huehnergard 2017: 20) .

332

JAOS 145 .2 (2025)

2.1. Why Is the Third Person Different?
As is clear from Tables 2 and 3 above, the relationship between the third person suffixes
and the independent pronouns does not correspond to what is expected based on the first and
second persons . Specifically, the third person suffixes are not derived from the pronominal
system . The reason for this is that the Semitic languages did not develop third person pronouns, but rather used distal demonstratives as third person pronouns . All branches, with the
sole exception of Arabic, still use the same form as both a distal demonstrative and a third
person independent pronoun . The same morpheme was also used as a third person possessive
suffix and an object suffix (see Table 4; Huehnergard and Pat-El 2012) . The distal demonstratives can be independent, in which case they can be nominative or, with the addition of
a suffixed *-ti, oblique, or suffixal, in which case they are identical to the nominative form
but cannot function as nominative . In other words, nominative third person can only be an
independent form . In comparison, first and second person pronouns have distinct forms for
subject and oblique, the latter marks both objects and possessive pronouns, which are only
distinct for the first person. See Table 5 (under the heading “oblique,” the leftmost form is
possessive and the rightmost object) .
Table 4 . Proto-Semitic Third Person Pronouns
independent
masc .sg

suffixal
-*su

*suʾa

fem .sg

*siʾa

-*sa/-*si

masc .pl

*sum(±uː)

-*sum(±uː)

fem .pl

*sin(±aː)

-*sin(±aː)

Table 5 . Proto-Semitic First and Second Person Pronouns
nominative

oblique

1cs

*ʾanā(ku)

*-ī/*-nī

1cp

*niḥ(nu)

*-ni/*-na

2ms

*ʾanta

*-ka

2fs

*ʾanti

*-ki

2mp

*ʾantum

*-kum(±uː)

2fp

*ʾantin

*-kin(±aː)

The pronominal system is, therefore, asymmetric: Distinct syntactic functions correspond
to distinct morphemes for first and second pronouns, but that is not true for the third person
pronoun . Rather, the different functions of the third person pronoun are marked syntactically:
When the morpheme is suffixed, it functions as the oblique form and when it is not, it functions as the nominative; for example, *yiqattal-su ‘he kills him’, but *yiqattal suˀa ‘he kills’.
This distribution blocks the possibility of a suffixed pronominal subject on the stative base,
because such a suffix will invariably be interpreted as nonsubject . The result is that third
person statives are built differently than 1 and 2 person statives .

Pat-el: The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme 333
2.2. The Origin of Third Person Verbal Suffixes of the Stative/Perfect
The stative/perfect markers show several peculiarities that are not easily explained .
Third Masculine Singular: The third masculine singular of the East Semitic stative is
unmarked (e .g ., damiq, maruṣ, wašab), while the West Semitic third masculine singular on
the perfect had a final vocalic suffix *-a (e .g ., samiʿ-a, ḥasun-a, faʿal-a) . 5 Although this final
vowel has been reduced in many daughter languages, e .g ., Hebrew, Aramaic, and modern
dialects of Arabic, it is easily reconstructible on the basis of verbs with object suffixes, e .g .,
Hebrew ʾākal ‘he ate’ but ʾǎkāl-a-ni ‘he ate me’. 6 Most scholars suggest that the *-a variant found in West Semitic must be reconstructed to Proto Semitic (Gelb 1965; Huehnergard
2019: 63; Hasselbach-Andee 2022) . The Akkadian evidence of a zero morpheme is explained
as an early loss of final short vowels (Huehnergard 2000: 591; Hasselbach 2013a: 320) . 7
Nevertheless, even if we assume that *-a is the original morpheme, it is not clear what its
function originally was .
Third Feminine Singular: This form, *-at, is identical to the fs marker on nouns and is
the least problematic morpheme in this paradigm .
Third Masculine Plural: The masculine plural morpheme *-ū is similar to the East
Semitic nominal plural morpheme -ū, as well as to the West Semitic plural morpheme -ū(na) .
Third Feminine Plural: The 3fs *-ā is not identical to the fp suffix on nominals, *-āt .
Its only other host is the 2fp and 3fp prefix conjugation in Akkadian (taprus-ā, iprus-ā) and
Classical Ethiopic (təgabbər-a, yəgabbər-a), which are thought to be secondary . The plural
morpheme on fp nominals is usually *-āt, not *-ā .
There is no agreement on the origin of the third person markers . Some scholars have
sought to connect them to pronouns . For example, Huehnergard (1987a: 222) suggests that
these endings are pronominal but were later completely lost and replaced by demonstratives .
Givón (2017: 74) similarly argues that these suffixes are “no doubt” relics of a defunct set
of independent pronouns . This is unlikely, as there is no evidence that the Semitic languages
had third person pronouns at all . Diem (1997: 71) proposes that the suffixes were derived
from the final vocalic ending of the demonstratives: 3ms *su’-a, 3mp *sum-ū, etc . This
proposal assumes that the stative suffixes are identical to the demonstrative final vowel, but
that is not the case . There is no evidence that the 3fs pronoun ended with *-at, and while
some languages show long vowels on the plural forms, it is not clear that those can be
reconstructed to proto-Semitic (Huehnergard 2019: 53–54). 8 The third person pronouns are,
therefore, unlikely to be the source of the vowels . Most scholars, however, have recognized
that the third person suffixes are not pronominal, but are likely related to nominal inflection
(Nöldeke 1884; Brockelmann 1908–13; Huehnergard 2019; Hasselbach-Andee 2022). The
third person proto-Semitic paradigm is hypothesized to be: 3ms *-a, 3fs *-at, 3mp *-ū, 3fp
*-ā .
How are these forms to be reconstructed? Many scholars believe that the 3ms suffix *-a
is an original case . Streck (2000) and Waltisberg (2011) argue that the final vowel -a, which
appears on masculine singular predicates in Amorite (West Semitic) and early Akkadian
5. The most common pattern in East Semitic is CaCiC, while in West Semitic it is CaCaC-a.
6. Some scholars claim that the vowel before the suffix on these forms is an inner Hebrew and Aramaic
development (Bauer and Leander 1918–22; Driver 1936); however, the final vowel was retained in some Yemeni
Arabic dialects, sometimes conditioned by suffixes, similarly to Hebrew, e .g ., in Minabbih (Behnstedt 2016: 191) .
7. This is also assumed in Hasselbach (2007a: 132 n. 51).
8. In fact, Huehnergard (2019: 54) suggests that these vowels were transferred to the third person pronouns from
the stative suffixes, not vice versa .

334

JAOS 145 .2 (2025)

names, is an oblique case ending that is common on nominal predicates cross-linguistically;
however, as Hasselbach (2013a) notes, the same vowel also appears on subjects of nonverbal
and verbal predicates, which is problematic for an oblique interpretation . 9
The most recent treatment of the third person problem is Hasselbach-Andee (2022), who
argues that third person suffixes on the stative are nominal . The two morphemes she finds
the most in need of explanation are the 3ms *-a and the 3fp *-ā . She connects the 3ms *-a to
an accusative case in a marked nominative system, which she reconstructed to proto-Semitic .
The 3fs *-at is built on the basis of the 3ms, by attaching the nominal feminine marker *-t on
the case of the 3ms *-a, producing BASE-a-t (Hasselbach-Andee 2022: 567) .
The 3fp *-ā was, according to Hasselbach-Andee (2022), the original fp on nominals,
at a period when the morpheme *-t had not yet been pressed into service as a gender morpheme . 10 The typical fp suffix on nominals, *-āt, was developed later, and was formed by
attaching the nominal gender morpheme *-t to the plural morpheme *-ā, once the former
had acquired a gender function . The 3fp suffix on the stative/perfect did not participate in
the innovation that yielded the nominal *-āt, because, according to Hasselbach-Andee (2022:
n . 66), the stative/perfect was already in the process of grammaticalization into a synthetic
form, and 3fp -*ā was therefore part of the morphology of the stative/perfect, not a freestanding fp morpheme .
Based on Hasselbach’s analysis and suggested reconstruction, the third person suffixes
on the stative/perfect are, therefore, older than the suffixes on nominals and reflect a more
archaic system . The following is a summary of this proposal:
3ms *-a < old accusative case from the original marked nominative system;
3fs *-a-t < nominal feminine morpheme -t attached directly on the masculine suffix (originally a case);
3mp *-ū < nominal masculine plural morpheme;
3fp *-ā < the original nominal feminine plural morpheme, before the development of the
nominal feminine morpheme -t .
There are, however, a number of problems with this explanation . If the 3ms *-a is an older
case, it is unclear why 3ms is the only form that reflects case . The 3fs morpheme terminates
with a consonant and can theoretically carry an oblique case, **-at-a . Furthermore, if the
3mp morpheme is identical to the nominal plural morpheme, it too could host case . Nominal plurals have an oblique case suffix *-ī, which is reconstructed to proto-Semitic . But the
attested 3mp is always *-ū, a nominative .
The hypothesis that the feminine morpheme *-t is a later addition, which is attached to
the suffix of the 3ms, is unlikely for several reasons . First, Hasselbach-Andee’s explanation
(2022) implies that the 3fs was originally zero . If it was not, it is unclear what morpheme was
used and how and when it disappeared . Second, in explaining the 3fp, Hasselbach-Andee
argues that at the time the stative/perfect developed, *-t did not yet mark gender, and that is
why the 3fp is *-ā, not **-āt . This explanation is inconsistent with the argument that the morpheme *-t on 3fs marks gender . 11 Third, in the nominal system, the gender morpheme precedes the case morpheme, not vice versa, e .g ., malk-at-a-m ‘queen (king.f.acc.abs)’. There is
9. Hasselbach (2013a) instead suggests that the final vowel -a is an accusative marker . She argues that Semitic
was originally a marked-NOM language; in such a system one of the functions of the accusative is marking the
predicate in a copulaic sentence . Hasselbach (2012) argues that this vowel developed from the accusative case
which marked nominal predicates in the original Semitic system (see also Hasselbach 2013a) .
10. Hasselbach-Andee (2022) assumes gender was a secondary function of this morpheme; originally it was a
marker of abstracts and singulatives .
11. In fact, regardless of the origin of 3fs morpheme, *-t must have been established as a marker of gender

Pat-el: The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme 335
no reason to assume that in proto-Semitic the gender morpheme was attached after the case .
For all these reasons, it is hard to accept Hasselbach-Andee’s reconstruction of 3fs .
The proto-Semitic mp morpheme on nominals is generally assumed to be *-ū-na
(Hasselbach 2007b: n . 47; Huehnergard 2019: 59) . Furthermore, -ū is the nominative form
of the masculine plural, which is inconsistent with the assumption that the 3ms is an accusative case in this reconstruction . Both these issues require an explanation .
Finally, if indeed nominal fs *-t is an old proto-, or pre-proto-Semitic morpheme, the
assumption that *-ā is an original nominal fp marker, which only later shifted to *-ā-t, is
questionable . If such a development indeed took place, it must have happened before Semitic
split from Afro-Asiatic, since the gender marker *-t is a feature of Afro-Asiatic and was
inherited by proto-Semitic . As noted above, the development of the 3fp is inconsistent with
the development of 3fs in Hasselbach-Andee’s reconstruction (2022) .
More problematic still, per Hasselbach-Andee’s reconstruction, each one of the suffixes
developed at a different period: 3fs developed after the development of 3ms and 3fp was
complete, but 3fp was not affected by the process that affected 3fs (the suffixation of the
gender morpheme *-t); the 3ms developed at a different time (earlier?) than 3mp, since these
morphemes do not reflect the same case system . Furthermore, the assumption that the stative
was already in the process of grammaticalization into a synthetic form is difficult to square
with examples in Akkadian, where the suffixes of the stative are attached to predicative substantives (exx . 2, 3); such usage is indicative that the morphemes were detachable and not
part of a synthetic form . Indeed, Huehnergard (1987a) has argued that the stative in Akkadian is synchronically a syntactic construction, because the formation and function of such
predicative substantives is analogical to the stative with predicative adjectives: išpar-at ‘she
is a weaver (weaver-3fs)’ . In other words, this construction is not yet grammaticalized as a
synthetic form in Akkadian, and by extension, in proto-Semitic . Thus, Hasselbach-Andee’s
account of why 3fp *-ā did not change to *-āt is not persuasive .
In the next section I will propose a different reconstruction, which considers these suffixes
as a system .
3. the oriGin oF third PerSon verBal SuFFixeS oF
the Stative/PerFeCt: a neW ProPoSal

I will examine the 3ms *-a, its relationship to 3fs *-at, and the relationship of the plural
markers, 3mp *-ū and 3fp *-ā, to the nominal system . I will argue that the original 3ms was
-0 (zero), and that the stative reflects a set of predicative adjectival suffixes, which stands in
opposition to a set of attributive adjectival suffixes .
3.1. 3ms *-a
The current consensus holds that the West Semitic 3ms suffix *-a should be reconstructed
to proto-Semitic (Nöldeke 1884; Huehnergard 2019: 63; Hasselbach-Andee 2022) . The lack
of any 3ms suffix in Akkadian and Eblaite is explained as a result of early loss of final short
vowels, especially a and u, a loss that is assumed to have happened before the attested period,
at least of Akkadian (Huehnergard 2000: 591) . This assumption largely rests on archaic personal names, where the stative ends with a . Whether these names support the originality of a
is unclear; the morpheme a is more often missing and sometimes it occurs with a fs subject . 12
already in proto-Semitic, if not earlier; Gragg (2019: 39) and Huehnergard (2023: 146) list nominal fs *-t as a shared
Afro-Asiatic feature .
12. Note, however, that Kouwenberg (2014) argues that names are unreliable evidence.

336

JAOS 145 .2 (2025)

There are limited environments in which final short vowels are expected to occur, vowels
whose absence will indeed verify the assumption of loss, because generally in Semitic there
are very few occasions for short vowels to occur in final open syllables . Huehnergard (2006:
n . 31) suggests two environments to support the hypothesis of loss: the nominative dual and
case vowels on construct nouns . I will argue that the evidence that final short vowels were
lost in proto-Akkadian is inconclusive at best and that 3ms stative is 0 in Eblaite where final
short vowels were not lost .
The nominative dual in all dialects of Akkadian and Eblaite is -ān, without a final vowel .
In West Semitic, Arabic has -āni, while Ugaritic has both -āmi and -āma (Huehnergard
1987b: 298–99). Under the assumption that Old Akkadian has not lost final -i, Huehnergard
(2006) reconstructs *-āna to proto Semitic and suggests that *-āni is a result of dissimilation
in Arabic and Ugaritic . Apriori, solely based on the evidence of the West Semitic dual, there
is no reason to prefer one form over another . The reconstruction of *-āna is preferred only if
the loss of case vowels on construct nouns is a result of the same process that produced *-ān .
As I will argue below, that is unlikely .
Bound nouns are considered the strongest support for the loss of final short vowels . They
are generally unmarked for case in Old Babylonian (mi-ik-sa-at A .ŠA-ia in ex . 6) and Old
Assyrian (ex . 7) .
6.

bēl-ī
l-išpur-ma
miks-āt
eqlī-ya
lord-my
mod-write.pret-coord
tax-fp.bnd
field.p-my
l-iddin-ū-nim-ma
lā
eberre
mod-give-p-vent-coord
neg
starve.1s
‘My lord should write so that they should give me the taxes of my fields and I will
not starve’ (Old Babylonian, AbB 13, 4: 19)
kiṣer
libb-em
la
tarašši-am
knot.cnst
heart-gen
neg
acquire.dur.2ms-me
‘Don’t be angry with me!’ (Old Assyrian, KBo. 9.9: r. 3’–4’)

In Old Akkadian, the genitive i is typically retained (ex . 8), while the other cases are generally lost, although examples of u are also attested (ex . 9) .
8.

in id-i
sipr-ī
in hand-gen messenger-my
‘by my messenger’ (Old Akkadian, CT 50, 70: 11)
šukān-u
KUG.BABBAR
one
ornament-nom
silver
‘one silver ornament’ (Old Akkadian, ELTS 40: A ii 12)

In Eblaite, case vowels on construct nouns are generally retained (exx. 10–13). While
there are scholars who argue that bound nouns in Eblaite have no case vowels and treat
examples with final case vowels as either frozen semi-logographic or as a representation of
the final consonant with a CV sign, Kogan and Krebernik (2021) dismiss these claims . In
Eblaite, when CVC is represented by CV-CV signs, both signs typically have the same vowel

Pat-el: The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme 337
(Kogan and Krebernik 2021: 678) . They collected many examples where the final CV sign
does not match the vowel of the preceding CV sign but corresponds systematically to the
relevant case . Therefore, they argue that these vowels represent a phonetic reality and are
indicative that Eblaite preserved case ending on bound nouns .
10 .

11.

12.

13.

ì-zu
gu-wi
tree .nom
copper .gen
‘copper tree’ (Kogan and Krebernik 2021: 845)
ru12-zi
UR.SAG.A
support.gen
hero
‘the support of the hero’ (Kogan and Krebernik 2021: 850)
ú-ḫu-wa-du
ma-ríki
brotherhood.fp.nom
GN
‘brotherhood with the city of Mari’ (Kogan and Krebernik 2021: 846)
bù-su-ga
NINDA
scarcity.acc
bread
‘scarcity of bread’ (Kogan and Krebernik 2021: 846)

As was noted above, the stative in Eblaite shows a -0 morpheme on the 3ms stative . If the
evidence of the case vowels on construct heads is relevant to this discussion, then the lack
of *-a on the Eblaite stative is inconsistent with the retention of the case on construct heads .
However, case vowels on bound nouns are final only orthographically, but prosodically the
construct functions as a single stress unit (see also Kouwenberg 2014: 188) . The heads of a
construct are unstressed proclitics, which are prosodically dependent on the following element (Huehnergard 2019: 53) . They essentially form a new stress unit with the following
noun, which results in the adjustment of word boundary, which locks the final syllable of the
head . This has implications for the morphology of such nouns . In Arabic, case is retained the
longest on bound nouns because short case vowels are protected in this position (van Putten
and Stokes 2018: 159) . Similarly, bound forms in Phoenician also preserve case longer than
unbound forms (Hasselbach 2013b) . In languages that lost case vowels, the status of the head
noun as a proclitic can be seen in the resurfacing of the gender morpheme -t . In Hebrew
and Arabic, for example, the gender marker *-at was reduced on independent nominals and
resurfaced on construct nouns . The reasons for this are both prosodic and morphosyntactic
(Faust 2014) .
The retention of the genitive case on construct heads in Old Akkadian and in some relics in Old Assyrian (e .g ., inūmi), can be explained as either morphological (i .e ., related to
its function) or phonological (i .e ., final short u/a were lost, but not final short i) . 13 Hasselbach (2013b) adopts the former explanation and argues that the case system first collapsed
on bound nouns, but that “Babylonian did not simply merge the singular nominatives and
13. The relative markers should also be included here, as they are construct heads as well. The feminine singular
and plural and the masculine plural forms should have short case vowels after the t in Old Akkadian (mp .gen θūti)
and Eblaite (fs .nom θatu, obl . θati) .

338

JAOS 145 .2 (2025)

accusatives because of the loss of final short /a/ and /u/ but first underwent a morphological
merger of the two cases” (p. 220; my italics). 14 In other words, Hasselbach argues that the
loss of a and u is not related to the alleged loss of final short vowels . Based on this view, the
lack of final short vowels on bound forms does not, therefore, support the hypothesis of final
short vowel loss in East Semitic . 15
But even if we prefer a phonological explanation, namely, that final short i was preserved longer than final short u/a (Hasselbach 2005: 104–5), this scenario is contradicted
by the subordination markers -u and -a . These are preserved in final position in Old Akkadian (ex . 14), Old Babylonian (ex . 15), and Old Assyrian (ex . 16) . They were eventually
lost whenever they were not protected by additional morphemes, but they were consistently
attested in all early Akkadian dialects .
14 .

15.

16 .

in
ši
uštābil-a
in
house
rel .ms .gen
consider .pret .1cs-sub
‘in the house that I considered’ (Old Akkadian, OAIC 10, apud Hasselbach 2012: 132)
ṭēm
Ibni-Marduk ša
ana
Bābilim
ašpur-u
report.bnd
PN
rel
to
GN
send.1cs.pre-sub
‘the report about Ibni-Marduk that I sent to Babylon’ (AbB 10, 79: 10–12)
ina
šamš-i
ṭuppa-ka
ašme-u
on
sun-gen .bnd
tablet .acc-you
hear .pret .1cs-sub
‘on the day when I heard your letter’ (Old Assyrian, CCT 2, 44a: 6–7)

The evidence so far is mixed . There are some cases in which final short vowels are
retained where according to the loss scenario they should not be . On the other hand, the
assumption that final short i is retained is complicated by the evidence from the dual *-āni,
which is attested without a final vowel in East Semitic . I suggest that it is likely that East
Semitic did not lose final short vowels in the proto-East Semitic stage or alternatively, that
the process has not yet been completed by the Old Akkadian period, and likely even later .
The hypothetical vowel on the 3ms stative remains the primary example of the loss of
final short vowels at this early stage of Akkadian; this assumption, however, privileges the
3ms suffix *-a in West Semitic . Given that the earliest attested dialects of Akkadian preserved several examples of final short vowels, we would expect some evidence of a final
*-a on the stative, but the only evidence for the existence of such a vowel comes from West
Semitic sources . In other words, there is no internal evidence to support the existence of an
original final vowel on the 3ms in East Semitic .
I suggest an alternative solution based on the assumption that the East Semitic -0 morpheme is original and the West Semitic morpheme *-a is secondary . Such an assumption
is consistent with the East Semitic evidence . Under this assumption, 3ms is the only mem14. A similar collapse is attested with the mp relative marker in Eblaite, where the oblique and the nominative
have the same form, θūti .
15. There are additional examples where a final short vowel occurs inconsistently. One such example is the 3ms
accusative pronoun šu’a (su4-a) in Sargonic royal inscriptions, which is treated as an exception (Hasselbach 2005:
103–4). First and second pronominal suffixes on the statives are sometimes written without their final vowel, e.g.,
*-āti 2fs ∼ -āt; original long vowels may also be eliminated, e .g ., *u-lā (‘negation’) > ula > ul or 3ms objects *-šū
> *-šu ∼ -š .

Pat-el: The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme 339
ber of the stative/perfect paradigm with no overt morphological marking . I suggest that
the morpheme *-a developed as a result of a reanalysis of the feminine gender morpheme
*-(a)t, which, as noted above, is at least proto-Semitic . This morpheme is what distinguishes
between masculine and feminine nouns—for example, ms awīl-0-um ‘free man’ vs. fs awīlt-um ‘lady’, ms šarr-0-um ‘king’ vs. fs šarr-at-um ‘queen’. Since the feminine suffix had
two allomorphs, -at (after CC) and -t (after C), it was possible for speakers to reanalyze the
vowel of the morpheme *-at as a distinct morpheme and transfer it to the masculine, creating a coherent and symmetric set where both feminine and masculine are clearly marked
after the base . This was possibly aided by the model of the second person singular where
the difference between masculine and feminine rests on a single phoneme: 2ms -k-a, 2fs -k-i
(see Table 2) . The change I am suggesting is summarized in Table 6 . If this is correct, then
initially, the third masculine singular on predicative adjectives was marked as zero, as is the
case in East-Semitic .
Table 6 . Suggested Development in West Semitic
Proto-Semitic
3ms

*qatal-0

3fs

*qatal-at

(re-analysis)
*qatal-a-t

West Semitic
qatal-a-0
qatal-a-t

3.2. 3fs -at
The 3fs morpheme *-at is identical to the gender morpheme on nominals . While substantives are not necessarily marked for gender, adjectives consistently are, whether they
are attributive or predicative . Since stative/perfect 3fs morpheme and the nominal gender
morpheme are identical in form and function, their historical relationship is fairly secure .
3.3. 3m plural -ū
Many scholars reconstruct the nominal masculine plural to *-ū-na on the basis of the
West Semitic nominal plural . The lack of the final syllable in Akkadian is explained as loss
of final -na . There is, in fact, no convincing evidence that East Semitic lost final -na (Bjøru
and Pat-El 2021) . In the only other environment where a similar syllable is expected, namely,
the 2–3fp prefix conjugation (e.g., 2fp *taqtul-na), the morpheme was probably replaced,
not lost (Huehnergard 2005: 169–70). 16 The 3mp -ū is also used as a gender-number marker
on the prefix conjugation of the second and third persons: Akkadian iparras-ū, Classical
Ethiopic yəqtəl-u / təqtəl-u, Arabic yaktub-ū-na / taktub-ū-na . In the nominal system, this
morpheme is found in both East Semitic and West Semitic . I therefore assume that the protoSemitic sound plural is *-ū.
In East Semitic, predicative adjectives are morphologically distinct from attributive adjectives: predicative mp -ū vs . attributive mp -ūt (before the case marker) . This distinction is
still used in East Semitic, but was largely lost in West Semitic, where the attributive mp

16. Recently, Suchard and Groen (2021) suggested that the original Semitic plural was -w in pre-proto-Semitic,
followed by case with masculine nominals (*W-u > ū), or the gender morpheme -at with feminine nouns (*W-at
> -āt), which accounts for the long vowel of the plural morphemes . They did not address the status of -na on
masculine plurals in Semitic .

340

JAOS 145 .2 (2025)

adopted the morphology of the predicative mp (see Table 7) . 17 Nevertheless, this morphological distinction should be reconstructed to proto-Semitic, on the basis of the mp relative
marker in Ugaritic, dt (dūt), which shows an attributive morphology with final suffix -ūt
(Pat-El 2022). 18
Table 7 . Predicative and Attributive Masculine Plural Morphemes
Attributive

Predicative

Akkadian

-ūt-

-ū

Arabic

-ū-na

-ū

Sabaic

Aramaic

-e

-in

Ugaritic

-ū-na

-ū

The 3mp on the stative/perfect is therefore identical to the mp predicative adjectival morpheme -ū and reflects its original function .
3.4. 3f plural -ā
The analysis of the 3fp morpheme is more complicated . The morpheme is used as a feminine plural marker on the prefix conjugation of the second and third persons in East Semitic
(iddin-ā, taddin-ā ‘they.f gave, you.cp gave’) and Classical Ethiopic (yənabbər-a, tənabbər-a
‘they.f sit, you.fp sit’); however, it is generally assumed that the original fp morpheme for
second and third persons was -na, as it is in most Central Semitic languages (e .g ., Arabic
yaktub-na ‘they.f write’). Under this hypothesis, the forms in Akkadian and Ethiopic changed
as a result of a simple analogy with the stative/perfect: 3mp qatVl-ū : 3fp qatVl-ā :: 3mp
yaqtul-ū : X = 3fp yaqtul-ā (Huehnergard 2005) .
Like mp adjectives, fp adjectives originally had a predicative/attributive distinction, a
distinction whose relics are still observable in attested Semitic languages (Table 8) . This
original predicative fp morpheme was retained in Akkadian, Aramaic, Sabaic, Ugaritic,
and Ethiopic, lost in Hebrew, and replaced by an innovation in Arabic; the predicative/
attributive distinction is therefore robustly attested in Semitic . We should, therefore, reconstruct the adjectival plural morphemes thus: attributive: mp *-ūt-, fp *-āt-; predicative: mp
*-ū, fp *-ā .

17. Aramaic developed secondarily a distinction between attributive-predicative, based on its defunct definiteindefinite morphology .
18. For example,
kl
bnš-m
dt
ḫbṯ
ṯmn
all
people-mp rel .p
run
there
‘all the people who run there’ (Tropper and Vita 2019: 489)

Pat-el: The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme 341
Table 8 . Predicative and Attributive Feminine Plural Morphemes
Attributive

Predicative

Akkadian

-āt-

-ā

Arabic

-āt-

-na

Sabaic

Aramaic

-āt-

-ā

Ugaritic

-āt-

-ā

The use of -ā to mark the feminine plural of a predicative form is, therefore, expected
(pace Hasselbach-Andee 2022). The analogy that produced the Akkadian and Ethiopic 2–3fp
morpheme in the prefix conjugation builds on the association of this morpheme with predicative adjectival forms .
4. an alternative ProPoSal

Given the discussion above, I suggest that the reconstructed set of suffixes should be 3ms
*-0, 3fs *-at, 3mp *-ū, 3fp *-ā . This set is similar, but not identical, to the reconstructed
gender-number markers on nominative nominals in Semitic, as is reflected in Akkadian (see
Tables 9, 10) . I suggest that the third person stative/perfect forms are adjectival bases with
gender-number predicative suffixes, but without case . If this is correct, the East Semitic
stative and the West Semitic perfect essentially have no pronominal subject expressed on the
third person forms; these forms are morphologically nominal . 19
Table 9. Akkadian Nominal Inflection
Akkadian nominal
inflection

masculine

feminine

nom . sg

šarr-0-um

šarr-at-um

acc . sg

šarr-0-am

šarr-at-am

gen . sg

šarr-0-im

šarr-at-im

nom . pl

šarr-ū

šarr-āt-um

obl . pl

šarr-ī

šarr-āt-im

Table 10 . Proto-Semitic Predicative and Attributive Adjectives
Predicative

Attributive

masc . sg

-0

-0-um

fem . sg

-at

-at-um

masc . pl

-ū

-ūt-um

fem . pl

-ā

-āt-um

Thus, first and second person suffixes are subject pronouns while the third person suffixes
are predicative nominal gender-number suffixes . This asymmetry is a result of structural
19. Buccellati 1996: 166 argues for a similar analysis for Babylonian Akkadian.

342

JAOS 145 .2 (2025)

constraints, because the Semitic languages did not develop third person pronouns but rather
used demonstratives to fill this function . These demonstratives mark their different functions syntactically: independent forms are subjects and clitic forms are oblique . When a new
verbal form, such as the stative/perfect, was constructed by cliticizing nominative pronouns,
there was no source material for the third person . In the context of a paradigm, third person forms are recognizable as predicative, but there are no specific markers to indicate the
subject, as there are with other persons . Syntactically, third persons are typically, though not
obligatorily, expressed via an overt subject, because they are less specific than other persons .
Thus, even without explicit subject clitics, the forms are still unambiguous .
While other explanations for the form of the third person of the stative/perfect have raised
good and important points, the current proposal accounts for all the forms found in the third
person stative/perfect paradigm . It further explains why the morphology of the third person
is different from that of other persons . In short, it offers a unified and simple solution . As the
following section aims to show, there is some supporting evidence from later development
in Aramaic for both the inability of the third person to function as a clitic nominative subject
and for the purely nominal inflection of the third person in verbalized adjectives .
4.1. Additional Evidence
The stative/perfect is not the only verbal innovation in the history of Semitic . In Late Aramaic dialects, represented below by Syriac, a new formation is starting to emerge based on
the active participle with clitic personal pronouns as its subject (Goldenberg 1983) . Unlike
the Akkadian form, in Syriac the adjectival base carries two sets of markers: a gender-number suffix attached directly to the base, and a subject clitic pronoun, marked for gendernumber-person, attached to it . The person suffixes are derived from the independent personal
pronouns (see Table 11) .
Table 11 . Syriac Participial Present
Singular
1ms

kātēb-0-nā (ʾənā)

Plural
kātb-īn-nan (ḥnan)

1fs

kātb-ā-nā (ʾənā)

kātb-ān-nan (ḥnan)

2ms

kātēb-0-att (ʾatt)

kātb-īn-tōn (ʾəttōn)

2fs

kātb-ā-atty (ʾatty)

kātb-ān-tēn (ʾəttēn)

3ms

kātēb-0 (hū)

kātb-īn (ʾennōn)

3fs

kātb-ā (hī)

kātb-ān (ʾennēn)

In example 17 below, the first suffix after the base, -ān- is the gender-number (fp) marker
of the predicative base, while the clitic -tēn is the person marker (2fp):
17.

kātb-ān-tēn
write.act.ptcl-fp-2fp
‘You (fpl) are writing’ (Syriac)

However, predicates whose subject is third person carry no person clitic; these forms have
a base with gender-number suffixes only; see example 18 .

Pat-el: The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme 343
18.

kātb-ān
write.act.ptcl-fp
‘they (fpl) are writing’ (Syriac)

These predicates differ from attributive adjectives, which carry a historical, but synchronically dysfunctional, definite article. This marker is known as “emphatic” in Aramaic studies . Synchronically, this marker serves to distinguish attributive adjectives from predicative
adjectives which lack it altogether . The adjectives in examples 17 and 18 are predicative
by nature of not carrying the emphatic morpheme . In practice, this means that third person
predicates are more likely to have nominal overt subjects, which is unsurprising cross-linguistically, but it is not a necessity; see example 19 below .
19.

lā-hwā
meṭṭōl
d-ṣāb-ēn-nan
b-hēn
neg
because
rel-want.act.ptcl-mp-1p
in-them.fp
hāwy-ān
l-an
be.act.ptcl-fp
to-us
‘It is not because we wanted them that they are ours’ (Syriac, Bardesanes, 571, 22–26)

While second and third person suffixes are clearly derived from independent personal
pronouns, the third person remains unmarked for person, and keeps its original adjectival
bases with only nominal gender-number suffixes; see Table 11 for the full conjugation . 20
The forms in parentheses are the independent personal pronouns . These bases are, of course,
shared with other members of the paradigm . 21 As was the case with the East Semitic stative,
in Syriac too, third person pronominal clitics cannot function as subjects . 22
Of course, the similarity between the formation of the West Semitic perfect and the Syriac
predicative participle has not escaped Semitists (Cohen 1975; Hasselbach 2012; Goldenberg
2013: 154–56). Both verbal forms are constructed on the basis of verbal adjectives with suffixed subject personal pronouns . 23 Like the perfect, the Syriac paradigm is a new formation
created by cliticization of subject pronouns, with the exception of the third person .
While in proto-Semitic the third person subject and the third person oblique were morphologically identical (Huehnergard and Pat-El 2012), in Syriac that is not the case, and the
third person oblique is morphologically distinct from the subject pronoun, e .g ., 3ms subject
hū, 3ms oblique -ēh . Nevertheless, third person clitics can only be interpreted as oblique . In
other words, the inability of the third person to serve as a clitic subject has nothing to do with
its morphology . The direct object of this new Syriac verb is expressed via a prepositional
20. Note that these are written forms; in pronunciation, n regularly assimilates under this condition: nC > CC,
even across morpheme boundaries .
21. Because the base is inflected for gender-number, Syriac has a marked first person feminine, unlike in
Akkadian, where the base is not inflected for gender-number .
22. A clitic form of the third ms independent pronoun in Syriac, -(h)ū, is regularly attached to noncanonical
predicates (prepositions, nouns, pronouns, interrogatives, etc .) to mark their status, even when the expressed subject
is not a third person ms (Goldenberg 1983) . It is clear, therefore, that the third person ms pronoun can be cliticized,
yet it still does not function as a subject .
23. Diem (1997: 47) argues that Syriac and Akkadian reflect a different process because the Syriac base carries
gender and number morphemes on all the bases, while the Akkadian stative does not . Diem also assumes that the
Syriac third person forms developed from a participle with a suffixed subject pronoun (qāṭel-ū < qāṭel hū, etc .), and
are not originally bare forms . This is, however, not correct . There is no evidence that the third person participle ever
had a clitic personal pronoun in Aramaic .

344

JAOS 145 .2 (2025)

phrase headed by the preposition l-, not via clitic objects as earlier verbs . Apriori there is no
morphological or syntactic reason to exclude the third person from serving as a clitic subject,
yet they are never used as such .
This state of affairs continues in modern dialects of Aramaic . In Northeastern Neo-Aramaic dialects, the Late Aramaic participial present shown above becomes the regular present
tense verb (see Table 12) .
Table 12 . Arbel Present Tense
Singular
1ms

palx-en

1fs

palx-an

2ms

palx-et

2fs

palx-at

3ms

palix

3fs

palx-a

Plural
palx-ex
palx-etun
palx-i

The basic difference between first and second persons, which are marked for person, and
third person, which is not, remains . For the most part, the stem, which inflected for gendernumber in Syriac, no longer shows such inflection in the first and second persons . The
only relic is the first person singular, where masculine and feminine have distinct suffixes,
resulting from their originally different stem (Khan 1999: 90) . In the plural, gender has collapsed, but from the remaining vowel (-e- for first and second, -i- for third), it is clear that
the masculine plural is the surviving form (< Syriac -īn) . Pronominal objects on these verbs
are expressed via prepositional clitics on l- . 24
20.

ma
’amr-in-nox
axon-i?
what
say.pres-1ms-to.2ms
brother-my
‘What shall I tell you, my brother?’ (Arbel, Khan 1999: 432, l. 1)

But is it never possible to use third person subject clitics? Apparently, the restriction is
operative only when the subject is nominative, but not when it is noncanonical . Some NeoAramaic dialects developed another verbal form, where the base is a passive participle and
the subject is an oblique pronoun cliticized to the verbal base via a preposition l- . Unlike the
paradigm in Table 12, the third person is marked morphologically in the same way as other
persons: 3ms -le, 3fs -la, 3cp -lu; see Table 13 .

24. The contact between the last consonant of the verbal stem and the l- of the object causes gemination, except
with the 3rd ms verb . Compare: 1cpl palxex ‘we open’ + le ‘it (obj)’ > palxix-xe ‘we open it’, but palix ‘he opens’
+ le ‘it (obj)’ > palix-le ‘he opens it’. The reason is that the stress on the 3ms falls on the stem, while in all other
forms it falls on the suffix .

Pat-el: The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme 345
Table 13 . Arbel Past Tense
Singular

21.

1cs

plix-li

2ms

plix-lox

2fs

plix-lax

3ms

plix-le

3fs

plix-la

Plural
plix-lan
plix-luxun
plix-lu

’iyya
qim-le...
zil-le?
geb-david
he
get.up.past-3ms
go.past-3ms
to-David
‘He got up and went to David’ (Arbel, Khan 1999: 424, l. 38)

The difference between the past and present in these dialects is that the past developed
from a passive participle with a noncanonical subject—originally a preposition with an
oblique pronominal suffix . Clearly, therefore, if a suffix developed from a subject pronoun,
the third person forms are zero, while if it developed from an oblique, the third person aligns
with other forms .
5. ConCluSionS

I have suggested here that the West Semitic perfect developed from an asymmetric paradigm, where the first and second person were marked with suffixed subject pronouns while
the third person was marked with nominal predicative gender-number markers but with no
person marker . I have shown that in two separate formations in Semitic, both constructed
out of a verbal adjective and a suffixed subject pronoun, the third person retains a nominal
inflection . Only when the subject morpheme developed from an original noncanonical construction was the third person marked like other persons .
The evidence presented here suggests that the third person zero suffix in the stative/perfect and later participle paradigms is not a result of loss or reduction but is rather an original
zero . These forms were always unmarked for third person subjects . The reasons for the
nonmarking are structural: The third person pronoun was originally a distal demonstrative,
whose clitic form cannot function as a nominative subject . Originally, this demonstrative,
in its capacity as a third person pronoun, had only independent forms; however, a syntactic
distinction between an independent form, subject, and suffixed forms, object and possessive, emerged in the proto-stage (Huehnergard and Pat-El 2012: 39) . The result was that
nominative forms were always independent, while oblique forms were suffixed . Thus, suffixed forms on deverbal nominal predicative bases could only be understood as non-subjects .
New verbal formations whose subject clitic developed from nominative pronouns, therefore,
exhibit third person zero subject .
But third person forms are not overwhelmingly zero in Semitic; they are zero in paradigms where person markers are derived from the suffixed subject pronominal paradigm,
where the third person slot is empty . The innovative perfect in Neo-Aramaic has a clear third
person marking and the paradigm is symmetric, because the origin of the person markers
is noncanonical subjects . This distribution suggests that the reason for the zero is primarily
structural, not semantic or pragmatic .

346

JAOS 145 .2 (2025)

My conclusions also highlight the need for an individual examination of internal language
processes before assuming a typological principle developed based on facts from other languages .
reFerenCeS

Bauer, Hans, and Pontus Leander. 1918–22. Historische Grammatik der hebräischen Sprache des Alten
Testaments . Niemeyer .
Behnstedt, Peter . 2016 . Dialect Atlas of North Yemen and Adjacent Areas . Brill .
Bickel, Balthasar, et al . 2015 . Exploring Diachronic Universals of Agreement: Alignment Patterns and
Zero Marking across Person Categories. In Agreement from a Diachronic Perspective, ed. Jürg
Fleischer, Elisabeth Rieken, and Paul Widmer. Pp. 29–51. Walter de Gruyter.
Bjøru, Øyvind, and Na‘ama Pat-El. 2021. The Indicative Markers in West Semitic and Their Relationship to the East Semitic Subordinative . In Bēl Lišāni: Current Research in Akkadian Linguistics, ed .
Rebecca Hasselbach-Andee and Na‘ama Pat-El. Pp. 41–69. Eisenbrauns.
Brockelmann, Carl. 1908–13. Grundriss der vergleichenden Grammatik der semitischen Sprachen .
Verlag von Reuther & Reichard .
Buccellati, Giorgio . 1968 . An Interpretation of the Akkadian Stative as a Nominal Sentence . JNES 27:
1–12.
———. 1996. A Structural Grammar of Babylonian . Harrassowitz .
Bybee, Joan L . 1985 . Morphology: A Study of the Relation between Meaning and Form . John Benjamins .
Cohen, David . 1975 . Phrase nominale et verbalisation en sémitique . In Mélanges linguistiques offerts
à Émile Benveniste. Pp. 87–89. Peeters.
Cristofaro, Sonia . 2021 . Typological Explanations in Synchrony and Diachrony: On the Origins of
Third Person Zeroes in Bound Person Paradigms. Folia Linguistica 55: 25–48.
Diem, Werner. 1997. Suffixkonjugation und Subjektspronomina: Ein Beitrag zur Rekonstruktion des
Ursemitischen und zur Geschichte der Semitistik . ZDMG 147: 10–76.
Driver, G . R . 1936 . Problems of the Hebrew Verbal System . T&T Clark .
Faust, Noam . 2014 . Where It’s [At]: A Phonological Effect of Phasal Boundaries in the Construct State
of Modern Hebrew . Lingua 150: 315–31.
Gelb, I. J. 1965. The Origin of the West Semitic “Qatala” Morpheme. In Symbolae linguisticae in honorem Georgii Kuryłowicz, ed. Stanislaw Drewniak. Pp. 72–80. Zakład Narodowy im. Ossolińskich.
Givón, Talmy . 2017 . The Story of Zero . John Benjamins .
Goldenberg, Gideon . 1983 . On Syriac Sentence Structure . In Arameans, Aramaic and the Aramaic
Literary Tradition, ed. Michael Sokoloff. Pp. 97–140. Bar Ilan Univ. Press.
———. 2013. Semitic Languages: Features, Structures, Relations, Processes . Oxford Univ . Press .
Gragg, Gene . 2019 . Afro-Asiatic . In The Semitic Languages, ed. John Huehnergard and Na‘ama Pat-El.
Pp. 22–48. Routledge.
Haiman, John . 1977 . Reinterpretation . Language 53: 312–28.
Hasselbach(-Andee), Rebecca . 2005 . Sargonic Akkadian . Harrassowitz .
———. 2007a. External Plural Markers in Semitic: A New Assessment. In Studies in Semitic and
Afroasiatic Linguistics Presented to Gene B. Gragg, ed. Cynthia L. Miller. Pp. 123–38. The Oriental
Institute .
———. 2007b. The Affiliation of Sargonic Akkadian with Babylonian and Assyrian: New Insights
Concerning the Internal Sub-Grouping of Akkadian . JSS 52: 21–43.
———. 2012. The Verbal Endings –u and –a: A Note on Their Functional Derivation . In Language and
Nature: Papers Presented to John Huehnergard on the Occasion of His 60th Birthday, ed . Rebecca
Hasselbach and Na‘ama Pat-El. Pp. 119–36. The Oriental Institute.
———. 2013a. Case in Semitic: Roles, Relations, and Reconstruction . Oxford Univ . Press .
———. 2013b. Phoenician Case in Typological Context. In Linguistic Studies in Phoenician, ed .
Robert D. Holmstedt and Aaron Schade. Pp. 196–222. Eisenbrauns.

Pat-el: The Semitic “Perfect” and the Problem of the Third Person Zero Morpheme 347
———. 2022. The Origin of the Third-person Markers on the Suffix Conjugation in Semitic. In “Like
’Ilu You Are Wise.” Studies in Northwest Semitic Languages and Literatures in Honor of Dennis G.
Pardee, ed. H. H. Hardy II, Joseph Lam, and Eric D. Reymond. Pp. 559–72. The Oriental Institute.
Huehnergard, John . 1983 . Asseverative *la and Hypothetical *lu/law in Semitic . JAOS 103: 569–93.
———. 1987a. “Stative,” Predicative Form, Pseudo-Verb. JNES 46: 215–32.
———. 1987b. Ugaritic Vocabulary in Syllabic Transcription . Scholars Press .
———. 2000. A Grammar of Akkadian . 3rd printing . Eisenbrauns .
———. 2005. Features of Central Semitic. In Biblical and Oriental Essays in Memory of William L.
Moran, ed. Agustinus Gianto. Pp. 155–203. Pontificium Institutum Biblicum.
———. 2006. Proto-Semitic and Proto-Akkadian. In The Akkadian Language in Its Semitic Context,
ed. Guy Deutscher and N. J. C. Kouwenberg. Pp. 1–18. Nederlandse Instituut voor het Nabije
Oosten .
———. 2017. Arabic in Its Semitic Context. In Arabic in Context: Celebrating 400 Years of Arabic at
Leiden University, ed. Ahmad Al-Jallad. Pp. 3–34. Brill.
———. 2019. Proto-Semitic. In The Semitic Languages, ed. John Huehnergard and Na‘ama Pat-El.
Pp. 49–79. Routledge.
———. 2023. Proto-Semitic and Egyptian. In Ancient Egyptian and Afroasiatic: Rethinking the Origins, ed. M. Victoria Almansa-Villatoro and Silvia Štobňová Nigrelli. Pp. 139–60. Eisenbrauns.
Huehnergard, John, and Na‘ama Pat-El. 2012. Third Person Possessive Suffixes as Definite Articles in
Semitic . Journal of Historical Linguistics 2.1: 25–51.
Kamil, Iris . 2023 . t-Forms of the Akkadian Stative . Brill’s Journal of Afroasiatic Languages and Linguistics 15: 262–90.
Khan, Geoffrey . 1999 . A Grammar of Neo-Aramaic: The Dialect of the Jews of Arbel . Brill .
Koch, Harold. 1995. The Creation of Morphological Zeroes. In Yearbook of Morphology 1994, ed .
Geert Booij and Jaap van Marle. Pp. 31–71. Kluwer Academic Publishers.
Kogan, Leonid, and Manfred Krebernik . 2021 . Eblaite . In History of the Akkadian Language, ed . JuanPablo Vita. Pp. 664–989. Brill.
Kouwenberg, N . J . C . 2000 . Nouns as Verbs: The Verbal Nature of the Akkadian Stative . Orientalia
69: 21–71.
———. 2014. The Akkadian Verb and Its Semitic Background . Eisenbrauns .
Nöldeke, Theodor . 1884 . Untersuchungen zur semitischen Grammatik . II . Die Endungen des Perfects .
ZDMG 38: 407–22.
Pat-El, Na‘ama. 2022. On the Akkadian Adjectival Masculine Plural -ūt Once Again . BSOAS 85: 165–
71 .
Putten, Marijn van, and Phillip W. Stokes. 2018. Case in the Qurˀānic Consonantal Text. WZKM 108:
143–79.
Siewierska, Anna. 2009. Person Asymmetries in Zero Expression and Grammatical Function. In
Essais de linguistique generale et de typologie linguistique offerts au Professeur Denis Creissels
à l’occasion de ses 65 ans, ed. Franc Floricic. Pp. 425–38. Presse de l’École Normale Supérieure.
Streck, Michael P . 2000 . Das amurritische Onomastikon der altbabylonischen Zeit, vol . 1: Die Amurriter: Die onomastische Forschung, Orthographie und Phonologie, Nominalmorphologie. UgaritVerlag .
Suchard, Benjamin D ., and Jorik F . J . Groen . 2021 . (Northwest) Semitic sg .*CVCC-, pl .*CVCaC-ū-:
Broken Plural or Regular Reflex? BSOAS 84: 1–17.
Tropper, Josef, and Juan-Pablo Vita . 2019 . Ugaritic . In The Semitic Languages, ed . John Huehnergard
and Na‘ama Pat-El. Pp. 482–508. Routledge.
Waltisberg, Michael . 2011 . The Case Function in Amorite: A Re-Evaluation . JSS 56: 19–36.
Watkins, Calvert . 1962 . Indo-European Origins of the Celtic Verb: 1. The Sigmatic Aorist . Institute for
Advanced Studies, Dublin .