The earliest stages of Arabic and its linguistic classification (Routledge Handbook

of Arabic Linguistics, forthcoming)

Ahmad Al-Jallad, Leiden University

1 Introduction

The first clear attestation of an Arabic word occurs in the Kurkh monolith inscription of
the neo-Assyrian monarch, Shalmaneser III (853 BCE). The text lists the names of a
coalition of leaders who opposed the expansion of the Assyrians into the Levant.
Among rulers such as Adad-idri of Damascus and Ahab the Israelite, we find mGi-in-di-
bu- kur
Ar-ba-a-a, that is, Gindibu the Arab (lit, of the land of Arby. The cuneiform
sources use the term Arab (A-ri-bi) to describe peoples living from Mesopotamia in
the east to the Anti-Lebanon mountains in the west, and from northwest Arabia to the
Sinai in the south (see Ephal 1982). Later Greek and Persian sources record the
presence of Arabs across the Fertile Crescent and North Arabia as well, although it not
always possible to determine what individual authors meant when they used the term
(see the various articles in Macdonald 2009). Only one text in the Arabic language can
be dated to this period. A single Ancient North Arabian inscription containing a prayer
to the gods of the Iron Age trans-Jordanian kingdoms of Ammon, Moab, and Edom,
Malkom, Kem, and Qaws, respectively, in what appears to be the Arabic language has
been discovered in south-eastern Jordan (Hayajneh, Ababneh, and Khraysheh 2015).
The text is accompanied by a Canaanite inscription, but is undated. Context, however,
would support a mid to late Iron Age II date. Aside from this short prayer, the Arabic
of the ancient Near East is only known from a handful of personal and divine names
transcribed in other languages (on these fragments, see Macdonald 2008).

Evidence for Arabic becomes more frequent in the 2nd century BCE with the arrival of
inscriptions in the Nabataean script. While the Nabataeans used a form of Achaemenid
Official Aramaic as a literary language, several features betray an Arabic substratum,
most notably in the areas of syntax and personal names. The epigraphy in the Safaitic
and Hismaic scripts, which extends from North Arabia to the awrn, also provides
considerable evidence for the earliest stages of Arabic. It is impossible to determine
when these writings began but their authors seem to have been especially productive in
the Nabataean and Roman periods (1st c. BCE 4th c. CE), as references to the political
events of these centuries are relatively abundant. A birds eye view of the situation
places the earliest stages of Arabic in the northwest Arabia and the southern Levant.

Geographic distribution of Old Arabic

At some point, Arabic moved south into the Peninsula. The term rb begins to appear
in the Sabaic inscriptions of ancient Yemen roughly around the turn of the Era. While
many scholars have equated the term rb with Qurnic arbun, which is understood
to mean nomads by traditional exegetes, there is no internal evidence in the Sabaic
inscriptions to support such an equation (Rets 2003:536-574). Moreover, there is no
evidence as to the language of the rb. No texts in the Arabic language have yet
appeared in pre-Islamic South Arabia proper, although several inscriptions from the
northern frontier, the so-called Haram area, seem to reflect an admixture of another
language (Stein 2008), perhaps Arabic but other North Arabian varieties are equally
likely. In south-central Arabia, the town of Qaryat al-Fw has yielded an interesting
epitaph exhibiting a mixture of Sabaic and non-Sabaic features. While the text has
been traditionally considered an example of Old Arabic, a recent linguistic
investigation suggests that it is better interpreted as a transitional dialect between
North Arabian and Sabaic, if not an artificial mixed register (Al-Jallad 2014). Another
example of Old Arabic was identified in Mulayah, but this has recently been shown to
be a form of Aramaic (Macdonald 2008). It is, therefore, unclear when Arabic replaced
the indigenous languages of the nomads and oasis towns of central and southern Arabia
(see Ancient North Arabian below) or the epigraphic languages of Ancient Yemen.
Regarding the latter, the works of the Arabic grammarians suggest that the Ancient
South Arabian languages continued to be spoken and perhaps even written well into
the 9th c. CE.

2 Historical Background and Perspectives: the debate on Arabics classification

The classification of Arabic has occupied a central position in the efforts of Semiticists
to understand the evolution of Semitic language family. Earlier scholars saw Arabic as
more closely connected with the languages situated in the southern half of the Arabian
Peninsula and Semitic languages of Ethiopia (Huehnergard and Rubin 2011: 260).
Together, these languages formed a sub-grouping called South Semitic. In addition to a
perceived geographic proximity, three features common to Classical Arabic and the
modern dialects, the Ancient South Arabian languages of pre-Islamic Yemen, the
Modern (non-Arabic) South Arabian languages, and Ethio-Semitic were taken as
evidence for a common South Semitic origin.

a. Plurals formed by pattern replacement rather than simply suffixation (broken

plurals), e.g. CAr, singular kalbun dog, plural kilbun or singular ilhun god,
plural lihatun.
b. The realization of Proto-Semitic *p as [f]: compare CAr fataa with Hebrew
pt, Aramaic pta, and Old Akkadian patum.
c. A verbal derivation formed by the insertion of a long vowel between the first
and second root consonant, the so called L-stem (form VI in Classical Arabica
grammar), fala.

As methods of language classification were refined in the 20th century, the subgrouping
of the Semitic languages was gradually revised. Instead of relying on geography and
arbitrary similarities, linguists began to focus on shared morphological innovations
(Hetzron 1974; 1975; 1976). Complex changes in morphology were less likely to be
borrowed or arise as the result of coincidence, and so such features could more
accurately suggest descent from a common ancestor.

This perspective immediately disqualified two out of the three South Semitic
features. The broken plurals, it turns out, were not an innovation at all, but rather
reflected the preservation of the original Proto-Semitic strategy of pluralization
(Huehnergard and Rubin 2011:272-3). Likewise, relics of the L-stem could be found
across the Semitic languages, indicating that it was not a unique ancestor of the South
Semitic languages which developed such a form, but that the other languages simply
lost it (Ibid., 2011:273). Finally, the sound change p > f is so typologically common
that it can hardly be used for classification. Its presence in the languages of Arabia and
Ethiopia probably points towards areal diffusion rather than a development in a
common ancestor (Ibid., 272). Moreover, there is conflicting evidence as to the
antiquity of this change within Arabic itself (see below), and we simply have no
evidence as to how this sound was actually pronounced in many of the ancient
epigraphic varieties.

From the vista of shared innovations, a key morphological development in the verbal
system defines the primary split in the Semitic language family: East and West. The
Proto-Semitic finite verb had two primary forms distinguished by stem ablaut a
perfective: yaqtul and an imperfective yaqattal (Huehnergard 2008: 151). This system
is preserved in Akkadian, while West Semitic grammaticalized a construction based on
a predicative adjective + pronominal clitic, giving rise to the suffix conjugation, the
perfective qatala/qataltu in Arabic (Huehnergard 1987). In most West Semitic
languages, the original preterite function of the yaqtul stem was marginalized,
preserved only in certain constructions (as in Arabic lam yaqul he did not say).

A sub-section of West Semitic languages exhibit another important innovation in the

verbal system: a new imperfective stem. Arabic, the Northwest Semitic languages
(Ugaritic, Aramaic, Hebrew, Phoenician, etc.), and the Ancient South Arabian
languages replaced the yaqattal stem with a new verb form comprising the preterite
plus an augment, -u in conjugations terminating in a consonant and -na in conjugations
terminating in a vowel (i.e. yaqtulu, yaqtulna). The languages that share this complex
innovation must have descended from a more recent common ancestor to the exclusion
of the Modern South Arabian languages and Ethio-Semitic, which continue the use of
the original imperfective *yaqattal. The yaqtulu languages were therefore removed
from the South Semitic sup-grouping and placed in a new category called Central
Semitic (on the features of Central Semitic, see Huehnergard 2005). Since the
remaining members of South Semitic did not share any morphological innovations, the
entire sub-grouping collapsed.

The position of Arabic in the Semitic family based on the principle of shared
innovations is as follows:
Classification of the Semitic Languages

Features Unique to Arabic

While there can be no doubt as to Arabics membership in the Central Semitic category,
until recently the characteristic features of Arabic itself were never explicitly laid out.
In a recent paper, Huehnergard (forthcoming) outlined some of the features which
distinguish Arabic from the other Semitic languages:

1. The pharyngealized realization of the emphatic consonants: The emphatic

consonants in Proto-Semitic were likely glottalized, as in the Modern South
Arabian languages and Ethio-Semitic (Kogan 2011).
2. The merger of Proto-Semitic *s [s] and *s [ts] to [s]: Proto-Semitic had three
voiceless sibilants: *s, an alveolar or apical voiceless sibilant [s], which
remains [s] in Classical Arabic; *s, a voiceless lateral fricative [], which
becomes [] or [] in Classical Arabic; and *s, a voiceless alveolar affricate [ts],
which also becomes [s] in Arabic, merging with *s.
3. The loss of the long form of the 1st person independent pronoun, anku: Proto-
Semitic had two forms of the 1st singular pronoun, an and anku, the latter
reflected in Hebrew ank, Akkadian anku, and Ancient South Arabian nk. No
trace of this pronoun survives in Arabic, which suggests that it was lost at the
Proto-Arabic stage.
4. The feminine singular demonstrative element, t-, as in Classical Arabic t, ht,
allat, and Old Arabic ty /t/.
5. The replacement of mimation with nunation: In Proto-Semitic, nouns which
were not in the construct state terminated in -n(a) in the dual and masculine
sound plurals and in -m everywhere else. Arabic leveled the -n ending,
producing what the Arabic Grammarians called tanwn, nunation.
6. Leveling of the at reflex of the feminine ending: Proto-Semitic had two
allomorphs of the feminine ending, -t and at. Arabic levelled the -at ending to
all situations, compare Arabic qtil-atun to Hebrew qlt < qil-t < *qilt
killing. Relics are preserved in isolated nouns, such as bintun girl and utun
daughter.
7. The 3rd feminine plural termination -na on the suffix conjugation: This
development is the result of leveling with the prefix conjugation, yaqtulna. The
same feature is found in Qatabanic and Hadramitic (Ancient South Arabian; see
Stein 2011:1060), which is best explained as a parallel development, as these
languages are attested much earlier than the period in which we can posit
contact with Arabic.
8. The mafl pattern as a paradigmatic passive: Proto-Central Semitic seems to
have had two forms of the G-stem passive, qatl and qatl, while the nominal
stem maqtl occurred in isolated forms. While adjectives often with a
passive/stative sense of the former two remain in Arabic (qatlun, kabrun), the
productive means by which to form a passive participle from the G-stem (form I)
is the pattern maqtl.
 9. The absence of a paradigmatic infinitive. According to some, Proto-Semitic had
a paradigmatic infinitive of the G-stem (form I) in the pattern qatl. The loss of
this feature and the variety of verbal noun patterns in Arabic would then be
interpreted as an innovation (but see old Arabic below).
10. The vowel melody u~i in the passive of the suffix conjugation. Internal passives
exist in other Semitic languages, but their vocalic pattern differs. In NWS,
Huehnergard reconstructs the pattern quttal.
11. The grammaticalization of the particle qad as a perfective morpheme, as in qad
faala he had done.
12. The preposition f, derived from the word mouth.
13. The loss of the anaphoric or remote demonstrative use of the 3rd person
pronouns. The third person pronouns were proper demonstratives in Proto-
Semitic and continued as such in most of the daughter languages, e.g., Hebrew
has-spr h-h that book; Dadanitic w l-h h and that belongs to him (Fars-
Drappeau 2005:66); Akkadian arrum that king. No such function is attested
in Arabic.
14. The presence of nunation on nominal heads of indefinite asyndetic relative
clauses: As Pat-El has shown recently (2014), Arabic exhibits an innovation in its
morphosyntax where nunation may occur on the head of asyndetic relative
clauses. Other Semitic languages use the construct form of a noun in this
syntactic position.

While not all of these developments carry the same weight for linguistic diagnosis, they
can with some confidence be reconstructed to the Proto-Arabic stage. The exception is
perhaps feature (1), where the evidence is ambiguous in Old Arabic (see below), and
feature (9), where it has been recently argued that the Madar system of Arabic is in
fact original and would therefore reflect an archaism rather than an innovation (Strich,
conference presentation). This view is supported by the presence and use of the
infinitives in Old Arabic, but the vocalic patterns are not always clear.
To these innovations identified by Huehnergard, we may add the following:

15. The subjunctive ending in -a: While Hebrew attests a verbal form ending in
and an a termination is equally found in Amarna Canaanite, in neither
language do verbs with this ending function as a subjunctive, but instead have a
cohortative function. Therefore, Huehnergard suggests that the subjunctive in
a could be characteristic of Arabic. Although Huehnergard did not place this on
his primary list of innovations, it seems clear that the subjunctive use of this
verb form is an important Arabic innovation.

16. The negative m: Huehnergard originally excluded the use of m as a negator

from the list of Arabic innovations because it occasionally occurs in NWS, e.g.
Hebrew ma-bb-yd r what evil is in my hand (i.e. there is no evil in my
hand) (1 Sam. 26:28). However, the negative meaning is clearly rhetorical in all
of the non-Arabic attestations. The innovation in Arabic is then in the
grammaticalization of this rhetorical device into a proper negative adverb.

17. Other prepositions and adverbs which are typical of Arabic may be added to f;
these are *an ablative, *inda locative, *attay until, and kdy (vocalization
unclear) thereafter (only found in Old Arabic).

18. Arabic uniquely uses the particle *an(na) as a complementizer and

subordinator, e.g. arda an yahaba he wanted to go.

19. The independent object pronoun base *(iy)y: despite attempts to connect
Arabic iyy with the NWS object markers , it is clear that the form is a unique
development in Arabic, and is probably related to the vocative marker y used
as a topicalizer (see Wilmsen 2013). Safaitic attests the form simply as y, which
 may suggest that the Classical Arabic form iyy comprises the presentative in
and y, with assimilation of the n.

Arabic and Ancient North Arabian Inscriptions

The relationship between Arabic and the languages attested in the Ancient North
Arabian (ANA) inscriptions has been the subject of some debate among scholars
(Macdonald 2000). The most notable difference between many of these texts and
Classical Arabic is the shape of the definite article, h- in most of the ANA inscriptions
and al in Arabic. Based on this feature, some scholars (Beeston 1981; Muller 1982)
have argued for the bifurcation of the languages of central and north Arabia into
Arabic and Ancient North Arabian. Knauf (2010) objected to this division and
instead argued that the ANA inscriptions were all to be considered an ancient form of
Arabic. His argument was based on the presence of broken plurals, a prefixed article,
and the merger of *s and *s. Following from our discussion of classification, both the
broken plurals and article are of little value to determine genetic affiliation. While the
*s and *s merger did occur in Proto-Arabic, it is after all a sound change and could
have been spread areally in Central and North Arabia. Moreover, this sound change
did not occur in Taymanitic.

Al-Jallad (2014, forthcoming a; 2015) argues that the linguistic unity of the languages
expressed by the ANA scripts should be demonstrated by the identification of shared
innovations, and not assumed. This approach fragmented the ANA corpus into several
independent branches, in turn indicating that even north and central Arabia were home
to considerable linguistic diversity in the pre-Islamic period.

Taymanitic: Taymanitic refers to a form of the South Semitic script used at the oasis
town of Taym in modern northern Saudi Arabia (Macdonald 2004: 490) and the
language it expresses. These inscriptions do not exhibit any of the aforementioned
Arabic innovations, but instead exhibits an interesting isogloss typical of the Northwest
Semitic languages, the change of w to y in word initial position: yr for *warum
month, moon and yd for wadaa to know. Other sound changes include the merger
of *z and *, *s3 and t, and of * and * (Kootstra, forthcoming). In general, the texts
are too short to provide a full linguistic assessment, but these few features remain
significant and preclude this language as being an early ancestor of Arabic.

Dadanitic: Dadanitic refers to the script and language of the oasis of Dadn. The
language of these inscriptions exhibits a few forms that seem to have been lost at the
Proto-Arabic stage. It retains the anaphoric use of the 3rd person pronoun, h; it does
not exhibit the innovative form *attay (= Classical Arabic att), but instead
preserves dky, probably */adkay/, and does not level the -at ending, e.g. mrh
*/marah/ < *marat woman vs. qrt */qart/ town, settlement compare with Arabic
qaryatun. Moreover, some dialects have a C-stem (form IV) with an h- prefix rather
than an - (i.e. hafala instead of afala), while Proto-Arabic seems to have undergone
the change h > in this verb form. Variation is also reflected in the definite articles,
where both h(n) and (l) are attested in the corpus. Other interesting features include
the special dissimilation of * to /t/ in the word three, lt instead of l and the dual
pronoun hmy */humay/. The grammar of Dadanitic is still poorly understood, and
while several of the aforementioned features exclude its belonging to the Arabic
category, more work is required to establish its correct position in the Semitic family
(see Macdonald 2004 for further discussion on some of these features).

Thamudic: Thamudic is a conventional term used to cover all of the unclassifiable

inscriptional material from the Arabian Peninsula and has nothing to do with the social
group known as Thamd from Cuneiform, Greek, and later Arabic sources
(Macdonald and King 2000). Most of these inscriptions are short and rather
uninformative from a linguistic point of view. Nevertheless, the significant challenges
they pose for decipherment can only speak to their remote linguistic character.
Judgment must be withheld until the entire corpus can be subjected to a thorough
linguistic study. At the present moment, scholars divide the Thamudic inscriptions into
four general categories according to the shapes of the glyphs.

Thamudic B: The Thamudic B inscriptions are concentrated in Northwest Arabia, but

can be occasionally found in Syria, Egypt, and Yemen. A single Thamudic B text
mentions the king of Babylon, which suggests that it was composed before the fall of
the kingdom in the middle of the 6th c. BCE. We have no information as to when these
inscriptions begin or end. Most of these texts consist of short prayers, the meanings of
which are still poorly understood, as illustrated by the sometimes bizarre translations
given: e.g. b-lh btr gzzt nm lt by (the power of) lh btr (I) sheared off (the wool of
sheep) (Hayajneh 2011: 770). A few linguistic facts, however, can be gleaned from the
texts we do understand. The suffix morpheme of the prefix conjugation in the first
person is t, as in Arabic and Northwest Semitic, as opposed to the -k of Ancient South
Arabian and Ethiopic, e.g. h rw b -k n rft O Rw, through you I am healed (ibid).
The dative preposition is nm, which appears to be an assimilated form of an original
*lim cf. Taymanitic lm, Hebrew lm. The consonant /n/ often assimilates to a
following contiguous consonant, t from earlier *Vnat and t from earlier *anta.

Thamudic C: The Thamudic C inscriptions are concentrated in the Najd, but can be
found elsewhere across western Arabia as well. None of these inscriptions contains
information that allows us to date them. These texts consist of short statements,
usually containing the word wdd, the meaning of which is uncertain (Tsafrir 1996).
One of the most common formulae is wdd followed by f and what appear to be personal
names. The personal pronoun n */an/ is attested, as well as two terms which appear
to be demonstrative pronouns, zn */zin/ and zt */zt/, masculine and feminine,
respectively. If this identification is correct, then it would appear that the phonemes
and z have merged to z, as in Taymanitic.
Thamudic D: These inscriptions are concentrated in northwest Arabia, and one occurs
alongside a Nabataean tomb inscription dated to the year 267 CE. The only thing of
linguistic substance in these inscriptions is the demonstrative zn, which like in
Thamudic C, could indicate that the sound change > z had operated.

Southern Thamudic: These texts come from the southwestern part of the Arabian
Peninsula and seem to contain only names, although some of these names contain
mimation and one example of a hl- */hal/ definite article. For more information, see
Ryckmans (1956).

Hismaic and Safaitic: Hismaic and Safaitic are the modern names of two scripts
which were used across Jordan and southern Syria. In so far as one can see, all of the
innovations typical of Arabic are attested in the inscriptions of these corpora. Most of
these are attested in the Safaitic corpus (Al-Jallad 2015), but this fact probably has to
do with the fact that the Safaitic inscriptions are generally longer and contain more
linguistic information than the Hismaic texts. Nevertheless, two long texts composed
in the Hismaic script from central Jordan attest a language that is unambiguously
Arabic (Graf and Zwettler 2004).

Old Arabic

The linguistic history of Arabic has been primarily told by modern Arabic dialectologist
and Classical Arabic philologists. For this reason perhaps, the pre-Islamic stages of the
language have been largely neglected. The strong bias towards the language of the
Arabic grammatical tradition placed the developmental timeline of Arabic between two
poles: Old Arabic as defined by the literary works of the Arab grammarians and the
modern spoken forms of the language (see for example Owens 2006; El-Sharkawi
EALL).
The term Old Arabic is used differently by epigraphists who work with material from
the pre-Islamic period, and this is the sense which I shall adopt in this essay. Old
Arabic does not refer to a homogeneous linguistic entity but instead to the entire
corpus of inscriptions produced before the Islamic Conquests (Macdonald 2008). The
focus on documentary evidence insures that the material included in this category was
not edited by later scribes/transmitters, who could have been influenced by the Arabic
grammatical tradition and the standard administrative language. As such, they provide
our clearest and most honest view of Arabics early history.

Sources for Old Arabic

Our knowledge of Old Arabic derives from the following sources:

Inscriptions in the Hismaic and Safaitic scripts:

The number of texts composed in both of these scripts nears 50,000 specimens and, as
such, they both provide us with a rather detailed view of Old Arabic. Since these
inscriptions span a considerable geographic distance and an unknown chronological
depth (but perhaps between the 2nd c. BCE and the 4th c. CE), one naturally encounters
a good degree of linguistic variation. The true extent of this variation is masked by the
purely consonantal nature of the writing system and the brief and formulaic style of the
texts. For the grammar of the Old Arabic of the Safaitic inscriptions, see Al-Jallad
(2015) and for Hismaic, Al-Jallad (in preparation).

In the Dadanitic script:

A single text, JSLih 384, composed in the Dadanitic script (see Macdonald 2008 for
bibliography and discussion), from northwest Arabia, provides our only non-Nabataean
example of Old Arabic from the Higz.

In the Nabataean script:

Only two texts composed fully in Arabic have been discovered in the Nabataean script.
The n Avdat inscription (Negev 1986) contains two lines of an Arabic prayer or
hymn embedded in an Aramaic votive inscription. The text is undated, but Negev
argues that it must have been composed prior to 150 CE. The second is the Namrah
inscription, 328 CE, which was erected about 60 miles southeast of Damascus. The text
is an epitaph of a king named Mrlqy br mrw /maralqays (bin) amr/, which recounts
his deeds and the year of his death (for bibliography, see Macdonald 2008). Most
examples of Arabic come from the substratal influence the language exercised on
Nabataean Aramaic. In the Sinai, one finds the Arabic passive participle makr,
spelled mdkwr in the Nabataean script, in place of Aramaic dkyr. The optative use of
the passive participle, which is otherwise unknown in Aramaic, is no doubt the result
of Arabic influence (Gzella 2011:601). Loanwords from Arabic are especially frequent
in the Nabataean legal papyri from Naal ever (Yardeni 2014). Macdonald (2010)
has taken this as evidence for an Arabic-language legal tradition among the
Nabataeans. Loanwords occasionally occur in the Nabataean inscriptions themselves,
but their formulaic nature reduces the possibility for such intrusions. Mixed Arabo-
Aramaic inscriptions are also known, the best example of which is JSNab 17, dated to
267 CE (see Macdonald 2008 for bibliography). This text is not only of value for the
linguistic light it sheds on Old Arabic but also for the evidence it provides for Arabic-
Aramaic bilingualism in the pre-Islamic period.

In the Nabataeao-Arabic script:

A growing corpus of texts carved in a script in between Classical Nabataean Aramaic

and what we consider the Arabic script from Northwest Arabia provides further lexical
and some morphological material for the later stages of Old Arabic in this region.
These texts not only provide important insights as to the development of the Arabic
script from its Nabataean forebear, but an important glimpse of the Old igz dialects
(Nehme 2013; forthcoming).
In the Old Arabic script:

Only three inscriptions in the fully evolved Arabic script are known from the pre-
Islamic period. These rather short texts come from 6th c. CE Syria, two from the
southern region on the borders of the awrn Jabal Usays (528 CE) and arrn (568
CE) and one from Zebed (512 CE), a town near Aleppo (see Macdonald 2008:470 for
a short discussion and bibliography). These short texts shed little light on the linguistic
character of Arabic and are more interesting for the information they provide regarding
the evolution of the Arabic script.

In the Greek script:

Fragmentary evidence in the Greek script, the Graeco-Arabica, is equally crucial to

help complete our understanding of Old Arabic. This category encompasses instances
of Old Arabic in Greek transcription from documentary sources. The advantage of the
Greek script is that it gives us a clear view of the vocalism of Old Arabic and can shed
important light on the phonetic realization of the Old Arabic phonemes. This material
has been comprehensively described in Al-Jallad (forthcoming a). Finally, a single pre-
Islamic Arabic text composed in Greek letters is known, labelled A1 (Al-Jallad and al-
Manaser 2015).

3 Critical Issues and Topics: the linguistic profile of Old Arabic

Considering these sources together, we can form a rather detailed picture of Old
Arabic. The following pages will outline some of the key phonological, morphological,
and syntactic features which characterize the earliest stages of the language.

Phonology

There is a virtual consensus among Semiticists that the Proto-Semitic emphatic series
was not pharyngealized but glottalized. While Huehnergard suggested that
pharyngealization was a Proto-Arabic development, there is some evidence from
Safaitic and the Graeco-Arabica that this might not have been the case in the earliest
stages of the language. In fact, Greek transcriptions show that the entire emphatic
series was originally voiceless in Arabic, which would agree with glottalization.
Moreover, vowels do not seem to be affected by their vicinity to emphatic consonants
until the 6th c. CE. These observations taken together could suggest that glottalization
was the emphatic correlate in Old Arabic (for more, see Al-Jallad forthcoming a):

Reconstructed values of the emphatic consonants in Old Arabic

Proto- Proto-Arabic Old Arabic (in Greek Classical Arabic

Semitic transcription)
*[t] * <t> []

*[t] * <t> [t]

*[ts] * <s> [s]

*[t] * <s> []
*[k] *q <k> [q]

It was probably the case that the reflex of *s retained its original value as a voiceless
lateral fricative []. This realization can be triangulated from two observations. The
Safaitic glyph corresponding to is never used to transcribe Aramaic [], indicating
that it had not yet achieved that value. The same sound is always transcribed as in
Greek (Al-Jallad forthcoming a, 3.8), which could also suggest that it did not have the
value that Sibawayh described, namely, a voiceless palatal fricative [], as velar and
post-velar fricatives are always given with the spiritus asper. Thus, it was probably the
case that the sound preserved its original lateral value.

While all later varieties of Arabic realize Proto-Semitic *p as [f], Old Arabic may have
retained a stop realization, albeit noticeably aspirated. This is suggested by the
transcription of the use of /p/ to transcribe a few Arabic names in Greek, such as
/alp/ = Classical Arabic alf (Al-Jallad forthcoming a, 3.4). Additionally,
Safaitic transcriptions of both Greek /ph/ and /p/ use the glyph f rather than b,
which could suggest that the former signified [ph] rather than [f] (Al-Jallad 2015,
3.1.1).

The alif-maqrah is a term for when word-final ys in the unpointed Arabic script
should be pronounced as // in Classical Arabic. In Old Arabic, this sequence is always
kept distinct from etymological //. Spellings in Greek such as /sufl/ for
Classical Arabic suggest that the alif-maqrah was pronounced as perhaps [ai] or
[e]. Safaitic and Hismaic attest forms such as fty (= Classical Arabic fatan youth) and
mny (=Classical Arabic manan fate), where the final y can only signal a final
diphthong or triphthong and not a long vowel (for more, see Al-Jallad forthcoming a,
5.1). Likewise, triphthongs seem to have obtained in all positions. Thus, verbs with a
glide as a third radical preserve the final triphthong: tw he came, s2ty he spent the
winter, bny he built. The consonantal quality of the final glide is proved by the
Graeco-Arabic inscription A1, where the verb he came is transcribed as
/atawa/.

Morphology

Perhaps one of the most striking morphological aspects of Old Arabic is the variation in
the presence of definite marking and its shape. The definite article spread areally
among the Central Semitic languages and it would seem that Proto-Arabic lacked any
overt marking of definiteness, as indicated by the Safaitic inscription HshNSMI 5: w lm
ybl s1fr */wa lam yoabbal sepr/ and may the writing not be obscured (referring to
the present inscription, see Al-Jallad 2015:4.8). Besides dialects with no definite
article, the Safaitic inscriptions exhibit about four different article forms, ordered by
frequency: h-, -, l-, and hn- (ibid.). The Old Arabic of the Nabataean inscriptions
exhibits almost exclusively the form l- . Unlike the Classical Arabic article, the Old
Arabic l almost never exhibits the assimilation of the coda to the coronals; the same
situation is attested in the Graeco-Arabica (Al-Jallad forthcoming, 5.5), but in A1 the
coda assimilates to the following d, */ad-dawra/ the region. Taking in the
entire Old Arabic corpus into consideration, it would appear that the l article was a
typically sedentary feature, as it is rare in the inscriptions produced by the nomads,
while the nomadic dialects varied considerably in definite marking, from the more
conservative -marking to the innovative, , l-, and h- articles.

The feminine ending at did not shift to ah in the earliest stages of the language. The
Safaitic and Hismaic texts attest an invariable -t ending, and the same appears to be
true of the earliest Nabataean Arabic, as evidence by spellings of names such as rtt
/reat/ = Classical Arabic riah and bdt /obodat/ = Classical Arabic ubudah.
While Greek transcriptions show a mixed situation, it is clear that by the 4th c. CE, the
ending had shifted to a(h) in non-construct position in the settled areas (Al-Jallad
forthcoming, 5.2.1).

The Graeco-Arabic inscription A1 proves the existence of a limited case system in the
Old Arabic of the 3rd or 4th c. CEa productive accusative case is present but there is
no evidence for a nominative or genitive. We have - /al-idmiyy/ the
Idmite (nominative) instead of **/al-idmiyyu/ and - /mis-si/ from Si
(genitive) instead of **/mis-sii/, but an accusative with a final /a/: ()
/wa yiraw baql-a/ and they pastured on fresh herbage or ...
/atawa ad-dawr-a/ he came to this place (Al-Jallad and al-Manaser 2015: 57-
58).

Disconnected pieces of evidence, however, suggest that a tripartite system of case

inflection was operative at least in the earliest stages of the language. The n Avdat
inscription attests two common nouns with a final -w in the nominative case (l-mwtw
*/al-mawtu/ death and grw */guru/ wound) and one noun in the accusative
terminating in - (tr */aar/ reward) (see Bellamy 1990, but disregard the
speculation on the presence of Classical Arabic metrics). This could point towards a
functional case system. Early Nabataean basileophoric and theophoric names based on
genitive constructions exhibit an /o/ vowel between the first and second term, which
could point towards a frozen nominative case, /taymo-mlek/,
/abdo-bal/, /abdo-rah/ (Al-Jallad forthcoming a, 5.5).
More evidence for case inflection is provided by the consonantal script itself. In
Safaitic, participles ending in a glide y are bi-radical in the nominative, dm /dm/
drawing dmy, but tri-radical in the accusative, dmy /dmey/ idem., suggesting the
presence of a final vowel in the latter syntactic position (Al-Jallad 2015, 4.6).
Vestiges of the genitive ending are frozen in Nabataean theophoric names, such as
tymlhy /taym(o)-allhi/ and bdlbly */abd(o)-al-bali/ (Negev 1990, s.v.).

While there is enough evidence to restore a three-part case system for Old Arabic,
although it was clearly lost in some areas before the Islamic period, the existence of
nunation is much more difficult to confirm. Rare vestiges of the feature are found in
the Safaitic inscriptions, mtn Libra (usually mt) and mltn dearth of pasture (usually
mlt), but both of these examples can be disputed (see Al-Jallad 2015, 4.5b-). No
evidence for the feature appears in Greek transcription or in the Nabataean script.

The existence of mood inflection is confirmed in the spellings of verbs with y/w as the
third root consonant. Verbs of this class in result clauses are spelled in such a way that
they must have originally terminated in /a/: f ygzy nr h */pa yagziya nara-hu/ that
he may fulfill his vow (Graf and Zwettler 2004). Sometimes verbs terminate in a n
which may reflect an energic ending, thus, s-nh join him! perhaps */eannoh/ (Al-
Jallad 2015, 4.14.2).

A few demonstrative pronouns are attested, but in general these are rare. The
commonest form is a proclitic h-, which does not inflect for gender or number (Al-
Jallad 2015, 4.8f). The masculine singular form and h are attested in Hismaic;
Safaitic attests , and the Harran inscription (568 CE) attests the form d, which can
only be *//. The feminine singular is more difficult to identify. A clear attestation
of a t-based feminine demonstrative occurs in the Namara inscription as ty */t/, and in
Safaitic as well, t h- s1nt this year. A feminine , however, is also attested, h- dr this
region (see Al-Jallad 2015, 4.9). No plural forms have yet been attested.

Relative pronouns are more frequently attested and exhibit a more unified form. In
Hismaic and Safaitic the masculine singular form is attested as */V/, and in two
inscriptions in Safaitic, agreement in definiteness is observed, producing the form h
*/haV/; feminine singular t */at/ (but rarely and t), and plural w */awV/
(Al-Jallad 2015, 4.10). The Namarah Inscription also exhibits dw, probably *//,
without inflection for case. Only the Old Arabic inscription in the Dadanitic script
(JSLih 384) exhibits a reflex of the allai type relative pronoun, the feminine singular
lt */allat/. I have argued elsewhere that the alla- base may be an isogloss of the old
Higz dialects (Al-Jallad 2015, 1.2).

Syntax

Perhaps one of the most marked differences between Old Arabic and later varieties is
the syntax of the infinitive. Instead of the an + subjunctive verb construction of
Classical Arabic or the serial verb constructions of the modern dialects, Old Arabic
employs a nominal form to express many of the meanings expressed by finite verbs in
later stages of the language (see Al-Jallad 2015, 16 for more examples):

yr qy ry
he returned to water dry season to pasture
he returned to permanent water in the dry season to pasture

wrd mn-tln tb
he came down from- Tln to raid
he came down from Tln to raid
mrd l- h- mlk grf ksr h-slslt
he rebelled against- the king Grf to break the chains of bondage
he rebelled against Agrippa the king to break the chains of bondage

The unmarked word order is verb first, and the subject can precede or follow its object,
perhaps reflecting nuances of focus or topic. No overt marker of existential predication
is attested; instead, as found marginally in Classical Arabic and other Semitic
languages, existential sentences are formed simply through the juxtaposition of the two
elements, for example, lg b- h- dr b- ry qbt there was snow in this region during the
rising of Scorpio; ty hl -h w my - sm he went to his family because there was water
in the north (Al-Jallad 2015, 12.1). Both definite and indefinite heads can form
asyndetic relative clauses, e.g. wgm l- bn dd -h msby sbyt -h y he grieved for his
paternal uncles son, who was captured, whom ayyi (the tribe) captured (Al-Jallad
2015, 17.1).

Old Arabic and the modern dialects

The relationship between Old Arabic and the modern dialects is open to investigation.
Several features attested in Old Arabic are found in the modern dialects but do not
appear as part of Classical Arabic. The Graeco-Arabica has put to rest one of the great
debates in the history of Arabic, namely, whether case inflection had disappeared in
some pre-Islamic dialects. The evidence from the Petra Papyri, 6th c. CE, confirms the
loss of this feature, at least when it is expressed by final short vowels:
/arbat Garwn/ the ruin of Garwn; /mt leyl/ the plot of land of
Layla (Al-Jallad et al. 2013). Had case inflection survived in these forms, we would
expect the first term of the genitive constructions to terminate in a case vowel (cf.
above Morphology). Other similarities include the demonstrative prefix h-, which is
found in modern vernaculars, e.g. hal-walad this boy and the ancient varieties. The
syntax of adnominal demonstratives finds parallels in the modern dialects, for example:
JSNab 17 l-qbrw d /al-qabro / is strikingly similar to Egyptian Arabic il-abri da. At
the morphological level, one may point towards the perfective use of the active
participle in Safaitic, which is shared with many modern dialects, e.g. Levantine Arabic
an rib I have drunk with Safaitic (Al-Jallad 2015, 5.5b):

sty nzt nfr mn- -rm

he wintered Toponym having fled from- DEF- Romans

he spent the winter at nzt having fled from the Romans.

The lexicon of Old Arabic is largely unexplored, but promises to be a fertile avenue of
future research.

4 Current Contributions and Research

There is currently only a single monograph-length study dedicated to the subject of Old
Arabic, Mascitelli (2006). The definition of Old Arabic in this work is rather
traditional, relying mainly on inscriptions that attest the definite article al. This
greatly reduces the scope of the study. Moreover, it includes several Ancient South
Arabian texts that most scholars would consider to be in a northern variety of Sabaic
rather than Arabic (96-102). Macdonald (2008) is a useful encyclopedia article
outlining the corpus of Old Arabic, but again focusing mainly on inscriptions that
contain the definite article al. Several outlines of the linguistic geography of Arabia
exist (Beeston 1981; Robin 1991a, b), but these are now outdated in light of the rapid
pace of new discoveries. For the emergence of Arabic as a written language, or rather
Arabic as a language written in the late Nabataean script, see the contribution of
M.C.A. Macdonald in Fiema et al. (2015). The subject is also the theme of the
Supplement to the Proceedings of the Seminar for Arabian Studies 40 (ed. M.C.A.
Macdonald 2010). A forthcoming monograph (Al-Jallad forthcoming b) attempts a
detailed, synthesized picture of Old Arabic as a dialect continuum based exclusively on
documentary evidence.
5 Future Directions

In addition to completing our understanding of the grammar of Old Arabic, which is

dependent upon new discoveries and advances in the interpretation of difficult texts,
much work remains to be done on the question of how Arabic developed into a written
language with an established administrative tradition by the time of the Islamic
conquests. The answer to this question begins with the better understanding of the
spread of the Nabataeo-Arabic script at the expense of the indigenous alphabets of
Arabia. The circumstances under which the Ancient North Arabian scripts disappeared
remain shrouded in mystery. Advances in our knowledge of the pre-Islamic varieties of
Arabic allow for the study of Arabics history on an entirely different scope. The
discovery of more and more authentic pre-Islamic texts will allow scholars to address
issues such as language contact and diglossia before the Islamic conquests by engaging
actual data rather than relying on much later literary accounts. Moreover, it is hoped
that historical Arabic linguists will utilize the growing body of documentary evidence
from the pre-Islamic period to unravel developmental trajectory of later forms of
Arabic both Classical Arabic and the modern dialects.

Bibliography

Al-Jallad, A. 2014. On the genetic background of the Rbbl bn Hfm grave

inscription at Qaryat al-Fw. Bulletin of the School of Oriental and African
Studies, 77, pp 445-465.

----. 2015. An Outline of the Grammar of the Safaitic Inscriptions. Leiden:

Brill.

----. forthcoming a. Graeco-Arabica I: The Southern Levant. In: A. Al-Jallad

(ed.) Arabic in Context, Celebrating 400 years of Arabic at Leiden. Leiden: Brill.
----. forthcoming b. Old Arabic: an evidence-based history. Leiden: Brill.

----. In preparation. The Hismaic Inscriptions.

Al-Jallad, A. and A. al-Manaser. 2015. New Epigraphica from Jordan I: a pre-Islamic

Arabic inscription in Greek letters and a Greek inscription from north-eastern
Jordan Arabian Epigraphic Notes 1: 5170.

Beeston, A.F.L. 1981. Languages of Pre-Islamic Arabia, Arabica (numro

special double: tudes de linguistique arabe), p. 178 186.

Bellamy, J. 1990. Arabic Verses from the 1st/2nd Century: The Inscription of En
Avdat, Journal of Semitic Studies 35.1: 73-79.

El-Sharkawy, M. Pre-Islamic Arabic. Encyclopedia of Arabic Language and

Linguistics. Managing Editors Online Edition: Lutz Edzard, Rudolf de Jong. Brill
Online, 2014. Reference. Universiteit Leiden. 09 July 2014.

Ephal, I. 1982. The Ancient Arabs: Nomads on the Borders of the Fertile Crescent 9th-
5th Centuries B.C. Jerusalem/Leiden: Magnes/Brill.

Fars-Drappeau, S. 2005 Ddn et Liyn: Histoire des Arabes aux confins des pouvoirs
perse et hellnistique (IVe - Iie avant lre chrtienne) (Travaux de la maison de
lorient et de la Mditerrane 42) Lyon: Maison de lOrient et de la Mditerrane.

Graf, D. F. and M. J. Zwettler. 2004 The North Arabian Thamudic E inscriptions from
Urayniba West. Bulletin of the American Schools of Oriental Research 55, 53-89.

Gzella, H. 2011. Northwest Semitic in General. In: S.Weninger, G. Khan, M. Streck,

and J. Watson (eds), The Semitic Languages: An International Handbook,
(Handbcher zur Sprach- und Kommunikationswissenschaft 36). Boston-Berlin:
De Gruyter Mouton. p. 425-451.
Hayajneh, H.. 2011. Ancient North Arabian. In: S.Weninger, G. Khan, M. Streck, and
J. Watson (eds.), The Semitic Languages: An International Handbook, (Handbcher
zur Sprach- und Kommunikationswissenschaft 36). Boston-Berlin: De Gruyter
Mouton. p. 756782.

Hayajneh, H., M.I. Ababneh, and F. Khraysheh. Die Gtter von Ammon, Moab und
Edom in einer neuen frhnordarabischen Inschrift aus Sdost-Jordanien. In: V.
Golinets, H. Jenni, H.-P. Mathys, and S. Sarasin Fnftes Treffen der
Arbeitsgemeinschaft Semitistik in der Deutschen Morgenlndischen
Gesellschaft vom 15.17. Februar 2012 an der Universitt Basel (Mnster,
Ugarit-Verlag) 79-105.

Hetzron, R. 1974. La division des langues smitiques. In: A. Caquot and D. Cohen
(eds.). Actes du Premier Congrs International de Linguistique Smitique et
Chamito-Smitque, Paris 16-19 juillet, 1969 (The Hague, Paris: Mouton)
181194.

Hetzron, R. 1975. Genetic Classification and Ethiopian Semitic. In: J. Bynon and T.
Bynon (eds.). Hamito-Semitica (The Hague: Mouton) 103-127.

Hetzron, R. 1976. Two Principles of Genetic Reconstruction. Lingua 38, 89-104.

Huehnergard, J. 1987. Stative, Predicative, Pseudo-Verb. Journal of Near

Eastern Studies 46: 21532.

Huehnergard, J. 2005 Features of Central Semitic. In: A. Gianto (ed.). Biblical and
Oriental Essays in Memory of William L. Moran (Rome: Pontificio Istituto
Biblico) 155-203.

Huehnergard. 2008. Afro-Asiatic. In The Ancient Languages of Syria-

Palestine and Arabia, ed. Roger D. Woodard. Cambridge: Cambridge
University Press. 22546
Huehnergard. Forthcoming. Arabic in its Semitic Context, In: A. Al-Jallad (ed.) Arabic
in Context, Celebrating 400 years of Arabic at Leiden. Leiden: Brill.

Huehnergard, J. and A. Rubin. 2011. Phyla and Waves: Models of Classification of the
Semitic Languages. In: S.Weninger, G. Khan, M. Streck, and J. Watson (eds), The
Semitic Languages: An International Handbook, (Handbcher zur Sprach- und
Kommunikationswissenschaft 36). Boston-Berlin: De Gruyter Mouton. p. 259
278.

Knauf, E. 2010. Arabo-Aramaic and Arabiyya. From Ancient Arabic to Early Standard
Arabic, 200 CE600 CE, in A. Neuwirth, N. Sinai, and M. Marx (ed.), The Qurn
in Context. Historical and iterary Investigations into Qurnic ilieu, Leiden (Texts
and studies on the Qurn 6), p. 197254.

Kogan, L. 2011. Proto-Semitic Phonetics and Phonology, in S. Weninger, G. Khan, M.

Streck, and J. Watson (ed.), Semitic Languages. An International Handbook, Berlin,
Boston (Handbcher zur Sprach- und Kommunikationswissenschaft 36), p. 54
151.

Macdonald, M.C.A. 2000a. Reflections on the linguistic map of pre-Islamic Arabia.

Arabian Archaeology and Epigraphy 11, 28-79.

----. 2004. Ancient North Arabian (Chapter 16). In: R. D. Woodard (ed.). The
Cambridge Encyclopedia of the Worlds Ancient Languages. 488-533. (Cambridge,
New York: Cambridge University Press).

----. 2008. Old Arabic. In: K. Versteegh (ed.). Encyclopedia of Arabic Language and
Linguistics.Vol. III (Lat-Pu) (Leiden: Brill) 464-474.

----. 2009. Literacy and Identity in Pre-Islamic Arabia. Surrey: Ashgate.

---- (ed.). 2010. The Development of Arabic as a Written Language. Supplement to the
Proceedings of the Seminar for Arabian Studies Volume 40.
Macdonald, M. C. A. and G. M. H. King. 2000. Thamudic. In: Encyclopedia of Islam,
Vol. 10 (Leiden: Brill) 436-438.

Mller, W. W. 1982 Das Frhnordarabische. In: W. Fischer (ed.). Grundri der

arabischen Philologie, Vol. I: Sprachwissenschaft (Wiesbaden: Reichert) 17-29.

Negev, A. 1986. Obodas the God. Israel Exploration Journal. 36:5660.

----. 1991. Personal Names in the Nabatean Realm. Jerusalem: Institute of

Archaeology, Hebrew University of Jerusalem.

Nehme, N. 2013. Epigraphy on the edges of the Roman Empire. A study of the
Nabataean inscriptions and related material from the Darb al-Bakrah, Saudi Arabia,
1st 5th century AD. Mmoire scientifique d'Habilitation Diriger des
recherches, Paris (Ecole Pratique des Hautes Etudes).

Nehme, L. forthcoming. Aramaic or Arabic ? The Nabataeo-Arabic script and the

language of the inscriptions written in this script, In: A. Al-Jallad Arabic in
Context, Celebrating 400 years of Arabic at Leiden. Leiden: Brill.

Owens, J. 2006. A Linguistic History of Arabic. Oxford: Oxford University Press.

Pat-El, N. 2014. The Morphosyntax of Nominal Antecedents in Semitic, and an

Innovation in Arabic. In: Proceedings of the OsloAustin Semitics Workshop. Ed. by L.
Edzard and J. Huehnergard. Abhandlungen fr die Kunde des Morgenlandes.
Wiesbaden: Harrassowitz., pp. 2847.

Rets, J. 2003. The Arabs in Antiquity: Their history from the Assyrians to the
Umayyads. London and New York: RoutledgeCurzon, Taylor & Francis.

Robin Ch. 1991a. Les langues de la pninsule Arabique. Revue du monde musulman
 et de la Mditerrane 61: 89-111.

Ryckmans, J. 1956. "Aspects nouveaux du problme thamuden", Studia Islamica 5, 5-

17

Stein, P. 2006. Zur Dialecktgeographie des Sabischen, Journal of Semitic Studies

49.1: 225-245.

----. 2008. The "Himyaritic" Language in pre-Islamic Yemen. A Critical Re-

evaluation, Semitica et Classica 1: 203-212.

----. 2011. Ancient South Arabian. In: S.Weninger, G. Khan, M. Streck, and J. Watson
(eds), The Semitic Languages: An International Handbook, (Handbcher zur
Sprach- und Kommunikationswissenschaft 36). Boston-Berlin: De Gruyter
Mouton. p. 1042-1072.

Strich, A. 2013. The infinitive in proto-Semitic and Arabic, paper read at Arabic
in Context: Celebrating 400 years of Arabic at Leiden University, November 2,
2013. Leiden, the Netherlands.

Tsafrir, N. 1996. New Thamudic inscriptions from the Negev. Le Muson 109, p. 137-
167.

Wilmsen, D. 2013. Grammaticalization of the Arabic Demonstrative iyy- as a

Pronominal Object Marker in Ditransitive Verbs: An answer to Bravmann.
Journal of Semitic Studies. 58.1:149-167.

Yardeni, A. 2014. A List of the Arabic Words Appearing in Nabataean and Aramaic
Legal Documents from the Judaean Desert, In R.F. Vishnia, R.Zelnick-
Abramovitz, Werner Eck (eds). Scripta Classica Israelica, Yearbook for the Israel
society for the promotion of classical studies,Rome Judaea, and its Neighbors: special
issue in honor of Hannah M. Cotton. Jerusalem: The Israel Society for the
Promotion of Classical Studies., pp. 301-324
7. Further Readings

Al-Jallad, A. forthcoming c. The linguistic landscape of pre-Islamic Arabia: Context for

the Quran. In M. Shah and A. Abdelhalim (eds.), The Oxford Handbook of the
Quran. Oxford: Oxford University Press.

Fisher, G. (ed.). 2015. Arabs and Empires before Islam. Oxford: Oxford University
press. (especially chapter 7).

Macdonald, M.C.A. 2015. On the uses of writing in ancient Arabia and the role of
palaeography in studying them. Arabian Epigraphic Notes 1: 150.

Robin, Ch. 1991b. Les plus anciens monuments de la langue arabe. Revue du monde
musulman et de la Mditerrane, 61: 113-125.