Subscribe to Safaitic

Thursday, October 28, 2021

The particle f- as a dative or direct object marker in Thamudic C?

Thamudic C inscriptions usually open with the phrase wdd 'love' and variants thereof followed by f and a personal name. Scholars have generally treated this phrase as a type of greeting, although the syntactic function of f has not been clear. Winnett (1970) suggested that it should be identified with the Arabic noun , the construct of fam "mouth", the entire phrase meaning 'he loved the mouth of so-and-so." Tsafrir (1996) outlined all the variants of this construction but did not offer an explanation for f. A Thamudic C/D inscription  -- the short text lacks any diagnostic features to distinguish between the two categories of Thamudic -- from the region of Tabuk published on Twitter by @ahmed666551 may shed some light on its use. Unlike most Thamudic C inscriptions, the wdd f phrase is broken up by the prepositional phrase b-y.

1) Thamudic C inscription published here:

Reading: wdd by f-mśr

Translation: 'There is love in me for Mśr'

The syntax of this short inscription indicates that wdd is a noun rather than a verb. The prepositional phrase by should be parsed as the locative preposition b 'in' and the first person common singular clitic pronoun, likely vocalized as ya. The inscription suggests that f should be treated as a dative, similar in function to li in Arabic, meaning 'to', 'for'. Its syntax mirrors a previously attested phrase with by in Thamudic D rbt śq by l-kn ʾmt śkrn 'there is much longing in me for Kn the maidservant of Śkrn' (UdhThamD 1 = JSTham 213; Macdonald 2018b). Thus, the present texts conclusively rules out the interpretation of f as a preposition cognate with Arabic 'in'.

Thamudic C inscriptions also attest the use of l- in the phrase zt l-PN 'this is for PN', so it is unclear what the difference between the two particles is. It is possible that wdd takes a direct object and that f is therefore a direct object marker. We must await the discovery of longer texts to arbitrate between these two possibilities. 


Macdonald, Michael C. A. 2018b. "The Ancient North Arabian and Ancient South Arabian Inscriptions. In The Darb al-Bakrah," edited by Laïla Nehmé . Riyadh: Saudi Commission for Tourism and National Heritage, pp. 227–84

Tsafrir, Nurit. 1996. “New Thamudic inscriptions from the Negev.” Le Muséon 109:79-93.

Winnett, F.V. & Reed, W.L. 1970. Ancient Records from North Arabia. with contributions by J.T. Milik and J. Starcky. (Near and Middle East Series, 6). Toronto: University of Toronto Press.

Sunday, October 10, 2021

Pre-Print: Outline of the Grammar of the Safaitic Inscriptions, second edition

The second edition of the Outline of the Grammar of the Safaitic Inscriptions is in production. You can read sections 4-20 here:

The new addition contains:

1) a more detailed phonological description 

2) newly attested forms in the pronominal and verbal paradigms

3) new negative adverbs, including ls /laysa/

4) new verb stems, including the Ct-stem

and more.

Saturday, October 2, 2021

Is a cognate of Classical Arabic "ǧanaza" attested in Hismaic?

Sadaqah et al. have recently published a new collection of Hismaic inscriptions from Wadi Umm Ṭulayḥa, Jordan in the Jordan Journal For History and Archaeology Volume 51, No. 3, 2021. In the longest text of this panel, the editors identified a verb gnz, which they translate as 'to collect/gather up', viewing it as a loan from Gəʿəz and South Arabian. The following is a photograph of the inscription from the original article and their reading.

If correct, then the verb could provide important evidence for linguistic contact between Safaitic and southern Semitic languages at some point in its pre-history. But a closer inspection of the photograph shows that gnz is a phantom. The medial letter, which is a straight line, may hold the value of n in Safaitic but in Hismaic, and in this particular hand, the n is a simple dot. 

Image 2: The alleged gnz

Image 3: The word bn, where the n is a dot

Rather, the straight line in Hismaic has the value of ś, that is = ش.  The word the editors took as gnz is in fact gśz. Now, gśz is clearly not a word, but we need not commit ourselves to the way the editors parsed the phrase. Hismaic does not employ any markers of word boundaries - these must be inferred by the reader. I would suggest that the g and ś should be read together as a single word, and then the remaining letters would form the final word of the inscription. These correctly read produce zmlt, contra the reading of the original edition, namely, mbnt. What the original editors took as a b is clearly a l once compared with the other l's of the inscription. What caused the editors to take it as a b is a stray line intersecting with the bottom end of the letter, perhaps an over-extension while carving, a slip of the blade, so to speak. The mark they took as an n appears to be a stray bit of damage, part of the same damage pattern that affects the shaft of the l as well. 

Image 4: zmlt 

Image 5: ʾl 

Now, the name zmlt has been previously attested in Hismaic (AMJ 143). The dating component of this text would therefore be snt gś zmlt 'the year of the troop of Zmlt', perhaps referring to when an individual or group called zmlt went to war. Similar modes of dating are attested in Safaitic, usually using ḥrb 'war' rather than a reference to the troop itself. 

There are a few other places where I would suggest a different reading than the original editors. I give my reading of the text below. With the exception of the dating formula and the corrected readings of the names, I follow the interpretation of the original editors. 

l ḏky bn śr ḏ ʾl ʾbt w dṯʾ snt gś zmlt
'By ḏky son of śr of the lineage of ʾbt and he spent the season of the later rains (here) the year of the troop of Zmlt'

ḏky: The name the editors suggested ḏġny is unattested and does not seem to come from a known root. What they read as ġ is more likely a k. They also took a stray mark as a n. The name ḏky 'clever' is well attested in Ancient North Arabian and has previously appeared in Hismaic, e.g. KJC 179. 

śr: The original reading was produced by mistaking stray marks for letters. A small bit of damage before the name was taken as a n, and a stray marking intersecting with the straight line ś caused its confusion with s. The name śr 'evil' is also well attested in Ancient North Arabian. It is possible that the name could be read as śb, but in general the b is less compact in the present author's hand.

ʾbt: The name of the lineage group is likely ʾbt rather than ʾġt. There is some damage on the stone that motivated the latter reading, but looking at it closely the original letter seems to be a bow. If indeed this damage is part of the letter - it is impossible to be certain from the published photographs, then it would produce the name ʾkt. Both of these names are previously attested. 

To conclude, this short, seasonal text does not offer us any lexical connection with South Arabia or Ethiopic. Rather, it is in line with the type of dating formulae employed mainly in Safaitic, where years could be known by major events they witnessed. In this case, the formation of a troop perhaps under the command of a man called Zmlt was a remarkable enough to use to date one's text.


Sadaqah, I.S., Tarawneh, M.B., Abudanah, F. 2021. "A Sort of Sepulchral Construction in Wadi Umm Ṭulayḥa, Southeastern Badia, Jordan?"  Jordan Journal For History and Archaeology Volume 51, No. 3: 97-109.

AJC, KJC = King, G.M.H. 1990. Early North Arabian Thamudic E. A preliminary description based on a new corpus of inscriptions from the Ḥismā desert of southern Jordan and published material. Ph.D thesis, School of Oriental and African Studies, University of London.

Wednesday, September 29, 2021

What present-day diglossia in inscriptions can suggest about the past

The Syro-Arabian ḥarrah, the so-called Black Desert, is home to tens of thousands of inscriptions, both ancient and modern. Public schooling introduced writing to the local nomads and they took to carving copious amounts of texts, just as the inhabitants of this place did twenty centuries ago. What is interesting to see, however, is that while their education led to the mastery of the alphabet, their grasp of the grammar of the written register is less secure. Often times, writers will carve vernacular forms and produce phonetic spellings. This is illustrated by the following text, which was documented in 2019 in Jordanian ḥarrah. The author invokes God to be pleased with him and his parents and terminates with a signature and date.

Image 1: A modern Arabic inscription, dated to 1997 (documented in 2019)

The second line makes use of a stock invocation but the spelling is interesting.

يارب العللمين ارحم ارحمين 

'O Lord of the worlds, the most merciful of the merciful'

In writing this line, he author accidentally reproduces early Arabic orthography, omitting the long alif word internally. He also writes the definite article phonetically, ʾrḥmyn instead of ʾlrʾḥmyn.

The text gets very interesting on the fourth line:

صاحب هظه الخط

'the owner of this writing'

The author uses the vernacular demonstrative pronoun [haðˁa(h)] rather than the high register [hāðā], suggesting that he had only a superficial grasp of the literary language and fell back on his spoken language. On the fifth line, too, he writes 'from the tribe of al-masāʿīd' phonetically: من عشيرت المساعيد.


The knowledge of writing but not of the written language manifests in an interesting way once authors turn to carving pious statements and quotations from the Quran, both of which require use of the high register. 

Image 2: A rendition of Q 112, dated to 1999 (documented in 2019)

The author of this inscription had committed Q 112 to memory. When he turned to engrave it on rock, he carved it phonetically as he had memorized it. The definite article when assimilated is not noted at all: allāhu ṣ-ṣamad is written الله صمد.  Verse 4 is particularly informative. In the Cairo Edition, the verse appears as in (a) while our author renders it as in (b):

a) ولم يكن له كفوا احد

b) لم يكوان لا هو كفوان احد

The divergences in spelling occur in grammatical forms that do not exist in the vernacular. The phrase la-hu 'for him' is not realized as such in the colloquial Arabic of the region. Instead, one would say lah. Perhaps the author was unable to parse it correctly and so he wrote it phonetically, treating the unreduced short vowels as long orthographically. The indefinite accusative kufuwan is further spelled phonetically, as this feature is not productive in any modern vernacular. The spelling of yakun is very curious. It suggests that the writer regarded ان as the correct way to write /n/ in final position.

Surah 112 is attested in another inscription from the area. This author seems to have had a slightly better grasp of the rules of writing, but still produced a number of orthographic anomalies. 

Image 3: Another rendition of Q 112 dated to 1992 (documented in 2019)

This writer did not represent the assimilated article in allāhu ṣ-ṣamad graphically. The spelling of lam yalid in line 2 (v. 3) as لم يلاد may suggest that he failed to distinguish between /i/ and /a/ in an unstressed position. Like our previous author, this writer produced an innovative spelling of lam yakun, adding an alif before the nūn. And predictably, he spelled the tanwīn of kufuwan with a nūn.


This small collection of texts provides a glimpse into the way Arabic was taught at the end of the 20th c. CE. The nomads of this region attended primary school and learned the Arabic script but not very much in the way of the grammar of Modern Standard Arabic. While they committed some Quran to memory, it seems that they were unable to parse the text, and so when they turned to writing it, they spelled it out phonetically. 

This situation stands in sharp contrast to pre-modern Arabic and Quranic inscriptions. While there are some divergences between the standard Quranic text and the verses that are sometimes carved on rock, in general those difference come down to memory, resulting in the substitution of words and paraphrasing. I do not know of one pre-modern text that exhibits the kind of orthographic anomalies and phonetic spellings witnessed in these modern texts. This may suggest that those who could write in pre-modern times were generally better educated in the standard register. Or perhaps the absence of such anomalies suggests that the spoken language in ancient times was much closer to the Quran and so authors were better able to parse the text, even if they only had the most basic training in writing. 

Thursday, August 26, 2021

More pre-Arabic texts from the ʿAsīr

The Twitter account @amrisultan published a number of pre-Islamic inscriptions from bādiyat banī ʿamr in the ʿAsīr region south of the Ḥigāz ( The pre-Islamic documentation from this region is relatively poor compared to other parts of the peninsula so these texts shed important light on the scripts and languages used there before the spread of Arabic as we know it. 

The script: The script is closest to Thamudic B, but with its own particulars. The texts are carved from right to left, similar to Thamudic B, rather than the vertical orientations of C and D. The downward-facing mīm's are compatible with Thamudic B, but the sunburst (line 7) is not a usual feature of this script. It is, however, attested in one Thamudic B inscription discussed previously on this blog, where I have argued that it should be interpreted as . In Thamudic D and some varieties of C, the sunburst renders ś. The H-glyph is attested once - it does not seem to have the value of ʾ as the two-horned alif is attested clearly in line 4 of photo 2. It could therefore be understood as a z or perhaps even a . The comb-glyph in photo 2, line 8 is also attested in Thamudic B and carries the value of ġ. The serpentine n, however, contrasts with the usual straight-line glyph for this phoneme in Thamudic B.

The language: Putting aside the ambiguities of the sunburst and H-glyph, the remaining glyphs attested in these texts have relatively stable phonemic values across the South Semitic scripts. These enable us to read the texts with some ease. The results, however, do not produce meaningful sentences. The only possible morpheme that can be isolated is the l, which begins several lines in Photo 2. This could suggest that it should be understood as an introductory particle related to the dative preposition. In every other respect, it seems that we are dealing with a language quite distinct from the attested Ancient South Arabian languages and Arabic and therefore only the discovery of more texts can lead to the proper interpretation of these ones. I will supply a preliminary reading below each photo. Note that the reading of the inscriptions in photo 2 is very tentative as the quality of the photographs obscures some of the important features of the glyphs. Uncertainties are given between { }-brackets and unreadable letters are marked by a dot.

Photo 1

1. y t y w l m l 
2. m l y b t ʿ

Note: The term btʿ has previously appeared in Thamudic B inscriptions to the north but its interpretation is unclear. 

Photo 2

1. ʿ l {l} n w ... y
2. w n b n {b / k}
3. l m l y H t {.} n
4. . .  w t b ʾ . l y 
5. m ʾ {k} r n 
6. l m m ʿ {b/k} m {.} n m
7. l m l y ʿ t {y}
8. l m l y ġ(?) w y 

Note: The reoccurrence of the word mly in lines 3, 7, and 8 as well as in Photo 1 may indicate that this a formulaic component of the composition.

Friday, August 20, 2021

Quranic etymology 1: ʾaḥwà (Q 87:5) and Safaitic ḥwwt

Quran 87:4-5 describes God's power over the seasonal cycle through the life and death of the land:


وَٱلَّذِىٓ أَخْرَجَ ٱلْمَرْعَىٰ
"he (God) is the one who brings forth pasture"
فَجَعَلَهُۥ غُثَآءً أَحْوَىٰ
"And then transforms it into dried up, dark refuse"

The use of the root ḥwy / ḥww to describe dried-up herbage in the context of drought is found in a Safaitic inscription as well, in the inscription WGRR 1 pictured below.

(courtesy OCIANA)

The author of this text Ġawṯ son of Hāneʾ grieves for Baʿl-Samīn, the weather god (wgm ʿl-bʿlsmn), in the year of dearth (snt mḥl). He then explains that the settlements were not rained up (lm tmṭr h-sknt) and describes the results poetically as: lwny rdn ḥwwt "it (the land) became dark, yellowish-red," no doubt referring to the color of the dried-up herbage after a long period of no rain. 

The Safaitic ḥwwt would appear to be the 3fs verb cognate with Classical Arabic iḥwawā "to become dark", utilizing the rare reduplicated pattern to signify a change in color or defect. Quranic ʾaḥwà is an an elative formation of the same root. 

The use of the same rare root in both corpora, and with the same seasonal reference, is another example of the shared cultural and linguistic background of the Ancient North Arabian inscriptions and the Quran.

For a discussion of the verb ḥwwt in Safaitic, see Al-Jallad, A. & Jaworska, K.A. A Dictionary of the Safaitic Inscriptions. Leiden: Brill, 2019, s.v.

Reference to the inscription:

Tuesday, August 17, 2021

Jebel Qarnayt and Pre-Arabic in the Ḥigāz

South of Ṭāʾif in the area of al-Shafā is a 2250 m. mountain called Jabal Qarnayt ( 

(A view of Jebel Qarynat looking south)

The name of the mountain does not seem to derive from Classical Arabic or the contemporary spoken varieties but rather finds its source in the Modern South Arabian family of languages. The term Qarnayt is in fact attested in the Mehri language, where it refers to a type of herbage. Perhaps this mountain was named after the abundance of grass growing at its base. 

Johnstone, Mehri Lexicon, p. 236

Toponymy is often a good source of information for pre-historic linguistic geography. The etymology of Jebel Qarnayt may therefore suggest that the geographical range of the Modern South Arabian languages was much larger in ancient times, receding as Arabic spread south and east from the Ḥigāz. But what direct evidence do we have for "pre-Arabic" in this region? The area of Jebel Qarnayt is unfortunately quite poor when it comes to epigraphic documentation but there is at least one Ancient Arabian text from mountain that has so far defied decipherment. 

Al-Ḥāriṯī, N. al-muʿǧamu al-ʾaṯariyyu li-muḥāfaẓati Ṭāʾif, pg. 29

The decipherment of the text is not helped by the poor quality of the photograph. But there does appear to be an h-stem on the fourth line reading hḥyt, which could be regarded as a causative of the root ḥyw 'to live', in the feminine singular, meaning 'to greet', cf. Arabic taḥiyyāt and in Sabaic hḥyw 'to cure,' 'to heal'. Above it there appears to be a signature ḥb ktb, that is, "Ḥb has written". The h-stem is consistent with the Mehri language as well as Sabaic. The meaning of the rest of the text is open to speculation; in my opinion further examples are required for its complete decipherment.

The name of the mountain combined with this single inscription provide us with a tiny glimpse at the linguistic situation of this region before the domination of Arabic as we know it. There is no doubt that a deeper investigation of the toponymy along with better epigraphic documentation will sharpen this image.


Al-Ḥāriṯī, N. 2001. al-muʿǧamu al-ʾaṯariyyu li-muḥāfaẓati Ṭāʾif. Taif: Liǧnatu l-maṭbūʿāti fi t-tanšīṭi s-siyāḥiyy.

Johnstone, T. M. \ Smith, G. Rex [Ed.]. 1987. Mehri Lexicon and English-Mehri Word-List. London: SOAS.

The particle f- as a dative or direct object marker in Thamudic C?

Thamudic C inscriptions usually open with the phrase wdd 'love' and variants thereof followed by f  and a personal name. Scholars ha...