Early Geez orthography in the ʔAbbā Garimā Gospels and in the Aksumite Epigraphic Geez

Maria Bulakh

doi:10.1017/S0041977X26101037

Early Geez orthography in the ʔAbbā Garimā Gospels and in the Aksumite Epigraphic Geez

Published online by Cambridge University Press: 06 March 2026

Maria Bulakh

Show author details

Maria Bulakh*: Affiliation:
HSE University, Moscow, Russia
*: Email: mbulakh@mail.ru

Article contents

Abstract
Introduction
Periodization of the Geez language and classification of Geez sources
Early Post-Aksumite inscriptions in RIÉ
The relevance of orthography for the dating of Geez sources
Orthography of AG I and III vs. orthography of Aksumite Epigraphic Geez
Conclusions
Funding information
Footnotes
References

Rights & Permissions

Abstract

ʔAbbā Garimā I and III (AG I and III) are the oldest manuscripts written in Geez (Old Ethiopic). According to a recent radiocarbon analysis they are dated to the Aksumite period. The present paper confronts the orthography of these manuscripts with that of the epigraphic corpus of the Aksumite period (the comparison is restricted to the roots containing gutturals and/or sibilants, since the preservation of these consonants is the hallmark of pre-seventh-century Geez). The investigation demonstrates remarkable agreement in spelling roots with gutturals/sibilants – an additional argument in favour of dating of AG I, III to the Aksumite period. The paper also discusses the general problem of periodization of the Geez language, focusing on the early post-Aksumite period (eighth–thirteenth century) and providing a preliminary survey of direct and indirect Geez sources pertaining to this period.

Keywords

Geez Old Ethiopic periodization of the Geez language Epigraphic Geez Geez orthography the Garima Gospels

Information

Type: Article
Information: Bulletin of SOAS , Volume 89 , Issue 2 , June 2026 , pp. 367 - 399

DOI: https://doi.org/10.1017/S0041977X26101037 [Opens in a new window]
Copyright: © The Author(s), 2026. Published by Cambridge University Press on behalf of SOAS University of London.

I. Introduction

In 2005 Alessandro BausiFootnote ¹ pointed out the significance of early manuscripts in Geez for the study of the Geez epigraphic corpus, stressing that any investigation of the earliest stage of Geez must take into consideration the evidence of such manuscripts. When Bausi wrote his seminal paper, the dominant opinion held that all extant manuscripts in Geez dated to the post-Aksumite period.Footnote ² The corpus of Geez inscriptions from the Aksumite period, found in both Ethiopia and Eritrea, and elsewhere, was regarded as the only direct source of our knowledge on the Geez language as used by native speakers.Footnote ³ Still, Bausi successfully demonstrated that even post-Aksumite manuscripts may display a number of features transmitted from the earlier stage.

The investigations of the famous ʔAbbā Garimā GospelsFootnote ⁴ carried out in the first two decades of the twenty-first century have led to a revision of their dating. While there is no general consensus on the dating of ʔAbbā Garimā I and III, their belonging to the Aksumite period, on the grounds of a radiocarbon analysis, is today considered a persuasive hypothesis.Footnote ⁵ In light of this approach, comparing the evidence of these two oldest Geez manuscripts with material found in the epigraphic corpus of Geez has become acutely relevant.

Among the features of epigraphic Geez, the consistent spelling of gutturals and sibilants has been observed for inscriptions up to the sixth century.Footnote ⁶ The gutturals ([ʔ], [ʕ], [h], [ḥ], [ḫ]) and voiceless sibilants ([s], [ś], [ṣ], [ṣ́]) were distinct phonemes at the time of the creation of the Geez syllabary,Footnote ⁷ but in the traditional Geez pronunciation, there is no distinction between [ʔ] and [ʕ]; [h], [ḥ], and [ḫ]; [s] and [ś]; [ṣ] and [ṣ́]. It is generally accepted that the merger between the respective sounds took place in the late Aksumite or in the early post-Aksumite periods, since the post-Aksumite manuscripts display interchangeability between the graphemes reflecting these soundsFootnote ⁸ (for brevity’s sake, they will be occasionally referred to in this article as “problematic” graphemes).

Now, a recent analysis of AG I and AG III has revealed remarkable consistency in the spellings involving the pertinent graphemes – a fact which supports the dating of these manuscripts to the Aksumite period.Footnote ⁹

If this dating is accepted, one should expect full coincidence in the spelling of the “problematic” graphemes between AG I and III, on the one hand, and the Aksumite epigraphic corpus, on the other. The purpose of this paper is to verify this hypothesis – that is, to confront the spellings of the roots containing gutturals and sibilants attested in the Aksumite inscriptions with the spellings of AG I and III.

In Section II, discussion of the terms “Aksumite” and “post-Aksumite” will be offered, since epigraphic material belonging to the post-Aksumite period has to be excluded from the present investigation. It is hoped that this discussion will contribute to the clarification of the general problem of periodization of the Geez language. Section III lists the inscriptions classified as Aksumite by the editors of RIÉ,Footnote ¹⁰ but which, with varying degrees of certainty, can now be reassessed as post-Aksumite. Section IV considers the question of whether orthography can be used as a solid basis for the dating of a given text witness. Section V presents the list of the relevant lexemes of Aksumite inscriptions confronted with the tokens of the same roots attested in AG I and III, accompanied by short discussions of a few specific cases. Section VI summarizes the results of the investigation.

II. Periodization of the Geez language and classification of Geez sources

Most grammatical descriptions of Geez contain brief accounts of the history of the language, employing such terms as “Aksumite period” and “post-Aksumite period”. These terms, however, seldom refer to the history of the language per se. Rather, they are used in the chronology of Geez literature.Footnote ¹¹ Now, the distinction between the periodization of the language and that of the literature in this language is far from trivial in the case of Geez, since there are no literary works in Geez reliably dated to the period between the seventh and thirteenth centuries.Footnote ¹² Thus, the post-Aksumite period of Geez literature can indeed be defined as starting in the fourteenth century.Footnote ¹³

The situation with texts in Geez (which encompass literary works as well as non-literary texts such as land grants and short inscriptions) is different. We do possess a few direct witnesses of the Geez language produced after the decline of the Aksumite empire, but before the Solomonic period (I will refer to this period as “early post-Aksumite”, see Table 1). They include the inscriptions on the tābots of Lālibalā churches,Footnote ¹⁴ the fragments of the Sermo de transfiguratione of Anastasius the Sinaite preserved on wooden panels in Beta Gabrəʔel and Madḫane ʕĀlam churches of Lālibalā,Footnote ¹⁵ the famous inscription on a metal cross from the church of ʕUrā Masqal,Footnote ¹⁶ and the inscription on the metal cover of the Gospels of ʔƎndā ʔAbbā Maṭāʕe.Footnote ¹⁷ Some inscriptions included in RIÉ likely date to the early post-Aksumite period (see Section III below). The heavily damaged fol. 208 of a Four Gospels manuscript of the Madḫane ʕĀlam church in LālibalāFootnote ¹⁸ contains a colophon and a donation by King Lālibalā, both partly illegible, and both presumably providing direct evidence of the early post-Aksumite period.Footnote ¹⁹ The creation of the B-text of the Gospel of Matthew (as a revision of the A-text) has been dated to the twelfth–thirteenth century ad,Footnote ²⁰ and thus, elements of the B-text found in the pre-Solomonic mssFootnote ²¹ can be counted, with some reservations, as direct witnesses of early post-Aksumite Geez.

Table 1.

Tentative periodization of the Geez language

Geez sources shift from mainly Aksumite inscriptions and coins in the first to seventh centuries to mostly manuscript and some epigraphic evidence from the eighth century onward. See long description.

Table 1 Long description

The table outlines a periodization of the Geez language by period, time span, and the kinds of evidence used as direct versus indirect sources. The Aksumite period, dated to the first through seventh centuries ad, relies directly on inscriptions and coin legends, plus some non-epigraphic material, while indirect evidence comes from Aksumite texts that survive only in later manuscripts. The early post-Aksumite subperiod, dated to the eighth through thirteenth centuries ad, has direct epigraphic evidence from a small set of inscriptions and church objects, alongside non-epigraphic sources, and indirect evidence from texts attributed to this era but copied later. The Solomonic subperiod, from the fourteenth century ad to the present, is supported directly mainly by Geez manuscripts and some epigraphy, with no indirect sources listed. Across the timeline, the evidence base moves from predominantly epigraphic and numismatic material to predominantly manuscript transmission. A key caveat is that a single surviving witness can inform more than one period, since later copies may preserve earlier language stages.

Obviously, a given text witness can simultaneously be a direct source for one period and an indirect source for an earlier period.

Although in previous works, I have used the term “Epigraphic Geez” to refer to the Aksumite epigraphic corpus only,Footnote ⁵³ in the present work, to avoid confusion, I will distinguish between “Aksumite Epigraphic Geez” (AEG), “early post-Aksumite Epigraphic Geez”, and “Solomonic Epigraphic Geez”. The term “early Geez” will be used in the present article as a synonym for “the Aksumite Geez”.

Furthermore, manuscripts and fragments of manuscripts datable to between the eighth and thirteenth centuries, even if they are witnesses of texts written in the Aksumite period, may contain some linguistic material specific to the period during which the particular manuscripts were created.

Some of the manuscripts relevant to the present discussion have received attention in special studies: the Aksumite collection, also known as Sinodos of Qəfrəyā (by far the best-studied early post-Aksumite manuscript);Footnote ²² the famous Gospels manuscript from the Madḫane ʕĀlam (EMML 6907);Footnote ²³ and a homiliary from Ṭana Qirqos (EMML 8509).Footnote ²⁴

In these and some other cases,Footnote ²⁵ pre-Solomonic manuscripts or fragments of manuscripts are regarded as witnesses of Aksumite literary texts.Footnote ²⁶ It is rare that the possibility of an early post-Aksumite composition comes into question. In this connection, one should mention a heavily damaged folio datable to 1160–1265 ad recently found outside Ethiopia, in the Monastery of St Antony in Egypt, which preserves fragments of two distinct texts in Geez. These fragments have been carefully edited in an article devoted especially to this sensational discovery.Footnote ²⁷ The text on the verso side of the folio has been identified as the treatise “On Silence” attributed to John, Bishop of Constantinople, presumably the same as John Chrysostom.Footnote ²⁸ This text is also known from later Ethiopian manuscripts as a part of the Collectio Monastica (Aethiopica). The text on the recto side of the folio has not been identified. The editors of the textFootnote ²⁹ mention the possibility of the folio containing post-Aksumite texts – either translated or directly composed in Geez. In both cases, the folio has become a witness not only for the language, but also for the literary texts (original or translated) of the twelfth–thirteenth century.Footnote ³⁰

In addition, a few land grants and other documents from the early post-Aksumite period are known in later copies.Footnote ³¹ Some evidence regarding the language of the period at which the originals were created can be gleaned from them.Footnote ³²

Thus, (direct and indirect) sources on the period between the fall of the Aksumite Empire and the beginning of the Solomonic period are not entirely absent. Obviously, they should not be overlooked in the discussion of the history of Geez language – all the more so since it was probably this period that saw the death of Geez as a vernacular.Footnote ³³

In view of the necessity of distinguishing between the primary and secondary witnesses to a given period in Geez history, it is sensible to follow the approach offered by A. Butts,Footnote ³⁴ who classifies Geez according to the types of sources from which the information on Geez grammar can be drawn. The four categories established by Butts are: Epigraphic Geez, Aksumite Geez, Solomonic Geez, and the modern pronunciation tradition. Among these categories, “Aksumite Geez” and “Solomonic Geez” in fact refer to one type of sources, namely, to the “literary form(s) of the language as attested in manuscripts”.Footnote ³⁵ The distinction between the two is defined in chronological terms: Aksumite Geez is held to span from the fourth to the beginning of the ninth century ad; Solomonic Geez starts in 1270 ad (the traditional, although by no means reliable, date of the Solomonic dynasty coming to power) and lasts into the present time. Thus, in Butts’ terminology, “Aksumite Geez” and “Solomonic Geez” are periods of Geez literature rather than of Geez language per se.

Butts’ approach can be accepted with a few additions and improvements. First, non-literary texts in Geez (donations, additions to the manuscripts, subscriptions on frescoes, icons, and manuscript illuminations, etc.) must be also taken into consideration as valuable linguistic sources. Second, the epigraphic corpus in Geez is not confined to the Aksumite period. As is well known, Geez inscriptions continued to be produced after the fall of the Aksumite state and well into modern times.Footnote ³⁶ Thus, one should distinguish between various periods of Epigraphic Geez. Third, the term “Aksumite period” is ambiguous insofar as it refers to different time spans in different works.Footnote ³⁷ It is reasonable to follow the dating of the Aksumite period commonly accepted nowadays by historians, namely, first–seventh century ad.Footnote ³⁸ Within this approach, some inscriptions classified in RIÉ as “Aksumite”,Footnote ³⁹ are in fact early post-Aksumite. In the first place this involves RIÉ 193 I, II, but also a few other potential candidates in the RIÉ corpus (see Section III).

Finally, it is important to distinguish between direct (epigraphy, contemporary manuscripts) and indirect (later copies) sources: while the evidence from the latter must be taken with a great deal of caution, it is not to be ignored entirely.

The present work distinguishes two periods of Geez language – Aksumite (first–seventh century ad) and post-Aksumite (eighth century ad – present time).Footnote ⁴⁰ In turn, the post-Aksumite period can be further divided into the early post-Aksumite (eighth–thirteenth centuries ad)Footnote ⁴¹ and Solomonic (fourteenth century ad – present time); see Table 1. Presumably during the early post-Aksumite period, Geez ceased to be a spoken language.Footnote ⁴² During the Solomonic period, the variety called Classical GeezFootnote ⁴³ came into existence.

III. Early Post-Aksumite inscriptions in RIÉ

The term “Aksumite Geez inscriptions” is not compatible with all inscriptions catalogued in the major source of Geez epigraphy, Recueil des inscriptions de l’Éthiopie des périodes pré-axoumite et axoumite (RIÉ). No fewer than three of them (see III.1–2) are more plausibly dated to the early post-Aksumite period. This dating is proposed on archaeological and historical grounds and is supported by their orthography/phonology. A fourth inscription (discussed in III.3) has been dated to the early post-Aksumite period, which is apparently in contradiction to its orthography. A still more uncertain case will be discussed in III.4.

III.1. RIÉ 193 I, II as post-Aksumite inscriptions

RIÉ 193 I, II are two partly vocalized inscriptions whose author calls himself ḥaṣ́āni Dāniʔel. In the recent studies,Footnote ⁵⁴ they are dated between the ninth and twelfth centuries ad. Importantly, the late dating agrees with the orthography of the inscriptions,Footnote ⁵⁵ which contain several examples of interchange of gutturals and sibilants, as presented in Table 2.

Table 2.

Orthography of RIÉ 193 I, II

RIE 193 spellings differ from earlier inscriptions, notably using s with a dot where earlier texts consistently use an alternate s form, and showing variant forms for led them and months. See long description.

Table 2 Long description

Spellings of three words in RIE 193 parts I and II are compared with earlier inscriptions. For I captured and captives, RIE 193 uses forms beginning with a dotted s, while earlier inscriptions use an alternate s form for the same words and do so consistently across cited examples. For I led them, RIE 193 presents a form read from a parallel source, while earlier inscriptions show related verbs meaning passed on, indicating a different but comparable verbal tradition. For months, RIE 193 is read as a form without a certain guttural consonant, whereas earlier inscriptions regularly include that consonant in the word for months. The overall pattern is that RIE 193 shows orthographic divergence from earlier, more uniform spellings. Readings for the led them and months entries are uncertain because the relevant lines are poorly preserved or illegible, so those comparisons are more tentative than the captured example.

III.2. RIÉ 194 as a post-Aksumite inscription

The unvocalized inscription RIÉ 194 is likely unrelated to ḥaṣ́āni Dāniʔel’s inscriptionsFootnote ⁵⁸ but may well be post-Aksumite,Footnote ⁵⁹ being inscribed upon a slab once belonging to an Aksumite stone throne. The post-Aksumite dating would also explain its orthography – exactly as ḥaṣ́āni Dāniʔel’s inscriptions, it contains several instances of interchanges between the “problematic” graphemes, summarized in Table 3.

Table 3.

Orthography of RIÉ 194

RIE 194 often uses dotted s and plain s where Aksumite aligned spellings prefer an alternate s form, while one verb form matches Aksumite usage. See long description.

Table 3 Long description

Spellings found in RIE 194 are compared with spellings that either diverge from or align with Aksumite sources, and with spellings in earlier inscriptions or the Abba Garima Gospels. For the verb to come, RIE 194 has forms meaning I having come and he came written with plain s, while an Aksumite aligned form meaning who came and earlier inscriptions meaning they came are consistently written with a dotted s. For the word king and the phrase my kingdom, RIE 194 writes these with plain s, whereas earlier inscriptions consistently use an alternate s form in the corresponding words, and no Aksumite aligned form is listed in the middle column for this item. For character, RIE 194 has one spelling, while the Abba Garima Gospels attest a related form meaning nature and character, again with no Aksumite aligned form provided in the middle column. Overall, the entries suggest RIE 194 tends to use plain s where other sources prefer the alternate s form, while the come forms show a closer match to Aksumite practice through use of the dotted s. The table provides examples rather than counts, so it indicates tendencies but not frequency or statistical strength.

III.3. RIÉ 232 as a post-Aksumite inscription

The funeral inscription RIÉ 232 from the village of Ham (southern Eritrea)Footnote ⁶⁰ contains the date of its production, which has been deciphered as corresponding to 873 ad by M. KroppFootnote ⁶¹ and, alternatively, as corresponding to 974 ad by A. Bausi.Footnote ⁶² At least one further argument in favour of its post-Aksumite dating can be brought forward: as Prof. Shiferaw Bekele (p.c.) pointed out to me, the name of the buried girl’s father, spelled as Mangaśā,Footnote ⁶³ must definitely reflect the Amharic (or another South-Ethio-Semitic) personal name Mängäša, built after the pattern mäC₁C₂äC₃a or mäC₁C₂äC₃iya (with palatalization s > š). The nominal patterns mäC₁C₂äC₃a < *maC₁C₂aC₃ā and mäC₁C₂äC₃iya < *maC₁C₂aC₃iyā are not recorded in Geez or Tigre, whereas the pattern mäC₁C₂äC₃ya (and related patterns) in Tigrinya is in all probability borrowed from Amharic.Footnote ⁶⁴ In other words, these patterns are purely South-Ethio-Semitic formations.Footnote ⁶⁵ The presence of such a name in RIÉ 232 must point to a prominent presence of South-Ethio-Semitic speakers in the vicinity of Ham village at the time of the production of the inscription, which is ill-compatible with the Aksumite dating.Footnote ⁶⁶

Thus, the contents of the inscription seem to indicate early post-Aksumite dating. At the same time, the orthography of RIÉ 232 shows none of the features characteristic of RIÉ 193 I, II, RIÉ 194, discussed above. On the contrary, it impeccably preserves all the “problematic” graphemes, quite numerous in the text: mangaśā Footnote ⁶⁷ (l. 1; cf. RIÉ IIIB 650–1),Footnote ⁶⁸ bawarḫa (ll. 1–2; cf. RIÉ IIIB 650–1), ʔame and ʔama (ll. 2, 12; cf. RIÉ IIIB 587), ʕəlat and mawāʕəlihu (ll. 3, 8; cf. RIÉ IIIB 674–5), rabuʕ (ll. 3–4; cf. RIÉ IIIB 657), ʔəl(l)a (ll. 5, 14; cf. RIÉ IIIB 553, 683–4), śahl (l. 5; cf. Ricci Reference Ricci2003, fig. 11; Bausi et al. Reference Bausi, Harrower and Dumitru2020b: 38),Footnote ⁶⁹ ṣəḥuf and ṣəḥəf (ll. 6 and 12; cf. RIÉ IIIB 669–70), wəsta (ll. 9, 13, 14–5; RIÉ IIIB 678–9), ʔəm (l. 7; cf. RIÉ IIIB 585–7), ʔanəst (l. 7; cf. RIÉ IIIB 590), śəgā (l. 9; cf. RIÉ IIIB 665–6), satiya (l. 10; cf. RIÉ IIIB 664), ʔana (l. 11; cf. RIÉ IIIB 589–90), ʔanaśəʔa and yətnaśəʔu (ll. 11, 13–4; cf. RIÉ IIIB 652–3), daḫāri (l. 12; cf. RIÉ IIIB 608–9), yaḥayəwu (l. 14; cf. RIÉ IIIB 625).

To these instances, the spellings śarq (l. 2), maḫātəwihu (l. 3), ḫarifu (l. 4), ʕāmat (l. 4), bāḥət(t)u Footnote ⁷⁰ (l. 5), balaʕa (l. 9), and yəṭəʕəmā (l. 10) can be added: these lexemes are unattested elsewhere in the epigraphic corpus,Footnote ⁷¹ but the spelling is in accordance with the well-established etymology.Footnote ⁷² The lexeme bəṣ́uʕ in line 6 may be related to bṣ́ʕ in two other RIÉ inscriptions (see RIÉ IIIB 601), and this spelling finds some support in the etymology.Footnote ⁷³ The spelling of the name of the month taḥsās (line 2) is not informative since no secure attestations of this word in Aksumite Geez are available (connection to ḫśś “to search” is highly hypothetical, see CDG 573). The semantically opaque lexeme ṣarane Footnote ⁷⁴ (ll. 4–5) has been discussed in detail by Bausi.Footnote ⁷⁵ The lexeme ḥədāṭ (l. 7) is not attested elsewhere in RIÉ,Footnote ⁷⁶ its etymology is obscure,Footnote ⁷⁷ but it is worth noting that the spelling of RIÉ 232 coincides with AG I and AG III.Footnote ⁷⁸

Moreover, the orthography of the word forms containing gutturals in RIÉ 232 displays remarkable similarity with the orthography of the fourth-century ad inscriptions.Footnote ⁷⁹ The guttural rules 2a and 2bFootnote ⁸⁰ are not reflected: taḥsās (l. 2), ḫarifu (l. 4), śahl Footnote ⁸¹ (l. 5), yaḥayəwu (l. 14). The sequence Hā appears only in those positions in which the vowel is historically long: maḫātəwihu (l. 3), ʕāmat (l. 4), daḫāri (l. 12). Furthermore, the writing balaʕa (l. 9), inconsistent with the norms of Classical Geez, finds two parallels in a fourth-century inscription.Footnote ⁸²

The spelling məwutān Footnote ⁸³ in l. 13 is non-informative since both məwutān and məw(ə)tān can be found in early and medieval Geez manuscripts. In particular, AG I consistently spells C₁əC₂uC₃ forms of the C₂=w roots with w in the second order, whereas AG III has the sixth order.Footnote ⁸⁴

At the same time, the spellings wəlata “daughter (st. constr.)” (l. 1) and ṣəḥəf “written” (l. 12) suggest a link to some early post-Aksumite manuscripts with “odd” vocalization (an expression used by D. Nosnitsin in a discussion of one such witness).Footnote ⁸⁵ Admittedly, irregular vocalization can be also detected in Aksumite Epigraphic Geez (gafʕan “he oppressed me/us”, RIÉ 188:7, instead of gafʕani or gafʕana; waqatal “and he killed”, RIÉ 188:7, instead of qatala; nadaq “stone”, RIÉ 189:36, instead of nədq, etc.).Footnote ⁸⁶

Potentially, another linguistic feature of RIÉ 232 can be pointed out which links it with a fourth-century inscription RIÉ 188: according to the interpretation proposed by Bausi,Footnote ⁸⁷ the word ṣarane of RIÉ 232 is to be identified with ṣarane in RIÉ 188:6, in both cases to be analysed as a common noun in a status constructus. Neither ṣarane nor its hypothetical status absolutus form *ṣaran is attested elsewhere in Epigraphic or Classical Geez.

All in all, the arguments brought forward by BausiFootnote ⁸⁸ suggest that we are faced with an early post-Aksumite inscription, which, however, consistently displays archaic features in its orthography.Footnote ⁸⁹ A possible explanation is that RIÉ 232 was created by a native speaker of Geez and that its consistent spelling of the “problematic” graphemes reflects the actual pronunciation of the inhabitants of the village of Ham in the tenth century ad. Within this approach, RIÉ 232 is the last direct witness of Geez as a living language.

III.4. RIÉ 202 as a post-Aksumite inscription?

RIÉ 202Footnote ⁹⁰ consists of one line and contains a reliable example of a misspelt root:

za-ṣ́aḥafkəwo “I who wrote it” (RIÉ 202:1).

This spelling contrasts with that of other inscriptions in the RIÉ corpus, where the root “to write” is consistently written ṣḥf (see RIÉ IIIB 669–70).

The inscription has been assessed by several scholarsFootnote ⁹¹ as post-dating the inscriptions of the sixth century. The only argument in favour of its post-Aksumite dating is its “post-Aksumite orthography”.Footnote ⁹² At the same time, it contains at least one archaic, or archaizing, feature: the vertical word-divider.Footnote ⁹³

IV. The relevance of orthography for the dating of Geez sources

Can one use the confusion of the “problematic” graphemes as a solid argument for dating a Geez inscription?Footnote ⁹⁴ And can one reliably date the phonological merger between the respective phonemes?

The facts are as follows.

The inscriptions reliably dated to the sixth century ad or earlier contain practically no instances of confusion between the “problematic” graphemes.Footnote ⁹⁵ In particular, consistent spelling of the “problematic” graphemes in Kāleb’s and WʕZB’s inscriptions of the sixth-century ad suggests that the relevant phonemes were clearly distinguished in the variety of Geez used by the authors of these inscriptions.Footnote ⁹⁶

Reliable instances of deviations from the rest of the RIÉ corpus, as far as the spelling of the “problematic” graphemes is concerned, are attested in RIÉ 193 I, II, 194, 202. Moreover, in RIÉ 194, the spelling of the root mṣʔ varies within one and the same inscription. For RIÉ 193 I, II, 194, some non-linguistic arguments in favour of the early post-Aksumite dating can be brought to bear.Footnote ⁹⁷ For RIÉ 202, no such arguments are known.

Confusion between ṣ and ṣ́ has been noted by Robin for two of the inscriptions in Ethiopic script found in the famous cave of Hoq on Soqotra island: bṣʕ in Hoq 2:34 [2] vs. bṣ́ʕ in Hoq 2:25 [2] and 2:27 [1].Footnote ⁹⁸ For these inscriptions, the general context suggests the Aksumite dating.Footnote ⁹⁹ However, the very fact of interchangeability ṣ/ṣ́ in Hoq 2:25, 27, 34 can be questioned: palaeographically, the reading bṣʕ in Hoq 2:34 [2] is not certain, and the final symbol may well be l rather than ʕ.Footnote ¹⁰⁰

All in all, in the absence of dated Aksumite or early post-Aksumite inscriptions with reliable instances of misspelt “problematic” graphemes, one cannot date the phonological merger with certainty. As a result, any dating of an inscription as post-Aksumite on the grounds of the confusion of the “problematic” graphemes can be criticized as circular reasoning. Moreover, we do not know if the merger of the respective phonemes could have taken place in some varieties of Geez prior to the sixth century or at the time of the creation of RIÉ 191 or 192.Footnote ¹⁰¹ Thus, orthography (as well as other linguistic features) can be used in the dating of the Geez inscriptions only as an additional argument (just as palaeography and punctuation).

The same applies, of course, to the dating of the manuscripts.

V. Orthography of AG I and III vs. orthography of Aksumite Epigraphic Geez

As observed in Section I, the Aksumite Geez inscriptions display consistency in the spelling of sibilants and gutturals. If AG I and III were indeed produced before the seventh century ad, one should expect full coincidence with the orthography of these inscriptions.

In general, this prediction is fulfilled for AG I, III. A fair number of roots with the “problematic” graphemes found in the epigraphic corpusFootnote ¹⁰² have been detected in AG I, III with the same spelling. The list in Table 4 contains these roots, as well those roots which are absent from AG I, III.Footnote ¹⁰³

Table 4.

The coincidence of spelling in Epigraphic Geez and the ʔAbbā Garimā Gospels

Most Epigraphic Geez roots listed have matching forms in the ʔAbbā Garimā Gospels, with only a minority absent in the manuscripts. See long description.

Table 4 Long description

The table compares individual Geez roots and particles attested in Epigraphic Geez inscriptions with their corresponding attestations in the ʔAbbā Garimā Gospels manuscripts (AG I and AG III). Each row gives an Epigraphic Geez form with an inscription reference and a matching AG form with folio and line references, or a dash when the root is not found in AG. Many entries show direct overlap, including common function words and markers such as the quotation particle ʔa-, the negation marker ʔi-, pronouns, and prepositions, as well as frequent nouns and verbs like “father,” “stone,” “world,” “king,” “write,” and “pray.” Several Epigraphic Geez items are reconstructed (marked with an asterisk in the source), while the AG column often provides an inflected or derived form rather than the exact citation form, indicating morphological variation alongside lexical continuity. A smaller set of roots appears only in Epigraphic Geez or is otherwise unattested in AG in this list, for example ʔsy “to feed” (uncertain), taʕaqṣa “to stumble,” and several nouns such as “honey,” “island,” “stele,” and “beer.” Some items are noted as limited to a particular inscription source, which affects how broadly they represent Epigraphic Geez usage. Meanings for a few roots are uncertain or debated, so absence or mismatch should be interpreted cautiously as it may reflect ambiguity, limited surviving text, or differing genres rather than true nonexistence.

Thus, on the whole, the orthography of AG I, III agrees with that of Aksumite Epigraphic Geez. Potential exceptions are discussed below. In each of these cases, the identification of the EG root is hypothetical. Therefore, none of them can count as a serious argument against dating of AG I, III to the Aksumite period.

V.1. Aksumite Epigraphic Geez (AEG) vs. AG I, III?

V.1.1. ʕṣ́b (?) in RIÉ 196 vs. AG I, III ʕəṣub “difficult”

The possible connection of ʕṣ́b in RIÉ 196 (an inscription consisting of one word only) with ʕəṣub “difficult”Footnote ¹¹⁸ is discussed and rejected by A. Bausi.Footnote ¹¹⁹

V.1.2. *ʕawādi “herald” or ʕwd “to surround” (?) in RIÉ 190:15 vs. AG I, III ʔawādi, ʔoda

In RIÉ 190:15, an unvocalized inscription written in South Arabian script, ʕwdm Footnote ¹²⁰ was interpreted as the verb “to surround” by MarrassiniFootnote ¹²¹ or, alternatively, as the noun *ʕawādi “herald” by Drewes.Footnote ¹²² Both the verb and the noun are spelled with the initial ʔ in AG I, III: ʔawādi,Footnote ¹²³ ʔoda.Footnote ¹²⁴ The root ʕwd is present in the ʔAbbā Garimā Gospels only in the noun ʕawd “court; threshing floor”.Footnote ¹²⁵

Still, it is far from certain that we are faced with disagreement between the spelling of EG and AG I, III: the passage in RIÉ 190:15 is obscure and the interpretation of ʕwdm as the noun *ʕawd “court” cannot be ruled out.Footnote ¹²⁶

V.1.3. *gys (?) in RIÉ 254:3 vs. geśa “to do things early in the morning” in AG I, III

The connection between g-yəs in RIÉ 254:3 and the CG gesa “to do things early in the morning” has been suggested by Littmann.Footnote ¹²⁷ The linguistic difficulties of this identification have been discussed by Drewes,Footnote ¹²⁸ who prefers to interpret g-yəs as a place name. Thus, the fact that the CG root is consistently spelled in AG I, III as gyś Footnote ¹²⁹ is just another argument against Littmann’s identification.Footnote ¹³⁰

V.1.4. ʔḥ[ṣ]ṣ “to diminish” (?) in RIÉ 248:3 vs. ḫṣṣ “to be little” in AG I, III

In RIÉ 248:3, the word ʔḥ.ṣ is tentatively reconstructed by Drewes as ʔḥ[ṣ]ṣ,Footnote ¹³¹ which, however, does not yield a coherent interpretation of the whole phrase. The orthography of AG I, III, where the root “to be little” is consistently spelled with the initial ḫ,Footnote ¹³² can thus be used as an additional argument against this highly hypothetical interpretation.

V.1.5. ṣaniḥā in RIÉ 345 vs. ṣ́nḥ “to wait” in AG I, III

The fragmentary inscription RIÉ 345 on a broken piece of pottery bears the letters zaṣaniḥā[… . It is difficult to refrain from identification of za- with the nota genitivi, and ṣaniḥā with the CG verb ṣanḥa “to wait”. For DrewesFootnote ¹³³ this is a female personal name derived from the same verb. This does not abolish the main problem: the verb “to wait” and its derivatives are consistently spelled with the initial ṣ́ in AG I, III.Footnote ¹³⁴ So far, among the examples collected in Section V.1, this is the most convincing example of orthographic differences between EG, on one hand, and AG I, III, on the other. Still, given the fact that the inscription is fragmentary and cannot be dated with certaintyFootnote ¹³⁵ it seems premature to draw any conclusions from this lone case.

V.2. AEG corroborated by AG I, III

Table 4 includes a lot of widely spread roots whose spelling is extremely stable even in the very late texts of the post-Aksumite period. What is crucial for the present discussion is the cumulative evidence of the comparison between AG I, III and AEG. In addition, a few remarkable examples of agreement in spelling roots with unstable orthography can be detected, such as gafʕa Footnote ¹³⁶ “to oppress” or naśata Footnote ¹³⁷ “to demolish”. Below, three roots whose orthography is of particular interest are discussed.

V.2.1. AEG malʔa, AG I, III mal(l)əʔa, ʔamal(l)əʔa “to set out, go forward”

It was Littmann, followed by Schneider and Drewes,Footnote ¹³⁸ who proposed identifying AEG malʔa with CG malləʕa, ʔamalləʕa “to set out”.Footnote ¹³⁹ On semantic grounds, this identification is irreproachable. It also finds some support in the etymology: compare Tna. tämaləʔä “to bring, to take along, take with o.s.; to pick up; to take s.th., to take s.o. a message”, ʔammaləʔä “to cause to bring, to take with o.s.”.Footnote ¹⁴⁰

The presence of mal(l)əʔa and ʔamal(l)əʔa “to set out, go forward” in AG I, III (where the root is written with final ʔ)Footnote ¹⁴¹ removes the only objection against Littmann’s interpretation, namely, the discrepancy between the orthography of AEG and that of CG.

V.2.2. sgrt (name of a military unit) and sagarāt “soldiers” in AG I, III

As in the preceding case, DrewesFootnote ¹⁴² considered the identification of the name of a military unit, sgrt Footnote ¹⁴³ with CG śagarāt “soldiers of the guard”,Footnote ¹⁴⁴ despite the orthographic discrepancy. Now, the spelling sagarāt in AG I, IIIFootnote ¹⁴⁵ supports Drewes’s hypothesis.Footnote ¹⁴⁶

This comparison is all the more plausible in view of another similar identification: ḥarā, likewise the name of a military unit, may well be connected to the common noun ḥarā “soldiers”.Footnote ¹⁴⁷

V.2.3. AEG *daḥara “to send forth”, AG I, III daḥara “to divorce”

The verb *daḥara is used in RIÉ 187:15.21.27Footnote ¹⁴⁸ as a transitive verb whose object is a noun/pronoun referring to a military unit. This usage finds no direct parallels in CG, but LittmannFootnote ¹⁴⁹ suggested a semantic link between the AEG verb, which he rendered as “entlassen” [to let go], and CG daḥara “to divorce, repudiate (a wife), send away (a wife)”.Footnote ¹⁵⁰ Littmann also considered the possibility of connection to CG daḫara “to bless”,Footnote ¹⁵¹ although the orthographic discrepancy forced him to reject this identification. The orthography of AG III, where daḥara “to bless” is once attested,Footnote ¹⁵² supports Littmann’s conjecture.

The drawback of the interpretation of *daḥara as “to let go”Footnote ¹⁵³ is that it is in disagreement with the context of RIÉ 187:15: the subsequent text implies that the troops were not dismissed but rather received an order to attack the Agʷezāt people (a similar consideration applies to RIÉ 187:21 and 27).

DrewesFootnote ¹⁵⁴ rejects the interpretation “to let go” and offers a purely contextual translation, “to rally” (intransitive in RIÉ 185 Ibis and transitive in RIÉ 187). Similarly, the translation “seguire” [to follow] offered by MarrassiniFootnote ¹⁵⁵ is apparently prompted by the context.

BausiFootnote ¹⁵⁶ drew attention to one more use of CG daḥara, namely, “to permit, to allow”,Footnote ¹⁵⁷ which he directly connected to the AEG *daḥara, as well as to CG daḥara “to divorce”. He generally agrees with Littmann’s interpretation of AEG *daḥara as “to let go”.

BulakhFootnote ¹⁵⁸ treats the AEG verb as a cognate to Arb. daḥara “to drive away, to expel”Footnote ¹⁵⁹ and various modern ES verbs traceable to Proto-ES *dḥr and displaying the semantics of “to bless” or, more specifically, “to give one’s blessing at sending one’s daughter to the husband’s house”. AEG *daḥara is translated as “to send forth”, and this is thought to be close to the original meaning of *dḥr. The translation “to send forth” has the advantage of agreeing with the semantics of the root in CG and of its cognates in Arabic and modern ES, as well as of fitting the contexts of RIÉ 187.

To summarize, the AEG *daḥara can be interpreted as “to send forth (troops)”, as an extension of the general semantics of “to send, to let go”.Footnote ¹⁶⁰ The same root is attested in CG with the meanings “to divorce, repudiate (a wife)”, “to bless”, and “to permit, to allow” (all of them traceable to the meaning “to let go”). Distinction between daḥara “to divorce” and daḫara “to bless; to allow” posited in Leslau’s dictionary is not confirmed by the ʔAbbā Garimā Gospels, where both verbs are written with ḥ.

VI. Conclusions

Comparison of the orthography of Aksumite Epigraphic Geez with that of AG I, III has not revealed any reliable instances of divergent spellings of sibilants or gutturals. Table 4 in Section V contains 224 roots of Aksumite EG whose spelling is in agreement with that of AG I, III. This evidence can hardly be undermined by the five potential cases of discrepancy discussed in Section V.1, none of which is fully convincing (in fact, only the example discussed in Section V.1.5 cannot be fully discarded).

These results may be used as an additional argument in favour of dating of AG I, III to the Aksumite period. In turn, if this dating is accepted, the orthography of AG I, III can be used in future studies to verify the hypothetical interpretations of numerous obscure passages in Aksumite inscriptions.Footnote ¹⁶¹

Funding information

This article is the fruit of the research project “Semitic lexicography in synchrony and diachrony” implemented as part of the Basic Research Program at the National Research University Higher School of Economics (HSE University), Laboratory for Middle Eastern and North African Languages and Cultures.

Acknowledgements

I am deeply indebted to Prof. Steve Delamarter, who kindly provided me with digital copies of the ʔAbbā Garimā Gospels and thus enabled me to initiate my investigation. I am likewise grateful to the Hill Museum and Monastic Manuscript Library for making public coloured photos of the ʔAbbā Garimā Gospels in the open access online depository (https://www.vhmml.org/readingRoom). Warm thanks go to Prof. Alessandro Bausi, who read the draft version of the paper, corrected errors and made a number of important suggestions. I would like to thank A.A. Tkachenko, who provided me with some otherwise inaccessible literature indispensable for this work. Last but not least, I am deeply grateful to Kevin McNeer for careful English proofreading.

Abbreviations of languages

AEG: Aksumite Epigraphic Geez
Amh.: Amharic
Arb.: Arabic
CG: Classical Geez
EG: Epigraphic Geez
ES: Ethio-Semitic
ESA: Epigraphic South Arabian
Gr.: Greek
Min.: Minaic
Qat.: Qatabanic
Sab.: Sabaic
Tna.: Tigrinya

Footnotes

¹ Bausi Reference Bausi2005.

² In Bausi Reference Bausi2005: 168, the possibility of an earlier dating for the ʔAbbā Garimā Gospels is only cautiously mentioned.

³ See Weninger Reference Weninger and Weninger2011: 1124, Bulakh Reference Bulakh2009: 394.

⁴ For general information on ʔAbbā Garimā Gospels (AG I, AG II, and AG III) see McKenzie and Watson Reference McKenzie and Watson2016. AG II is invariably dated to the post-Aksumite period and hence is excluded from the present discussion.

⁵ See Zuurmond and Niccum Reference Zuurmond, Niccum, Ehrman and Holmes2013: 238, with n. 29, McKenzie Reference McKenzie, McKenzie and Watson2016a: 1–3, McKenzie Reference McKenzie, McKenzie and Watson2016b: 40–41, Bausi Reference Bausi and Bausi2015: 47, Bausi Reference Bausi and Crevatin2018a: 75, Piovanelli Reference Piovanelli2018: 9–10. The possibility of the Aksumite dating is more cautiously mentioned in Butts Reference Butts, Huehnergard and Pat-El2019: 118, Nosnitsin Reference Nosnitsin and Kelly2020: 286; see also a special article by S. Kim (Reference Kim2022). The results of the radiocarbon analysis have been published by J. Mercier (Reference Mercier2021: 29).

⁶ DAE IV 80 (on references to DAE and RIÉ see n. 10); Kogan Reference Kogan and Weninger2011: 103–4, 105, 117. As the earliest example of confusion between s and ś, the spelling ʔs¹rḥ “(that) I prosper” in the sixth century RIÉ 191:6 (written in South Arabian script, in which s² is expected to render Geez ś) is usually adduced (see Drewes Reference Drewes and Kaye1991: 385–6, Bausi Reference Bausi2005: 152–3, with n. 11; see also RIÉ IIIB 251; the variation in the spelling of the numeral for “3” in RIÉ 190:34, 40 is explained by regressive assimilation, see RIÉ IIIB 247–8).

Strictly speaking, the spelling ʔsrḥ cannot count as an instance of a synchronic confusion between s and ś: the root is not attested at all elsewhere in Epigraphic Geez. Rather, it is an example of non-etymological orthography, which is a much less reliable piece of evidence (even in spite of the convincing etymology, the root s²rḥ being well attested in ESA with the semantics very close to that of Geez: s²rḥ “to deliver, preserve”, SD 134; Min. s¹s²rḥ id., LM 87; Qat. s²rḥm “safety, prosperity”, LIQ 171). Non-etymological spellings cannot be uncritically used to date the respective phonological merger. However, coupled with the scant evidence from the Arabic sources for the merger between Geez s and ś in the seventh century (see Ullendorff Reference Ullendorff1955: 113–4, Podolsky Reference Podolsky1991: 22, Kogan Reference Kogan2025: 630), the spelling ʔsrḥ may indeed be seen as the indication that the process of delateralization of ś begun in the sixth century (even so, it could not be widespread: the 40-line inscription RIÉ 191 contains no other examples of misspelt ś; likewise, the lateral ś is consistently preserved in RIÉ 192, a sixth-century inscription post-dating RIÉ 191.

⁷ Gragg Reference Gragg and Kaye1997: 173. For their realization in the Aksumite period, see Weninger Reference Weninger2010.

⁸ Meyer Reference Meyer, Mendisu and Johannessen2016: 153.

⁹ The preliminary results are published in Bulakh Reference Bulakh2024. Since then, the analysis of AG I, III has been completed by the present author. Not a single case of vacillation of the “problematic” graphemes has been detected.

¹⁰ Recueil des inscriptions de l’Éthiopie des périodes pré-axoumite et axoumite (RIÉ) is the main source on Epigraphic Geez and, as such, it was consistently used as a source of material for the present paper. Data from Geez inscriptions that became publicly known after the appearance of the first volume of RIÉ were also taken into consideration (for an overview see Bulakh Reference Bulakh2021, to which, however, the inscriptions from cave Hoq have to be added, see Strauch Reference Strauch2012a), but the majority of lexical items in them is either shared by RIÉ or of obscure meaning. References to the inscriptions of RIÉ are given without volume numbers (e.g. RIÉ 232 is the reference to the inscription of Ham); references to the pages of the volumes of RIÉ are given after the number of the volume (e.g. RIÉ I 324 is the reference to the comments on the inscription of Ham). The same principle is employed with the corpus of Deutsche Aksum-Expedition (DAE).

¹¹ E.g. Weninger Reference Weninger1999: 3–4, Weninger Reference Weninger and Uhlig2005: 732, Weninger Reference Weninger and Weninger2011: 1124–5, Bausi Reference Bausi2005: 167–8, Tropper Reference Tropper2002: 1, Tropper and Hasselbach-Andee Reference Tropper and Hasselbach-Andee2021: 1–2, Meyer Reference Meyer, Mendisu and Johannessen2016: 152–3.

¹² Yet literary texts possibly created within this time span are not entirely absent. In particular, non-translated homilies contained in EMML 8509 can be dated to the thirteenth century or earlier (as with the manuscript itself, see Villa and Zarzeczny Reference Villa, Rafał, Sernicola and Villa2026: 293), but there is no special reason to date them to the Aksumite period.

¹³ Tropper Reference Tropper2002: 1–2, Tropper and Hasselbach-Andee Reference Tropper and Hasselbach-Andee2021: 2–3. The same period is referred to in Weninger Reference Weninger and Weninger2011: 1125 as “post-Classical” = “post-13^th century”. In Bausi Reference Bausi2005: 167, the post-Aksumite period is said to continue “since the 13^th century at the latest”.

¹⁴ Gigar Reference Tesfaye1984; Bosc-Tiessé Reference Bosc-Tiessé2010. Unlike Geez sources found in the north, for the inscriptions on the tābots of Lālibalā, both geographical and chronological context suggests that they were not created by native speakers of Geez.

¹⁵ Bausi Reference Bausi2019b: 71 with further references.

¹⁶ Call number UM-object 1 according to the online catalogue of Ethio-SPaRe project (headed by Dr. D. Nosnitsin from Universität Hamburg for the period from November 2009 to May 2015), https://mycms-vs03.rrz.uni-hamburg.de/domlib/receive/domlib_artitem_00000109, last accessed 11.02.2025. See Derat Reference Derat2018: 38 (with the edited text).

¹⁷ See Derat Reference Derat2018: 38–9, with the edited text of the inscription.

¹⁸ On the pre-Solomonic dating of the manuscript, see Derat Reference Derat2010: 22, Reference Derat2018: 59–60.

¹⁹ For the edited text and translation of the end of the donation, see Derat Reference Derat2010: 21–4, Reference Derat2018: 59–61. For the edition and translation of the mutilated colophon, see Bausi Reference Bausi2018c: 440, Bausi Reference Bausi2022: 134.

²⁰ Zuurmond Reference Zuurmond2001: 16–7; Zuurmond and Niccum Reference Zuurmond, Niccum, Ehrman and Holmes2013: 238, 241.

²¹ EMML 1832, EMML 6907, see Zuurmond and Niccum Reference Zuurmond, Niccum, Ehrman and Holmes2013: 238.

²² For a detailed description, see Bausi et al. Reference Bausi, Brita, Di Bella, Nosnitsin, Sarris and Rabin2020a; for a detailed analysis of its linguistic features, see Bausi Reference Bausi2023.

²³ Butts Reference Butts2020. Observations on the orthography of this and a few other ancient post-Aksumite Gospel manuscripts can be found in Zuurmond Reference Zuurmond1989:II 50–8.

²⁴ Villa and Zarzeczny Reference Villa, Rafał, Sernicola and Villa2026.

²⁵ A short overview of pre-fourteenth-century Ethiopic manuscripts and fragments can be found in Nosnitsin Reference Nosnitsin and Kelly2020: 286–8. New evidence is constantly coming to light (see e.g. Nosnitsin Reference Nosnitsin, Heeßel, Tsukanova and Waltisberg2024, a detailed study of a newly discovered pre-fourteenth-century Geez fragment).

²⁶ The dating of the texts is based on the general assumption that the bulk of biblical and apocryphal literature was translated into Geez in the fourth–sixth centuries as well as on dates of circulation of the Greek originals. See, e.g. Erho and Henry Reference Erho and Henry2019: 179, on the dating of the Geez translation of an early apocryphon “Jannes and Jambres” (fragmentarily preserved in a pre-fourteenth-century bifolium Schneider ms. frag. 19, deposited at the Walda Masqal Centre of the Institute of Ethiopian Studies, Addis Ababa).

²⁷ el-Antony, Blid and Butts Reference el-Antony, Maximous and Butts2016.

²⁸ el-Antony, Blid and Butts Reference el-Antony, Maximous and Butts2016: 32–3.

²⁹ el-Antony, Blid and Butts Reference el-Antony, Maximous and Butts2016: 46.

³⁰ Another fragmentary piece of evidence on early post-Aksumite Geez may be preserved in fols. 122–3 of ms. EMML 7229 (see Getatchew Haile Reference Haile2021). While the ms. itself is dated to the seventeenth century, the palaeography of these two folios, clearly belonging to a considerably older manuscript, suggests their dating to the pre-Solomonic (Zagʷe or even late Aksumite) period. The text of the folios, which is a fragment of a commentary on the Gospel of Matthew, was published by Getatchew Haile. The editor suggested that the text itself is a translation of a commentary by one of the early church scholars (ibid. 223), but no Vorlage has yet been found. There are no clues as to whether the two folios and/or their text belong to the Aksumite or post-Aksumite period.

Still another intriguing piece of evidence is the palimpsest Greek NF M 90, preserved in St Catherine’s Monastery of Mount Sinai. It contains three undertexts in Geez, which probably belong to the early post-Aksumite period (the Ethiopic writing is dated between 600 and 1300 in Delamarter and Getatchew Haile, https://sinai.library.ucla.edu; cf. also Rapp Reference Rapp, Rapp, Rossetto, Grusková and Kessel2023: 43; according to Erho (Reference Erho, Gippert, Maksimczuk and Sargsyan2025: 406), it does not significantly antedate the Greek scriptio superior, which is dated to the thirteenth–fourteenth century ad, but is certainly older than 1270 ad).

³¹ See Derat Reference Derat2018: 42–3, 49–53, 56–8, 60–1.

³² In the same way, some features of Aksumite Geez can be detected in relatively recent manuscripts preserving Aksumite texts, see Villa Reference Villa2019: 190–221.

³³ The death of vernacular Geez cannot of course be dated with any precision. It is a common opinion (see e.g. Ullendorff Reference Ullendorff1955: 13; Tropper Reference Tropper2002: 1; Tropper and Hasselbach-Andee Reference Tropper and Hasselbach-Andee2021: 2–3; Meyer Reference Meyer, Mendisu and Johannessen2016: 137, 159; Bausi Reference Bausi, Bausi and Sokolinski2016: 52; also, with some reservations, Butts Reference Butts, Huehnergard and Pat-El2019: 119) that Geez was not a spoken language in the Solomonic period. Coupled with the assumption that Geez was in use as a vernacular until the fall of the Aksumite empire, this yields the early post-Aksumite period as the time span in which Geez ceased to be a living language. A more positive statement is that starting from the early post-Aksumite period, Geez texts reliably written by scribes for whom Geez was not a mother tongue steadily increase in number (see n. 14).

³⁴ Butts Reference Butts, Huehnergard and Pat-El2019: 118.

³⁵ Butts Reference Butts, Huehnergard and Pat-El2019: 118.

³⁶ RIÉ IIIB 6–7, Fiaccadori Reference Fiaccadori and Uhlig2007; see further Bosc-Tiessé et al. Reference Bosc-Tiessé, Derat, Fritsch and Abullif2010: 44–5, Nosnitsin Reference Nosnitsin2011: 36–8.

³⁷ In Weninger Reference Weninger1999: 4, the end of the Aksumite Empire is dated to the tenth century ad; in Weninger Reference Weninger and Uhlig2005: 736 (see also el-Antony, Blid and Butts Reference el-Antony, Maximous and Butts2016: 46), the Aksumite age is given as 330–900 ad. In Tropper and Hasselbach-Andee Reference Tropper and Hasselbach-Andee2021: 2 (see also Tropper Reference Tropper2002: 1), the Aksumite period is defined as “the height of the Aksumite kingdom in the time between the fourth and seventh centuries”, whereas the decline of the Aksumite empire is dated at around 960.

³⁸ Phillipson Reference Phillipson2012: 47, 209–10, Bausi Reference Bausi, Uhlig, Appleyard, Bausi, Hahn and Kaplan2017: 99, Lusini Reference Lusini, Gehler, Rollinger and Strobl2022. According to Munro-Hay (Reference Munro-Hay1991: 95), the post-Aksumite period began in the (second half of the) seventh century ad. Even more radically, in Binyam and Krebs Reference Binyam and Krebs2024, the end of the Aksumite period is dated at the sixth century ad., and the beginning of the post-Aksumite, respectively, at the seventh century ad.

While the earliest datable Geez inscription (RIÉ 180) is from the third century ad (Sima Reference Sima and Uhlig2005), earlier dating cannot be excluded for some other inscriptions, hence the dating “at the earliest in the 1^st/2^nd century A.D.” for the oldest attestations of Geez (Avanzini Reference Avanzini and Uhlig2007: 160).

³⁹ For the editors of RIÉ, the Aksumite period lasted to the mid-twelfth century, see RIÉ IIIB 5.

⁴⁰ The term “post-Aksumite period” also refers to different spans of time in different works. Thus, it can denote a period from the eighth–eleventh centuries ad (Bausi Reference Bausi, Uhlig, Appleyard, Bausi, Hahn and Kaplan2017: 99). Still, in works on Geez and Geez literature, post-Aksumite is mostly used to cover the whole period which followed the decline of the Aksumite state (see e.g. Tropper and Hasselbach-Andee Reference Tropper and Hasselbach-Andee2021: 2–3).

⁴¹ The date 1270 ad as the beginning of the Solomonic period is a mere convention and it would hardly be justifiable to set it as marking a borderline in the periodization of Geez language.

⁴² See n. 33.

⁴³ Classical Geez (= standard Ancient Ethiopic in Bausi Reference Bausi2005: 15 = standard Geez in Villa and Zarzeczny Reference Villa, Rafał, Sernicola and Villa2026) is documented in the traditional grammars – sawāsəw – as well as in the grammatical descriptions produced by Western scholars.

⁴⁴ The most significant contributions to Aksumite numismatics are Munro-Hay Reference Munro-Hay1984; Munro-Hay and Juel-Jensen Reference Munro-Hay and Juel-Jensen1995; Hahn and West Reference Hahn and West2016; Hahn Reference Hahn2021 (see further Bausi Reference Bausi2003). The linguistic evidence of the Aksumite coinage is only occasionally taken into account in the studies on Geez grammar (see Weninger Reference Weninger2010: 80–81).

⁴⁵ That is, manuscripts containing the Versio Antiqua of the Bible, as well as other Aksumite texts (see, e.g., Piovanelli Reference Piovanelli1993; Lusini Reference Lusini2009: 9–10 with further references; Villa Reference Villa2019: 190–221). The Aksumite Collection (see Bausi et al. Reference Bausi, Brita, Di Bella, Nosnitsin, Sarris and Rabin2020a) belongs here, too.

⁴⁶ See Section III.

⁴⁷ See n. 14, 15.

⁴⁸ See n. 16.

⁴⁹ See n. 17.

⁵⁰ See n. 18, 19, 21, 27, 30.

⁵¹ See n. 31.

⁵² Fiaccadori Reference Fiaccadori and Uhlig2007.

⁵³ Erroneously assuming that all inscriptions included in RIÉ belonged to the Aksumite period, following the statement in RIÉ I 35 (“avec la fin d’Axoum se perd aussi l’usage épigraphique”). This statement is reassessed in RIÉ IIIB 6–7, where a useful list of some post-Aksumite inscriptions, compiled by R. Schneider, is adduced. In the framework of the periodization employed in the present paper, some of the inscriptions included in RIÉ are early post-Aksumite (see Section III).

⁵⁴ Hatke Reference Hatke2013: 164–5; Derat Reference Derat2018: 111; Lusini Reference Lusini, Gehler, Rollinger and Strobl2022: 333.

⁵⁵ But is in contradiction with their palaeography; see discussion in RIÉ IIIB 271.

⁵⁶ According to RIÉ 193 II:11, only the symbol -m- can be clearly discerned in this line.

⁵⁷ According to RIÉ 193 II:12, the line is fully illegible.

⁵⁸ See Derat Reference Derat2018: 109–10.

⁵⁹ Lack of vocalization in RIÉ 194 is not a persuasive argument in favour of an early dating. Unvocalized texts could exist side-by-side with the fully vocalized ones, as unvocalized liturgical rubrics in AG II, III testify (see Kim Reference Kim2022, 3.4.2, 3.6.2).

⁶⁰ Curtis and Bausi Reference Bausi2005.

⁶¹ Kropp Reference Kropp1999.

⁶² Bausi Reference Bausi and Quenzer2021: 15; for an overview of previous research see Bausi Reference Bausi, Bausi, Camplani and Emmel2019a: 85–9. Still another deciphering, with a date corresponding to 1090 ad, was proposed by A. Drewes (RIÉ IIIB 329), but has been convincingly shown to be untenable (Bausi Reference Bausi and Quenzer2021: 13–4).

⁶³ RIÉ 232:1.

⁶⁴ Leslau Reference Leslau1941: 27 (in particular, the forms with palatalization quoted by Leslau are obvious Amharisms).

⁶⁵ For its presence in South Ethio-Semitic, see Leslau Reference Leslau1995: 413 (Amharic); Wetter Reference Wetter2010: 235–6 (North Argobba); Leslau Reference Leslau1956: 110 (Gafat).

⁶⁶ Yet, such a situation is not entirely impossible. Trade connection with the southern areas must have existed in the Aksumite era, and the presence (and, obviously, the high status) of a daughter of a southerner in the village which was lying at the crossroads of the Aksumite trade routes may not be excluded, after all.

⁶⁷ Unless we are faced with an attempt to reflect the Amharic š (see discussion above in this section).

⁶⁸ In the brackets, here and elsewhere in this list, the references to RIÉ IIIB or other sources are given, where the spelling of the same root in the Aksumite inscriptions is adduced.

⁶⁹ See n. 81.

⁷⁰ The transcription bāḥtu offered in RIÉ 232:5, with the long ā in a closed syllable, is difficult to accept. The etymology suggests a vowel after ḥ and the gemination of t (see Bulakh Reference Bulakh2023a: 273–4).

⁷¹ See RIÉ IIIB 553, 595, 603, 604, 627, 628, 667, 673.

⁷² For śarq, see CDG 534, HALOT 1361–2. For ḫtw, see CDG 268. For ḫrf, see CDG 264, Kogan Reference Kogan and Weninger2011: 197. For ʕāmat, see Kogan Reference Kogan2015: 197, Bulakh Reference Bulakh2023a: 254. For blʕ, see Kogan Reference Kogan2015: 445 and Bulakh Reference Bulakh2023a: 14. For ṭʕm, see Kogan Reference Kogan and Weninger2011: 238 and Bulakh Reference Bulakh2023a: 257.

⁷³ See CDG 88. Even more importantly, the root is consistently written as bṣ́ʕ in AG I and III (bəṣ́uʕ in AG III f. 18ra l. 5 etc.).

⁷⁴ Contra Kropp (Reference Kropp1999: 168), hardly to be connected to ṣ́rr “to be hostile”. See the criticism in Bausi Reference Bausi and Quenzer2021: 21, n. 40.

⁷⁵ See Bausi Reference Bausi and Quenzer2021: 20–4.

⁷⁶ See RIÉ IIIB 622.

⁷⁷ See CDG 259.

⁷⁸ ḥədāṭ in AG I f. 111rb ll. 11–2 etc.

⁷⁹ See Diem Reference Diem1988: 249.

⁸⁰ In the terminology of Diem (Reference Diem1988: 240).

⁸¹ The transcription śahəl in RIÉ 232:5 is to be rejected. We are clearly encountering the pattern C₁aC₂C₃.

⁸² tamakaḥa (RIÉ 189:8) and ʔardaʔani (RIÉ 189:45). See Bulakh Reference Bulakh, Bulakh, Kogan and Romanova2013: 207.

⁸³ Contra RIÉ 232:13 (where the word is read as məwtāān): the second order of w is clearly visible on the photo (RIÉ II, Pl. 165), and the shape of the letter is exactly the same as in wayaḥayəwu in line 14. Cf. also Bulakh Reference Bulakh2023b: e16, n. 27.

⁸⁴ məwutān (AG I f. 16ra l. 19 etc.) vs. məwətān (AG III f. 24ra l. 14 etc.).

⁸⁵ See Nosnitsin Reference Nosnitsin, Heeßel, Tsukanova and Waltisberg2024: 303. Some relevant examples from the fragment treated by Nosnitsin are wə-təsəm(m)əyo “and she will call him” (instead of the expected wa-təsam(m)əyo) in fol. 1r, l. 1, ward “it descended”(instead of warada) in fol. 1v, l. 10, wa-ʔəʔmərna “and we learned” (instead of wa-ʔaʔmarna) in fol. 1v., ll. 15–6, śəg “flesh” (instead of śəgā) in fol. 1v, l. 19 etc. Likewise, the appearance of ə instead of a and some other cases of non-standard orthography can be found in EMML 8509, discussed by M. Villa and R. Zarzeczny (Reference Villa, Rafał, Sernicola and Villa2026: 298, 299–300).

⁸⁶ Besides, the inscription RIÉ 189 provides several instances of the shift a > ə in the position between two guttural consonants in a closed syllable: ʔəḥmārihomu “their boats”, RIÉ 189:22, instead of ʔaḥmārihomu etc. (see Diem Reference Diem1988: 258, Bulakh Reference Bulakh, Bulakh, Kogan and Romanova2013: 208, Bulakh Reference Bulakh and Sokolinski2016: 114–5 with further references). This phenomenon, however, can hardly be directly linked with non-standard vocalization in the Ham inscription.

⁸⁷ See Bausi Reference Bausi and Quenzer2021: 20–4.

⁸⁸ Bausi Reference Bausi and Quenzer2021: 15-20. An important argument in favour of the post-Aksumite dating is the shape of the inscription, undoubtedly derived from a Nubian model (see Kropp Reference Kropp1999: 166; Bausi Reference Bausi and Quenzer2021: 18, n. 33).

⁸⁹ Admittedly, the orthography can be archaizing rather than reflecting the contemporary state of the language. Still, in case of a funeral inscription, which – unlike a manuscript – could not have been copied word-for-word from an ancient original, archaic orthography is rather unexpected.

⁹⁰ As argued in RIÉ IIIB 301–3, the same inscription was published by Fiaccadori (Reference Fiaccadori1981).

⁹¹ See RIÉ IIIB 302–3; Hatke Reference Hatke2022: 174–5.

⁹² Hatke Reference Hatke2022: 174–5.

⁹³ See discussion in RIÉ IIIB 302.

⁹⁴ As is done by various scholars of Geez epigraphy (Robin Reference Robin and Strauch2012: 440, Hatke Reference Hatke2022: 174–5).

⁹⁵ See Section I, with n. 6.

⁹⁶ It would be difficult to explain this consistency in terms of archaizing orthography (on which see Section III.3) because RIÉ 191 and 192 are written in South Arabian script. No established orthographic tradition can be posited for the employment of South Arabian script for Geez texts (see Sima Reference Sima2003/2004: 278; see also Robin Reference Robin and Robert2022a: 191).

As for the sixth-century Geez inscriptions from South Arabia (RIÉ 195 I, II), archaizing orthography cannot be ruled out.

⁹⁷ See Sections III.1, III.2.

⁹⁸ Ch.J. Robin’s edition in Strauch, Foreign Sailors on Socotra (Reference Strauch2012a: 54–5, 56–7, 64).

⁹⁹ First, the mere fact of its presence in this important hub of the Indian Ocean trade makes any date later than the seventh century ad unlikely – otherwise one has to claim that in the eighth century ad or later, Christian Ethiopians continued to play an active part in the long-distance trade from the Red Sea at least as far as Soqotra (the presence of Muslim Ethiopians on Soqotra is of course easier to imagine in the post-Aksumite period; however, there is no reliable evidence that Ethiopian Muslims in the early medieval period employed the Ethiopic syllabary for writing down their speech – rather, they would have left inscriptions in Arabic). Second, other inscriptions in the cave of Hoq are dated to the first–early fifth centuries ad (Strauch Reference Strauch and Strauch2012b: 540–1). Third, one of the Geez inscriptions, Hoq 2:30, contains a proper name (?) ˁygns, which links this inscription to HS1, a pre-Christian inscription on a stele in Ḥənzat, a settlement in Tigray (see Bulakh and Gebre Selassie Reference Bulakh and Selassie2022: 154–5).

¹⁰⁰ See photo in Strauch Reference Strauch and Strauch2012b: 64.

¹⁰¹ RIÉ 192 is clearly linguistically divergent from the rest of Epigraphic Geez (see Bulakh Reference Bulakh, Bulakh, Kogan and Romanova2013: 214).

¹⁰² The main source is RIÉ (see the glossary in RIÉ IIIB 582–687); other sources (cf. n. 10) were also consulted.

¹⁰³ Proper names and words with obscure semantics are not included; specific cases of the use of h instead of ʔ in pronouns and verbal prefixes in RIÉ 192 (see Bulakh Reference Bulakh, Bulakh, Kogan and Romanova2013: 214, Robin Reference Robin and Robert2022a: 192, n. 81) are likewise ignored as irrelevant to the present discussion.

¹⁰⁴ An asterisk marks the forms reconstructed from those actually employed in the inscriptions; forms without an asterisk are direct quotations. The transcriptions of the reconstructed forms are in some cases modified in accordance with the general principles of Geez transcription outlined in Bulakh Reference Bulakh and Sokolinski2016. The translations are given, for the reader’s convenience, in English (after CDG). EG refers here to the Aksumite inscriptions in Geez as well as to RIÉ 232, on whose status cf. Section III.3 (those roots which are attested in RIÉ 232 but not found elsewhere in EG are specially marked as such).

¹⁰⁵ In Table 4, as elsewhere in the paper, the “new” foliation of AG I, III is used (see Kim Reference Kim2022, with Tables 2, 4 on the correspondence between “old” and “new” foliation; unfortunately, the “new” foliation offered by S. Kim does not take into account one folio between 141 and 142 [John 2:15–3:12]; in the present publication, this folio is referred to as 141A, as in Table 3). A dash means the absence of the root in AG I, III.

¹⁰⁶ However obscure the meaning of the verb tbʕl is (see Nebes Reference Nebes2017: 364–5, Bausi Reference Bausi2018b: 292–3), its relationship with the common Semitic root bʕl is not in doubt.

¹⁰⁷ Perhaps related to Tna. bäḥaq̆ä “to be leavened (dough), become fermented” (Kane Reference Kane2000: 1105), phonetically rather close to Geez and possibly a result of a hypercorrection ʔ > q. Furthermore, as rightly observed by W. Leslau (Reference Leslau1979: 135), it is difficult to refrain from the comparison to the modern ES root bkʕ with the same meaning: Tna. bäḵʕä “to rise (dough), to ferment (batter)” (Kane Reference Kane2000: 1175), Amh. bokka “to ferment (batter)” (Kane Reference Kane1990: 923); on the cognates in Gurage cf. Leslau Reference Leslau1979: 135, Bulakh Reference Bulakh2005: 328. The phonetic dissimilarities suggest that the two roots – bḥʔ in Geez and bkʕ in modern ES – are independent borrowings from the same source (on possible Cushitic sources, see Bulakh Reference Bulakh2005: 328).

¹⁰⁸ On bṣʕ and bṣ́ʕ in the Aksumite inscriptions on Soqotra, see Section IV.

¹⁰⁹ There are no reasons to posit a connection to dḥn in RIÉ 190:13, whose exact parsing and meaning remain unclear (see RIÉ IIIB 608).

¹¹⁰ But note that the root gʕz “to be free”, considered by Dillmann (LLA 1187; see also Voigt Reference Voigt2016: 69) to be related to gʕz “to travel”, is well represented in AG I, III (e.g. gəʕuzān “free (masc. pl.)”, AG I f. 33ra l. 17).

¹¹¹ If one accepts the emendation ḥə<l>qat proposed in RIÉ IIIB 623.

¹¹² Unless related to qoros “a measure”, cf. AG I f. 123vb ll. 1–2, AG III f. 119rb l. 5.

¹¹³ The verb is attested in RIÉ 192:(A)31 (written in ESA script) as lys³ddm “let him banish them”. The ESA grapheme s³ is used, along s¹, in the “pseudo-Sabaic” inscriptions to render Geez s (see RIÉ IIIB 247–8).

¹¹⁴ If we are faced with the same root in EG. In fact, EG sḥt is attested only once (RIÉ 192:16), in a context which rather favours an alternative interpretation: the EG verb may be a borrowing from Sabaic s¹ḥt “to defeat” (SD 125). This hypothesis is plausible in view of an unusually high number of reliable Sabaic loans in the same inscription (bnt “women” in l. 44, cf. Sab. bnt, SD 29; mḫmr “favour” in l. 8, cf. Sab. ḫmr “grant a favour”, SD 61; rḥmnn “the Merciful One” in l. 8, obviously going back to the divine name rḥmnn in the Jewish and Christian Himyaritic inscriptions starting from the fifth century ad, see Robin Reference Robin and Robert2022b: 144, Gorea Reference Gorea2023: 92 with n. 3; yglb “he overcomes (him)” in ll. 7–8, cf. Sab. γlb “victorious”, SD 53). These words are evaluated as Sabaic borrowings by Drewes (RIÉ IIIB 260–1, 571, 604, 618, 627; in case of yglb, as an Arabic root borrowed via Sabaic, see RIÉ IIIB 260–1; cf. also Bulakh Reference Bulakh, Bulakh, Kogan and Romanova2013: 213). The same inscription is also characterized by the presence of the negative marker dʔ, likewise borrowed from late Sabaic (see Schneider Reference Schneider1974: 780, RIÉ IIIB 607, Frantsouzoff Reference Frantsouzoff, Darles, Khalidi and Arbach2021: 238, n. 14, with further references).

¹¹⁵ Unless related to salaṭa “to fulfil an obligation”, cf. sələṭ (imv. masc. sg.), AG I f. 118va l. 16 etc.

¹¹⁶ In Harrower et al. Reference Harrower, Dumitru, Perlingieri, Nathan, Zerue, Lamont, Bausi, Swerida, Bongers, Woldekiros, Poolman, Pohl, Brandt and Peterson2019: 1544, the transcription is SaHaLəNa due to a typographical error, but ś is clearly seen on the photo (ibid., Figure 6). The correct transcription is given in Bausi et al. Reference Bausi, Harrower and Dumitru2020b: 38.

¹¹⁷ The vocalization ṣəllul and the translation “aveuglé” offered in RIÉ IIIB 670 (see also discussion in Sima Reference Sima2003/2004: 273, n. 24) is unwarranted. Rather, the plene writing ṣllw in RIÉ 185 Ibis:24 (South Arabian script) should be taken at face value, which leads to the reconstruction of the noun *ṣalalo. In CG, ṣalalo is only attested with the meaning “soot” (CDG 555, LLA 1257), but this is undoubtedly a development of a more general “darkness”.

¹¹⁸ AG I f. 36ra, l. 14 etc.

¹¹⁹ Bausi Reference Bausi2005: 152, n. 11; cf. also RIÉ IIIB 289. An obvious obstacle for this interpretation is the lack of a rationale behind writing a one-word inscription of such content. The alternative reading proposed by Bausi (Reference Bausi2005: 152, n. 11) and Drewes (RIÉ IIIB 289) suggests right-to-left reading bṣ́ʕ and is therefore rather problematic. At the same time, the reading ʕṣ́b in itself is not secure: on the photo adduced in RIÉ II, Pl. 145, the first grapheme rather resembles a ḫ.

¹²⁰ The final -m is a purely graphic phenomenon (see Sima Reference Sima2003/2004: 276–7, Robin Reference Robin and Robert2022a: 192–3).

¹²¹ Marrassini Reference Marrassini2014: 241, with n. 8.

¹²² RIÉ IIIB 596.

¹²³ AG I f. 9va ll. 1–2.

¹²⁴ AG I f. 99va l. 16. Cf. also ʔawdo “around him” (AG I f. 62ra l. 7).

¹²⁵ ʕawd in AG III f. 79ra l. 14–15 etc. Its spelling is in agreement with EG, cf. ʔnss ʕwd *ʔənsəsā ʕawd “beasts of burden” (RIÉ IIIB 596; corresponds to Gr. κτηνῶν νωτοφόρων in RIÉ 270:14, 270bis:11), where ʕawd possibly means “court” (see Nöldeke Reference Nöldeke1913: 697), or “threshing floor”. On the distinction between two semantically close Geez roots, ʕwd and ʔwd, cf. Bulakh Reference Bulakh2024: 85, with further references.

¹²⁶ The same is valid for minuscule Sabaic ʕwd, interpreted in Stein Reference Stein2010: 589 as “herald”.

¹²⁷ DAE IV 69.

¹²⁸ RIÉ IIIB 352; cf. also RIÉ IIIB 620.

¹²⁹ geśa in AG I f. 36va l. 11 etc. This spelling is in agreement with the etymological data from ESA texts on wood: Sab., Min. gys² “Morgen; morgens” (Stein Reference Stein2023: 171–2, 479, 484).

¹³⁰ As an alternative to both Littmann’s and Drewes’s interpretations, one could tentatively identify g-y-s with gāys, gāyso “shield, spear, haft of spear” (CDG 209, LLA 1424), which would well fit the context of RIÉ 254. Contra Leslau (CDG 209), gāyso is clearly a borrowing from Greek, being attested only in Jos. 8:18 and 8:19, where it renders Gr. γαῖσος “javelin” (LSJ 335). The only obstacle is that the Geez term may well be merely a transcription of the Greek wordform γαίσῳ, with its employment limited to the above-mentioned passages of the Book of Joshua.

¹³¹ RIÉ IIIB 348, 624.

¹³² ḫəṣuṣ “diminished” in AG I f. 32vb l. 9 etc.

¹³³ RIÉ IIIB 404–5; see also Drewes Reference Drewes2000: 208.

¹³⁴ yəṣ́an(n)əḥu “they (masc. pl.) wait”, AG I f. 117va l. 8 etc.

¹³⁵ The pottery piece was found “dans le niveau supérieur de la couche archéologique” (RIÉ I 447).

¹³⁶ gafʕa/gafʔa according to CDG 183, LLA 1212.

¹³⁷ naśata/nasata according to CDG 404, LLA 634.

¹³⁸ DAE IV 27, 86; Schneider Reference Schneider1974: 782; RIÉ IIIB 642.

¹³⁹ LLA 153, CDG 342.

¹⁴⁰ TED 332–3; Täkkəʔä Täsfay 399. The alternative forms with ʕ – tämalʕä “to take away, ʔamälʕä “to escape” (TED 338) – are not attested in Täkkəʔä Täsfay’s dictionary (the same is valid for ʔamälləʔä “to go away”, adduced in TED 332–3) and can be ignored as unreliable. Thus, the evidence from Tigrinya agrees with that of AEG and AG I, III, and safely excludes comparison with Arb. malaʕa “run swiftly (she-camel)” (BK 1149) offered in CDG 342.

¹⁴¹ ʔamal(l)əʔa (AG I f. 129rb ll. 8–9), mal(l)əʔu (imv. masc. pl.) (AG I f. 59vb l. 19).

¹⁴² RIÉ IIIB 478–9.

¹⁴³ Attested in RIÉ 192:27–8, 42, 48, see RIÉ IIIB 573, 661. Drewes (RIÉ IIIB 478–80) tends to regard sgrt in RIÉ 192:48 as a common noun but still considers it to be linked to the name of the military unit in RIÉ 192:27–8, 42.

¹⁴⁴ CDG 527, LLA 267. Morphologically plural, the singular is not attested (LLA 267).

¹⁴⁵ AG I f. 53vb l. 5 etc.

¹⁴⁶ Incidentally, it makes untenable Leslau’s (CDG 527) hypothetical connection of this noun with the well-known root śgr “to catch in a net, ensnare”.

¹⁴⁷ See RIÉ IIIB 475–6. The spelling ḥarā is well attested in AG I, III (AG III f. 13ra l. 8 etc.).

¹⁴⁸ In RIÉ 187:26, the same verb is restored by the editors of RIÉ (none of the root consonants are visible).

¹⁴⁹ DAE IV 27.

¹⁵⁰ CDG 128, LLA 1083–4.

¹⁵¹ CDG 129, LLA 1111.

¹⁵² daḥarəwomu “bless (imv. masc. pl.) them (masc.)!” (AG III f. 96vb ll. 18–9).

¹⁵³ With the direct object denoting the troops, the whole construction must indicate, apparently, the act of demobilization, cf. also Bausi Reference Bausi, Lourié and Mouraviev2009: 53, n. 23.

¹⁵⁴ Drewes Reference Drewes and Kaye1991: 390–1, RIÉ IIIB 225, 608.

¹⁵⁵ Marrassini Reference Marrassini2014: 227.

¹⁵⁶ Bausi Reference Bausi, Lourié and Mouraviev2009: 51–3, n. 23.

¹⁵⁷ In LLA 1108, the respective verb is spelled with ḫ; Leslau, in CDG 129, treats “to allow” as a sub-meaning of “to bless”, both adduced under daḫara. In the witnesses of the text edited by Bausi, spelling with both ḥ and ḫ is found (Bausi Reference Bausi, Lourié and Mouraviev2009: 52).

¹⁵⁸ Bulakh Reference Bulakh2023a: 274–5.

¹⁵⁹ Lane 855, BK I 673.

¹⁶⁰ In RIÉ 185 Ibis:6, *daḥara may have the same meaning, without any overt direct object (“to send forth (emissaries, messengers)”). However, the reading dḥrm in RIÉ 185 Ibis:6 is not secure: Uhlig (Reference Uhlig2001: 15), followed by Sima (Reference Sima2003/2004: 271), reads gnym (a form which is supported by the other three versions of the same text).

¹⁶¹ For instance, Drewes’s reconstruction of the verb *ḫallawa “to watch, guard” in RIÉ 192 B:6 (where only the letter ḫ is legible) is difficult to accept, inter alia, because of the spelling yəḥel(l)əwu “they (masc. pl.) watch” in AG I f. 88va l. 2 etc.

References

Sources and literature

BK – Biberstein Kazimirski, Albert de. 1860. Dictionnaire arabe–franҫais, Vols. I–II. Paris: Maisonneuve.Google Scholar

CDG – Leslau, Wolf. 1987. Comparative Dictionary of Geʿez (Classical Ethiopic): Geʿez–English/English–Geʿez, with an Index of the Semitic Roots. Wiesbaden: Otto Harrassowitz.Google Scholar

DAE IV – Littmann, Enno. 1913. Deutsche Aksum-Expedition. Band IV: Sabäische, Griechische und Altabessinische Inschriften. Berlin: Verlag von Georg Reimer.Google Scholar

EMML – Ethiopian manuscripts microfilmed for the Ethiopian Manuscript Microfilm Library, Addis Ababa and for the Hill Monastic Manuscript Library, St John’s Abbey and University, Collegeville, Minnesota.Google Scholar

HALOT – Koehler, Ludwig and Walter Baumgartner. 2001. The Hebrew and Aramaic Lexicon of the Old Testament. Leiden: Brill.Google Scholar

Lane – Lane, Edward W. 1863. An Arabic–English Lexicon. London: Williams and Norgate.Google Scholar

LIQ – Ricks, Stephen D. 1989. Lexicon of Inscriptional Qatabanian. Rome: Editrice Pontificio Istituto Biblico.Google Scholar

LLA – Dillmann, Christian Friedrich August. 1865. Lexicon Linguae Aethiopicae. Lipsiae: T.O. Weigel.Google Scholar

LM – Arbach, Mounir. 1993. Le maḏābien: lexique, onomastique et grammaire d’une langue de l’Arabie méridionale préislamique. T.I. Lexique maḏābien. Doctoral thesis, Université de Provence Aix Marseille I. Centre d’Aix, Aix-en-Provence.Google Scholar

LSJ – Liddell, Henry George and Robert Scott. 1996. A Greek–English Lexicon, Revised and Augmented throughout by Sir Henry Stuart Jones, with the Assistance of Roderick McKenzie. Oxford: Clarendon Press.Google Scholar

RIÉ I – Bernand, Étienne, Abraham J. Drewes, and Roger Schneider. 1991. Recueil des inscriptions de l’Éthiopie des périodes pré-axoumite et axoumite. Tome I. Les documents. Paris: Diffusion de Boccard.Google Scholar

RIÉ II – Bernand, Étienne, Abraham J. Drewes, and Roger Schneider. 1991. Recueil des inscriptions de l’Éthiopie des périodes pré-axoumite et axoumite. Tome II. Les Planches. Paris: Diffusion de Boccard.Google Scholar

RIÉ IIIB – Drewes, Abraham J., Manfred Kropp, and Harry Stroomer. 2019. Recueil des inscriptions de l’Éthiopie des périodes pré-axoumite et axoumite. Tome III – Traductions et commentaires. B. Les inscriptions sémitiques. Wiesbaden: Harrassowitz.Google Scholar

SD – Beeston, Alfred F.L., Mahmud A. Ghul, Walter W. Müller, and Jacques Ryckmans. 1982. Sabaic Dictionary. Louvain: Peeters.Google Scholar

Täkkəʔä Täsfay – Täkkəʔä Täsfay. 1999. Zämänawi Mäzgäbä Q̱alat Təgrəñña [A modern dictionary of Tigrinya]. ʔAsmära: ʔAḥtämti Ḥədri.Google Scholar

References

el-Antony, Fr. Maximous, Jesper Blid, and Butts, Aaron Michael. 2016. “An early Ethiopic manuscript fragment (twelfth–thirteenth century) from the Monastery of St Antony (Egypt)”, Aethiopica 19, 27–51.CrossRef Google Scholar

Avanzini, Alessandra. 2007. “Gəˁəz inscriptions in Ethiopia/Eritrea in antiquity”, in Uhlig, Siegbert (ed.), Encyclopaedia Aethiopica, Vol. 3. Wiesbaden: Harrassowitz, 159–62.Google Scholar

Bausi, Alessandro. 2003. “Numismatica aksumita, linguistica e filologia”, Istituto Italiano di Numismatica, Annali 50, 157–75.Google Scholar

Bausi, Alessandro. 2005. “Ancient features of Ancient Ethiopic”, Aethiopica 8, 149–69.CrossRef Google Scholar

Bausi, Alessandro. 2009. “La Versione Etiopica Delle Risposte Canoniche Di Timoteo I Attribuite A Pietro Di Alessandria (Cpg II, Nr. 2520)”, in Lourié, Basil and Mouraviev, Aleksei (eds), Universum Hagiographicum: Mémorial R.P. Michel van Esbroeck, S.J. (1934–2003). Piscataway, NJ: Gorgias Press, 41–57.CrossRef Google Scholar

Bausi, Alessandro. 2015. “Ethiopic manuscripts”, in Bausi, Alessandro et al. (eds), Comparative Oriental Manuscript Studies. An Introduction. Hamburg: Tredition, 46–9.Google Scholar

Bausi, Alessandro. 2016. “On editing and normalizing Ethiopic texts”, in Bausi, Alessandro, with assistance from Sokolinski, Eugenia (eds), 150 Years after Dillmann’s Lexicon: Perspectives and Challenges of Gəʕəz Studies. Wiesbaden: Harrassowitz Verlag, 43–102.Google Scholar

Bausi, Alessandro. 2017. “History of Aksum”, in Uhlig, Siegbert, Appleyard, David, Bausi, Alessandro, Hahn, Wolfgang and Kaplan, Steven (eds), Ethiopia. History, Culture and Challenges. Münster: LIT; East Lansing, MI: Michigan State University Press, 98–103.Google Scholar

Bausi, Alessandro. 2018a. “Translations in Late Antique Ethiopia”, in Crevatin, Franco (ed.), Egitto crocevia di traduzioni. ΔΙΑΛΟΓΟΙ 1. Trieste: EUT, 69–99.Google Scholar

Bausi, Alessandro. 2018b. “The recently published Ethiopic inscriptions of King Ḥafilā (AΦΙΛΑϹ): a few remarks”, Bibliotheca Orientalis 75/3–4, 285–95.Google Scholar

Bausi, Alessandro. 2018c. “The enigma of a medieval Ethiopian dynasty of saints and usurpers [Review of Derat, Marie-Laure: L’énigme d’un dynastie sainte et usurpatrice dans le royaume chrétien d’Éthiopie du XI^e au XIII^e siècle. Turnhout: Brepols 2018]”, Orientalistische Literaturzeitung 113/6, 439–47.CrossRef Google Scholar

Bausi, Alessandro. 2019a. “‘Tempo e storia’ nella tradizione etiopica cristiana”, in Bausi, Alessandro, Camplani, Alberto, and Emmel, Stephen (eds), Time and History in Africa. Tempo e storia in Africa. Milan: Biblioteca Ambrosiana, 79–112.Google Scholar

Bausi, Alessandro. 2019b. “A few remarks on hagiographical-homiletic collections in Ethiopic manuscripts”, Manuscript Cultures 13, 2019 (= Jost Gippert and Caroline Macé (eds), Homiletic Collections in Greek and Oriental Manuscripts. Proceedings of the Conference “Hagiographico–Homiletic Collections in Greek, Latin and Oriental Manuscripts – Histories of Books and Text Transmissions in a Comparative Perspective”. Center for the Study of Manuscript Cultures, Universität Hamburg, 23–24 June 2017), 63–80.Google Scholar

Bausi, Alessandro. 2021. “‘Paleografia quale scienza dello spirito’: once more on the Gǝʿǝz inscription of Ham (RIÉ No. 232)”, in Quenzer, Jörg B. (ed.), Exploring Written Artefacts: Objects, Methods, and Concepts, I. (Studies in Manuscript Cultures 25.) Berlin and Boston, MA: De Gruyter, 3–33.CrossRef Google Scholar

Bausi, Alessandro. 2022. “Ethiopic colophons: an update”, Comparative Oriental Manuscript Studies Bulletin 8/1, 121–97.Google Scholar

Bausi, Alessandro. 2023. “Linguistic phenomena from the Aksumite Collection (CAe 1047)”, Kervan – International Journal of Afro-Asiatic Studies 27, 155–77.Google Scholar

Bausi, Alessandro, Brita, Antonella, Di Bella, Marco, Nosnitsin, Denis, Sarris, Nikolas, and Rabin, Ira. 2020a. “The Aksumite Collection or Codex Σ (Sinodos of Qǝfrǝyā, Ms C3-IV-71/C3-IV-73, Ethio-SPaRe UM-039): codicological and palaeographical observations. With a note on material analysis of inks”, Comparative Oriental Manuscript Studies Bulletin 6/2, 127–71.Google Scholar

Bausi, Alessandro, Harrower, Michael, and Dumitru, Ioana A.. 2020b. “The Gə‘əz inscriptions from Beta Samā‘ti (Beta Samati)”, Bibliotheca Orientalis 77/1–2, 35–56.Google Scholar

Binyam, Yonatan and Krebs, Verena. 2024. “Ethiopia” and the World, 330–1500 ce. Cambridge: Cambridge University Press.Google Scholar

Bosc-Tiessé, Claire. 2010. “Catalogue des autels et meubles d’autel en bois (tābot et manbara tābot) des églises de Lālibalā: jalons pour une histoire des objets et des motifs”, Annales d’Éthiopie 25, 55–101.CrossRef Google Scholar

Bosc-Tiessé, Claire, Derat, Marie-Laure, Fritsch, Emmanuel, and Abullif, Wadi Awad. 2010. “Les inscriptions arabes, coptes et guèzes des églises de Lālibalā”, Annales d’Éthiopie 25, 43–53.Google Scholar

Bulakh, Maria. 2005. “Semantic shifts in the lexical field of taste in Geez”, Scrinium 1, 325–53.CrossRef Google Scholar

Bulakh, Maria. 2009. “Nota Genitivi za- in Epigraphic Geez”, Journal of Semitic Studies 54, 393–419.CrossRef Google Scholar

Bulakh, Maria. 2013. “Epigraficheskiy Geez” [Epigraphic Geez], in Bulakh, Maria S., Kogan, Leonid E., and Romanova, Olga I. (eds), Jazyki Mira: Semitskie Jazyki, Efiosemitskie Jazyki [Languages of the World: Semitic Languages, Ethiosemitic Languages]. Moscow: Academia, 199–215.Google Scholar

Bulakh, Maria. 2016. “Some problems of transcribing Geez”, in Alessandro Bausi, with assistance from Sokolinski, Eugenia (eds), 150 Years after Dillmann’s Lexicon: Perspectives and Challenges of Gəʕəz Studies. Wiesbaden: Harrassowitz Verlag, 103–37.Google Scholar

Bulakh, Maria. 2021. “Recent progress in Ethiopian epigraphic studies: the final volume of Recueil des inscriptions de l’Éthiopie des périodes pré-axoumite et axoumite”, Orientalistische Literaturzeitung 116/4–5, 323–8.CrossRef Google Scholar

Bulakh, Maria. 2023a. “The gutturals in North Argobba: an etymological evaluation”, Journal of Semitic Studies 68/1, 235–325.CrossRef Google Scholar

Bulakh, Maria. 2023b. [Review of Josef Tropper and Rebecca Hasselbach-Andee, Classical Ethiopic. A Grammar of Gəʕəz], Journal of Semitic Studies 68/1, e11–e23.CrossRef Google Scholar

Bulakh, Maria. 2024. “Early Geez phonology as reflected in ʔAbbā Garimā I”, Journal of Near Eastern Studies 83/1, 77–89.CrossRef Google Scholar

Bulakh, Maria and Selassie, Yohannes Gebre. 2022. “New readings and interpretations on the inscribed stele from Ḥənzat (HS1)”, Aethiopica 25, 125–59.Google Scholar

Butts, Aaron Michael. 2019. “Gəʕəz (Classical Ethiopic)”, in Huehnergard, John and Pat-El, Na‘ama (eds), The Semitic Languages. Second Edition. London and New York: Routledge, 117–44.CrossRef Google Scholar

Butts, Aaron Michael. 2020. “A marginal case of secondary opening in Gəˁəz”, Journal of Semitic Studies 65/2, 495–509.CrossRef Google Scholar

Curtis, Matthew C. and Bausi, Alessandro. 2005. “Ham”, in Uhlig, Siegbert (ed.), Encyclopaedia Aethiopica, Vol. 2. Wiesbaden: Harrassowitz, 980–1.Google Scholar

Delamarter, Steve and Getatchew, Haile. “Sinai Greek NF M 90, unknown, unidentified horologium. Liturgy; unidentified homily on themes from Deuteronomy 6:5, Matthew 22:37, Deuteronomy 20:2, Matthew 19:21. Homily; unidentified. Unknown”, Sinai Palimpsests Project, a publication of St. Catherine’s Monastery of the Sinai in collaboration with EMEL and UCLA, https://sinai.library.ucla.edu, accessed on 25 .03.2024.Google Scholar

Derat, Marie-Laure. 2010. “Les donations du roi Lālibalā: Éléments pour une géographie du royaume chrétien d’Éthiopie au tournant du XII^e et du XIII^e siècle”, Annales d’Éthiopie 25, 19–42.CrossRef Google Scholar

Derat, Marie-Laure. 2018. L’énigme d’une dynastie sainte et usurpatrice dans le royaume Chrétien d’Éthiopie du XIe au XIIIe siècle. (Hagiologia 14.) Turnhout: Brepols Publishers.Google Scholar

Diem, Werner. 1988. “Laryngalgesetze und Vokalismus. Ein Beitrag zur Geschichte des Altäthiopischen”, Zeitschrift der Deutschen Morgenländischen Gesellschaft 138/2, 236–62.Google Scholar

Drewes, Abraham Johannes. 1991. “Some features of epigraphical Ethiopic”, in Kaye, Alan S. (ed.), Semitic Studies in Honor of Wolf Leslau on the Occasion of his Eighty-Fifth Birthday, November 14^th, 1991, vol. I. Wiesbaden: Otto Harrassowitz, 382–91.Google Scholar

Drewes, Abraham Johannes. 2000. “Noms propres dans les documents épigraphiques de l’Éthiopie, II”, Semitica 50, 199–210.Google Scholar

Erho, Ted M. 2025. “Ethiopic palimpsests”, in Gippert, Jost, Maksimczuk, José, and Sargsyan, Hasmik (eds), Palimpsests and Related Phenomena across Languages and Cultures. Berlin and Boston: De Gruyter, 393–431.Google Scholar

Erho, Ted M. and Henry, W. Benjamin. 2019. “The Ethiopic Jannes and Jambres and the Greek Original”, Archiv für Papyrusforschung und verwandte Gebiete 65/1, 176–223.CrossRef Google Scholar

Fiaccadori, Gianfranco. 1981. “Per una nuova iscrizione Etiopica da Aksum”, Egitto e Vicino Oriente 4, 357–67.Google Scholar

Fiaccadori, Gianfranco. 2007. “Gəˁəz inscriptions in Ethiopia/Eritrea in medieval and modern times”, in Uhlig, Siegbert (ed.), Encyclopaedia Aethiopica, Vol. 3. Wiesbaden: Harrassowitz, 163–5.Google Scholar

Frantsouzoff, Serge A. 2021. “Les inscriptions axoumites royales en écriture sudarabique: un réexamen”, in Darles, Christian, Khalidi, Lamya, and Arbach, Mounir (eds), Contacts between South Arabia and the Horn of Africa, from the Bronze Age to Islam. In Honor of Rémy Audouin. Toulouse: Presses universitaires du Midi, 232–8.Google Scholar

Haile, Getatchew. 2021. “A fragment of the late Aksumite or early Zagʷe period on the commentary on the Gospel of Matthew”, Aethiopica 24, 223–32.Google Scholar

Tesfaye, Gigar. 1984. “Inscriptions sur bois de trois églises de Lalibala”, Journal of Ethiopian Studies 17, 107–26.Google Scholar

Gorea, Maria. 2023. “From the Aramaic raḥmānāʾ to raḥmānān and al-raḥmān”, Millennium 20/1, 91–106.CrossRef Google Scholar

Gragg, Gene B. 1997. “Geʕez phonology”, in Kaye, Alan S. (ed.), Phonologies of Asia and Africa (Including the Caucasus), Vol. I. Winona Lake, IN: Eisenbrauns, 169–86.Google Scholar

Hahn, Wolfgang. 2021. Münzgeschichte der Akßumiten Könige in der Spätantike. Unter Mitarbeit von Robert Keck. (Veröffentlichungen des Instituts für Numismatik und Geldgeschichte der Universität Wien 21.) Vienna: Österreichische Forschungsgesellschaft für Numismatik.Google Scholar

Hahn, Wolfgang and West, Vincent. 2016. Sylloge of Aksumite Coins in the Ashmolean Museum, Oxford. Oxford: Ashmolean Museum.Google Scholar

Harrower, Michael J., Dumitru, Ioana A., Perlingieri, Cinzia, Nathan, Smiti, Zerue, Kifle, Lamont, Jessica L., Bausi, Alessandro, Swerida, Jennifer L., Bongers, Jacob L., Woldekiros, Helina S., Poolman, Laurel A., Pohl, Christie M., Brandt, Steven A. and Peterson, Elizabeth A.. 2019. “Beta Samati: discovery and excavation of an Aksumite town”, Antiquity 93, no. 372, 1534–52.CrossRef Google Scholar

Hatke, George. 2013. Aksum and Nubia: Warfare, Commerce, and Political Fictions in Ancient Northeast Africa. New York, NY: New York University Press, Institute for the Study of the Ancient World.Google Scholar

Hatke, George. 2022. “Again the question of Judaism in Ancient Ethiopia”, Wiener Zeitschrift für die Kunde des Morgenlandes 112, 163–87.Google Scholar

Kane, Thomas L. 1990. Amharic–English Dictionary. Wiesbaden: Harrassowitz.Google Scholar

Kane, Thomas L. 2000. Tigrinya–English Dictionary. Springfield, VA: Dunwoody Press.Google Scholar

Kim, Sergey. 2022. “New studies of the structure and the texts of Abba Garima Ethiopian Gospels”, Afriques [En ligne], 13.Google Scholar

Kogan, Leonid. 2011. “Proto-Semitic lexicon”, in Weninger, Stefan (ed.), The Semitic Languages. An International Handbook. Berlin and Boston: De Gruyter Mouton, 179–258.CrossRef Google Scholar

Kogan, Leonid. 2015. Genealogical Classification of Semitic. The Lexical Isoglosses. Berlin and Boston: De Gruyter Mouton.CrossRef Google Scholar

Kogan, Leonid. 2025. “On the etymological awareness of the traditional Arabic lexicography: Aethiopica in Lisān al-ʕarab and Tāj al-ʕarūs”, Journal of the American Oriental Society 145/3, 623–34.CrossRef Google Scholar

Kropp, Manfred. 1999. “‘Glücklich, wer vom Weib geboren, dessen Tage doch kurzbemessen!…’ Die Altäthiopische Grabinschrift, datiert auf den 23. Dezember 873 n. Chr.”, Oriens Christianus 83, 162–76.Google Scholar

Leslau, Wolf. 1941. Documents Tigrigna (Éthiopien Septentrional). Grammaire et textes. Paris: Librairie C. Klincksieck.Google Scholar

Leslau, Wolf. 1956. Étude descriptive et comparative du Gafat (Éthiopien méridional). Paris: Librairie C. Klincksieck.Google Scholar

Leslau, Wolf. 1979. Etymological Dictionary of Gurage (Ethiopic). III. Etymological Section. Wiesbaden: Harrassowitz.Google Scholar

Leslau, Wolf. 1995. Reference Grammar of Amharic. Wiesbaden: Harrassowitz Verlag.Google Scholar

Lusini, Gianfrancesco. 2009. “Gli Atti apocrifi di Marco”, Aethiopica 12, 7–47.CrossRef Google Scholar

Lusini, Gianfrancesco. 2022. “The decline and collapse of the Kingdom of Aksum (sixth–seventh ce): an environmental disaster or the end of a political process?”, in Gehler, Michael, Rollinger, Robert, and Strobl, Philipp (eds), The End of Empires. (Universal- und kulturhistorische Studien. Studies in Universal and Cultural History.) Wiesbaden: Springer VS, 321–36.CrossRef Google Scholar

Marrassini, Paolo 2014. Storia e leggenda dell’Etiopia tardoantica. Le iscrizioni reali aksumite: con un’appendice di Rodolfo Fattovich su La civiltà aksumita: aspetti archeologici e una nota editoriale di Alessandro Bausi. (Testi del Vicino Oriente antico 9, Letteratura etiopica 1.) Brescia: Paideia.Google Scholar

McKenzie, Judith S. 2016a. “General introduction”, in McKenzie, Judith S. and Watson, Francis (eds), The Garima Gospels: Early Illuminated Gospel Books from Ethiopia, With Preface and Photographs by Michael Gervers and Contributions by Matthew R. Crawford, Linda R. Macaulay, Sarah S. Norodom, Andres T. Reyes, and Miranda E. Williams. (Manar al-Athar Monograph 3.) Oxford: Manar al-Athar, 1–9.Google Scholar

McKenzie, Judith S. 2016b. “The discovery and study of the Garima Gospels”, in McKenzie, Judith S. and Watson, Francis (eds), The Garima Gospels: Early Illuminated Gospel Books from Ethiopia, With Preface and Photographs by Michael Gervers and Contributions by Matthew R. Crawford, Linda R. Macaulay, Sarah S. Norodom, Andres T. Reyes, and Miranda E. Williams. (Manar al-Athar Monograph 3.) Oxford: Manar al-Athar, 31–41.Google Scholar

McKenzie, Judith S. and Watson, Francis (eds). 2016. The Garima Gospels: Early Illuminated Gospel Books from Ethiopia, With Preface and Photographs by Michael Gervers and Contributions by Matthew R. Crawford, Linda R. Macaulay, Sarah S. Norodom, Andres T. Reyes, and Miranda E. Williams. (Manar al-Athar Monograph 3.) Oxford: Manar al-Athar.Google Scholar

Mercier, Jacques. 2021. “Les deux types d’édicules associés aux canons d’Eusèbe. Apport des évangiles d’Abba Gärima (c. 450–650) à leur histoire et leurs symboliques byzantines et latines”, Cahiers Archeologiques 58, 29–54.Google Scholar

Meyer, Ronny. 2016. “The Ethiopic script: linguistic features and socio-cultural connotations”, in Mendisu, Binyam Sisay and Johannessen, Janne Bondi (eds), Multilingual Ethiopia: Linguistic Challenges and Capacity Building Efforts. (Oslo Studies in Language 8.1.) Oslo: University of Oslo, 137–72.Google Scholar

Müller, Walter W. 2012. “Äthiopische Inschriftenfragmente aus der himjarischen Hauptstadt Ẓafar”, Aethiopica 15, 7–21.CrossRef Google Scholar

Munro-Hay, Stuart. 1984. The Coinage of Aksum. New Delhi: Manohar Publications.Google Scholar

Munro-Hay, Stuart. 1991. Aksum. An African Civilisation of Late Antiquity. Edinburgh: Edinburgh University Press.Google Scholar

Munro-Hay, Stuart and Juel-Jensen, Bent. 1995. Aksumite Coinage. A Revised and Enlarged Edition of The Coinage of Aksum. London: Spink.Google Scholar

Nebes, Norbert. 2017. “The inscriptions of the Aksumite King Ḥafil and their reference to Ethio-Sabaean sources”, Zeitschrift für Orient-Archäologie 10, 2017, 356–69.Google Scholar

Nöldeke, Theodor. 1913. [Review of Deutsche Aksum-Expedition. Hg. von der Generalverwaltung der Kgl. Museen zu Berlin. Bd. IV. Sabaische, griechische und altabessinische Inschriften. Von Enno Littmann. Mit 6 Tafeln, 1 Karte und 109 Textabblidungen. Berlin 1913], Zeitschrift der Deutschen Morgenländischen Gesellschaft 67, 694–706.Google Scholar

Nosnitsin, Denis. 2011. “The antiquities of Däbrä Zäyt Qǝddǝst Maryam (East Tǝgray, Ethiopia)”, Aethiopica 14, 33–46.CrossRef Google Scholar

Nosnitsin, Denis. 2020. “Christian manuscript culture of the Ethiopian-Eritrean highlands: some analytical insights”, in Kelly, Samantha (ed.), A Companion to Medieval Ethiopia and Eritrea. Leiden: Brill, 282–321.CrossRef Google Scholar

Nosnitsin, Denis. 2024. “An ancient manuscript fragment from northern Ethiopia: chants for John the Baptist”, in Heeßel, Nils P., Tsukanova, Vera and Waltisberg, Michael (eds), Der Perlentaucher. Festschrift für Stefan Weninger. Wiesbaden: Harrassowitz, 295–315.Google Scholar

Phillipson, David W. 2012. Foundations of an African Civilisation: Aksum & The Northern Horn 1000 BC–AD 1300. Addis Ababa: Addis Ababa University Press.CrossRef Google Scholar

Piovanelli, Pierluigi. 1993. “Les aventures des apocryphes en Éthiopie”, Apocrypha 4, 197–224.CrossRef Google Scholar

Piovanelli, Pierluigi. 2018. “Aksum and the Bible: old assumptions and new perspectives”, Aethiopica 21, 7–27.Google Scholar

Podolsky, Baruch. 1991. Historical Phonetics of Amharic. Tel-Aviv: Baruch Podolsky.Google Scholar

Rapp, Claudia. 2023. “A Cache of Palimpsests and Christian Manuscript Culture across the Medieval Mediterranean: First Results of the Sinai Palimpsests Project”, in Rapp, Claudia, Rossetto, Giulia, Grusková, Jana, and Kessel, Grigory (eds), New Light on Old Manuscripts. The Sinai Palimpsests and Other Advances in Palimpsest Studies. (Veröffentlichungen zur Byzanzforschung 45.) Vienna: Austrian Academy of Sciences Press, 39–53.CrossRef Google Scholar

Ricci, Lanfranco. 1991. “Iscrizioni Paleoetiopiche”, in Kaye, Alan S. (ed.), Semitic Studies. In honor of Wolf Leslau. On the Occasion of his Eighty-Fifth Birthday. November 14th, 1991. Vol. II. Wiesbaden: Harrassowitz, 1291–311.Google Scholar

Ricci, Lanfranco. 2003. “Iscrizioni rupestri dell’Eritrea – [Koḥaytò]*: Bukhwābùkh; Sariddoleḍ

${\grave{\bar{a}}}$(h): Ṣihillò; ‘Tomba dell’Egiziano’; Cristianesimo”, Rassegna di Studi Etiopici, Nuova Serie, Vol. 2 (45), 51–76.Google Scholar

Robin, Christian Julien. 2012. “South Arabia, Ethiopia and Socotra”, in Strauch, Ingo (ed.), Foreign Sailors on Socotra. The Inscriptions and Drawings from the Cave Hoq. Bremen: Hempen Verlag, 437–46.Google Scholar

Robin, Christian Julien. 2022a. “Le guèze maquillé en sabaʾique des inscriptions royales aksūmites (Éthiopie antique)”, in Robert, Jean-Noël (ed.), Hiéroglossie III: Persan, syro-araméen et les relations avec la langue arabe. Collège de France, 25 juin 2018. Paris: Bibliothèque de l’Institut des Hautes Études Japonaises, Collège de France, 171–205.Google Scholar

Robin, Christian Julien. 2022b. “Arabie antique: variations dans la manière de nommer le Dieu unique”, in Robert, Jean-Noël (ed.), Hiéroglossie III: Persan, syro-araméen et les relations avec la langue arabe. Collège de France, 25 juin 2018. Paris: Bibliothèque de l’Institut des Hautes Études Japonaises, Collège de France, 139–70.Google Scholar

Schneider, Roger. 1974. “Trois nouvelles inscriptions royales d’Axoum”, in IV Congresso Internazionale de Studie Etiopici (Roma, 10–15 aprile 1972), Tomo I (Sezione Storica). Rome: Accademia Nazionale dei Lincei, 767–86.Google Scholar

Sima, Alexander. 2003/2004. “Die ‘sabäische’ Version von König ˁE̅zānās Trilingue RIE 185 und RIE 185bis”, Archiv für Orientforschung 50, 269–84.Google Scholar

Sima, Alexander. 2005. “GDR(T)”, in Uhlig, Siegbert (ed.), Encyclopaedia Aethiopica, Vol. 2. Wiesbaden: Harrassowitz, 718–9.Google Scholar

Stein, Peter. 2010. Die altsüdarabischen Minuskelinschriften auf Holzstäbchen aus der Bayerischen Staatsbibliothek in München. Band 1. Die Inschriften der mittel- und spätsabäischen Periode. 1. Teil: Text. Berlin: Wasmuth.Google Scholar

Stein, Peter. 2023. Die altsüdarabischen Minuskelinschriften auf Holzstäbchen aus der Bayerischen Staatsbibliothek in München. Band 2. Die altsabäischen und minäischen Inschriften. Mit einem Anhang: Unbeschriftete Objekte und Fälschungen. Wiesbaden: Reichert.Google Scholar

Strauch, Ingo (ed.). 2012a. Foreign Sailors on Socotra: The Inscriptions and Drawings from the Cave Hoq. (Vergleichende Studien zu Antike und Orient 3.) Bremen: Hempen Verlag.Google Scholar

Strauch, Ingo. 2012b. “Concluding remarks. The discoveries in the cave Hoq: a short evaluation of their historical meaning”, in Strauch, Ingo (ed.), Foreign Sailors on Socotra: The Inscriptions and Drawings from the Cave Hoq. Bremen: Hempen Verlag, 540–4.Google Scholar

Tropper, Josef. 2002. Altäthiopisch. Grammatik des Ge‘ez mit Übungstexten and Glossar. Münster: Ugarit-Verlag.Google Scholar

Tropper, Josef and Hasselbach-Andee, Rebecca. 2021. Classical Ethiopic. A Grammar of Gəʕəz. (Languages of the Ancient Near East.) University Park, Pennsylvania: Eisenbrauns.Google Scholar

Uhlig, Siegbert. 2001. “Eine trilinguale ˁEzana-Inschrift”, Aethiopica 4, 7–31.CrossRef Google Scholar

Ullendorff, Edward. 1955. The Semitic Languages of Ethiopia. A Comparative Phonology. London: Taylor’s Foreign Press.Google Scholar

Villa, Massimo. 2019. Filologia e linguistica dei testi Gəʕəz di età Aksumita. Il Pastore di Erma. Studi Africanistici. Serie Etiopica 10. Naples: Università degli studi di Napoli “L’Orientale”.Google Scholar

Villa, Massimo and Rafał, Zarzeczny. 2026. “Again on MS EMML 8509 from Ṭana Qirqos: unconventional orthography and textual content”, in Sernicola, Luisa and Villa, Massimo (eds), From Aksum to Lalibäla. The Myth of the “Dark Ages” of Eritrean and Ethiopian History (7th–13th centuries). Naples: UniorPress.Google Scholar

Voigt, Rainer. 2016. “Language, script and society in the Axumite kingdom”, ITYOP̣IS – Northeast African Journal of Social Sciences and Humanities, Extra Issue II, 59–80.Google Scholar

Weninger, Stefan. 1999. Geʕez. (Languages of the World/Materials 01.) München: Lincom Europa.Google Scholar

Weninger, Stefan. 2005. “Gǝʕǝz”, in Uhlig, Siegbert (ed.), Encyclopaedia Aethiopica, Vol. 2. Wiesbaden: Harrassowitz, 732–5.Google Scholar

Weninger, Stefan. 2010. “Sounds of Gǝʕǝz – how to study the phonetics and phonology of an ancient language,” Aethiopica 13, 75–88.CrossRef Google Scholar

Weninger, Stefan. 2011. “Old Ethiopic”, in Weninger, Stefan (ed.), The Semitic Languages. An International Handbook. Berlin and Boston: De Gruyter Mouton, 1124–42.CrossRef Google Scholar

Wetter, Andreas. 2010. Das Argobba: Eine deskriptive Grammatik der Varietät von Shonke und Tʾollaha (Zentraläthiopien). (Grammatical Analyses of African Languages 38.) Cologne: Rüdiger Köppe Verlag.Google Scholar

Zuurmond, Rochus. 1989. Novum Testamentum Aethiopice: The Synoptic Gospels, I: General Introduction; II: Edition of the Gospel of Mark. (Äthiopistische Forschungen 27.) Stuttgart: Franz Steiner.Google Scholar

Zuurmond, Rochus. 2001. Novum Testamentum Aethiopice: The Synoptic Gospels, III: The Gospel of Matthew. (Aethiopistische Forschungen 55.) Wiesbaden: Harrassowitz Verlag.Google Scholar

Zuurmond, Rochus and Niccum, Curt. 2013. “The Ethiopic version of the New Testament”, in Ehrman, Bart D. and Holmes, Michael W. (eds), The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis, 2nd ed. (New Testament Tools, Studies and Documents 42.) Leiden and Boston, MA: Brill, 231–52.CrossRef Google Scholar

Table 1. Tentative periodization of the Geez languageTable 1 long description.

Table 2. Orthography of RIÉ 193 I, IITable 2 long description.

Table 3. Orthography of RIÉ 194Table 3 long description.

Table 4. The coincidence of spelling in Epigraphic Geez and the ʔAbbā Garimā GospelsTable 4 long description.

Article contents

Early Geez orthography in the ʔAbbā Garimā Gospels and in the Aksumite Epigraphic Geez

Abstract

Keywords

Information

I. Introduction

II. Periodization of the Geez language and classification of Geez sources

III. Early Post-Aksumite inscriptions in RIÉ

III.1. RIÉ 193 I, II as post-Aksumite inscriptions

III.2. RIÉ 194 as a post-Aksumite inscription

III.3. RIÉ 232 as a post-Aksumite inscription

III.4. RIÉ 202 as a post-Aksumite inscription?

IV. The relevance of orthography for the dating of Geez sources

V. Orthography of AG I and III vs. orthography of Aksumite Epigraphic Geez

V.1. Aksumite Epigraphic Geez (AEG) vs. AG I, III?

V.1.1. ʕṣ́b (?) in RIÉ 196 vs. AG I, III ʕəṣub “difficult”

V.1.2. *ʕawādi “herald” or ʕwd “to surround” (?) in RIÉ 190:15 vs. AG I, III ʔawādi, ʔoda

V.1.3. *gys (?) in RIÉ 254:3 vs. geśa “to do things early in the morning” in AG I, III

V.1.4. ʔḥ[ṣ]ṣ “to diminish” (?) in RIÉ 248:3 vs. ḫṣṣ “to be little” in AG I, III

V.1.5. ṣaniḥā in RIÉ 345 vs. ṣ́nḥ “to wait” in AG I, III

V.2. AEG corroborated by AG I, III

V.2.1. AEG malʔa, AG I, III mal(l)əʔa, ʔamal(l)əʔa “to set out, go forward”

V.2.2. sgrt (name of a military unit) and sagarāt “soldiers” in AG I, III

V.2.3. AEG *daḥara “to send forth”, AG I, III daḥara “to divorce”

VI. Conclusions

Funding information

Acknowledgements

Abbreviations of languages

Footnotes

References

Sources and literature

References

Save article to Kindle

Save article to Dropbox

Save article to Google Drive

Reply to: Submit a response

Your details

You have entered the maximum number of contributors

Conflicting interests