I. Population diversity and multilingualism in Roman Cyrenaica

The population of Roman Cyrenaica comprised three main linguistic groups: Greek, Latin and ‘Libyc’.

Greeks settled in these lands at the end of the seventh century BC from the island of Thera and their number increased over time through regular population movements. As a result, Cyrene was first and foremost a Greek polis and a centre of Hellenism in North Africa, which gave rise to major scholars and intellectuals of Classical to Late-Antique Greek thought, such as Aristippus, Callimachus, Eratosthenes and Synesius. The vast majority of inscriptions are therefore written in Greek, which remained the dominant language of the region for a millennium and a half.

In the absence of a Roman colonial foundation, the spread of Latin seems to have been the result of two main factors, beginning in the first century BC: the settlement of Italians for commercial reasonsFootnote ⁴ and the imposition of an administrative apparatus in Latin from the time of the formation of the province, supplemented a little later by a military presence which, however, remained relatively discreet during the High Empire.Footnote ⁵ In 7/6 BC, the edicts of Augustus (C.101, ll. 4–6) listed only 215 Roman citizens, which is very few. This small Latin-speaking groupFootnote ⁶ always remained largely in the minority compared with the Greek-speakers. We also know of new population influxes, the composition of which remains unknown, for instance in the aftermath of the Jewish revolt of AD 115–17 when 3,000 veterans were sent to repopulate the region, giving the population of the areas in which they settled a more military – and perhaps Latin – flavour (SEG.17.584, but we are not sure of their precise origin).

The last group are the so-called ‘Libyan’ peoples, who were present in these areas long before the arrival of the Greeks. They originally belonged to tribes, some of whom had acculturated to the Greek way of life, while others, further away from urban centres, retained a nomadic or semi-nomadic lifestyle. These populations had a specific language called ‘Libyc’, most probably an ancestor of Berber (Chaker Reference Chaker2008). In the western African provinces (from the Atlantic to Tripolitania),Footnote ⁷ Libyc is known in the epigraphical record at least until the third century AD mainly for funerary and votive texts,Footnote ⁸ along with Punic (until the fifth century AD), thus creating a unique and very diverse linguistic landscape (Coltelloni-Trannoy Reference Coltelloni-Trannoy2015; Coltelloni-Trannoy and Veïsse Reference Coltelloni-Trannoy and Veïsse2007; Millar Reference Millar1968). However, no written documents in Libyc have come to light in the region, so the local form of the language – which we can assume was in a dominant position at least among the rural populations – remains completely unknown: in Cyrenaica, Libyc is therefore a ‘ghost language’.Footnote ⁹

Only a few anthroponyms transliterated into Greek make it possible to trace it, although these names do not determine any ethnicity for their bearers (Camps Reference Camps2002; Chevrollier Reference Chevrollier2020–2024, 216–17; Dobias-Lalou, this volume; Marini Reference Marini2018, 119–40; Masson Reference Masson1974; Rebuffat Reference Rebuffat2018). The Libyans and those who belonged to mixed groups chose to engrave their inscriptions in Greek or Latin certainly for reasons of prestige (Bérenger Reference Bérenger and Roure2023, 58) and in order to emphasise their adoption of elite writing practices. This also testifies to a nascent literacy and a new epigraphic habit among this population, based on the influence of Graeco-Roman traditions.

However, the question of the absence of inscriptions in Libyc in Cyrenaica remains unexplained.Footnote ¹⁰ In the other Roman provinces of Africa (Egypt aside), the influence of Phoenician and/or Punic on the development of writing practices in Libyc proved to be very important.Footnote ¹¹ This raises the question of why the influence of Greek, which had been present in Cyrenaica since the seventh century BC and which also had an alphabetic script as well as epigraphic practices, apparently did not have the same impact as Phoenician or Punic on the development of a local epigraphic production in Libyc. This might be explained by the presence of urban centres populated by newcomers (Greeks in Cyrenaica, Punics in the West) who would have ‘spurned’ the local language: the fact that Libyans were not constituted as a ‘state’ would have prevented the language from being standardised and institutionalised, and from imposing itself in the face of other idioms. It is in fact interesting to note, as a comparison, that no Libyc inscriptions have been found so far in the Punic territory which became the first Roman province of Africa in 146 BC (Ghaki Reference Ghaki, Gori and Viti2022, 156 and fig. 3): Libyc must have been quite undervalued by the urban elites and was spoken by segments of the population who had remained distant from Hellenism and Romanness. We can simply conclude from this that the Libyc language existed in Cyrenaica, but that neither the practice of writing nor the habit of engraving it on stone or other materials had taken root there.Footnote ¹² On the other hand, the lack of archaeological research in areas far from the Graeco-Roman cities and their immediate surroundings may explain the supposed absence of Libyc texts in present-day eastern Libya. These considerations would take me too far in these pages, but research into Libyc in Cyrenaica, and more broadly into the populations who lived in the vicinity of the Greeks poleis, is a subject that needs to be developed further, but to do so it must take into account the whole of North Africa, from the Atlantic to the Nile (Desanges Reference Desanges1962), as well as parts of the Sahara.

Therefore, it is important to bear in mind that the society of ancient Cyrenaica was in fact trilingual, and there is no reason to believe that some members of the local elite did not speak Libyc as well. But in the inscriptions, bilingualism is exclusively Greek-Latin in Cyrenaica:Footnote ¹³ it is therefore impossible to study Latin/Greek/third language plurilingualism in the same way as in other regions of the ancient world, such as the Levant with Phoenician, the Near East and notably Palmyra with Aramaic, Asia Minor with the various Anatolian languages, Sicily with Punic and southern Gaul with Gallic.Footnote ¹⁴

Besides, these three groups intermingled considerably thanks to matrimonial alliances. Some mixed families included individuals with Libyan and Greek names, and even with tria nomina with Greek or Latin cognomen. At the same time, Italian immigrant families acquired citizenship of one of the local Greek poleis, whereas ciuitas Romana was extended to the members of the elite of the local cities, especially during the Julio-Claudian and Flavian periods, so that it becomes difficult to distinguish between ‘Greeks’, ‘Italians/Romans’, or ‘Graeco-Libyans’. Demography, marriages, population movements, social mobility and civic status are all factors that have had an impact on identity, and so indirectly on the language spoken and used in the inscriptions during the Principate.

Studying multilingualism in Cyrenaica on the basis of epigraphy therefore constitutes a methodological bias, even if inscriptions remain the only means of expression that survived and can give an idea of the place of Latin. There is also a documentary bias since, among the inscriptions studied here, there is an over-representation of official documents, which are more likely to be written in Latin, since they emanate from Roman authorities. A few figures will demonstrate this.

II. The Latin inscriptions of Cyrenaica: geographical distribution and typologies

Over the nearly 30 years since the publication of G. Paci's article, the corpus of Latin texts has doubled. Paci indicated that he was relying on just over 100 texts, whereas IR Cyrenaica 2020 now comprises 237 Latin or bilingual inscriptions,Footnote ¹⁵ to which must be added five texts published subsequently:

1. 1) A bilingual funerary inscription for the freedman G. Cascellius Mommus, dated to the first century BC (Mei and Antolini Reference Mei and Antolini2019, 59–60, no. 3), from the southern necropolis of Cyrene.

2. 2) A restitutio agrorum cippus bearing the name of the imperial legate L. Acilius Strabo, from the time of Claudius or Nero, found in situ at the start of the road leading from Cyrene to Balagrae (Mei and Antolini Reference Mei and Antolini2019, 60–61, no. 4).

3. 3) A boundary inscription demarcating the border between the province of Cyrenaica and a village outside it, by the same legate L. Acilius Strabo (Alshareef et al. Reference Alshareef, Chevrollier and Dobias-Lalou2021, 60–64, no. 4).

4. 4) A terminatio from the area of Bersis, in the territory of Taucheira, under emperor Titus (Chevrollier et al. Reference Chevrollier, Dobias-Lalou, Elhaddar and Al-Abdali2023).

5. 5) A first/early-second-century-AD bilingual funerary inscription from Cyrene with the name of P. Iodius Apol(lonios?), son of Manius (Dobias-Lalou, Reference Dobias-Lalouin print). Iodius might derive from the Jewish name Ioudios/Ioudas; the integration of such a name as a gentilicium would be an unicum.

This study therefore covers a total of 242 texts geographically distributed as shown in Table 1.

Table 1. Geographical distribution of the Latin inscriptions of Cyrenaica.

Cyrene is the city with the most Latin inscriptions, but Ptolemais has around 50, which is consistent with its rise during the High Empire to the rank of capital of the Pentapolis after the reforms of Diocletian. Boundary-markers and milestones are over-represented among the inscriptions from the chora; but if we exclude them, there are only two texts from the Altar of the Philaeni (M.1–2),Footnote ¹⁶ graffiti in the fort at Esc-Sheleidima (M.46, M.49) and an isolated funerary inscription from Limnias-Lamludah (M.258), to which I return in section VI.

The online corpus makes it easy to pinpoint the findspots of the inscriptions and to carry out micro-topographical analyses, especially as it is linked to the Heritage Gazetteer of Libya.Footnote ¹⁷ Such research shows that some areas of Cyrene were home to more Latin documents than others. This is the case, for example, with the forum-Caesareum and the basilica (Luni Reference Luni and Mastino1992; Reference Luni, Gasperini and Marengo2007). The Caesareum was built on the site of the ancient gymnasium of Cyrene in the early Imperial era. From the Flavian period onwards, the architectural complex was equipped with a three-aisled basilica and became the forum of the Roman city. In the centre of the monument, bordered on the south, west and east by porticoes and on the north by the basilica, a temple was built in the Antonine era. It was dedicated either to Bacchus or to Antoninus Pius, according to various hypotheses. The quadriportico was restored under Hadrian after the damage caused by the Jewish revolt, and the basilica was given an apse on its west end. Among the inscriptions from those two buildings,Footnote ¹⁸ three are in Greek, two are bilingual but seven are in Latin (e.g., C.4, C.7, C.9, C.10, C.18). The forum area is therefore the place where Latin inscriptions dominate in number throughout the site. Other places were also home to many Latin inscriptions, such as the Augusteum on the agora (C.105 to C.112, with the exception of C.110, in Greek), in this case because of the building's particular link with the imperial cult.

By contrast, Latin inscriptions in the sanctuary of Apollo, the core of the Greek rituals in the city, are unusual and only two were discovered among the multiple shrines located inside the sanctuary.Footnote ¹⁹ These are two dedications to Luna and Mars respectively (C.300–301) recovered from the temple of Isis, both belonging to a series dedicated to the gods of the seven planets and days of the week. The online corpus and its additional resources are therefore also relevant to understanding the ‘ideology’ that the Romans wanted to convey in such or such monument.

The Latin inscriptions of Cyrenaica can roughly be divided into the categories listed in Table 2.

Table 2. Typologies of the Latin inscriptions of Cyrenaica.

¹ Including, for instance, the price edict of Diocletian P.144.

² M.244 and P.199 have been recorded here as milestones.

³ A.73 (brick stamp), C.122 (validation of a weight), C.268 (sculptor's signature), C.479 (seal legend), P.378 (mason's mark?). The category also includes the series of graffiti from the Berenice ‘hostel’ B.4, B.5, B.6, B.27, B.33, B.34, B.35, B.40.

G. Paci (Reference Paci1994, 254) pointed out that the majority of Latin inscriptions were official texts and were related to the activities of the provincial government; the publication of the IR Cyrenaica 2020 corpus confirms this general assessment, since around half of the documents belong to this typology.

To get into more details, Table 3 lists the official inscriptions mentioning Roman magistrates whenever the texts show that they had a concrete action – starting in 27 BC, Cyrene, joined with Crete, became a public province of praetorian rank governed by a proconsul assisted by legates residing in Cyrene and quaestors residing in Gortyn. These texts include building inscriptions relating to the construction or restoration of public and religious monuments; dedications in honour of emperors, members of the imperial family, or governors; consecrations of statues or temples to deities; other types of official document of uncertain nature; and ephebic inscriptions (in one case). This small corpus is dominated by bilingual inscriptions, here again over-represented by the restitutio agrorum boundary-markers and the milestones, which were both in Latin, the administrative language, and in Greek, so that they could be understood by everyone.

Table 3. Inscriptions pertaining to the activities of Roman magistrates in Cyrenaica. This table includes only those inscriptions in which Roman officials have an action of their own (either by dating the inscription, initiating a construction or a dedication, authorising it on behalf of the emperor, etc.). By contrast, proconsuls who are only mentioned in official documents that do not emanate directly from them have not been included. This is the case: 1. of P. Sextius Scaeua, mentioned in the edicts of Augustus (C.101); 2. of Saluius Carus, who appears in the letter of Hadrian to the Cyrenaeans (C.163, ll. 7–8); 3. of the legate M. Tittius, honoured by the Jewish politeuma in AD 24/5 (B.75). The latter documents are in Greek as they were supposed to be sent and read by the local inhabitants.

¹ Q. Paconius Agrippinus had been quaestor of Crete and Cyrene under Claudius (Baldwin Bowsky Reference Baldwin Bowsky2006) before returning to the province under Vespasian as imperial legate.

I shall return in section III to the chronological conclusions we can draw from Table 3, but I would like to stress first that this group of inscriptions also raises the question of the governor's accessibility to the local population. We know that governors had to deal with multilingualism in the provinces they ruled, particularly in matters of justice (Bérenger Reference Bérenger and Roure2023), in the context of a bilingual Latin-Greek empire (Corbier Reference Corbier and Villard2008; Rochette Reference Rochette and Bakker2010). Of the 52 proconsuls of Crete and Cyrene whose geographical origin can be determined with varying degrees of certainty, only six came from the Greek-speaking East, while 46 originated from Italy (35) and the western provinces (11), meaning that they may have had to hire interpreters, although we can assume that the education of senators included Greek (Bérenger Reference Bérenger-Badel2004; Reference Bérenger and Roure2023, 59–62). In Cyrenaica, it is clear that both in the Augustan era (edicts of Augustus: C.101) and in the Antonine period (dossier of imperial rescripts of Hadrian and Antoninus Pius: C.163), it was necessary to have texts relating to justice or directly addressed to the people engraved in Greek so that the local inhabitants could understand them. Even if the elite could be expected to be bilingual, it is likely that not all Cyrenaeans understood the inscriptions erected in Latin by their governors. Cyrenaica was therefore in a situation of Latin-Greek diglossia, with the Latin language in a hierarchically superior position but in a large minority in practice.

III. From Pompey to the Valentinian dynasty: a brief chronological overview of the corpus

The earliest Latin inscriptions date from the year 67 BC and the rule of Pompey the Great and his legate Cn. Cornelius Lentulus Marcellinus. Eight inscriptions belong to this period (Reynolds Reference Reynolds1962a):

• 1–3/ Three inscriptions in dialectal Greek in honour of Cn. Cornelius Lentulus Marcellinus (C.132[?], C.271 and C.280).

• 4/ An epikrima (‘decree’, from Apollonia, A.8), in Greek, relating to a dispute between Cyrene and Apollonia (according to J. Reynolds). However, André Laronde (Reference Laronde1987, 457–59) suggested that it may instead refer to a sanctuary of Apollo. If Reynolds’ proposal is correct, it could be the earliest mention of Apollonia as an autonomous city.

• 5/ A decree written in Greek and then translated into Latin,Footnote ²⁰ for Alexis son of Alexander, in which the legate's name and that of the eponymous priest of Apollo are both used to date the document (C.688). This translation was certainly commissioned by the legate to honour a philorhomaios at the time of the establishment of the first Roman administration in Cyrenaica.

• 6/ A fragmentary account of contribution to an aqueduct probably mentioning Pompey (C.687), in Latin.

• 7–8/ A decree, in Latin, potentially concerning an allocation of lands, or a census list of landowners, from Ptolemais, under the authority of Cn. Cornelius Lentulus Marcellinus and probably mentioning Pompey the Great as well (P.100, possibly the Latin version of P.101, in Greek).

The use of both Greek and Latin in this dossier, which can be dated to the foundation of the Roman province, needs to be addressed. The first four inscriptions were either initiatives of the poleis or intended for a local audience, and were therefore written in Greek – as were, for example, the inscriptions in honour of Cyrene's first Roman patrons.Footnote ²¹ Documents in Latin, on the other hand, are written in this language because they fall under Roman authority. In the decree in honour of Alexis son of Alexander, the use of Latin is also explained by the involvement of the association of the ciues Romani qui Cyrenis negotiantur.Footnote ²² These businessmen may have included publicans, who are documented from the earliest years of Roman rule in Cyrenaica.Footnote ²³ This group of inscriptions shows that the Roman authorities understood the importance of maintaining the Greek language in the process of provincialisation, and that the use of Latin was never exclusive, neither at the time of the formation of the province nor later, as expected for a prouincia of the Roman East.

Table 3 clearly shows that, from a chronological perspective, there is a clear break between a long first century AD dominated by Latin in official documents, and a second century in which Greek inscriptions outnumber the Latin ones, bilingual inscriptions left aside. It also demonstrates that a first period of expansion of Latin epigraphy in Cyrenaica occurred at the beginning of the Julio-Claudian era with numerous inscriptions dated to the reigns of Augustus and Tiberius – other texts, not included in Table 3 because magistrates are not involved, could be added such as the dedications to Augustan deities (numina Augusta, C.106–109) and to Ceres by a promagister of a company of publicans (C.455). The strong presence of Latin texts in the first century AD on the agora, where the first monuments dedicated to imperial worship were erected, where Roman governors were particularly active and where members of the familia Caesaris were honoured, had already been stressed by S.M. Marengo (Reference Marengo1988; cf. also the building inscription of the nomophylakeion under Domitian: C.92).

The second part of the Julio-Claudian period and the Flavian era are characterised by the presence of the already mentioned serial documents such as milestones and boundary-markers dated to the reigns of Claudius, Nero and Vespasian.

In the second century AD, official documents were more likely to be written in Greek (Table 3). It is entirely possible that the legacy of Hadrian, who played an active role in restoring Cyrene on Hellenic, and more specifically Dorian, cultural grounds (Rosamilia Reference Rosamilia, Gallo and Gallotta2021 based on C.163) after the difficulties caused by the Jewish revolt that broke out during the principate of his predecessor (Chevrollier Reference Chevrollier and Nantet2019), explains the (re)expansion of Greek in a region where Latin had finally never really taken root among the population. However, this observation is not free of exceptions. Table 4 lists the inscriptions relating to this phase in the city's history: it includes inscriptions in honour of Hadrian, building inscriptions relating to repairs after the tumultus Iudaicus and milestones on the road to Apollonia. The table clearly shows that Greek inscriptions were linked with the actions of benefactors or of the city of Cyrene, whereas Latin inscriptions were the result of imperial initiatives. The interest of this survey is to illustrate that, despite the emperor's insistence on a return to the city's Greek roots, the inscriptions concerning the buildings on which the imperial authority or imperial financing were involved remained in Latin. The corpus of inscriptions pertaining to the recovery of Cyrene is therefore quite balanced between Greek and Latin and could be considered a moment of transition in the epigraphic habit of the region. Inscriptions mentioning Roman officials also show that governors appear to have been less active in the second century and that their activities were carried out in deeper coordination with civic authorities and local magistrates than before.

Table 4. Inscriptions pertaining to Hadrian's actions after the Jewish revolt.

¹ This inscription has not been included in IR Cyrenaica 2020. The only fragment I know of is in Greek, but it is possible that the text was originally bilingual. Ongoing research in Joyce Reynolds' notebooks may provide further information. This entry is therefore temporary.

The most recent surviving Latin inscriptions date from the fourth century AD:

• • the Price edict of Diocletian dated AD 301 (P.144).

• • imperial honours for Diocletian at the altar of the Philaeni on the border with Tripolitania (M.1).Footnote ²⁴

• • milestones from the Tetrarchy (P.199, A.69, A.70).Footnote ²⁵

• • imperial honours for Maximin Daia, Constantine and Licinius, from Ptolemais, dated AD 311–13 (P.118).

• • imperial honours (but possibly a milestone) for Constantine and his sons, from the Christian church at Ras al-Hilal, east of Cyrene, dated AD 333–37 (M.244).

• • imperial honours for Gratian (?), Valentinian and Valens dated AD 375–78 (C.157):Footnote ²⁶ [Diuina] stirpe progenitos [D(ominos) n(ostros tres) imp(eratores) Caes(ares) Gratianum et Valentinian]um et Valentem pios, [felices, semper Augustos], ‘Born of [divine] stock, our [three masters, the emperors Caesars Gratian, Valentinian] and Valens, pious, [fortunate, eternal Augusti]’.

• • other fourth- or fifth-century AD inscriptions not precisely datable (M.2: inscription from a governor-praeses at the altar of the Philaeni;Footnote ²⁷ P.141: building inscription [uncertain]).

The latest well-dated Latin text from Cyrenaica is therefore the dedication to Gratian, Valentinian and Valens. On the contrary, much later Greek inscriptions are attested, such as copies of the imperial edict of Anastasius I dating from AD 491–518 and found in fragments at Taucheira (T.219), Ptolemais (P.116) and Apollonia (A.30). Other types of texts in Greek, mainly in a Christian context (but not only, e.g., P.120), were known until the Arab conquest in the mid-seventh century AD (see the index of IR Cyrenaica 2020 as well as Dobias-Lalou Reference Dobias-Lalou, David and Geny2012; Dobias-Lalou and Elhaddar Reference Dobias-Lalou and Elhaddar2018).

The survey shows that the first Latin texts date to the foundation of the province in 67 BC. A flourishing of Latin inscriptions can be observed during the reigns of Augustus and Tiberius, followed by a fairly significant maintenance until the reign of Hadrian, and then a certain decline which is more noticeable in the third century.Footnote ²⁸ Finally, several important documents in Latin are still known up to the early Valentinian dynasty.

IV. A complex and diversified linguistic landscape: literacy and hybridity

From a general perspective, it is particularly difficult to know how many people were able to read and write, and at what level, in the ancient world, even in a single language.Footnote ²⁹ Statistics on this point are bound to remain unattainable – Harris (Reference Harris1989) estimates that 10–20% of the population was literate –, particularly because the sources refer to the educated elite, and never to the illiterate. Even if the Imperial Roman period saw a renewed importance of public writingFootnote ³⁰ and a definite expansion of literacy for everyday activities, as shown, inter alia, by the Egyptian papyri (Depauw Reference Depauw and Riggs2012), the graffiti of Pompeii (Franklin Reference Franklin and Beard1991), the Vindolanda tablets (Bowman Reference Bowman, Bowman and Woolf1994) or the specific case of the instrumentum domesticum (Harris Reference Harris, Solin, Salomies and Liertz1995; Woolf Reference Woolf, Johnson and Parker2009), the range of possibilities in the practice of writing and reading was as huge as the supports of writing were diverse.

The case of epigraphy is specific in this matter. Inscriptions were not necessarily understood by everyone, especially Latin inscriptions in the Greek-speaking East. This raises the question of the capacity of the local stone-cutters to write/engrave inscriptions in Latin (textual approach), and of the recipients to read the messages they conveyed (meta-textual approach) (Bodel Reference Bodel, Bruun and Edmondson2015; Cooley Reference Cooley2002; Corbier Reference Corbier2006; Harris Reference Harris1983). Moreover, the ability to comprehend Latin inscriptions also required skills in decoding abbreviations and a contextual knowledge of Roman realities such as the cursus honorum: in the specific case of official inscriptions, literacy was therefore also based on an almost specialised ‘epigraphic culture’ or ‘epigraphic literacy’, and not just on an understanding of the language – a problem which also arises in the cases of transcription of Latin words in Greek, as we shall see below, and of translation from Latin to Greek in the bilingual inscriptions.

Cases of bilingualism in Cyrenaica have been explored in several studies by C. Dobias-Lalou (Reference Dobias-Lalou, Biville, Decourt and Rougemont2008a), mainly based on the documents relating to Cn. Cornelius Lentulus Marcellinus in 67 BC (P. 100–101), the restitutio agrorum boundary-markers (also Dobias-Lalou Reference Dobias-Lalou and Brunet2008b) and the edicts of Augustus (C.101), looking at loanwords (Dickey Reference Dickey2023), borrowings, variations and a number of linguistic features between Greek and Latin, and there is no point in dwelling on these aspects here. I will just focus very briefly on three topics: the poor command of Latin (such as incorrect cuts, abbreviated translations or grammatical errors) that can be seen in some documents; a few examples of Latinisms in the Greek inscriptions of Cyrenaica; and the relationship between dialect, koine and Latin in the region.

In the corpus, there are indications that local stone-cutters did not necessarily understand the Latin they were engraving. Sometimes, the layout involves an awkward division of the words, certainly due to a poor command of the new language. The dedication of the theatre at Apollonia at the time of Domitian is a good example thereof (A.48). The inscribed blocks, arranged in two registers, were distributed in seven bays interrupted by pilasters. The words are cut irregularly, especially the first ones, the last letter of a word being sometimes inscribed on the next block: Imp(erator) Caesa|[r diui] Vespasian|i f(ilius) ⟦Domitianus⟧ | [Aug(ustus)] etc., ‘Emperor Caesar Domitian Augustus, son of the deified Vespasian’. A fragment of an imperial titulature from Berenice shows the same irregularity (B.42). It must have been a consequence of miscalculation of word length and letter size by stone-carvers; the blocks were nevertheless used on the theatre façade, a sign that the inhabitants of Apollonia may have had little understanding of the meaning of the dedication, or, at any rate, were not offended by such clumsy Latin.

In other cases, translations can be approximate or incomplete: a dedication to Hadrian (C.102) gives his full imperial titulature in Latin, but not in Greek: Nerva is not designated as diuus, Hadrian's titles (Augustus, pontifex maximus, consul) as well as his tribunician power have been omitted and the dedicating power, in this case the city of Cyrene, is not mentioned – although we cannot rule out a desire simply to shorten the Greek section.

Grammatical mistakes can also occur. One example is the already mentioned bilingual epitaph of the libertus G. Cascellius Mommus, in which the cognomen is rendered as Μωμμους instead of Μωμμος, which was expected in Greek for an anthroponym in the nominative: here the name has been transliterated purely phonetically, without understanding that Mommus was in the nominative case in Latin. Other transcription errors, spelling mistakes and translation variations can be found in the corpus, but it would be tedious to list them all.

Influences from Latin epigraphic habit can be found in the Greek inscriptions of Cyrenaica. The clearest case is that of abbreviations: δ(ήμῳ) Ῥ(ωμαίων) for p(opulo) R(omano) in several cippi of restitutio agrorum; ψ(ηφίσματι) β(ουλῆς) for d(ecreto) d(ecurionum) in P.145; λεγ(ιῶνος) for leg(ionis) in M.215; and more generally for Latin praenomina or for imperial titulatures inscribed in Greek. More generally, Greek inscriptions show new linguistic features under the influence of Latin, such as strong variations in the Greek transcription of Latin anthroponyms (Dobias-Lalou Reference Dobias-Lalou, Bureau and Nicolas1998a and Reference Dobias-Lalou, Biville, Decourt and Rougemont2008a, 159–60). However, this was in no way original in the context of the Eastern Roman world, where Greek simply adopted Latin traditions or was influenced by them.

Some documents issued by the imperial chancellery were obviously written in Latin and then translated into Greek locally. In the dossier of imperial letters and documents (C.163), there is a noticeable difference between those emanating from Hadrian, the Philhellenic emperor, which are in good Greek because he may have dictated them directly in that language, and those from Antoninus Pius, in which Latinisms are much more prevalent (Dobias-Lalou Reference Dobias-Lalou, Biville, Decourt and Rougemont2008a, 156 n. 2) – this may also be due to the fact that different ab epistulis Graecis were involved, Pius' not being quite as good at producing idiomatic Greek as Hadrian's.Footnote ³¹ Indeed, there are quite a few Latinisms in the Greek inscriptions from Cyrenaica, for example in the boundary-markers of restitutio with the removal of the Greek article under the influence of Latin (Dobias-Lalou Reference Dobias-Lalou, Biville, Decourt and Rougemont2008a, 164).

It is not uncommon either to find transliterations of Roman realities directly into Greek,Footnote ³² for example, κɛ́νσωρ (censor, M.199), κορουλής (curulis, C.223), κɛντυρίων (centurio, M.5, M.11, M.18), λɛγίων (legio, C.573; cf. λɛγ(ιῶνος) Εἰταλικῆς for leg(ionis) Italicae, M.215), βɛνɛφικιάρις (beneficiarius, P.380), βιξιλλατίων (uexillatio, M.55), κουράτωρ (curator, M.49?, M.55), δɛκ(ουρίων) (decurio, C.195?), στράτωρ (strator, M.3, M.48?), βɛτράνος/ὀɛτράνος (ueteranus, C.158, C.573, M.28, M.215, P.265), ἰνδικτιών (indictio, in the Christian inscriptions M.134, M.135, M.241) or πρηπότης which captures the Latin praepotens (B.76, l. 6). Here again, these transliterations are in no way original to other Greek inscriptions from the Roman East.

The simultaneous presence in Roman Cyrenaica of an epichoric dialectal Greek that is still very much present whenever an initiative emanates from the city (Dobias-Lalou Reference Dobias-Lalou1987b; Reference Dobias-Lalou1994), of the koine – which became the majority language in the region's inscriptions in the first and second centuries AD –, of general developments specific to the Greek language (iotacisms, etc.), as well as of Latin, created a particularly diversified and fertile linguistic landscape during the Empire. Even if the use of a particular idiom is generally restricted to specific categories of texts (Latin mainly for official documents, for example, as seen above in section II), this complexity nonetheless makes the interpretation of certain linguistic features sometimes problematic (Dobias-Lalou Reference Dobias-Lalou, Biville, Decourt and Rougemont2008a, 156–59, on the edicts of Augustus C.101). Some texts could even be described as ‘hybrid’. A catalogue of priests of Apollo which includes the cursus honorum of P. Sestius Pollio is interesting in this respect. This individual, who belonged to the only senatorial family from Cyrenaica,Footnote ³³ became legate of the province around AD 100, then held the eponymous priesthood of Cyrene in AD 111/2:

In this text, the Roman magistracies (quaestor = ταμίας, aedilis = ἀγορανόμος, legatus = πρ̣ɛσβɛυτὴς) are translated with the Greek official equivalent, except for the adjective κορούλης (curulis) which is transliterated from Latin although a well-attested Greek equivalent existed.Footnote ³⁴ It is also worth noting that the Greek text uses the very local Cyrenaean word ἄμναμμος for ‘grandson’ (Dobias-Lalou Reference Dobias-Lalou1998b), although the rest of the text is in koine. This cursus illustrates the many ways in which different realities can be transcribed, depending on how each individual wishes to reflect his identity – and it also mirrors the hybrid ‘epigraphic habit’ of Cyrenaica itself.

The possibilities are actually manifold. Greek terms are found in Latin texts, for instance choria (Greek χωρία) in A.54, ca. AD 75 (Dobias-Lalou Reference Dobias-Lalou, Biville, Decourt and Rougemont2008a, 165). In the opisthograph stele bearing the dedication of the cisterns by the proconsul [Q.] Pomponius Naeuianus and the quaestor Silius Haterianus around AD 165, the Latin text on side (a) is translated into koine, as it emanates from the Roman authorities, whereas on side (b), dialectal Greek is used, as it is the city of Cyrene that is presented on this side of the stone as the instigator of the construction (C.166–67). Even if C. Dobias-Lalou observes that reciprocal influences are fairly limited, these few examples nevertheless show how inscriptions can develop a ‘hybrid’, almost ‘trilingual’ (koine/dialect/Latin), character that can be further accentuated by influences from other parts of the Roman world.

V. Vocabulary peculiarities and external influences

The serialisation of documents of a similar or identical typology, thanks to the online corpus, makes it possible to identify here and there a few peculiarities in the Latin inscriptions of Cyrenaica and to detect influences from the epigraphic habit of other Roman provinces. A first example is the dating formula used by one of the province's governors.

A milestone discovered in 1947 in the southern necropolis of Cyrene, at the start of the road leading to the sanctuary of Asklepios at Balagrae, gives the name of the proconsul of the year AD 46/7:Footnote ³⁵

The dating of Caesernius Veiento's government is introduced by the formula annus followed by the anthroponym in the genitive and then the title proconsul, also in the genitive. This formulation is an unicum in the Cyrenaican documentation on governors and, to my knowledge, is known only in inscriptions from the African provinces. Examples can be found in the second century AD in Hippo Regius,Footnote ³⁶ in the ciuitas Vcres,Footnote ³⁷ in Thugga,Footnote ³⁸ in the municipium Mactaritanum,Footnote ³⁹ in modern KairouanFootnote ⁴⁰ or in Limisa.Footnote ⁴¹ The question arises as to whether this formula has any particular significance. For R. Syme (Reference Syme1968, 100), the different forms of dating were variable and of little importance. According to J. Kolendo (Reference Kolendo and Panciera1982, n. 34), the purpose of the dating form is only to identify a particular event. M. Dondin-Payre (Reference Dondin-Payre1990, 343 and n. 40), on the other hand, believes that these different formulas were not chosen at random, that they went beyond simple dating and that they therefore possessed a more important meaning by expressing the power of the proconsul and his direct participation in the dedication. For X. Dupuis (apud AE 2004, 1675) the formula, when associated with an imperial titulature as in our inscription, suggests that the governor's activity should be dated to the end of his term of office. These debates are not resolved by the Cyrenaican inscription, which nevertheless shows that the official epigraphy of the region was influenced early on by that of the African provinces. It is possible that Caesernius Veiento held an official position in Africa before becoming governor of the province; some African members of his consilium, for instance the accensus or scribes, could have accompanied him to Cyrene and thus reproduced the epigraphic habit of Africa Proconsularis.

A second example of the influence of epigraphy from the western provinces can be found in the epitaphs of soldiers from the Roman army. Two funerary stelae from Cyrene read as follows:

The general habit of mentioning years of service by stipendia is replaced in these two texts by the plural aera, which was particularly common in the Hispanic provinces in the first century AD.Footnote ⁴² The two soldiers precisely belonged to a cohors Hispanorum – most probably the cohors I Hispanorum equitata Cyrenaica – which had been levied in Hispania and sent to Cyrenaica, perhaps after the first Jewish revolt instigated by Jonathas and crushed by the proconsul Catullus around AD 73–75.Footnote ⁴³ T. Pompeius Ligyrus and M. Aemilius Macer Turanicus must therefore have been Spaniards who had brought with them the epigraphic habits of Hispania in Cyrenaica.Footnote ⁴⁴

VI. Private epigraphy and multiple identities: a case study of the Latin funerary inscriptions of Cyrenaica

Examination of Latin epitaphs provides interesting information about Cyrenaican society in the High Empire and the concrete funerary practices of the population, even if the sample is very limited with just over 20 texts (Table 2).

First of all, we need to consider a number of military funerary inscriptions from Cyrene, Ptolemais and Taucheira (C.168, C.552, C.726, P.67, P.220, P.326, T.18, T.365), which were erected by soldiers or veterans of legions or auxiliary units of the Roman army. These deceased were either Cyrenaica-born soldiers who had returned to their homeland after their years of service, or soldiers from all over the Empire who had come to serve in the province, for example from the Iberian Peninsula (C.552, C.726; see above in section V). In the case of the Cyrenaicans, the use of Latin is probably a consequence of their time in the army, of the influence of the regions of the Empire where they were posted and of their new status as Roman citizens. For soldiers from the western provinces, Latin was the obvious choice, and a few texts even present distinctly Latin funeral forms, such as the invocation Dis Manibus (sacrum) (P.67, T.365).

A comparison between pre-Roman and imperial funerary inscriptions in Greek also reveals changes in the wording, which reflect the borrowing of Latin habits in Greek texts. The traditional local practice is to give only the year of death, the name often followed by the patronymic and then the age of death in the following sequence: L (= ἔτους) + anthroponym + L (= ἐτῶν) (the examples are numerous and can easily be found in the corpora IG Cyrenaica and IR Cyrenaica 2020). A few exceptions display also a verb and/or the name of the person who had the tomb built, but some formulas found elsewhere in the Greek world, for example ἐνθάδɛ κɛῖμαι/κɛῖται, are absent from pre-Roman Cyrenaican epitaphs – except in epigrams, see IG Cyrenaica Verse 006. On the contrary, such formulas multiplied in the Roman period, without ever surpassing the minimum standards mentioned above. They can be found in several epitaphs such as C.185, C.607, ll. 1–6 (ἐνθάδɛ ἐτάφη), in the Christian inscriptions M.235 and C.568 and in several epigrams as well, e.g., IG Cyrenaica Verse 010, 011, 020, 043, 047. This might well be an influence of h(ic) s(itus) e(st) recorded in a few Latin funerary inscriptions (C.552, C.613, P.379[?]), but to draw conclusions is all the more challenging as we can sometimes observe opposite influences: the bilingual funerary inscription C.614 presents a Latin form aligned with the traditional Greek one (anthroponym followed by the age of death), without Roman-style funerary formulas.

If we take these considerations a step further, we may well ask whether certain epitaphs that are particularly well-developed but written in Greek have been influenced by Latin epigraphy, particularly in the military field. Two veteran epitaphs (C.573 and another recently published in Chevrollier, Dobias-Lalou and Hussein Reference Chevrollier, Dobias-Lalou and Hussein2021, 29–32, no. 3) show a Latin-style military ‘cursus’ very different from the simple local forms. Similarly, should we understand the formulas concerning the protection of tombs (C.605, C.735, P.295, P.404) as a Latin influence of the iura sepulchrorum?Footnote ⁴⁵ The limited corpus makes it impossible to be positive on all these points: the influence of the army, in particular, is difficult to quantify and, in fact, may have remained quite marginal, as Cyrenaica was only guarded by auxiliary troops which were never very numerous (Paci Reference Paci1994, 253).

The majority of Latin funerary inscriptions pertain to individuals possessing the duo (C.728) and the tria nomina (C.512, C.623, P.283, perhaps A.17 and those for the soldiers mentioned above), showing that they were mainly immigrant families from Italy or the western provinces. Others concern freedmen (P.281) who also had their origins abroad. It is more difficult to understand the status of bearers of the tria nomina with a Greek cognomen but who chose to inscribe their epitaph in Latin (M.258, P.390) or of certain individuals who were probably immigrants but who opted for Greek as their epigraphic language (C.731).

In the particular case of bilingual inscriptions, the order of the languages is of some importance. In official texts (boundary-markers, milestones, official building inscriptions and dedications), Latin naturally comes before Greek, since these are documents issued by public authorities. On the restitutio agrorum cippi, Greek often appears on a minor side, when the texts are inscribed on two (or three) different sides of the stone. Bilingual funerary inscriptions also generally have Greek preceded by Latin: see the epitaph of Mommus cited above, a very fragmentary text from Taucheira (T.18) or the inscription of P. Iodius Apol(-), who was however most probably of local origin (Dobias-Lalou, Reference Dobias-Lalouin print). In the northern necropolis of Cyrene, a burial of two brothers, freedmen of the same master (tomb N.11, cf. Thorn and Thorn Reference Thorn and Thorn2009, 28), features two inscriptions: C.554 is bilingual with the Latin coming before the Greek, while C.555 is in Latin only and is for their sister who died later (her name was added by another hand). The Latin names and the use of this language show that this family probably originated in Italy and had settled in Cyrenaica for a long time, if we consider that several members of the household died there. Nevertheless, the use of Latin has endured and always comes first.

Again in the northern necropolis (Thorn and Thorn Reference Thorn and Thorn2009, 141), the Octauii tomb features bilingual inscriptions (C.613) that are exact translations of each other, except for the formula h(ic) s(itus) e(st), which is not translated into Greek (see also above on the formula). The case of the Blaesii tomb (C.614) seems a little different: only one of the inscriptions, that for L. Blaesius Rusticus, is bilingual, while those for two other members of the lineage are in Latin only. In all these examples, the use of Latin shows that it was indeed this language that prevailed among the commissioners of the epitaphs. The order of the languages, with Latin first and Greek in translation below, is not insignificant: it was certainly intended to emphasise the western (Italian?) origins of these families, some of whom had been living in the region for a long time, but also, as a consequence, to demonstrate a certain prestige and social differentiation through the use of an idiom that was certainly little understood or at least poorly mastered by the majority of the population. However, translation into an epigraphic language (Greek) more widely understood locally was seen as essential from the moment that funerary practices required passers-by to be able to read aloud the inscriptions in the necropolis – this also related to the complementarity between literacy and orality in the ancient world, but these considerations are well beyond the scope of this paper.

Throughout its history, Cyrenaica has been home to funerary monuments that are unique to the region, including the faceless funerary busts from the Classical and Hellenistic periods, the ‘Roman-Libyan’ funerary portraits that replaced the former from the Roman period onwards (Belzic Reference Belzic, Boschung and Queyrel2019) and the anthropomorphic stelae as well. The latter take the form of small, inscribed or anepigraphic limestone stelae on which a head, shoulders or a bust can be recognised, however sculpted in an unrealistic manner. They come from the hinterland, from the countryside surrounding the cities, and bear witness to populations less steeped in Graeco-Roman culture but who nevertheless sought to imitate it through sculpture and epigraphy. Funerary formulas are generally kept to a minimum, and include the name of the deceased, his or her age, sometimes the date of death and a verb (e.g., ἐτɛλɛύτησɛ).Footnote ⁴⁶ Many of these stelae have inscriptions in Greek, often badly engraved, indicating an imperfect command of writing and/or of the language. But it is particularly remarkable that two such stelae are inscribed in Latin. The first (P.390) comes from near Ptolemais and reads as follows: C(aio) Papirio Diomedi Cornelia Polla amico, ‘To G(aius) Papirius Diomedes, Cornelia Polla (scil. erected this), to her friend’. The woman has the traditional Roman onomastic formula while the deceased has the tria nomina with a Greek cognomen. The word amicus could refer to an informal marriage between the two. The second one (M.258) was discovered in the area of Limnias (modern Lamludah), east of Cyrene, and bears the following text: C(aius) Iulius Epafroditus uixit annos LXX, ‘G(aius) Iulius Epaphroditus, lived 70 years.’ Here again, the individual has the tria nomina with a Greek cognomen. These two anthropomorphic stelae bear witness not only to the penetration of Latin into the countryside, but also to cross-cultural influences, in this case between a funerary monument in the Libyan tradition and the choice of Latin. It is possible that G. Papirius Diomedes and G. Iulius Epaphroditus were freedmen working on agricultural estates belonging to wealthy Roman citizens, or freedmen or descendants of veterans who adopted a local form – the anthropomorphic stela – for their tombs. The very vague date attributed to these documents (first–second centuries AD) makes it impossible to say for sure. But the choice of Latin by these two individuals with Greek cognomina or by their relatives for their epitaphs on a monument probably inspired by Libyan customs and erected quite far from the poleis says something about the ethnic and linguistic diversity of the region under the Empire, as well as the reciprocal cultural and linguistic influences between Libyan, Greek and Latin traditions.Footnote ⁴⁷

These examples chosen from the field of funerary epigraphy illustrate the interplay between languages, as well as lexical and formal borrowings, even if the small Cyrenaican corpus does not allow us to go beyond mere hypotheses. In the bilingual inscriptions, the choice to inscribe the Latin first certainly reflects a desire for prestige, while the anthropomorphic stelae of Libyan tradition engraved in Latin reveal the multiple identities of the inhabitants of ancient Cyrenaica. In this sense, private funerary epigraphy demonstrates the ‘ethnic and linguistic porosity’ (quoted from Sarrazanas Reference Sarrazanas and Roure2023, 79, about Philippi) between groups of different traditions, as well as the complexity of multilingualism in ancient Cyrenaica.

Concluding remarks

The Latin inscriptions of Cyrenaica come mainly from the cities and are documented between 67 BC and around AD 375, with periods of greater presence, for example in the Augustan-Tiberian and Hadrianic periods, although they never surpassed those written in Greek – like in the rest of the Greek East, Latin never became the dominant language of communication between the Roman government and the inhabitants of the provinces. Most of these texts are official documents from the Roman authorities: activities of proconsuls, dedications to emperors, documents relating to the imperial cult, boundary-markers and milestones.

From a general point of view, this survey confirms the conclusions of G. Paci (Reference Paci1994). However, my intention in this article was also to examine the Latin inscriptions of Cyrenaica in the light of more recent concerns which include, among others, the literacy of the population, the bilingualism and multilingualism of the region, the linguistic hybridity that can be seen in certain texts, the (unmeasurable) importance of the Libyc cultural background, the influence of neighbouring provinces on the language and the expression of identities through the choice of an epigraphic language. In this respect, funerary texts in Latin or bilingual, although scarce, are particularly interesting to study as the choice of a specific epigraphic language bears witness to the identities claimed by individuals, as well as to the cultural and linguistic permeability between the different linguistic groups that inhabited the region in Antiquity. The Latin inscriptions, which have been made considerably easier to consult thanks to the publication of the IR Cyrenaica 2020 corpus, allow us to delve into the heart of Cyrenaican society in the Roman era, which remained largely Greek-speaking, but in which Latin played a significant role for more than four centuries.

This study also echoes the questions raised by historians about the presence of Greek in the Latin-speaking western provinces (Chausson, Hostein and Rossignol Reference Chausson, Hostein and Rossignol2022) and, more broadly, about the many languages spoken in the Empire and their relationship with Latin and Greek (Bérenger-Badel Reference Bérenger-Badel2004, 48–50; Coltelloni-Trannoy and Moncunill Martí Reference Coltelloni-Trannoy and Moncunill Martí2022; Harris Reference Harris1989, 175–90; Neumann and Untermann Reference Neumann and Untermann1980). The specific case of Cyrenaica, with Greek in a numerically dominant position, Latin in a numerical minority but with a hierarchically superior position and both against the background of a ‘ghost language’ (Libyc), can also help us to better understand these crucial issues in Roman history.