The Languages and Epigraphies of Iron Age and Roman Gaul* (2024)

Alex Mullen

5.1 The Background

Roman Gaul covered multiple communities, varied landscapes, and climates. At various times it stretched across what is now France, Belgium, Luxembourg, most of Switzerland, parts of the Netherlands and Germany, and even northern Italy, though Gallia Cisalpina was incorporated into Italia in the first century bce. The creation of Transalpine Gaul, ‘Gaul-on-the-other-side-of-the-Alps’, and the focus of this chapter, began in southern Gaul with the conquest of what would become Gallia Narbonensis. The south was drawn into the Roman sphere of interest during the Punic Wars, and was viewed differently from the rest of Gaul, at least in the earlier generations of Roman rule. Deemed ‘more like Italy than a province’,1 at some point in the early years of the second half of the first century bce Narbonensis was granted the ius Latii, which allowed access to citizenship via magistracies, and in 14 ce all citizens of Narbonensis were given the right to stand for magistracies at Rome, thus opening up the possibility of senatorial careers. Until the end of the first century ce, Narbonensis provided more equestrians and senators than anywhere in the Empire outside Italy. Conversely, the proposition of Claudius, himself born at Lugdunum, similarly to promote citizens from the Tres Galliae was opposed, with the opposition leaning on the deep-seated fear of the terror Gallicus, stemming from the trauma of the sack of Rome in 390 and more recent incidents, so only the Aedui were at first granted this privilege.2 Although even in Narbonensis some social groups, isolated communities, and more mountainous areas apparently clung to existing ways of life, the transformations there were generally faster and more widespread. The differential outcomes for Narbonensis versus the regions to its north can be linked to its long history of interaction with Roman traders, negotiators, soldiers, and other groups, as Rome sought to establish a foothold in the Iberian Peninsula; to its lengthy imbrication with what has been described as the Mediterranean koine, a dense network of trading, socio-political, and cultural interactions from groups who criss-crossed the Mediterranean and settled around its edges; and the network of colonies with nodes only on average 115 kilometres apart.3

Caesar naturally focused on the area that concerned him: the Tres Galliae, or Gallia Comata. The three parts he described, Aquitania, Celtica (later Lugdunensis), and Belgica were to become a feature, in different permutations, of the Roman division of Gaul until the reorganization under Diocletian. Caesar’s ethnographic opening to the Gallic Wars relays a crucial piece of information: the locals differed in language, institutions, and laws: hi omnes lingua, institutis, legibus inter se different (BGall. 1.1). Without doubt even the tripartite organization was a simplification, as Caesar indicates with his singling-out of various peoples of the three regions as being culturally different from, and often far from harmonious with, their neighbouring communities.4 During the Gallic Wars, the communities of Gaul were unable to unite to defeat the Roman aggressor, and indeed Caesar was able to recruit from Mediterranean Gaul for the Gallic campaigns. Indeed, probably the concept ‘Gaul’ did not mean much to the majority of its inhabitants, especially in the Iron Age and early Roman periods.5 The archaeological remains indicate a high degree of regional variation in the so-called La Tène material culture.6 The question whether the Gaulish language unified this area—and indeed whether the broader Celtic language family linked this area with others beyond it—must be approached with caution: the fact we can describe the linguistic resources as ‘Gaulish’ (the Celtic language of Gaul) and ‘Celtic’ (the linguistic family to which Gaulish belongs) does not mean the ancient communities could do the same, and we should be extremely cautious of the potentially anachronistic view of ‘one language, one nation’.7 Speakers of Celtic languages might have understood one another more easily than they did speakers of non-Celtic languages such as Iberian, but we should not automatically assume any deeper links. This world was composed of dozens of complex, sometimes fractious, and relatively migrant communities, whose ‘tribal’ names on the eve of conquest were transmitted to us by the Roman elite. These early texts may well have misrepresented (deliberately or not), and fixed at points in time, some of the groupings; the precise composition and interaction of the communities are not fully understood and require cautious interpretation of archaeological remains in combination with epigraphic, numismatic, and literary testimonies.8

Given the complexity and regionality across Gaul in the later Iron Age and under Empire, it is perhaps unsurprising that there have been few attempts to write detailed synthetic histories.9 Most academic works on a wide scale are understandably restricted to a province or region of Gaul,10 and there has been a lack of integration of the study of languages (and to a lesser extent epigraphy) into the historical analyses. Experts in Gaulish language and epigraphy have traditionally worked in Celtic/Linguistic departments, and epigraphic research into Latin inscriptions, and the far fewer Greek ones, has tended to focus on the interpretation and publication of individual texts or regional collections. A gulf has formed between the focused epigraphic outputs and the more historically driven debates on epigraphic culture more broadly.11 It is time to revisit this following the boom in digital epigraphy and recent cross-disciplinary cooperation.

5.2 Multilingualism and the Advent of Literacy in Gaul: Etruscan, Iberian, and Mediterranean Entanglements

To understand the development of the epigraphies of Gaul, one has to start in c.600 bce with the foundation of Marseille on its south coast, when we increasingly see Mediterranean Gaul as a meeting ground for speakers of Greek, Etruscan, Iberian, Italic, Phoenician, Punic, ‘Ligurian’, Gaulish, and other Celtic languages,12 many of whom were likely to have been bi- or multilingual. The epigraphic landscape can be roughly divided, at least until the time of Augustus, into two broad zones: to the west of the river Hérault, the Levantine or North-eastern Iberian semisyllabary, a Palaeohispanic script, dominates;13 to the east, the picture is influenced more by the presence of the Greek script (Fig. 5.1).

The first non-indigenous groups to settle widely in southern Gaul were the Phokaian Greeks, who peppered the coastline and river valleys with settlements of various sizes and types from Nikaia (Nice, Alpes-Maritimes) to Emporion (Empúries, Catalonia) and beyond.14 The Greeks in southern Gaul, mirroring the apparent practice at the metropolis, Phokaia (Asia Minor), left few epigraphic remains.15 The pre-imperial Greek epigraphic record of southern Gaul, set out in Inscriptions grecques de la France (IGF), which is largely, but not entirely, restricted to lapidary inscriptions, numbers only approximately 25 out of a total of 169, of which most can be linked to Greek settlements, the coast, and trading contexts. An update to this catalogue, including non-lapidary inscriptions, did not change the picture considerably,16 though a few ‘non-Greek’ sites appear.17 This seems to support what we see in other evidence from the region—namely, that Phokaian Greek interaction with local populations, including the Iberians, may not generally have been of a type that facilitates particularly extensive and deep linguistic interactions and many colonial Greeks may have focused primarily on their settlements and Mediterranean, rather than continental, entanglements. New finds may, however, change this picture, and the Aristaion near Olbia on the south coast, whose numerous texts on pottery have still not been comprehensively published, demonstrates apparently close interactions between Greek- and Celtic-named sanctuary visitors.18

These Mediterranean-facing interactions can be traced on the Ionic Greek Pech Maho (Aude) tablet (IGF 135), dated to around 425 bce, a private document from a commercial context, which involves Iberian-named individuals, and perhaps others from local population groups, as witnesses.19 Strikingly, it also contains, on the other side of the metal, an Etruscan text that can be dated to 450 bce and appears not to be directly related.20 The place name Marseille is attested (Mataliai), as is a possibly Latin-named individual Utavu (Octavius?) and an Etruscan individual Venelus. The Etruscan presence in the region is most securely attested at Lattara (Lattes, near Montpellier),21 a settlement established around 500 bce, which appears to have welcomed a range of population groups.22 Sixth- and fifth-century inscriptions in South Etruscan script from Lattes include short texts, including an Etruscan version of a Celtic name found twice in the genitive: Ucial. One North Etruscan-script text on an Attic cup, found at Ensérune and dated to 425–375 bce, also contains a Celtic name in the genitive: Smeraz.23 Etruscan texts, and some texts that seem to be in Etruscan script but are not assignable to any language, are also recorded at Marseille, Saint-Blaise, La Monédière, the wreck Grand Ribaud F (off the coast near Olbia), and possibly at Béziers.24 The discovery at Lattes of two alphabets of a later date, from around 200 bce, with one followed by the term knax, which also appears on the educational papyri from Egypt, indicates that instruction in the Greek alphabet, at least, may have been occurring in coastal areas of south-western Gaul.25 Whether the instruction was ever formalized is impossible to tell, and it is unclear whether it involved Greeks and/or other population groups. At a settlement as multicultural as Lattes, we need more than this meagre evidence to understand a potentially complex picture.26

Contacts between the Iberian world and southern Gaul can be illustrated by the transfer of production practices and techniques beginning in the Bronze Age.27 From the fourth century bce, these included writing. In south-western Gaul, Palaeohispanic epigraphy appears as ownership marks on imported ceramics and stamps on dolia, on a dozen sheets of lead, coins, and, very rarely, stone. The epigraphy is largely a feature of the third to first centuries bce and rapidly disappears under the Empire. Of the dozen or so Iberian-script-producing settlements, one provides substantial numbers of texts: Ensérune has produced nearly 400,28 and offers both Iberian and Celtic personal names (e.g. katubaŕe Hesperia HER.02.373).29 A further three, Elne, Château-Roussillon, and Vieille-Toulouse, offer similar numbers, that is 33, 34, and 35, respectively, and Pech Maho provides 53. Further to the south, close to what becomes the boundary between Gallia and Hispania in the mountainous Cerdagne region, ten or so main sites plus some dispersed finds yield 46 inscribed rocks with more than 180 texts, which appear to respond to specific local issues.30 The presence of the North-eastern Iberian script in these locations is one of the supporting arguments for Javier de Hoz’s hypothesis that Iberian was a lingua franca, stretching from the south-eastern Iberian Peninsula into south-western Gaul,31 particularly in the trading realm.32

In the middle of the second century bce (between 175 and 150/130), direct evidence of close contacts and trade between Iberian-, Latin-, and Gaulish-speaking groups in Vieille-Toulouse (Haute-Garonne) can be found on amphorae marked with North-eastern Iberian characters in red paint.33 Several interpretations have been advanced for the sequences linking Latin, Celtic, and Iberian names with quantities, dates,34 prices,35 lot numbers, and, more recently, tolls.36 One is written with the Latin alphabet and gives the Latin name Q. Ofeli, followed by a sequence of Iberian metrological signs, combining knowledge of two writing systems. The texts are on Graeco-Italic amphorae, but are associated with cookware of Massaliote origin. On the nearby site of the Quartier Saint-Roch, other inscriptions dating to around 125–100 bce in levels with inscriptions in North-eastern Iberian script include Greek names also attested at Marseille.37 Even if these short inscriptions cannot allow a firm conclusion about the nature of the interactions, North-eastern script was certainly being used at Vieille-Toulouse at a relatively late period for writing Iberian,38 and for transcribing local names,39 alongside Latin and Greek.

Coin legends from Languedoc, dated to the second and first centuries bce, similarly illustrate the interactions between the Gaulish, Greek, and Iberian worlds. The bronze ‘ΛΟΓΓΟΣΤΑΛΕΤΩΝ’ issues probably originate from Ensérune, and some have two legends, one in the North-eastern Iberian script and the other in Greek (e.g. biuŕbi/ΛΟΝΓΟCΤΑΛΗΤΩΝ Feugère and Py (2011) IBL-2367/2368, 150 to 75 bce). On two of these types we find on the reverse Celtic names in Greek script (ΒΩΚΙΟC IBL-2363; ΛΟΥΚΟΤΙΚΝΟC IBL-2369, both 150 to 75 bce). These coins, restricted to Languedoc and the Aude Valley up to Toulouse, are imitations of the Massaliote bronze coins with tripods.40 Other coins found on the coast of Languedoc contain Celtic names of men identified through the Greek term βασιλεύς (e.g. ΒΙΤΟΥΚΟC IBL-2408; ΚΑΙΑΝΤΟΛΟ IBL-2425, both 125–75 bce). The ‘Gaulish king’ set have an iconography that links them to the Iberian issues of kese/Tarragona and perhaps of iltirta/Lleida (examples with a bounding lion, though this could equally be inspired by Massaliote coinage),41 and their metrology and denominations (as, semis) can be linked to the Roman system.42

With the exception of a few isolated cases, such as that of Vieille-Toulouse, however, the North-eastern Iberian script in south-western Gaul fades from around the time of increased Roman interest in the area from the second century bce. Palaeohispanic epigraphy disappears totally in the space of a few decades, and within Latin epigraphy in south-western Gaul very few inscriptions mention Iberian-named individuals, with a series of notable exceptions in the Pyrenees.43 The Gaulish-speaking communities from this period follow a different trajectory, adopting the Greek (‘Gallo-Greek’) and then the Roman script (‘Gallo-Latin’) to produce their own epigraphy, and appearing in large numbers in imperial-period Roman epigraphy. Similar differences in the epigraphic trajectories of communities—for example, Iberian- and Celtic-speaking—can also be witnessed in the Iberian Peninsula (Chapters 2–3).

5.3 Writing Gaulish: Gallo-Greek and Gallo-Latin Epigraphies

At about the time that direct linguistic evidence for Iberian is waning in the south-western zone, Gaulish-speaking communities in the south-eastern zone, and especially in the lower-Rhône delta, are entering the epigraphic stage with Greek script to write Gaulish, an epigraphy known as ‘Gallo-Greek’ (see Fig. 5.3). This was traditionally attributed to the result of a process of ‘Hellenization’ and of contacts between ‘elites’ at Marseille.44 More recent research has argued that the distribution of texts, the time period, the lack of Greek linguistic features in the texts, and the evidence of Italic languages within them mean that we should see the phenomenon of Gallo-Greek as part of a more complex process of local–Mediterranean interactions, with influences from a range of actors, including colonial Greeks of course, but also various Greek and other linguistic groups from the Italian peninsula and elsewhere. The epicentre of Gallo-Greek lies on the confluence of several key fluvial routes, with the Rhône running through the centre, with Marseille outside the epigraphic zone. The analysis of the forms of the script indicate that handwritten Greek may have been the model, which may also encourage us to envisage a trading context for the contact and a possible motivation for borrowing.45 The time period in which Gallo-Greek is securely attested has been pushed later by Michel Bats to start in the last quarter of the second century bce.46 Rather than seeing Gallo-Greek as the result of Hellenization of indigenous groups, we might rather see the timing of this phenomenon as in part the result of internal changes within the indigenous societies and the intensifying Roman role in the region.

Gallo-Greek comprises largely short lapidary inscriptions, graffiti on pottery, usually just containing names and a small number of repeated verbs, and coin legends (see Fig. 5.3). Greek–Macedonian staters arrived in Gaul between the 320s and 270s bce. The increasingly widespread adoption of coinage in local communities from the second half of the third century can be linked to the changing economic landscape of the region. In the north of Gaul the earliest local coinage is gold and imitates, among others, the staters of Philip II of Macedon, initially copying ΦΙΛΙΠΠΟΥ accurately.47 In the south the local coinage is silver and, from the third century bce, around the area of Narbonne, imitates the drachma of Emporion, and further towards the Atlantic, on the Aude-Garonne axis, copies the coinage from Rhoda (both Greek colonies on the north-east coast of Spain, though both types have their distribution in Gaul not Iberia). In the south-west the so-called monnaies à la croix spread to cover much of Occitanie to the mouth of the Garonne, starting at the end of the third century bce.48 However, as we have seen, it is only from the second century that local language legends appear, with the first Celtic names on legends in Mediterranean Gaul in the region of Béziers. Soon afterwards a Roman denarii-based local coinage takes off in the Rhône Valley, with many deploying Gaulish legends, using both Greek and Latin script, a period of activity that does not last long after the conquest.49 Coinage sits slightly apart from broader epigraphic expression, since it is created in a restricted number of locations, and the coins themselves are made mechanistically after the creation of the dies, by a potentially small number of people. Having said that, it also reaches larger numbers than most epigraphic types and is clearly used to advertise elites and their communities.

The origins of a fully-fledged epigraphy for Gaulish probably begin properly with graffiti on ceramic and may be associated with trading contexts and internal settlement requirements. The lapidary development can be associated with additional factors in the evolving cultural environment of the later second and first century bce, including increasing numbers of ‘non-indigenous’ individuals, partly owing to incomers from the Italian peninsula. At least some of these incomers might have brought knowledge of literate practices and a growing awareness of the epigraphic habit (Section 5.4), which, as we have seen, did not appear to have been much embraced by the Phokaian colonists of the area, but also a motivation for local expression in this new world order. The non-Gaulish names are Roman/Italian type and as yet none of the non-Gaulish names in the published Gallo-Greek inscriptions is definitely Greek or Iberian.50 Hints of language contact too can be associated with the Italian peninsula, perhaps most obviously in a funerary stele with triangular summit from Velleron,51 which contains a Latin name but also a code-switch into Latin for the closing ualete ‘farewell’ within its Gaulish text, mirroring texts from multicultural centres in Italy that fix the oral greeting in stone.52 In another code-switch or borrowing we find a praetor, ΠΡΑΙΤωΡ (RIG I G-108, RIIG BDR-14-01), attested in a stone inscription at Vitrolles (Bouches-du-Rhône) (Fig. 5.2). In its incomplete state and without contextual clues (it was found built into an eighteenth-century bell tower) the text is hard to interpret, but the use of praetor perhaps indicates an engagement and investment, at least superficially, in Roman modes of organization. The praetorship was a position of power in the Roman world, and the nature of this apparently local version remains unknown.53

The main recurring formula throughout the Gallo-Greek texts from the south is ΔΕΔΕ ΒΡΑΤΟΥ ΔΕΚΑΝΤΕΝ, probably meaning ‘gave a tithe in gratitude’. Though it can be analysed as Gaulish, it can be linked to influences from a Mediterranean koine, and again communities from central-southern Italy provide all the elements and appropriate circumstances for contact-induced adoption.54 One of the three lapidary expressions of the formula from Glanum (Bouches-du-Rhône) (RIG I G-65) is dedicated by ΚΟΡΝΗΛΙΑ ‘Cornelia’ to the ΡΟΚΛΟΙCΙΑΒΟ ‘the far-hearing goddesses’. This might be linked to the close-by and beautifully carved Latin inscription to the Auribus (IAG 18) of the first or second century ce, complete with a carving of a pair of ears within a wreath.55 Loreia is the pia ministra who dedicates to the Auribus; the name Loreia is otherwise attested only at Narbonne (CIL XII 4731) and in Italia and the Alpes Poeninae. A dedication to the Bonae Deae was also found nearby that to the Auribus in house VII.56 Since this cult appears at Arles, again with a pair of ears within a wreath (CIL XII 654), and was particularly practised in Italy, and specifically Latium, we may see links between the regions and even continuity of practice, illustrated in the Gallo-Greek record by a woman named Cornelia (probably local and perhaps having received Roman citizenship) and in the Latin record by another, perhaps from, or with associations to, Italy.

Glanum was one of the most prolific sites for lapidary Gallo-Greek, alongside Nîmes and Cavaillon, all of which became significant Roman settlements. Recent work by the RIIG team has identified the importance of the individual (and perhaps the influence of the Iron Age-period cult of the hero),57 developments in urbanism (particularly urban sanctuaries), and the religious sphere in the creation and evolution of Gaulish monumental epigraphy.58 What is clear is that the lapidary habit is not taken up in all settlements in the Lower Rhône basin, and that between regions, though the texts are nearly always short and formulaic, there seems to be variation in the choice of objects and execution.59

Attestations of Gallo-Greek beyond the lower Rhône basin are fewer and concentrated in central-eastern France, with the number of graffiti on pottery outweighing considerably the lapidary expressions. Two sites provide the majority of graffiti, with Alesia offering twenty-nine examples and forty-one from Bibracte, according to a recent review—both well known as sites of local resistance in the Gallic Wars.60 The find-spots for Gallo-Greek in this region take in domestic, public, artisanal, religious, and funerary contexts, in declining order of frequency. The relatively widespread contexts for the use of Gallo-Greek stand in opposition to a tiny set of texts types, with the vast majority marks of property, and a very restricted geographical spread across Gallia Comata. It seems that commerce and length of contacts with the Mediterranean world were key drivers for the adoption of Gallo-Greek in the centre-east, just as they appear to have been in the Rhône basin, and that it may have spread from artisans to the wider population.61 Gallo-Greek in this region disappears around 70 ce.

‘Gallo-Latin’ is used for Gaulish written in Latin script and was traditionally dated narrowly to the century or so following Caesar’s conquests (RIG II.1, II.2, III),62 but its dating has been shown in recent years to extend later than previously assumed, and Lejeune’s depiction of a three phase evolution—Gallo-Greek until Caesar, then Gallo-Latin until Claudius, and finally Latin—cannot be sustained.63 In the Tres Galliae, Gallo-Latin is found alongside both Gallo-Greek and early Latin (Section 5.6): the non-Mediterranean Gallo-Greek phase can be dated to c.100 bce to 70 ce and Gallo-Latin begins in c.50 bce, meaning for the most part of the attestation of Gallo-Greek, Gallo-Latin also exists. A Gallo-Latin phase might not be attested in southern Gaul at all, since the evidence is largely restricted to legends on local coins in Latin script, some of which are likely to have been envisaged as Latin,64 and otherwise only to a funerary inscription from Coudoux BOVDIL/ATIS · LEMISVNIA (RIG II.1 L-2) and a citation within an ansate frame on a carved image of a shield within a sculpted relief on the arch from Orange (RIG II.2 L-18), about which Lambert remarks ‘il ne s’agit pas d’une véritable inscription gallo-latine, mais de la citation, en contexte latin, d’un mot gaulois’ (RIG II.2, p. 36).65

It is possible that the non-lapidary material for Gallo-Latin has not been as systematically recognized and published as for Gallo-Greek. One problem is that a few Latin letters on pottery, even if they contain a Gaulish name, will often not be assigned to Gallo-Latin but rather to Latin (and often do not speedily reach publications), whereas more of the inscriptions with Greek letters are published as Gallo-Greek if they contain Celtic elements (in part owing to the relatively small number of Greek texts from the region).66 Even though our grasp of the full extent of Gallo-Latin epigraphy is uncertain, it seems to be more geographically diffuse than Gallo-Greek beyond southern Gaul (Fig. 5.3). It also appears on a greater range of materials than that in Greek script, including spindle whorls, bricks, firedogs, and tiles, and boasts a wider range of functions, including magical (e.g. Larzac and Chamalières: RIG II.2 L-98, L-100), calendrical (Coligny (Ain) and Villards-d’Héria (Jura): RIG III), production lists (Section 5.7), accounts (Rézé: Lambert and Stifter (2012)), and jovial and/or amatory invitations to drink/fetch drinks and/or for liaisons (Banassac: RIG II.2 L-50; Autun: L-112; Auxerre: L-121). Gallo-Latin contains several lengthy texts, to which nothing currently compares in Gallo-Greek, for example the tile from Châteaubleau (RIG II.2 L-93). There are also, however, large numbers of marks of property on ceramic, with the Aedui offering significant numbers at Bibracte (along with multiple Gallo-Greek examples) and Autun.

Referring to what he calls the Middle Gaulish period—namely, the first two centuries ce—David Stifter goes as far as saying that

The functional range of written Gallo-Latin is indeed impressive, but it is perhaps strikingly so, given how few examples of Gaulish writing we have in total. We are working with published examples of Gallo-Latin, which number in the high hundreds. This could be contrasted with the tens of thousands of Latin texts over the same period and area, and yet no one today would say that literacy, even in Latin language, is a skill of a sizeable portion of the Roman provincial population. It was functionally broad based but still not common (Chapter 1). In the case of Gaulish, we might see a similar phenomenon but with a considerably smaller proportion of literates.

The very small number of ‘monumental’ Gallo-Latin inscriptions is particularly striking. Given their nature, these are more likely than texts on other media to have been recovered and published, so the small number may reflect reality. The Gallo-Latin corpus contains vastly fewer monumental inscriptions than the output of the Latin epigraphic habit over the same period,68 but significantly also proportionally fewer lapidary inscriptions than the Gallo-Greek corpus, in which the lapidary inscriptions seem to occupy a more dominant role. Arguably, though Gallo-Latin was used on a range of objects, from the Augustan period onwards stone inscriptions were seen as so intrinsically ‘Roman’ that those choosing to engage in the epigraphic habit for epitaphs and honorific inscriptions would use Latin. However, the content, function, and contexts of the handful of ‘monumental’ examples of Gallo-Latin may be instructive (Table 5.1). It is worth noting that, while the cluster in the Aeduan region (Fig. 5.2) lies in a region that is relatively active in terms of the Latin epigraphic habit, Vieux-Poitiers, Sazeirat, and Genouilly, and their vicinities do not offer a single Latin lapidary text, Plumergat simply one milestone,69 and Lezoux an almost entirely ceramic output.70

It is striking that many of these twenty-one lapidary or monumental bronze Gallo-Latin inscriptions can be assigned to the religious domain, in which indigenous languages can be particularly tenacious, a sociolinguistic dimension to which we shall return in Sections 5.6 and 5.8. One major issue in understanding these items in their epigraphic and linguistic contexts, however, is the uncertainty over their dating. Since linguists and epigraphists traditionally tended to think about language developments in phases (an indigenous-language phase followed by a Latin phase) and not to pay much attention to bilingualism, these have been dated to between the Caesarean period and the first century ce, after which point the communities would have been ‘Romanized’ enough to be using Latin epigraphy. However, our current perspective appreciates that periods of bilingualism can be lengthy, and that epigraphic expression in Gaulish might not in any case imply a lack of ‘Romanization’, but in some cases could reflect the self-confidence of a community that might want to use Roman modes of expression to convey their local language (which may, or may not, still be widely used).

Close attention to the dating of the material from Alesia has led Cazanove and Estarán Tolosa (2023) to argue that the move to an entirely Latin lapidary epigraphic landscape possibly happened later than traditionally assumed, perhaps in the second century ce.71 Alesia was part of the small Mandubii group, and later subsumed into the Aedui,72 and famous for being the location of the last stand of Vercingetorix in 52 bce. It was not, however, destroyed and became a prosperous, substantial (97 hectares), and well-furnished Gallo-Roman city. A contextual analysis of the epigraphic record shows that Gallo-Greek, both monumental and non-monumental, was being produced until at least the Neronian period; that the earliest Latin inscription may be Tiberian (based on a marble dedication probably to Tiberius (and Rosmerta?) (CIL XIII 2876), though most of the Latin votives of the sanctuary of Apollo Moritasgus date to the second century ce;73 and that the dating of the only Gallo-Latin inscription from the site, the dedication to Ucuetis (RIG II.1 L-13, RIIG CDO-01-19, CIL XIII 2880),74 cannot be fixed to the date of the so-called Ucuetis monument (late first century ce), and may even date to the second century ce, as does the Gaulish-inscribed bronze patera of relatively nearby Couchey (RIG II.2 L-133). The conclusion is that the transition to an entirely Latin epigraphic practice, dated by Lejeune to the Flavian period, might be placed in the second century. We need to avoid dogmatism on the dating of these epigraphic outputs and build our knowledge based on the constantly growing evidence. Indeed, we should be cautious about the broad period 1–100 ce dating of the objects in Table 5.1. A group of four Gallo-Latin inscriptions from the larger Aeduan region, including the Ucuetis inscription whose possible second-century dating was discussed above,75 might be taken together as adopting the frames and features of Latin epigraphy and are likely to be roughly contemporary, and therefore we might need to keep the possibility of a later date in the second century open.

5.4 The Earliest Latin

The inescapable fact of the study of Latinization in the western provinces that were brought under Roman control in the earlier period is that very little Latin epigraphic material is securely datable to the earliest phases: few items survive that can be definitively set before the end of the Republic. For Narbonensis, this is part of the reason why we cannot be sure about some of the facts concerning the establishment of the Roman province, its extent, and the dating and status of its early foundations.76 It also contributes to our difficulty in tracing the early phases of the spread of spoken Latin in the province.

Woolf kicks off his section on ‘the rhythm of change’ on epigraphy in Gaul with the sentence: ‘Virtually no Latin inscriptions of the Republican period are known from Gaul.’77 There are, indeed, relatively few early Latin inscriptions: our data set provides only 186 records of dated Republican Latin inscriptions for Gaul, of which just 12 are on stone (Fig. 5.4). However, it is also the case that several lapidary inscriptions—for example, those showing ‘Italian formulae’ (see further below)—might well date to the later years of the Republic, but these tend to be assigned broad period dates that extend to the end of the Julio-Claudian period. A review of all possible inscriptions and their archaeological contexts to produce a corpus akin (though smaller) to that of Borja Díaz Ariño’s Latin Republican inscriptions of Hispania (ELRH) would be extremely valuable.

The earliest epigraphic material, as with the Iberian Peninsula, is dominated by tituli picti, incised texts, and stamps on amphorae or other vessels.78 Ceramic is given as the material type in 166 records, and the vast majority are texts related to production and trade. Of the three that may be of other text types, one, on a Campanian vessel from Entremont, dates to the second century bce and has been presented as being in cursive Latin with Greek letters.79 The text is not in fact in cursive Latin, but cursive-inspired Latin capitals, nor does it certainly contain Greek letters. It can be read: BAL • MAṚDIVS • HP • C̣[—] T BAL.80 In this period it might reasonably be interpreted as Gaulish, depending on what we make of the possible abbreviation HP. However, Entremont has not provided the epigraphic output in the local language that we might have expected for a site of relative importance in the region, no doubt because of its destruction in the late second century bce, which, incidentally, might encourage us to consider that the main phase of output of Gallo-Greek should indeed be dated to the first century bce. The second of the inscriptions, from Saintes and on a local ceramic dish, is also perhaps better viewed as a previously unrecognized Gallo-Latin inscription, perhaps a dedication to a deity. It reads MATVGENOS▴COSỊṆTOS▴BELINỊ[—] (ILA, Santons, 1005.13) and is dated to before the Augustan period based on archaeological context. The third and final example on ceramic is of an extremely unusual type, being an inscription on a ceramic mould used for making slingshot projectiles. There are three ovoid depressions for the fabrication of the projectiles, and in the middle one an inscription of four letters appears, written before firing, and therefore would have appeared retrograde on the ballistics. The object was found in Paris, and the reading is printed as ẸVLG by AE (2000, 974), with no secure interpretation yet offered.81

The small set of a dozen stone inscriptions that can definitively be assigned to before the Principate is strikingly homogeneous. Two can be assigned Gaulish contexts—at Ventabren (an inscription traditionally counted as Gallo-Latin) (RIIG BDR-13-03) and at Nimes (an inscription found on the same abacus of a column as a Gaulish inscription) (RIIG GAR-10-03)—and the remaining ten are all public in both function and content, citing military and political roles (praetor, aediles, consul, imperator, and so on), with the exception of just two funerary inscriptions to freed-persons (CIL XII 5208 and the conjoining fragments 4597 and 5252). Naturally it is easier to date precisely those texts that provide clues about known characters, so it may be that the official and public inscriptions take up an inflated percentage of this set. The earliest of the stone set, dated to c.118 bce, is the milestone found at the Pont-de-Treilles in the Aude, originally attached to the newly built via Domitia, which reads Cn(aeus) Domitius Cn(aei) f(ilius)/Ahenobarbus/imperator/XX (CIL I² 2937). The construction of the via Domitia inland from the existing via Heraclea, in the late second century bce,82 entailed the display of several inscribed milestones, which would have played a no-doubt significant role in advertising monumental texts to a large passing audience.83 The official inscriptions may constitute an over-exaggerated proportion of this list, but their presence was no doubt influential.

In the earliest phase, the epigraphic evidence on ceramic, along with the masses of non-text bearing archaeological material, allow us to trace the increase in the density and spread of the economic connections with the Italian Peninsula.84 Narbonne in particular, the earliest founded Roman colony in Gaul in 118 bce, acted as a front-runner in the networks of interaction.85 As Christol has described, it was not only a gateway for Italian trade with storage and redistribution facilities, but a hub of economic activity for entrepreneurs from Italy, who are mentioned in Cicero’s Pro Fonteio.86 The influence of Italian practices in the first decades of the uptake of the Roman lapidary epigraphic habit can clearly be traced in the colonies of both Narbonne and Béziers.87 In the funerary epitaphs this can be illustrated by the use of arbitratu (plus the genitive of the name of the person responsible for setting up the funerary monument),88 V or VIV plus the number of years lived, and the use of the so-called theta nigrum, which (at least in these contexts) indicates the deceased, all three of which are features of the epigraphy of Italy. The theta nigrum has a quite specific distribution, common at Rome during the end of the Republic and the first decades of the imperial period,89 and is found in the provinces in this same period primarily at Narbonne and Ludgunum Convenarum (Saint-Bertrand-de-Comminges). This is not a widespread usage, and examples in Noricum, Pannonia Superior, and at Arles seem to be part of a slightly different use of the theta—namely, as part of an ‘age-at-death formula’.90 Christol assesses the epigraphy of the Republican period, including more broadly dated items that were not discussed above, and finds that the region of Narbonne and Béziers can be viewed as ‘un appendice de l’Italie’, with fewer traces of local onomastics in the naming practices than elsewhere in the south of Gaul.91

Another feature of Roman rule beginning in the early years of the Principate may have played a strong role in spreading the epigraphic habit to the corners of Gaul. The imperial cult was introduced under Augustus, and associated inscriptions were prominently displayed, quickly becoming ‘a focal point for Gallic loyalty and ambition’ for the members of the local elite.92 On 1 August 12 bce, with the German campaigns underway led by Drusus, a sanctuary and altar to Rome and Augustus were set up at the confluence of the Rhône and Saône at Lugdunum, apparently following a controversial census in Gaul, possibly linked to the raising of taxes to support the war. The ara was perhaps designed to foster Gaulish–Roman unity against the Germanic enemy,93 essential in the wake of the Varian disaster in the Teutoberg forest in 9 ce.94 An annual festival was instituted, which meant that local Gallic elites from sixty civitates regularly convened in a place ornamented with monumental inscriptions, creating a powerful concilium Galliarum (Cass. Dio 54.32.1) led by an elected sacerdos. Being a member of this concilium, and particularly the sacerdos, was ‘a position at the apex of the provincial cursus honorum’,95 and a not insubstantial honour, which seems to have had caché even in civitates far from the centres of Roman civic life in Gaul.96 The epigraphic commemoration of the engagement was prominent at Lyon,97 and, as a result, became widely reflected in the provinces, facilitating the wide dissemination of the Roman expression of civic status.

5.5 Regionality in Latin

From the Augustan period onwards, literacy and the epigraphic habit, and alongside them undoubtedly also Latin language, increased markedly across the whole of Gaul, marking a change from the piecemeal fashion of the adoption of writing in the Iron Age, underscoring the ‘importance of Roman power as a context for the change’.99 Striking about the uptake, here as across the western provinces, is that there appears to be no time lag between Mediterranean and non-Mediterranean Gaul, despite the gap between periods of conquest, confirming the Augustan period as a watershed moment.100 But this often-repeated statement about the generalized spread of an epigraphic habit obscures one of the most important dimensions of the uptake of these phenomena—namely, regionality.

The all-object epigraphic output of the Gallic provinces ranks output in descending order for Narbonensis, Belgica, Aquitania, and Lugdunensis, with the epigraphic ‘density’ figure for Narbonensis nearly four times that of Lugdunensis (Chapter 1, Table 1.2, Fig. 1.5). We have seen, however, how easily skewed this whole epigraphic data set can be (Chapter 1).101 A focus on the epigraphic habit specifically provides more consistency of documentation across the provinces, since this has been a mode of epigraphic expression receiving interest of collectors and scholars since Antiquity. Although there were different practices of reuse of these monuments, which have resulted in varied preservation patterns—some were unwittingly preserved in city walls or churches, others were smashed up for religious reasons, for foundations of buildings, or for use in lime kilns—at a zoomed-out scale at least the patterns are more reliable. In Becoming Roman, Woolf made the case, with caveats, for using just CIL,102 and our data set, which includes a large number of items from the regional series such as ILN and ILA, and the annual collections in AE, provides deeper coverage still.

Narbonensis again leads the pack, this time more markedly so, with an epigraphic habit density score of 85, six times more than Lugdunensis and Aquitania at 14, and nearly four times more than Belgica at 30 (Chapter 1, Table 1.3, Fig. 1.6). Again we see correlation of the uptake of the lapidary habit with social factors that are repeated across the western provinces.103 Across Gaul, river valleys and areas with good access to appropriate stone are generally more epigraphic than mountainous and remote regions; this is linked, of course, to crucial human factors of density of settlement, and positioning of road and fluvial networks. The epigraphy to the south of Lyon perhaps most neatly shows this riverine–epigraphic relation. But it is clear that some regions—for example, the Parisian basin or western Belgica—though far from unconnected or remote, are only weakly epigraphic. Equally the highly mountainous area to the south of Saint-Bertrand-de-Comminges bucks the expected trend, with numerous lapidary inscriptions dating to the first century bce to the third century ce, the vast majority of which are of religious nature. This region is known for its high-quality marble, and much of the inscriptional output deploys the local resource; indeed some dedications explicitly mention the marmorarii (e.g. ILTG 3 and 16 from a rock sanctuary at Saint-Béat). It is not implausible that the risks and importance of the mountainous passages created a dedicatory realm that perhaps continued deeper Iron Age rituals and therefore took in local communities, a suggestion borne out by the local-named deities and dedicators (both Vasco-Aquitanian and Celtic).104 One altar demonstrates the desire for divine assistance in the use of the landscape, with its votive inscription, dedicated by Natalis, Martialis, and Sintus together with their collegii, illustrated with four scenes representing the protection of Silvanus over quarries, forests, and flocks (ILTG 23),105 and a couple of inscriptions open with dis montibus (e.g. ILTG 123). Military installations, colonial foundations, centres of commercial and road networks, areas of trade and, in some regions, production also attract the epigraphic habit. There may also be some mapping onto the previous uptake of literacy by pre-Roman communities and longevity of Roman–local contacts, for example in the area of the Aedui and Treveri. But it should be remembered that throughout the Tres Galliae before the conquest the carving of stone was an exceptional practice.106

To take the example of Belgica in more detail, there is a great internal disparity, with the eastern part much more densely epigraphic than the west, and particularly the north-west, in which several major settlements provide next to no lapidary evidence, for example Castellum Menapiorum.107 This is no doubt in part the result of proximity to military communities. Though Belgica was not a frontier province, its fortunes were closely dependent on those of the frontier and its large garrisons. Security issues might entail uncertainty and deprivations, including via increased taxes and requisitioning. But over the long term the economic ecosystem that the army stimulated, especially after the end of the conquest of Britannia, is no doubt partly responsible for the rural development of northern Gaul from the Flavian period onwards, which accelerated under Trajan;108 the prosperity generated through production and trade;109 and the rise of lapidary forms of expression and the spread of literacy. This factor renders the relative paucity of epigraphic finds from Reims, the caput provinciae, somewhat less surprising, since geographically it lies just within the less-epigraphic, less-militarily influenced western half of Belgica.110 The Remi offer mostly funerary inscriptions, preferring to leave religious practices to sculpted stone and other means, and the examples congregate in the capital. Conversely, the epigraphy of the Leuci is dominated by its rural hinterland, with the central place, Toul, offering very little, and a greater proportion of religious inscriptions and reliefs. The Treveri produce more inscriptions, both religious and funerary, than any other group in Belgica, both in the civitas capital and in the surrounding area. Christine Hoët-Van Cauwenberghe discusses the nexus of factors resulting in the low levels of lapidary epigraphy in the western part of Belgica: acidic soils eating away inscriptions, difficult-to-carve local stone, the recutting of stones for city walls, and various destructions of sites (and collections) in ancient times and more recently (particularly during the world wars). Nevertheless, these, or similar reasons, also pertain to a greater or lesser extent to eastern Belgica, so cultural and social factors must also be in play. The evidence of writing equipment and non-monumental writing, in which the mistakes indicate spoken Latin, suggests that western Belgica was adopting literacy and Latin, including among ‘the lower levels of society’,111 just not so much the epigraphic habit (see also Chapter 10).

When the imperial-period dated inscriptions are isolated and assessed from within the epigraphic habit subset, each province shows a different complexion, and most do not neatly match the standard curve of epigraphic-habit decline and fall for the provinces, as set out by MacMullen (1982) (see Chapter 1) (Fig. 5.5a–c), with Baetica’s line coming closest (Fig. 5.5b).112 Narbonensis presents several peaks along its line, owing to clusters of well-dated milestones. If these are ignored, the output presents a relatively consistent level across the centuries, with a decline—that is, however, not steep—in the third century and a slight recovery and continuation at a lower but consistent rate from the fourth century onwards. Belgica demonstrates a significant output from 100 ce to around 250 ce, with the curve rising towards 200 ce and then falling. After 250 ce for around a century there is a relative lull, but still the output is higher than in the first century and in around 350 ce the curve again begins to rise, a factor no doubt linked to the rise in prosperity and status of Trier in the third century onwards, at a time when many other major cities of Gaul were in decline. This matches the findings of Wightman, who, before the assistance of digital techniques, argued that Belgica, like the Germanies (Fig. 5.5c), produced the bulk of its epigraphic output in 140–260 ce. Aquitania demonstrates a relatively consistent output for the first century and a half ce, with a rise after 150 ce, which holds for about a century, after which there is a relatively steady decline until the early years of the fourth century; thereafter output is negligible. A recent survey has confirmed the intensity of the decline, with only seventeen funerary inscriptions dated between the fourth and fifth centuries and describing the epigraphic practice as ‘varied, scattered and scarce in comparison with previous centuries’.113 Lugdunensis shows relatively little output throughout the first century ce, with most of the dated stone inscriptions falling between 100 ce and the last decades of the third century. After this time the output is low. While the method to generate these dated epigraphic-habit curves might be questioned, if we can assume that the methods of dating inscriptions, however problematic, have been at least relatively consistently applied across Gaul, what they make clear is that there is regional variation in the chronological complexion of these outputs across these regions, which would benefit from more detailed further inspection.

Regionality is visible not just in the rural–urban divide and different periods of peak activity depending on the socio-historical and political realities of the communities, but also in the choice of monuments and iconography. Narbonne, for example, seems to have had a penchant for large masonry mausolea and stele, whereas Nîmes preferred the altar for its funerary inscriptions.114 There was no doubt a phenomenon of emulation at play here, with elites, the aspirant ‘middle classes’, freedmen, and others wanting to make the choice that was socially advantageous and appropriate. It is possible that, in some parts of the Lower Rhône Valley, perhaps in the early imperial period, the use of the stele was associated with pre-Roman community practices, and may have been deliberately selected or avoided. Available craftsmen and materials influenced choices too, and connections between the stone-carving practices in iconography, ordinatio, and style of lettering can be made across regions, and have been shown in studies of the Rhineland.115 All this created what Carroll calls a ‘language’ of the tombs and a ‘communal identity’ for the users.116 Lyon was distinctive, not for its choice of monument or iconography, but for the extreme verbosity of its inscriptions, in which the iconography seems to have consisted of an imposing sheet of letters, destined to impress its viewers (literate and illiterate alike) with its display of literacy. The text types of the inscriptions could also vary between regions and settlements within Gaul, as we have described already for Belgica. Table 5.2 indicates again not only the differential uptake of the epigraphic habit between communities in the central-eastern region of Gaul, with some communities such as the Tricasses, Leuci, and Segusiavi producing next to nothing, but also differences in the function of the texts, with some records weighted strongly to the funerary, and others with a larger religious component.

Who is producing these texts also depends on the region in question, with some areas led by the output of the military, and others primarily bolstered by trading communities. On just one page of CIL we can find at Narbonne a mensor (XII 4490), mensularius (4491), frugi mercator (4492), navicularius (4493, 4494?, 4495), nummularius (4497), reflecting the importance of trade in the epigraphic habit of this city.117 Scholars have also used the evidence of names and naming practices to identify the status of those inscribing and their possible origins. Again significant differences are revealed: in the early inscriptions from Narbonne and Béziers there are considerably fewer peregrine formations and local-origin names than in the area of Nîmes.118 An examination of the naming practices shows large percentages of tria and duo nomina, in both the capital and the surrounding areas of the Treveri (72 per cent), whereas only 22 per cent of the Remi deployed this distinctively Roman format.119 Chapter 6 reveals some of the complexity of naming practices in the northern Continent.

Regional choices can also be seen in the epigraphic formulae used. In Bordeaux, annis XXX defunctus/a was used by inhabitants of the city and its hinterland to refer to the age at death of those who died, and the form seems to have been followed also for foreigners who died in the area.120 Lyon and other sites in the Rhône Valley offer a significant number of the formula sub ascia dedicavit, with or without the hammer-shaped tool that was used for working stone. The exact meaning of the ascia has been debated, but its frequency may suggest that in this region the symbol,121 like the theta nigrum or the rosette in other areas, may have been enough to alert the viewer to the funerary nature of the stone. At Dijon in central-eastern Gaul and within a surrounding radius of 100 kilometres, a preference for pyramidal stelae can be identified, many displaying the distinctive and succinct formula monumentum plus name of the deceased, which has been attributed to a continuation of local practices.122 Raepsaet-Charlier has set out regional patterns in religious inscriptions for the Tres Galliae and the Germanies.123 For Belgica, the most common formulae are, in descending order, deo/deae, in h(onorem) d(omus) d(iuinae), genio, Num. Aug.; for Aquitania, deo/deae, Num. Aug., sacrum, pro salute; for Lugdunensis, deo/deae, Augustus/a, sacrum, Num. Aug. The specificities of the formulae found in the eastern part of Lugdunensis have even led experts to rethink the boundaries of, and associations between, communities. The formation Aug(usto) sacr(um), plus deo/deae and a local deity, usually following but sometimes preceding, is almost exclusively found in the region of Autun, Auxerre, and Chalon-sur-Saône.124 The use of the formula, alongside other evidence, encouraged scholars to redraw the borders of the Aedui for the second and third centuries,125 as a result moving the epigraphy of the Auxerre region from its traditional home (for example, in Hirschfeld’s CIL) with the Senones into the Aeduan corpus. Other features, such as the distinctive deities involved—namely, Ucuetis and Clutoida—and the dedication to Rosmerta without Mercury (the pair common in Belgica, and to a lesser extent Germania Superior) supported this reassessment. However, human movements and interactions, not least religious exchange, mean that a dedication can turn up in a major Lingonian sanctuary with the Aeduan formula.126 Understanding regionality requires both knowledge of the broad contours of epigraphic habits across the Empire, with which our digital resources can assist, but also more focused and contextualized knowledge.

5.6 Gaulish–Latin Bilingual Texts

Since the publication of Adams (2003a), ancient multilingualism has received increasing attention, undoubtedly more than any other area of ancient sociolinguistics, and, with the stream of recent publications and conferences, we might even consider that we are in a ‘multilingual moment’.128 In this context we can begin to claim that we know something of the nature of the bi- and multilingualism in Gaul.129 It is worth reminding readers about the range, if not quality, of evidence that is available to tackle this theme. Not only do we have the direct evidence of epigraphic bilingual texts, which will be the main focus of this section, but also loanwords, which can be borrowed from Gaulish into Latin and vice versa and can be seen in either contemporary or later evidence, and which can suggest linguistic and cultural contacts;130 epigraphic evidence of different languages and/or scripts being used contemporaneously at the same site;131 the evidence of the effects that a substrate has on the nature of Gallic Latin and its Romance developments;132 and the copious evidence of Celtic-origin place names, personal names, and deity names, which provide a thesaurus of the language, and evidence, in some cases, of the ongoing understanding of the meaning of the names (for example, in the case of translation names).133

The bilingual texts of Gaul, however, intrigue and frustrate in equal measure: their individual interpretations are often uncertain and they are so scattered (Fig. 5.6) and at times so apparently idiosyncratic that it is hard to draw conclusions from them about bilingualism in society more broadly.134 In the south, no clear textual evidence of Gaulish–Greek bilingualism can be found within the Gaulish texts, nor in the Greek inscriptions, and when hints of language contact appear they can more easily be traced to the Italian Peninsula, as we saw in Section 5.3.135 Beyond Narbonensis, however, the Tres Galliae provide a series of bilingual texts. Among the biscriptal texts we find examples of the same word given in both Gallo-Latin and Gallo-Greek on the same coin legends, all with non-Mediterranean distributions. There are even two examples of legends in mixed alphabets ΔEIOYIGIIAGOC (RIG IV 134, north-eastern Gaul, group unknown), ATIIVLOIB SOLIM YC (55, Carnutes?). These two may be the result of confusion or mistake, but the other biscriptal examples seem to be deliberately presenting both the Greek and Latin script, and, since they date to the last centuries bce, there may be an appreciation of the association of Gaulish with Gallo-Greek, coupled with the need to reach a wider audience, which increasingly used Latin script. Other examples of biscriptal texts, this time on pottery, and probably owners’ or makers’ marks, are found in Bibracte (CIL XIII 10017.79, 80) and possibly Roanne (RIG II.2 L-81b/I G-595), locations in the Tres Galliae that are trading nexuses and offer both Gallo-Greek and Gallo-Latin texts (and possibly Greek in the case of Roanne). Autun may also provide a biscriptal Gaulish text on a sherd of terra nigra, though its interpretation remains uncertain (Labaune et al. 2015). In areas where both scripts were used for Gaulish, perhaps the writer was covering all bases, and we might extrapolate that this kind of practice suggests that writing was perhaps not narrowly restricted at these settlements.

Two lapidary biscriptal texts may be linked to one another through similar motivations. At Saint-Germain-Sources-Seine a striking naiskos-style monument of white oolitic limestone with a male bust carved within a recess displays a Gaulish inscription within its tympanum (RIG I G-271; II.1 L-12; RIIG CDO-02-01) (Fig. 5.7). It was found in a secondary context at the sanctuary of the Sources de la Seine, about 40 kilometres north of Dijon, a sanctuary that has provided very numerous stone, bronze, and wood sculptures, and Latin dedications to the goddess Sequana. It contains a Gallo-Latin text over four lines; probably a dedication, but the interpretation is uncertain. This is followed by a line of Gallo-Greek, which sits on the bottom frame of the triangular tympanum and reads ‘Dagolitus made this’. The artisan named on this monument has recently been shown to have belonged to a sculptor’s workshop that has been named ‘Source 2’ and that can be dated to the second half of the first century ce.136 The switch into Greek script/language for the artist’s signature matches a pattern that we see not only in the next biscriptal Gaulish text but often in imperial Latin inscriptions in Gaul.137

An unusual biscriptal monument from Genouilly again displays an artist’s signature, this time with switch of language into Greek (RIG II.1 L-4, L-5, I G-225; RIIG CHE-01-01).138 The text reads

and contains, in the following order, Gallo-Latin (Gaulish in Latin script), Gallo-Greek (Gaulish in Greek script), Greek (language and script), Gallo-Latin (Gaulish in Latin script). The stone is roughly hewn, reminiscent of prehistoric standing stones. The context encourages a date in the final two-thirds of the first century ce. It is closely related to a smaller stele found nearby (RIG L5; RIIG CHE-01-01: [—]RVONDV). It is unclear precisely how we should understand the components on the larger monument and the second stele, how they fit together, and how they relate to one another chronologically. It is clear, however, that all the texts on the larger monument are connected. Another element that links these two biscriptal monuments may be local traditions of dedication, since neither presents a Roman rendering (unlike, for example, the four Gallo-Latin inscriptions from the larger Aeduan region).139 The menhir-style at Genouilly would perhaps strike a viewer as tapping into long-standing local traditions and not following the more worked and ordered Roman style, and similarly the organization of the naikos-style monument, with the writing crammed into the triangular section and the line of Gallo-Greek written on its frame, seems to be disregarding usual Roman layouts, despite a certain degree of skill on the part of the stonecutter, which might suggest knowledge of Roman stone-carving practice. Although not much remains on the stone, the man depicted wears a so-called Blussus cloak—namely, the variation of the Gallic tunic with a thick, multilayered cloak, a common representation of members of the local community in this area of Gaul.

Bi-version bilingual texts, which we find a few times across Gaul in the imperial period with Greek and Latin,140 do not seem to be attested for Gaulish and Latin, or at least our understanding of the texts and contexts prevent certainty.141 Perhaps the clearest possible example is the famous Mercury of Lezoux (RIG II.1 L-8; RIIG PDD-01-01): a large sculpture of local stone, which Lejeune described as ‘d’un art local naïf’, depicting a person with ‘la stature et le visage d’un paysan arverne’,142 and restored in an ‘imaginative’ way by Plicque. It was found in the remains of a small sanctuary of Mercury near the south-western entry to the artisanal quarter and has been dated to the first or second centuries ce simply on the grounds that this is when the pottery was at its height. The original statue may have originally displayed three separate inscriptions, one in three lines on the stomach of the figure in a tabula ansata and carved in capitals of an average of 4 centimetres, another spread across the back of the statue, unframed unlike the first, and of capitals of around 10 centimetres, of which the letters are now visible only on the left-hand side, and a third of just a few letters, which apparently ran down one of the folds of the cloak, but is now lost. The Latin text reads Mercurio et Augusto sacrum. The text on the back has been interpreted as Gaulish, but the reconstruction of the dedicatory verb IE[VRV] is not secure,143 and it remains unclear what relationship the three (?) texts may have had to one another. If it is a bi/tri-version bilingual text of some description, it is the first of several bilingual texts in this section that is without doubt squarely to be linked to the religious domain. The two biscriptal texts we have just discussed are harder to place functionally but possibly also fit into the same domain. We shall see that this domain seems to preserve both oral and written Gaulish (Section 5.8).

Translingual texts,144 mixed-language texts,145 and texts with bilingual phenomena involving Latin and Gaulish are again easier to recover from the Tres Galliae than they are from Narbonensis. Roughly half of the attested examples of texts with bilingual phenomena can be deemed of ‘monumental’ type.146 One is the unusual bronze plaque from Le Vieil-Évreux (RIG L-16; RIIG EUR-01-01), a unique object within the Gaulish corpus, with the exception of the calendars, which report a local form of measuring time, displayed in a distinctively Roman way (RIG III).147 The site formed in Roman times a vast urban complex with both major Roman-style buildings such as baths and a theatre and smaller structures such as a fanum. The plaque was excavated in the so-called basilica and has been dated to the first century. The execution of the letter forms is the most sophisticated, judged from a Roman point of view, of all the inscriptions of the Gaulish corpus.

Unfortunately, thanks at least in part to the fragmentary nature of the text, it is impossible to deliver a secure interpretation, though it is likely to be a dedication. What is clear, however, is that the inscription is Gallo-Latin, since various combinations of letters are Celtic and the presence of the barred D to represent the so-called tau gallicum is diagnostic of Celtic language words.148 There are clearly words that must be interpreted as Latin however, for example filia in the penultimate and civis in the final line, which might be taken as intra-sentential code-switching (perhaps tag-switching), or borrowings, into Latin. Crispos in the first line may be a Latin name with the Gaulish morphological ending (-os).

A Gaulish inscription on a block of granite from Arrènes near Sazeirat (Creuse) (RIG II.1 L-7, RIIG CRE-01-01), dating to the first or second century and found in the construction of a railway in the nineteenth century, may show a further example of intra-sentential code-switching.

It can be interpreted as ‘Sacer son of Perocos/Sacer and Peroco dedicated the porticus’, followed by the common Latin votive formula. Lejeune spotted that the incisions were less deep for VSLM and that the rest of the text did not use interpuncts, and therefore suggested a later addition for the Latin formula, but it seems likely that we have a text with a code-switch from Gaulish into Latin. The form duorico has elicited much debate, with many commentators deeming it a calque of Latin porticus, and therefore assuming a bilingual context. A similar example from Narbonensis may reflect, if not a bilingual text, the context of epigraphic interaction, since a Latin text may have replaced a Gaulish one. This can be reconstructed for the lost abacus of a capital of a column from Nîmes (RIG I G-203, RIIG GAR-10-03), known only from a manuscript of Séguier, which offers a votive inscription on one face, and what appears to be part of a dede bratou dekantem inscription in Gaulish on another. In the absence of the object itself and with very little further contextualization, it is impossible to determine whether the two inscriptions were contemporary or even related, but Lejeune was confident that part of a Gallo-Greek inscription had been removed to make way for a Latin text. The fact that VSLM and the Gaulish formula are functionally equivalent may suggest that the texts are linked by continuity of practice.

Another religious context that provides a complex-to-interpret bilingual text was uncovered in 1711 under the choir of Notre-Dame de Paris (RIG II.1 L-14, RIIG PAR-01-01),149 and can be securely dated to the reign of Tiberius thanks to the Latin dedication by the nautae Parisiaci on the third block of five in the latest reconstruction:150 Tib(erio) Caesare/Aug(usto) Ioui Optum(o)/Maxsumo s(acrum)/nautae Parisiaci/publice posierun[t]. On the other faces of the same block and the faces of the sculpted blocks below and the two above are a series of deities, many of which have labels in the nominative, alternately Latin and Gaulish depending on the detity: Volcanus, Iovis, Castor, Fort-, alongside Cernunnos, Esus, Tarvos Tigaranus, Smer-. There are also two other scenes with SENANṬV[ . ]ẸṬLỌṆ[ . . . . ] and EVRISES as the labels, both probably Gaulish and in the second case perhaps referring to ‘dedicants’.151 The result is an imposing monument, apparently honouring a new pantheon in images and words. The order and orientation of the blocks have been intensely debated, as has the hierarchy between those depicted. From a linguistic and cultural point of view this monument is unparalleled in the pre-Roman or Roman western provinces: the deities are represented with their corresponding iconographies and names, on separate faces. Perhaps this should be seen as a continuous message with constant code-switching between the deities in a newly mixed ‘Gallo-Roman’ pantheon, a reflection of a phase of steady integration of at least parts of the two religious schemes, which was beginning in the late Republic. On the other hand, we could see the elements as inhabiting their own space and kept separate within their frames, linked but meant to be kept apart. Or perhaps this is a guide to one version of the so-called interpretatio Romana, in which the Gaulish and Roman traditions are overlaid, as we see in the descriptions of Caesar and in multiple sculpted representations and inscriptions across imperial-period Gaul.152 If so, the tradition with the upper hand here, if any, is not obvious.

In one reconstruction of the pillar, the blocks are oriented so that the Latin dedication on the third block faces frontwards and has the paring Mars and Venus beneath it, Jupiter above it, and his son Pollux above that, creating a Latin side of the pillar, which would have created the initial vista for the viewer.153 If this positioning is accepted, the result would be two contiguous faces of the monument with Latin focus and two with Gaulish, with the Latin taking the lead if facing forwards, though it is unclear which other conjoining side(s) would be privileged, since we do not know how the monument was displayed, how it was accessed, or how it was lit. How did the creator(s) of the monument intend for the readers/viewers to experience it? Was it intended to be read as a whole by the same, bilingual or monolingual visitors, literate or illiterate? Or were different sides of the whole meant for different audiences? The broader context suggests that a pick-your-side option is less likely, but we are on far from secure ground in attempting to understand the reception of this unusual monument. There are, however, some bilingual dossiers within Gaul that allow us to say more about their creation and reception, particularly the graffiti from La Graufesenque and the spindle whorls, the first of which we shall consider here.

5.7 Case Study of Bilingualism in Context: Mass Production and the Stimulation of Writing Practices in Gaulish and Latin

The site that produces the most examples of bilingual epigraphy from Gaul is the massive Samian ware production centre at La Graufesenque, situated on the outskirts of Millau (ancient Condatomagos) in south-western France (Aveyron).154 The substantial corpus of bilingual administrative texts, containing both Latin and Gaulish, is unparalleled in the Roman West, and makes La Graufesenque one of the most interesting sites in the Empire for considering the function of Latin and indigenous languages and the cultures of writing in the High Empire. However, while the materials from the site are impressively copious, they should not be seen as ‘unique’ or unusual, but rather they fit into an epigraphic culture and bureaucratic practice that, though relatively poorly attested elsewhere, was no doubt relatively widespread among at least the larger provincial productive centres.

Production of proto-sigillata ware began at La Graufesenque at the end of the first century bce, followed by the manufacture of genuine terra sigillata starting in the first decade ce. The pottery flourished in the second half of the first century ce, before production tailed off in the first half of the second century.155 The epigraphic material falls into two main categories: the stamps and signatures on the vessels and the so-called graffiti.156 The stamps and signatures are found on vessels destined for external consumption, and the vast majority were stamped.157 Latin alone is used as the language of the stamps and signatures at La Graufesenque.158 This is not an insignificant point, as several stamps and signatures generated in other locations, particularly on other types of ceramic—for example, terra nigra/rubra and perhaps most notably on the Venus figurines (RIG II.2 L-22) and the moulds of the figurine maker Sacrillos (L-23)—do contain Gaulish, usually forms of AVOT, which is the Gaulish equivalent of Latin fecit, alongside names.159 There are over a hundred records with AVOT on pottery in the LatinNow database, some of which, owing to the way the examples were published, represent one record for the same stamped impression, which was repeated on multiple different items. Since these Gaulish-language texts appeared also on fibulae, this was one way that written Gaulish was making its way onto the tables and the bodies of the inhabitants of Gaul (Chapter 7).

At La Graufesenque, in addition to the stamps and signatures, around 250 graffiti have been uncovered, mainly dating to the second half of the first century ce and edited most recently by Marichal (1988) and Lambert (RIG II.2).160 These are largely firing lists incised onto pottery, which were fired in the kiln along with the pottery they recorded (Fig. 5.8). The complete lists contain a heading (contextualizing information such as date, number of firing/kiln) and key information in fixed order, including potters’ names, vessel types, dimensions, and quantities. Other texts are also attested, for instance the record of the purchase of a slave.161 The material falls loosely into three groups: inscriptions that are mostly in Gaulish, those that are mostly in Latin, and many that are so mixed that they resist this linguistic classification and might be categorized as mixed-language texts. One section of a large Samian ware plate (Drag. 18) provides two texts, one in Latin on one side and one at least partly Gaulish on the other (RIG II.2 L30e, 80–120 ce).162

The importance of this material for understanding language contact and related issues in the Empire encouraged Adams to devote a whole chapter to it in his magnum opus on bilingualism.163 He concluded that ‘there is not a single, mixed language in evidence’, no ‘institutionalised mixed language’, no ‘popular fusion’.164 The texts are not in a mixed language but rather in a ‘mixture of languages’, a subtle, but important, difference. He argued that the languages can be differentiated and that the potters were able to appreciate two separate languages at the site,165 though in practice it proved difficult to assign the dominant language to various texts.166 The graffiti were analysed as the bilingual output of individuals and interpreted with reference to the bilingual phenomena described by modern bilingualism theory, namely code-switching, interference, and borrowing.167

The names of the personnel attested in both the firing lists and the stamps are potentially useful for understanding the composition of the bilingual community.168 Up to 45 per cent of the stamped names are Latin, alongside up to 37 per cent Celtic, and as many as 14 per cent impossible to assign to only one linguistic category and classified as Latin/Celtic (perhaps testament to the fact that so-called cover names were popular in such a context).169 The smallest group of names were Greek or Graeco-Latin or cover names with Greek (c.3 per cent).170 A comparison with the names in the firing lists suggests roughly similar percentages, but with slightly higher percentages of Latin (48 per cent) and Greek names (7–9 per cent) and fewer Celtic (34 per cent) and Latin/Celtic (8 per cent).171 It is, therefore, not the case, as Adams states, that in the stamps ‘not only is the -us ending almost universal,172 but Celtic names themselves are virtually eliminated. […] The language choice is determined by the expected readership.’173 This misunderstanding, based on a misconstrued comment in Marichal (1988),174 contributed to Adams’s choice of diglossia to label language use at the site, describing the ‘grading of the two languages in terms of their status and function, with Latin treated more as an international or imperial language, but Gaulish as provincial and unsuited for use in the wider world. Here again is a trace of the polar diglossic opposition H-L.’175

Diglossia, strictly defined, refers to societal bilingualism in which the linguistic varieties can be assigned ‘High’ and ‘Low’ values and are functionally compartmentalized.176 If we stick to this established sociolinguistic definition (and do not use it as simply a synonym of societal bilingualism, as some have tended to do), diglossia may be unsuitable for describing bilingualism at the pottery. It would be hard to find a corpus of material that alternates between languages more than the firing lists, which, in a diglossic community, would be in only one language (e.g. Gaulish), with the stamps in another (Latin). Furthermore, since in situations of diglossia the two varieties are kept strictly apart, bilingual phenomena such as code-switching are unlikely, and L varieties are in any case not usually committed to writing. The languages at La Graufesenque are absolutely not kept strictly apart.

While we must not view the graffiti and stamps as directly representing the language spoken at La Graufesenque,177 arguably no other site in the western Empire provides better evidence for exploring the interaction of languages and bilingualism. The code-switching, interference, borrowing, translations of names, and mixtures of names and languages indicate the exact opposite of the ghettoizing of languages and communities that some scholars have suggested,178 or the strict division of labour between the languages as required under Adams’s diglossia. The community at La Graufesenque was doubtless bilingual, to the extent that both Latin and Gaulish could be used alternately and in tandem in the internal bureaucracy. The majority of the potters were probably Gaulish-speaking, and would have picked up linguistic competency in Latin at levels varying from a smattering (vessel names, sizes, and so on) to fluency; those workers from Italy (or indeed elsewhere in the Empire) who joined the potters, whether Latin- or Greek- and Latin-speaking, would no doubt also have learnt anything from a low-level Gaulish to fluency in order to communicate widely at the site and perhaps beyond. It is relatively likely that levels of proficiency in Latin at the site may have increased over its history, though it seems impossible to demonstrate this through the graffiti.179

The best analysis of the bilingualism of the textual output of this potting community, Adams (2003a), overlooks a significant aspect of its composition. Adams discusses reasons motivating a switch in language in the graffiti and describes the texts as if they represent the direct written record of bilingual individuals. However, he has neglected the important technical point that Marichal explains in his discussion of the ‘coutume ouvrière’—namely, that the firing lists were probably created through an amalgamation of notes and dockets.180 These overall lists are at least partly created from written and oral information handed over at the kiln: short notes stating the number of vessels from a certain source and perhaps potters’ stamps being shown with verbal communication of the rest. This would account for various linguistic inconsistencies in the documents, which are the result of mechanical copying and compiling in a highly bilingual environment, where, for example, the same individual can appear as both Tritos and Tertius. Marichal’s repeated references to ‘slavish copying’ undermines the detailed analysis of verbal bilingual output presented by Adams. If the lists are in part created through a collection of notes in either Gaulish or Latin or both, the bilingual phenomena and language mixture attested will naturally be at a remove from spoken bilingualism. What the bilingual documents clearly demonstrate, however, is that the site should be viewed in terms of language contact and interaction. There would be little point in operating a bilingual internal bureaucracy if the community were not, at least in part, bilingual.181

Whilst excavations at La Graufesenque have yielded unusually copious remains of its bureaucratic workings, that bureaucracy should not be seen as exceptional, a view that the site’s constant singling out in works concerning ancient sociolinguistics might tend to encourage.182 Several graffiti probably related to production have been identified at some other Samian production centres in Gaul and surrounding provinces.183 From Gaul fifteen or so were found at Chémery and can be dated to 60–150 ce.184 Two firing lists and perhaps a partial firing list were found at Lezoux, dating to the Tiberian period and therefore earlier than those at La Graufesenque.185 Montans also provides a couple that can be roughly dated to 1–175 ce.186 A single, but detailed, account was discovered at Blickweiler and can be dated to around 140 ce.187 Fragments of what may be firing lists have also been found in the Iberian Peninsula, for example Segobriga yields a fragment of a plate of terra sigillata with text incised ante cocturam on both sides. On the exterior it reads PAN(N)AṢ [/X̣XXII and on the interior ]S CCCCCX.188 This fits with evidence that pottery production, including that of Samian ware, was occurring in the area.189

These, together with further meagre examples from other sites, are slim pickings compared to the corpus of a couple of hundred texts from La Graufesenque, and it is possible that at other mass-production sites perishable material such as wood may have primarily been used for the documentation, but the snapshots suggest that the bureaucratic set-ups might have been similar to that in the Aveyron. It is apparent that the knowledge and need to write these lists had spread over a large area. But, based on the evidence from all the sites, including La Graufesenque, there seem to have been no manuals to help the scribes. At La Graufesenque the alternating endings and spellings for vessel names and the appearance of Gaulish, and even snippets of Greek, in the internal bureaucracy indicate that there was no fixed model from which to copy. This also appears to be the case at the other sites where, when we have enough evidence to investigate, for example, at Montans, we find alternation between parapsidi and parabsides. Similarly, the representation of personal names across the sites shows differences, for example potters’ names are presented in the genitive at Blickweiler,190 whereas the nominative seems normal at La Graufesenque and Montans. And yet, the written material from the sites is relatively homogenous in terms of layout, its use of Latin symbols for numerals, which are a recurrent feature in even the most fragmentary documents, and, most significantly, its relatively skilled use of cursive Latin.

The similarities of practice are then not so great as to encourage us to think that manuals and firm rules circulated, but close enough to support the model of the movement of skilled literate workers.191 Indeed, across the Empire we find similar firing lists for other types of ceramic production, for example the manufacture of tiles and non-Samian tableware.192 The following example on a plate from Vayres (Gironde) (RIG II.2 L-27) dates to around 150 ce and was found in a large production centre producing céramique commune.193 The text seems to be entirely composed in Gaulish, including the vessel names, and follows the standard format. Roman symbols are deployed for the numerals, but these could, of course, be read ‘bilingually’, or in Gaulish.

At Saint-Bonnet à Yzeure (Allier) (RIG II.2 L-73) we find a first-century ce list of names on céramique blanche, from a workshop producing Gallo-Belgic wares. The exact purpose of the list is unknown—and it would be strange for the names in a firing list all to begin with the same letter as here—but it perhaps fits into this genre of administrative document. There are ‘oddities’ about the list, including the same letters with different forms and the alternation of -u(s) and -os in the endings of the names, but these could be explained by the nature of composition of such documents, as has been presented above for La Graufesenque. The difficult line 3, which has been read aoa xiiu[ by Lambert, might instead be read as Apa XLIV, or even Ada XLIV. Appa is attested as a name at Autun (AE 1901, 59) and there are numerous Celtic names beginning Ada-, so we could, at a push, consider an abbreviation. I see no reason, at least from the published information, to worry about the authenticity of the document.194 It is hard to imagine how, and why, a forger would come up with such a text.

A possibly first-century plate with another text (RIG II.2 L-72) was found in the nineteenth century in the Allier, this time from a funerary context in Vichy, and has been described as an account on terra nigra, but with uncertainty over context and interpretation.195 Lambert’s edition is as follows:

In its fragmentary state it is hard to be sure which language the graffito represents,196 though the symbol for the denarius confirms that it is some form of account rather than a firing list. Along with the day book of Atelia, which records the activities of the slaves of Atelia (Marichal 1988, no. 169), an account (no. 171), and a possible abacus (no. 110) from La Graufesenque,197 it demonstrates that pottery was used for a range of documentary purposes. There was a Samian ware kiln site at Terre Franche, near Vichy, from which this practice could have spread.

Ceramic production in Gaul and its subsequent trade spread millions of vessels around the western Empire every year and generated a need for documentation. Deru remarks that ‘the world of craftsmen is on the periphery of literate culture’,198 but the evidence seems to be pointing in a different direction. Not only was this world responsible for the majority of the earliest Latin inscriptions from Gaul (Section 5.4), and may have been a driver in the writing of Gaulish (Section 5.3), but we have also traced documentary practices across the ceramic-producing centres, and the stamps and signatures on vessels that made their way into the homes, barracks, graves, and sanctuaries across the provinces.199 Artisans left signatures on their mosaics, on their decorated Samian ware, and their figurines. It is a mode that may have helped to democratize access to literacy and it was certainly a locus for the written use of Gaulish (Section 5.6).

5.8 The Persistence of Gaulish: Social Contexts for Continued Use

The societal bilingualism that we see reflected through the bilingual documentary output from La Graufesenque is likely to have been a feature of Roman Gaul for a couple to several generations after the conquest, depending on the region and context. Bilingualism can, of course, depending on the relative ethnolinguistic vitality involved, be a very long-lived or a short-lived transition while language shift takes place. The Celtic vernacular of pre-Roman Gaul eventually experienced complete language death (the Breton language is the result of inward migration of British Celtic-speaking communities).200 The questions of the longevity of Gaulish, and its sociolinguistic patterning, have long exercised commentators. Given the links between identities, cultures, and languages, these conversations have sometimes been coloured by modern constructs of identity and nationhood, and/or confused by an unquestioning link between ‘Romanization’ and the uptake of Latin, coupled with simplistic views about the nature of ‘Romanization’ in Gaul. The usual presentation can be summarized as follows: the same selected ancient literary testimonia are wheeled out and presented either with little caution as to whether Gaulish is actually concerned, or with an abundance of scepticism, though apparently in both cases without detailed study of the original sources.201 The conclusion is usually broadly the same: Gaulish was still spoken in restricted contexts in the fifth/sixth centuries ce, long after it had ceased to be written. The epigraphic evidence, apart from, more recently, the bilingual texts, is usually overlooked, because tradition had set Gallo-Latin as a feature of the Caesarian to Flavian period.

The whole subject requires a complete overhaul. Much of the Gallo-Latin lapidary material was found in the premodern period, and, with few text-internal clues to help with dating, in contrast to Latin material, its dating followed from long-held views of the nature of the development of Roman Gaul. However, the discussion in Section 5.3 revealed the possibility that some of these lapidary expressions of Gallo-Latin may date to the second century ce. But, more significantly, it is the well-dated non-lapidary epigraphy that is increasingly encouraging us to see written Gallo-Latin as extending through to the third century ce, with even some minor continuation in the fourth, especially if the tile of Châteaubleau (RIG II.2, L-93) can be dated to the later third or fourth century ce, which seems not unlikely, given the combination of archaeological context and the ‘late’ features of the language. It is a comparatively lengthy text (eleven lines of cursive), which seems to be related to a marriage, and its diphthongization, final position -ou and -ie from -ū and -ē, suggests a reflection of late spoken Gaulish.202 The full range of epigraphic remains leads us to think that the period of Gallo-Latin should be revised to span largely from the mid-first century bce to the end of the third ce, with our best guess that perhaps complete disappearance in written form occurred in the (early?) fourth century ce.203

But, while the recent epigraphic analysis suggests we need to stretch the dating of Gallo-Latin, conversely work on the literary testimonia suggests we might not be able to place as much weight on them as we have in the past, since they have been shown by Alderik Blom and others to have a complicated relationship with Gaulish.204 The terms used: lingua Gallica/Celtica, Gallice/Celtice, sermo Celticus/Gallicus, and so on must be considered carefully in context,205 and the tone, tropes, literary allusion, and inference carefully assessed on a contextualized case-by-case basis. In some instances it seems more likely that a regional form of Latin, Gallic Latin, is implied rather than the Gaulish language itself, or that, if Gaulish is in question, it cannot necessarily be taken to demonstrate current use. To take perhaps the most obvious example, we can consider the Dialogi de Vita Martini of Sulpicius Severus (c.363–425 ce). In an interchange about this saint from Pannonia, Postumianus responds to Gallus’ hackneyed captatio benevolentiae that his uncultured Latin—sermo rusticus—might offend urbane ears with an amusing rhetorical response that he can celtice, aut, si mauis, gallice loquere ‘speak Celtic or, if you prefer Gaulish’ (1.26) as long as he gets on with discussing Saint Martin. Perhaps the Gallice might be a deliberate play on his name, but, in any case, the exhortation is not serious and Postumianus states that Gallus is perfectly competent. Among the educated, Gallice is more likely to refer to a local form of Latin, and indeed this self-consciously provincial Gallus later sets up a distinction between the language of the rustici Galli (i.e. himself) and the scholastici (i.e. Postumianus) and uses as the shibboleth a vernacular Latin term tripetias for standard Latin tripodas, rather than a Gaulish word. As Blom has recently remarked: ‘this dialogue, therefore, is greatly concerned with different styles and registers of Latin speech, commending the simple style that St Martin himself supposedly used.’206

The original sources of the examples must also be carefully considered. There has been a tendency to repeat the same set of material without due caution, with the deracinated examples living lives of their own. This can be seen in a passage of the Passion of Saint Symphorianus, known from an account probably composed around the middle of the fifth century ce. The well-known exclamation of the martyr’s mother as he was led to his execution in Autun, probably in the late third century, has generally been interpreted as a mix of Gallic Latin and Gaulish. However, this passage had been reconstructed by Thurneysen from just two manuscripts from a much wider set, and it seems that from the ‘tangle of early medieval textual confusion: evidence for Gaulish has not so much been found in the Passio as created from its raw material’.207 Most of the manuscripts in fact say that the mother delivered her words earnestly, sedula voce, rather than the by-now standardly quoted Gallica voce, and the careful edition of Chevalley prints the opening of her Latin speech as nate, nate Symphoriane, in mente habe Deum tuum ‘Son, Symphorianus my son, remember your God!’208 The first word nate could be Gaulish or, more probably, Gallic Latin, perhaps to lend some regional verisimilitude,209 but Symphorianus is described in the opening of the passion as ‘son of the noble man Faustus’ and ‘well-versed in scholarship and morals’, and in third-century Autun such a family is likely to have primarily spoken Latin. We could argue for a return to a traditional tongue on the part of the mother in a situation of extreme emotion, but there is little Gaulish left after the Quellenforschung. Discussions of the possible evidence for Gallic Latin, and possibly Gaulish, in this text must consider its composition in the fifth century and the manuscript tradition through which it travelled.

This is not to say that the standardly cited examples are irrelevant for consideration of the survival of Gaulish, but that, as always, the picture might not be straightforward: the sources might reveal more about Gaulish-influenced Latin, or about later commentators’ attitudes towards an already dying or dead Gaulish, than anything factual about the language and its use. It may, however, be no coincidence that in these ancient sources apparently evoking Gaulish, the oral and religious contexts reoccur several times. Religion in Gaul seems to encourage the spread of Latin, via the imperial civic religion, which ‘had the powerful visual element of writing and a capillary spread thanks to the infrastructure and propaganda established by Augustus and his successors’,210 but simultaneously local religious and ritual practice seem to have clung to the local language, a cross-culturally familiar phenomenon.211 In the Life of Severus Alexander (222–35), from the to-put-it-mildly problematic Historia Augusta,212 it is a Druidess, a leader of the old practices, who shouts at him in what is described as Gallicus sermo as he is leaving for battle: mulier Druias eunti exclamauit Gallico sermone: uadas nec uictoriam speres nec te militi tuo credas ‘A Druidess cried out to him in Gaulish as he left: “Go, but have no hope of victory, nor trust yourself to your soliders”’ (60.6).

We possess around a dozen examples of what appear to be defixiones or similar magico-ritual texts written in Gaulish, or in a mix of Latin–Gaulish, or in Gaulish plus magical features such as voces magicae.213 The dates of many of these examples are uncertain, and they are usually dated to the first two centuries ce, though the lead tablet found folded on the chest of a skeleton in a cemetery in the quarter of Saint-Marcel, Paris, if it is written in Gaulish, would be Constantinian (RIG II.2 L-105). The Rom tablet, whose language and interpretation are uncertain and may be in parts deliberately obfuscatory, might contain ‘a sizeable Gaulish element’,214 with links to the Late Gaulish features of the Châteaubleau inscription.215 This tablet has been dated on account of its script to the end of the third, beginning of the fourth, ce. A small tablet of gold with several lines of capitals framed by a square with triangular summit found at Baudecet near Gembloux (Belgium) (L-104) was found in a ritual context and has been dated to the second or possibly the early third century ce.216 There is no scholarly consensus on the linguistic and precise interpretative analysis,217 though a religious (broadly conceived) function is undisputed and the barred D is suggestive of a Gaulish-influenced epigraphic environment. Some have reconstructed a Gaulish medical incantation, with the same diphthongization (final -ū to -ou) that was deemed a feature of Late Gaulish in the aforementioned Châteaubleau tablet.218 If we skip forward to shortly after 400 ce, the Gaulish spells and charms that have been identified in the medical remedies compiled by Marcellus of Bordeaux219 have been argued to have no value for everyday spoken Gaulish, but instead reflect ‘ritual language, which might include garbled Gaulish elements’, which might add credence to the suggestion that Gaulish clung to the ritual domain.220 On this subject we might add the fourth-century so-called Grammaticomastix poem of Ausonius of Bordeaux,221 which in part depends on the Vergil-associated Catalepton II, famous for its evocation of the tau gallicum.222 Fiorenza Granucci argues that Ausonius is playing with a Gaulish maledictio in this poem, and, while this may not reflect any real knowledge of Gaulish, it nevertheless may reveal a later Roman perspective on the functions of Gaulish.223

The calendars of Coligny and Villards d’Héria, the first dating probably to 175–200 ce and the second to the second to third centuries ce,224 are, as far as we can understand, dependent on a local lunar organization of time, and are written in Gaulish, though features suggest that a Latin speaker has been involved. The choice of monumental bronze plaques and the nature of the execution, layout, and abbreviations make it clear that a Roman presentation has been adopted, but Roman fasti are not known on bronze. As Jay Fisher describes, the calendar reflects ‘a hybrid of multiple voices, oral and written, that were used to express identity in Roman Gaul in the late second century’ and ‘is at once resistant to and complicit with Roman ideas about Gallic culture’.225 In a similar context, two Gaulish words are also deployed in an otherwise entirely Latin funerary inscription found at Gélignieux (Ain), which provides for a funeral banquet that should be held on the fourteenth (petrudecameto) of every month of thirty days (tricontis) (CIL XIII 2494).226 The appearance of the formula memoriae aeternae at the beginning of the monument encourages a date in the third century ce. Another lapidary text with Gaulish religious elements, though probably dating to the first century ce, concerns a vergobretus, a senior magistrate, from Limoges, and offers a Roman-style lapidary euergetistic dedication, but with the term decamnoctiacis, a Celtic term referring to a festival lasting ten nights for the deity Grannos:227 Postumus Du[m]/norigis f(ilius) verg(obretus) aqu/am Martiam decam/noctiacis Granni d(e) s(ua) p(ecunia) d(edit) (AE 1989 521, 1991 1222) (note that we also find counting by nights in the Gaulish calendar).228 Indeed, the position of the vergobretos/vercobretos almost certainly developed from an existing position of religio-political power in the Iron Age communities (Caes. BGall. 1.16.5), and is attested in several Gaulish and Latin inscriptions.229

The ritual world often clings to local languages, in a reflection of its indigenous and deep-rooted nature (whether confected or not). The possibly gendered connection to the maintenance of local languages has also occasionally been made, indeed with specific reference to Symphorianus’ mother and Severus Alexander’s mouthy Druidess.230 The epigraphic texts that can be assigned to Late Gaulish repeatedly feature women.231 A Samian ware (Drag 37) mould made by a potter from Lezoux who was active during the late second and early third centuries seems to bear a Gaulish text, whose interpretation is very uncertain, but features a woman named Cal(i)a (L-70), and contains a code-switch into Latin at the end ([C]aleni oficina). The black vase inscription of La Place de Séraucourt in Bourges, found in the nineteenth century, has been dated by script to the third or fourth century (L-79), and is probably a religious dedication, though a funerary text cannot be excluded. It reads Buscilla sosio legasit in alixie magalu ‘Buscilla dedicated this for Magalos in Alixion’, and the subject is again a woman.232 Another inscription, this time, and unusually,233 painted on a vase from a fourth-century tomb in Étréchy (Cher), again seems to be Gaulish, and, though the text is uncertain, the several terms with feminine endings suggest women are again involved (L-80).234 The set of two dozen inscribed spindle whorls, half of which were found at Autun, the rest from adjacent regions, can be attributed to a female-dominated activity, that of spinning. An assessment of the known archaeological information and contextual clues led to a dating for these objects of 90–235 ce (the date given by linguists and epigraphists of the third/fourth century ce is simply based on their not unproblematic view that code-switching must show a language is dying).235 Like the graffiti from La Graufesenque, some of the spindle whorl texts are Latin, some Gaulish, and some a mix, though, unlike the firing lists, the code-switching and possibly translingualism, may well be deliberate, playing with language for effect. Given that women repeatedly feature in these and more securely later-dated Gaulish texts, we might wonder if they are keeping Gaulish alive in domestic, ritual, and perhaps other female-dominated contexts.

5.9 Concluding Remarks

In this chapter we have ranged over a vast amount of material, revealing what gains can be made through an approach that crosses the disciplinary boundaries between Celtic and Classical, and that combines sociolinguistics, epigraphy, archaeology, and history. We began with the earliest epigraphic texts from Mediterranean Gaul in the sixth to the second bce, which remained relatively restricted in terms of spread, numbers, and their trade-associated functions, but nevertheless witness the involvement of the local communities in various epigraphic practices and form the context for the adoption of Gaulish epigraphies. Parallels can be drawn with the Celtiberian region of the Iberian Peninsula, where, though Roman involvement may not have been the trigger for the epichoric writing, it was one of the contributing factors in its development (Chapter 2). The nature of epichoric epigraphies for Gaulish was assessed, and their dating reconsidered, alongside the scanty (and no doubt under-identified) remains of the Latin of Republican Gaul, in which the role of trade and officialdom are salient. The reflections of regionality and bilingualism, which become increasingly visible after the epigraphic boom beginning in the first century ce, were explored in detail, and the evidence for the end points of Gaulish epigraphy and language remodelled, playing down the strength of the evidence from literary sources and playing up the evidence from Gallo-Latin epigraphy itself.

The advantages of a more archaeologically informed epigraphy, which has encouraged attention on non-lapidary epigraphy and more caution with interpretations and dating, and the digital epigraphic data sets, have allowed us to build a more detailed understanding of sociolinguistic and epigraphic developments over time and space. However, we are constantly reminded of the limits of our evidence and our ability to interpret it. The dated Republican Latin inscriptions form a tiny set, but how many are lurking in the undated shadows? We have set out examples of bilingual texts involving Gaulish, but, with so few scattered examples, it is hard to derive much more than piece-by-piece insights; only when we consider the several hundred texts from the bilingual pottery of La Graufesenque, or the couple of dozen inscribed spindle whorls from eastern Gaul, can we begin richer analyses. Time and again we are reminded of the unavoidable bias towards those who can write and, in particular, chose to write in permanent ways. We learn about those who engage with Roman modes, and, although the language, images, and presentation can be regional, this is not a medium through which we are likely to see resistance, discontent, and unrest. Everyday writing is starting to form a more reasonable proportion of our epigraphic corpora, but it is still skewed by centuries of recovery and focus on the ‘monumental’. The elite and their freedmen are overrepresented, as are military men, officials, traders, and craftsmen in certain areas, whereas enslaved people and women are in most contexts underrepresented, and the vast majority of the non-elite rural population are barely heard.236 We discover from the pockets of Gaulish lexis in later dialects (of Alamannic German and Gallo-Romance) that Sprachinseln of Late Gaulish may have survived in the Jura, the Swiss Alps, parts of the Massif central, Wallonia, and, more controversially in Armorica, but we would not have known this from the epigraphy, since Gaulish texts are not extant.237 Or perhaps we might: if only the Latin texts had been consistently and carefully edited and digitized in a way that linguistic features of language contact could be easy to recover. There is still work to be done.

Metrics

View Metrics

More from Oxford Academic