10 Religion, Language Maintenance, and Language Shift Dedications, Cult Places, and Latinization in Roman Gaul Olivier de Cazanove and María José Estarán Tolosa 1 Introduction In religions whose very core is practice, such as those of the Roman world, where ritual effectiveness is aimed at as much through the appropriate accomplishment of gestures and acts as through the correct presentation of formulae,1 where, in short, ‘to do is to believe’,2 the question of language is fundamental. However, we have direct access to only a vanishingly small amount of Latin epigraphic data that explicitly describe the details of the rituals of Roman religious practice.3 And, when it comes to local religious practices in the provinces, we have to be content with indirect and fleeting glimpses, sometimes hard to interpret, of the persistence of spoken vernacular languages.4 Estarán’s work on this chapter was funded through the Fellowship Ramón y Cajal no. RYC2018­024089­I (Fondo Social Europeo/Agencia Estatal de Investigación). We would like to thank all those who provided valuable information and advice on different aspects of this chapter: Fabienne Creuzenet, Monique Dondin­Payre, Emmanuel Dupraz, Sylvia Estienne, Nicolas Laubry, and Mathieu Ribolet. We would like to thank Alex Mullen in particular for her valuable suggestions, translations, careful and constant monitoring and revision of the manuscript, from its inception to the final ver­ sion. We are, of course, responsible for all the opinions expressed here. 1 Scheid (1990; 2016); Beard, North, and Price (1998), 42–54; Rüpke (2018). On the same inter­ pretative line, with regard to the provincial religion of the Roman Gaul, see Derks (1998); Van Andringa (2002). 2 Scheid (2005). 3 The textual sources transcribe only quite exceptionally authentic ritual formulas, prayers, or protocols—for instance, those produced for the Arval Brotherhood or for the Secular Games. This is the case for Rome (Scheid 2007) and elsewhere in Italy (Iguvine Tables) (Weiss 2010). 4 While Augustine’s testimony on the need for preaching in Punic is clear (Lepelley (2005), and particularly well highlighted in the case of Antoninus of Fussala (Lancel 1983; McLynn 2010), the same cannot be said about the testimony of Irenaeus of Lyon, who claims to use the ‘barbarian dia­ lect’ (βάρβαρον διάλεκτον) for the day­to­day affairs of his diocese (Irenaeus, Adversus Haereses I, praef. 3). This passage can be interpreted in various ways: most recently, for Rochette (2021), 231–48, the ‘barbarian dialect’ is nothing other than Latin, in a context of erudite Graeco­Latin bilingualism. Olivier de Cazanove and María José Estarán Tolosa, Religion, Language Maintenance, and Language Shift: Dedications, Cult Places, and Latinization in Roman Gaul In: Social Factors in the Latinization of the Roman West. Edited by: Alex Mullen, Oxford University Press. © Olivier de Cazanove and María José Estarán Tolosa 2023. DOI: 10.1093/oso/9780198887294.003.0010 Religion, Language Maintenance, and Language Shift 207 Cross­culturally, the documentation of the religious domain can be a key factor in understanding certain linguistic phenomena, including language shift, maintenance, and revival.5 These phenomena are dependent on historical and sociolinguistic circumstances, ethnolinguistic vitality,6 and the strength of the links between religion, language, and other core values in every multilingual community. In the western Roman provinces involved in the process of Latinization, it seems that the epigraphic documents relating to local religious practices, broadly conceived to include the realm of magic, whether publicly or privately displayed, provide indications that the linguistic domain related to these practices was per­ haps one of the last to succumb to Latin.7 In tandem, across all the provinces, the imperial cult and more official and formal aspects of Roman religion were dis­ seminated and practised primarily, it seems, through the language of Rome, and may even have been a vector of Latinization. In this chapter we shall illustrate this double aspect of religion and language choice through a series of case studies from Gaul. We wonder what language might have been spoken in the sanctuaries of Roman Gaul, in the decades and centuries following the Caesarian conquest. Certainly, Latin would have been used for the celebration of public and official worship. But which language would have been deployed for private devotion? We do not have documents preserved that allow us to reconstruct a secure depiction of the linguistic choices in private religious practice, although we have a series of texts that reflect aspects, particularly of a ‘darker side’, of these devotions: the magical texts on metal, which are sometimes deposited in sanctuaries. Some of these are still written in Gaulish long after the conquest. However, in the current state of our knowledge, we are not in a position to know whether the type of lan­ guage that is preserved is one also used orally in the locally focused religious sphere and perhaps more widely in private spheres of communication or whether, on the contrary, its nature is specific to these ritual texts alone, and is no longer an oral medium. Defixiones and related texts represent, of course, a very particular epigraphic context, which must be treated with the utmost caution: the realm of ritual and magic can contain specific and sometimes usual linguistic practices,8 5 Pandharipande, David, and Ebsworth (2020). On the strong link between religion and its associ­ ated languages, the so­called religious classicals, see Fishman (1991), 360; Mesthrie (1999), 42; Gogonas (2012), 116. 6 For this concept, see Haugen (1972), Giles, Bourhis, and Taylor (1977). For the application of the concept in ancient sociolinguistics, see Mullen (2013a), 69–73, Mullen and James (2012). On objective linguistic vitality (the set of socio­structural factors that determine the strength of a certain ethnolin­ guistic group), see Giles and Viladot (1994); on the subjective (judgements and perceptions that dif­ ferent ethnolinguistic groups have in relation to these socio­structural factors), see Harwood, Giles, and Bourhis (1994). 7 For the concept of ‘private inscriptions’—that is, ‘not intended for public display’—we follow Beltrán’s definition (2015b), esp. 89–90, and Beltrán and Díaz (2018). 8 In particular, the use of incomprehensible or difficult to understand language: Blom (2012); Marco Simón and Gordon (2010). 208 Cazanove and Estarán Tolosa which are not the main focus of this chapter. Another issue directly related to the topic at hand, and which we will not go into here, is what has been called interpretatio Romana, which affects the Latinization/Romanization of local divinities through a series of particular mechanisms.9 In this chapter we focus instead on linguistic choice more broadly in religious dedications, and their relevance for understanding the Latinization of Gaul. This constitutes an interesting case study for several reasons (Fig. 10.1). First, Gaulish survived in epigraphic sources until at least the third century ce, and the literary sources push this chronology even later, into the fifth century (though by Figure 10.1 Map of Gaul with the main sites mentioned in the text. (Drawn by O. de Cazanove.) 9 As Ando (2005) pointed out in an important paper, these mechanisms go far beyond the purely linguistic level and simple translation; and they teach us at least as much about the specifically Roman attitudes to polytheism and their theology as about the ‘interpreted’ local gods. Religion, Language Maintenance, and Language Shift 209 this time the language was probably virtually extinct).10 This makes it the only Palaeoeuropean language to coexist in writing alongside Latin until well into the imperial period. Second, several of the later Gaulish texts and several of the few texts on stone seem to be magical and/or religious in character. Both factors make this region a privileged setting for the topic we explore in this chapter: the assess­ ment of the role of religion, broadly conceived, in linguistic change and mainten­ ance in Gaul during the process of Latinization. We of course recognize that the information we have is not anywhere near as complete as that available in sociolinguistic studies of present­day societies and that our hypotheses are based on very fragmentary information. Nonetheless, abundant work since the beginning of the twenty­first century, following the pioneering work of J. N. Adams,11 demonstrates that the field of research of ancient sociolinguistics and its application to past societies is viable,12 and offers fruitful interpretative avenues. The published Gaulish corpus contains around 350 coin legends and well over 700 inscriptions.13 Of these, approximately forty are inscriptions of certain religious or magical content, and many others could also be included in this count.14 They are attested between the second century bce for the earliest Gallo­Greek texts15 (the earliest Gallo­Latin are later, con­ temporary possibly with the Caesarian campaigns),16 and the third to fourth centuries ce for the latest Gallo­Latin examples.17 These provide the bad data that we must try to navigate.18 The Latin corpus is much larger, and constantly grow­ ing: at least 5,600 inscriptions of a religious nature from the Tres Galliae, Narbonensis and the Germanies, to which must be added about 110 defixiones. 2 Religion and Latinization in Narbonensis: The Epigraphic Record of Pre-Augustan Tolosa and the Earliest Temple with Central Plan in Gaul The first case study is taken from the western edge of Narbonensis. This province, created in the last quarter of the second century bce, received Italian immigrants early on who became involved in the growing economy of the region, especially focused on wine. The presence of these individuals presumably encouraged the appearance of the first documents written in Latin, whose dating Christol rightly 10 For the survival of Gaulish, see Blom (forthcoming). 11 Adams (2003a). 12 Knooihuizen (2006). 13 Mullen and Ruiz Darasse (2020), 759. 14 De Tord (2020), 843–901. 15 On this terminus ante quem, see Mullen and Ruiz Darasse (2020), 757. 16 Mullen and Ruiz Darasse (2020), 758. 17 The latest text certainly in Gaulish is probably the tile from Châteaubleau (Mullen and Ruiz Darasse 2020, 758); for indirect testimony on the survival of Gaulish, see Meissner (2009). 18 Labov (1994). 210 Cazanove and Estarán Tolosa pointed out may be earlier than generally thought.19 In this territory, writing in the local language coincided with the rise of Rome, though the local communities used the Greek alphabet, giving rise to so­called Gallo­Greek epigraphy, of which more than 400 inscriptions and 70 coin legends have been published. In a territory so apparently open to external Mediterranean influences and that had already entered the orbit of Rome,20 the Augustan ‘new imperial cul­ ture’ perhaps in a sense had its way paved. The appearance of dedications to the emperor and his family is a good indicator of this ‘new religious order’ described by Fishwick, Woolf, and others.21 Augustus is present in the Jardins de la Fontaine in Nîmes as early as 25 bce, in the spring sanctuary of the topical god Nemausos,22 from where a few dedications in Gaulish language to the local divinity come, perhaps from only a few decades before.23 A generation later, in the forum of Nîmes, the Maison Carrée was built (the architectural and stylistic analysis of the temple confirms a dating in the first decade of our era), with a dedication to Gaius and Lucius Caesar, according to the reading proposed by Séguier more than 250 years ago, based on the holes for the missing bronze letters.24 This typical Roman pseudoperipteral temple was installed in the heart of Nîmes by its own elites.25 The efforts of senior locals to demonstrate their Romanity were likely to have been important to the spread of Latin among at least some Gaulish speakers. These displays, which we combine, for convenience, under the label ‘imperial cult’, undoubtedly played a role during this process, at least in the official and public linguistic domains. In Belgic Gaul, monuments to the prematurely deceased Gaius and Lucius Caesar are raised on the forums of cities that are just starting to be urbanized—for instance, Reims­Durocortorum and Trier.26 Therefore, the first expressions of dynastic loyalty written in the official language could be seen and read in cities just emerging out of the ground and still under construction. To have reached this point in the linguistic choice of the local elites, it must be remembered that Latin had been spoken in Narbonensis for decades by Italian settlers and merchants. An example of these first steps is provided by the oldest Latin dedication, which can be linked to the religious sphere in Gaul, which is quite different in nature from the inscriptions relating to the imperial cult. It 19 Christol (1999). 20 See Mullen (2013a) for the interactions between southern Gaul and Mediterranean communities. 21 Fishwick (1987–2005); Woolf (1998). 22 CIL XII 3148–3149; Rosso (2006), 193. 23 Roth­Congès and Gros (1983); Lejeune (1994). 24 Christol and Darde (2009). 25 Amy and Gros (1979). The Maison Carrée is a provincial transposition, on a small scale, of the great pseudoperipteral temples of the Urbs: Apollo in Circo and Apollo Palatinus: Gros (1976), 119–22. 26 The overall impression that emerges from a recently published series of studies on the birth of city capitals in Gallia Comata (Reddé and Van Andringa 2015) is that the establishment of urban grids and then the progressive urbanization of the capitals came after Augustus’ long stay in Gaul from 16 to 13 bce, the census of Drusus and the dedication of the altar in Lyon in 12 bce. Religion, Language Maintenance, and Language Shift 211 comes from the margins of the province but from a settlement particularly interesting with regard to the multilingual epigraphic record: Tolosa (Fig. 10.2).27 Here we shall not enter into exhaustive detail about the long debate concerning the exact location of pre­Roman Tolosa before its move to the current site of Figure 10.2 Sites along the Garonne River: the oppidum of Vieille­Toulouse, the settlement of Saint­Roch, Roman Tolosa. 27 Pailler (2002b); Pailler, Darles, and Moret (2015); Provost, Pailler, et al. (2017). 212 Cazanove and Estarán Tolosa Toulouse around 10 bce. The oppidum of Vieille­Toulouse, a little less than 7 km to the south, long seemed the best candidate. Then, the discovery of numerous wells filled with wine amphorae in the Saint­Roch district of Toulouse suggested that this site, much closer to the Roman city, was in fact the settlement of the Volcae Tectosages, and would have controlled the lakes where the famous aurum Tolosanum was kept under water before being stolen in 106 bce by the consul Q. Servilius Caepio.28 Also in the Saint­Roch district, the extensive excavations of the former Niel barracks (between 2009 and 2011) revealed more clearly the commercial function of this valley site.29 The hypothesis of a polycentric organ­ ization was then advanced, with several distinct hubs: one or two hilltop oppida, and an emporium on the bank of the river, would have finally given rise to a sin­ gle Roman city. Today, the refinement of chronologies leads to a clearer picture. The site of Saint­Roch is abandoned shortly after 100 bce, whereas the oppidum of Vieille­Toulouse goes on, and even enjoys a period of strong development from the middle of the first century bce. In fact, in a house with sophisticated architec­ tural features, portico, garden, and masonry basin, which dates to between 40 and 10 bce, some inscribed bone tokens were found, as well as writing materials, sty­ luses, and seal boxes,30 which show a degree of literacy and the use of Latin.31 The two main settlements that predate Roman Tolosa were, from the second century bce, decidedly multilingual places. The Graeco­Italic amphorae of Vieille­Toulouse, dated between 175 and 130 bce,32 exhibit various tituli picti consisting of names of individuals that, with a single exception,33 are written in Palaeohispanic script, followed by what seems to be a numeral and a unit of measurement. The names in these texts are of Latin, Celtic, and Iberian origin. Whatever the precise interpretation of these texts, what interests us here is that individuals of diverse origins and languages (Gauls, Iberians, Greeks, Italians) were all involved in the wine trade, in which Tolosa was a key player.34 In the Saint­Roch district, inscriptions (on Dressel 1a amphorae and Campanian A black gloss ware) were also found,35 dating to between 125 and 100 bce, which illustrate this environment of mobility between merchants and multilingualism, since they occur in levels with Iberian inscriptions and Greek names also documented in Marseilles.36 This period coincides with intense activity at the site and with the dating of most of the coinage. This set of evidence contains texts in Iberian, Greek, and Latin, in their respective writing systems, written in the same period. These findings reflect the coexistence and collaboration of merchants of 28 Moret (2008). 29 Verrier (2019). 30 See Mullen, this volume, n. 73, for the debate on the functionality of seal boxes. 31 Ugaglia (2013), 38–46; Gardes (2017). 32 See Gorgues (2010), with previous bibliography. 33 Estarán (2016), I11. 34 On this point, see Vidal and Magnol (1983); Lejeune (1983); Gorgues (2010); De Hoz (2011), 195; Estarán (2016). 35 Moret, Ruiz Darasse, and Verrier (2015). 36 Gorgues (2010); Moret, Ruiz Darasse, and Verrier (2015). Religion, Language Maintenance, and Language Shift 213 diverse origins, languages, and cultures in this environment, attracted by the strength of this centre. Half a century later, Vieille­Toulouse offers a more homogeneous linguistic documentation. The Latin inscription CIL XII 5388 was found in 1879 on the Vieille­Toulouse site and dates to 47 bce,37 with the consular date, incomplete but restorable (Fig. 10.3). It involves the dedication of the building of a temple (aedes), of a base (basim), and a solarium (we will return to this term) made by dedicants, mostly of lower social status. For a long time, it was considered the oldest Latin inscription in Gaul, though this place now belongs to the more recently dis­ covered Domitius Ahenobarbus milestone.38 The restoration of missing sections and the original dimensions of the stone are not certain; three visualizations are provided in Figure 10.3. Figure 10.3 Three possible reconstructions of the inscription of Vieille­Toulouse CIL XII 5388; the lower is that cited in the main text. (Drawn by O. de Cazanove.) 37 Close to the ‘Borde Basse’ farm: Provost, Pailler, et al. (2017), 170–3. The findspot of the inscrip­ tion is shown in Fig. 10.2. 38 CIL XVII/2 294; AE 1952, 38; Duval (1951). 214 Cazanove and Estarán Tolosa Following the third reconstruction, the text of the inscription can be recon­ structed as follows: [P(ublio) Vatinio Q(uinto)] Fufio co(n)s(ulibus) [heisce magistreis a]ẹdem basim et solari[um] [—faci]ụnd(um) coir(auerunt) [—]us Cirrat(us) Diallus Atti P(ubli) s(eruus) [—]ṣ Ṣurus Hilar(us) Flaui M(arci) s(eruus) l(eiber) c(oerauit) [—]ior(—?) L(?) L(?) s(eruus) l(eiber) c(oerauit?) Philodam(us) Volusi C(ai) s(eruus) [—]nus Trebi L(uci) s(eruus) P{h}i[n]dar(us) Curiati Ap(pi) s(eruus) Under the consulship of Publius Vatinius and Quintus Fufius, (the magistri) saw to the construction of . . . a temple, a base and a sundial: (1st column, not preserved: four individuals missing) (2nd column, incomplete) . . . us Cirratus; . . . s Surus; . . . slave of the Lucii Ior . . . (he saw to the construction once freed); . . . nus, slave of Trebius Lucius (3rd column) Diallus, slave of Publius Attius; Hilarus, slave of Marcus Flavius (he saw to the construction once freed); Philodamos, slave of Caius Volusius; Phindarus, slave of Appius Curiatius. (Translation: de Cazanove) Although fragmentary, it is highly plausible that this inscription concerns a collegium, which mostly consisted of slaves.39 Some of them appeared to have been freed between the time when construction was contemplated (perhaps as a result of a vow) and the completion of the works. We have similar inscriptions from Carthago Nova, Tarragona, Capua, and Minturnae, as well as on Delos40 and Cora.41 In this particular inscription, the members of the collegium organized the construction of structures relating to the practice of religion, including a solarium, which usually refers to the terrace of a tomb, but may, in this example, refer to a horologium solarium,42 a sundial. As stressed by the recent literature on this topic,43 it seems that we are dealing with Italian merchants settled in Toulouse (and first in Narbonne, where the same names are partially attested in the Augustan period), involved in the wine trade 39 On the collegia, see Díaz (2004). 40 Díaz (2004). 41 CIL X 6512, 6513, 6514; Palombi (2012). Here, magistri of the collegium of Bona Mens and slaves offer a substantial sum (probably a total of more than 8,000 sesterces) to build the monument on which the inscription was to be placed. Three of the ten magistri are slaves who were freed between the fundraising and the supervision of the works. Hence the recurrent formula seruus leiber coerauerit, found three times (in full) in Cora and at least twice (abbreviated) in Vieille­Toulouse. 42 Horologium solarium: AE 1972, 168 (Macerata, Villa Potenza). This can be part of the equipment of a place of worship such as that of the temple of Apollo at Pompeii (CIL X 802). A good parallel is offered by an inscription from Mayen (near Koblenz) in Germania Superior (CIL XIII 11978a). Despite the fact that the inscription is mutilated, the terms aedes and horologium are easily restorable in the inscription from Germania Superior. 43 Tran (2014); Bonsangue (2016); Agusta Boularot (2017), 316–19. Religion, Language Maintenance, and Language Shift 215 along the ‘Gallic Isthmus’—a trade well documented by the huge quantities of wine amphorae and Campanian ceramics found on the two sites that antedate Roman Toulouse (Saint­Roch quarter and Vieille­Toulouse). This inscription, consequently, must be understood as one of the first publicly displayed expres­ sions of the munificence of individuals settled in the province, mainly slaves and freedmen enriched by the thriving wine trade. Their mutual cooperation allows them to perform acts of euergetism, such as the one seen here, in which a temple is erected in honour of an (unfortunately now anonymous) divinity. Some 200 metres south of the place of discovery of the inscription, the Baulaguet temple was found and excavated between 1972 and 1974.44 It has a cen­ tral plan, perfectly recognizable, with a square cella and a peripheral gallery around it—in short what is conventionally called a Gallo­Roman temple, or a fanum (Fig. 10.4). It is in fact the oldest known temple of this type.45 The strati­ graphic and archaeometric data locate its construction after the beginning of the first century bce, in any case before 20 bce, then an occupation in the second half of the first century bce, and finally an abandonment towards the end of the same century. It is, therefore, exactly contemporary with the inscription. It is conse­ quently tempting to attribute the inscription to the small place of worship, and to identify the aedes mentioned in the text with the Gallo­Roman temple of Baulaguet. Even if it was not the case, it is worth noting that there was more than one fanum temple in Vieille­Toulouse, and that it was therefore the main 44 On the Baulaguet temple, see Vidal (2002), 113–18; Provost, Pailler, et al. (2017), 170–2. 45 Vesly (1909) was the first to recognize the recurrence of a particular type of temple in Roman Gaul, characterized by a square cella embedded in a gallery of the same form. He called this type fanum—mistakenly, because this term, as is well known, is generic. But, if the designation itself is inappropriate, the existence of a specific architectural category of religious buildings is beyond doubt. Over the past century, the number of identified fana has grown steadily, numbering now around a thousand; see Fauduet (2010), 22–41. Moreover, they are not only square, but also polygonal or circu­ lar, but always with a centred plan. Their elevation, with a tower­shaped cella overhanging a peripheral gallery with a single pitched roof, can be reconstructed mainly thanks to the cellae still preserved at more than 20 m high at Autun (so­called temple of Janus: Parlasca 1998), and Périgueux (‘tour de Vésone’). Their areal distribution is exclusively related to the north­western provinces (Gaul, the Germanies, Britannia: Derks 1998, 134–85). For this reason, they have sometimes been called ‘tem­ ples of indigenous tradition’ or ‘Romano­Celtic’—again, both qualifications are inappropriate, because they cannot be found to have a secure local (‘Celtic’) background, despite numerous attempts to assert it. Most recently, Poux and Fichtl (2019), in an article on fana, theatres, and villas with a subtitle ‘Trois emprunts protohistoriques aux origines de l’architecture gallo­romaine’, try to combine both local and non­local origins, while omitting the key example of the temple at Vieille­Toulouse. The Italian paral­ lels that they invoke (in particular the Lucanian temples, on which see Cazanove 2009a) are in fact ‘faux­amis planimétriques’ (Cazanove 2020, 222–3, fig. 1)—i.e. their ground plans may look similar, but the super structures were completely different. The fana (or ‘temples with central plan’, as it would be preferable to call them) are in fact identifiable in Gaul only from Roman times onwards. Therefore, rather than focusing on the search for protohistoric origins (which cannot be excluded but cannot be proven), it would be better to study the extreme standardization of this plan and its diffusion from the beginning of the Roman period. In any case, the importance of the Gallo­Roman temples with a cen­ tral plan cannot be underestimated. Their number and visibility certainly make them the main marker of the western provincial religious landscape. 216 Cazanove and Estarán Tolosa Vieille Toulouse Toulouse N 0 1 2 Foundation pebbles Bricks Demolished area Edge of foundations Columns Restored section 3 4 5 metres Undisturbed clay Figure 10.4 The Baulaguet temple in Vieille­Toulouse. (After a plan by M. Vidal (Gallia (1976), 483).) architectural model available: a second temple with cella and peripheral gallery was seen near the first one, a few years later, in a trial trench.46 The combination of evidence is striking: one of the oldest Latin inscriptions of Gaul, and the oldest mention of an aedes from this area, probably referring to the oldest known Gallo­Roman temple, which lies at the southern edge of the distribution area of the ‘fanum­type’ temples. Does this mean that the probably Italian­origin dedicators of Vieille­Toulouse had a local architect build a temple 46 Ugaglia (2013), 57–8. Religion, Language Maintenance, and Language Shift 217 ‘of native tradition’ precisely because it was the only model available? This seems unlikely, for two reasons: first, it is no longer believed that the (improperly) so­called Romano­Celtic temples have protohistoric antecedents; at least they are not archaeologically attested.47 All the examples date from after the Roman conquest. Second, the building techniques of the Baulaguet temple link to Late Republican Italy: columns and intercolumnations made of bricks, floors of opus signinum. Just a short distance from Toulouse, the recent excavations of the Cornebarrieu baths, dating from the first half of the first century bce, offer another striking example for the early adoption of Roman lifestyle, for new tech­ nical solutions (in part similar to those implemented in Baulaguet) and perhaps evidence for mobility of craftsmen. The Cornebarrieu balneum looks astonish­ ingly like a series of Italian baths, as found at Musarna, Sperlonga, Crotone, and so on.48 The Vieille­Toulouse issue thus opens new perspectives on the origin and the first developments of the temples with central plan, which will become the archi­ tectural form par excellence of the places of worship in the north­western prov­ inces. Pursuing this line of enquiry would take us beyond the topic of this volume;49 however, the puzzling issue of the temples with central plan deserves to be reopened. In any case, it is striking that, in Vieille­Toulouse, the temple appears very early and at the same time (and with the same agents) as the Latin language. Latin is a lingua franca of the Italian (and possibly other) immigrants (it could also, of course, be the mother tongue of some of them),50 and was used in order to consecrate and dedicate religious infrastructures in provinces already in the Republican period. Specifically, in the case of Vieille­Toulouse, this linguistic choice is contemporary with the substantial Roman­style urban and material cultural transformations in the oppidum of the Volcae Tectosages in the first century bce. 47 See n. 45. 48 Viers and Veyssière (2012). 49 Only a few key examples can be indicated here: in Montferrand, which corresponds to the site of Elesiodunum mentioned by Cicero in the Pro Fonteio, on the wine route between Narbonne and Toulouse, a temple with a central plan was seen in aerial photography and cannot therefore be dated, though a series of related finds date back to the last century of the Republic: Dressel 1 amphorae and sling bullets inscribed in Latin, with clear allusions to the famine of the siege of Perugia, in 41 bce (Passelac 2006; 2016). At Magalas, 40 km north­east of Narbonne, excavation uncovered a small tem­ ple dating to 30–20 bce, within an Iron Age ditched enclosure. The reconstruction proposed in the interim report (Ginouvez et al. 2016) makes it a hybrid building, with a fanum roof overlooking that of the peripheral gallery. But this reconstruction relies on very few elements. It would be equally pos­ sible to propose parallels with the peripteral temples of the Iberian Peninsula, or with that of Izernore in the Jura mountain. And, finally, in Cisalpine Gaul, the recent publication of a temple at Marano di Valpolicella has relaunched the debate on the existence of cult buildings of the fanum type in Italy, even if none of the cases invoked appears conclusive (Bruno and Falezza 2015). 50 Estarán (2019b). 218 Cazanove and Estarán Tolosa 3 Aspects of Religion and Latinization in Gallia Comata: The Persistence of Gaulish in the Epigraphy of Alesia The second panorama, this time from the core of Gallia Comata, will allow us to consider, thanks to the archaeological and epigraphical documentation, the per­ sistence of Gaulish in the first and second centuries ce, the resultant Gallo­Latin bilingualism, differential Latinization, and local Latins. We will focus on the rela­ tionships between language and the religious sphere, understanding the latter in a broad sense, including both publicly visible and private and intimate devotions, and taking Alesia’s epigraphic record as a case study. Alesia is the oppidum of the small Gallic people of the Mandubii (Fig. 10.5), caught between two apparently more important groups, the Lingones to the north, and the Aedui to the south. The settlement can be archaeologically identi­ fied from the years 100 or 80–70 bce (La Tène D1b–beginning of La Tène D2a), thanks to a few stratified contexts within the future Gallo­Roman town (excava­ tions at the place called En Curiot, others in front of the monument of Ucuetis and around the ‘forum’),51 and now, more notably, thanks to the new excavations at La Croix Saint­Charles, on the outskirts of Mont Auxois, which revealed a ditched enclosure for banquets of meat and imported wine.52 Alesia is well known because of the siege of Caesar, which ended with the surrender of Vercingetorix in the autumn of 52 bce. And, finally Alesia became a prosperous Gallo­Roman town, though it never reached the rank of civitas capital. Nevertheless, its urban adornment (theatre, temple, basilica, forum) and its extension (97 ha) rival those of Langres (Andemantunnum), capital of the Lingones. Alesia and the Mandubii seem originally to have been included in the territory of the Lingones, and then in that of the Aedui in 69 ce, according to general consensus.53 The Lingones may have been punished for finding themselves on the wrong side of the Roman polit­ ical situation, that of Vitellius and Julius Sabinus. Alesia offers a privileged field of observation for studying the Latinization of religious written expressions and the persistence of the Gaulish language. Indeed, the epigraphic corpus of the city counts seventeen or eighteen inscriptions in Gallo­Greek: two on stone, fourteen graffiti on ceramic, and one or two on lead; one inscription in Gaulish written in Latin characters (the famous dedication of Martialis son of Dannotalos).54 Of the well­over 200 texts in Latin, the vast majority are found on ceramic and comprise mostly stamps on terra sigillata. About fifteen Latin religious dedications are datable to the second to third centuries ce, with one or two possible exceptions, as we shall see.55 51 52 54 55 Creuzenet (2010); Creuzenet and Olivier (1994); Bénard and Bénard (1997). Cazanove et al. (2012), 133–40; Golosetti et al. (2019). 53 Reddé (2003). Numbers taken from RIIG (accessed 3 May 2021). Information from the LatinNow database; see https://gis.latinnow.eu/. Figure 10.5 Alésia plan (J. Vidal (Gallia (2012), 143). Reproduced with kind permission.) 220 Cazanove and Estarán Tolosa The epigraphy presents an unusual combination: the writing of Gaulish in the Greek alphabet in the interior of Gallia Comata (in a location far from the main area of Gallo­Greek epigraphy in the lower Rhône Valley) and the use of the Latin alphabet to write both the local language and Latin. It is, therefore, one of few sites in Gaul to provide evidence of Gallo­Greek, Gallo­Latin, and Latin. Michel Lejeune postulated a linear evolution to try to describe the epigraphic habits: Thus, for local epigraphy, we arrive at a chronological distribution between three phases, approximately datable: a) inscriptions in Gaulish language and Greek alphabet (still the custom in the Claudian period, to which probably belongs the text of Samotalos [. . .], as well as various minor texts); b) inscriptions in Gaulish language and Latin alphabet. The main testimony is the dedication of Martialis (a brief transitional phase, covering one or two generations, corresponding roughly to the Flavian period); c) inscriptions in Latin language and script (exclusive use from the Antonine period). Nothing informs us about the condi­ tions and duration of a purely oral survival of Gaulish in Alise during the second century ce, and possibly beyond.56 This rather schematic view, which tries to depict the different epigraphies as not overlapping temporally, is contradicted by what is observed on neighbouring sites. Lejeune must have been aware of this himself, having published a stele of the sanctuary of the sources of the Seine (about 15 km further east) with a Gallo­ Latin dedication in the pediment, and the Gallo­Greek signature of the artisan on the prominent band below,57 proof that the two epigraphic practices overlap (Fig. 10.6). The artisan, Dagolitous, belonged to a recently identified sculptor’s workshop (‘Seine 2’), dating from the third quarter of the first century ce.58 Furthermore, from Alesia itself, there is perhaps evidence that Latin is being used early on in the epigraphic record. There is a fragment of a marble inscription in the Alise Museum with a possible dedication to Tiberius (and Rosmerta?), with elements of the emperor’s title, which would make it the first Latin text from Alesia and, moreover, of religious content (Fig. 10.7).59 For the period of use of Gallo­Greek at Alesia, we can rely on some archaeo­ logical data. Some graffiti can be dated, with the help of stratigraphy and ceramic typology: one is pre­Augustan, some are Augustan, Tiberian, and Claudian,60 while the most recent one belongs to the ceramic deposits found in the temple area (Fig. 10.8).61 A few metres away there was a portico. Its remains have been 56 Lejeune (1979), 259 (our translation). 57 Lejeune and Martin (1956); RIG II.1, L­13. 58 Lamy (2015), 231. 59 CIL XIII 2876; Provost (2009a), 367. 60 RIG I, G­554 (Augustan), G­555 (Tibero­Claudian); from the strata predating the Ucuetis monument: Creuzenet and Olivier (1994). 61 Le Gall and Sénéchal (1974), 208–9; Lejeune RIG I, G­268, fig. 325; Provost (2009a), 373 (Claudian or Neronian). The same deposit also contained an inscription in Latin: a jug bears the graffito Virtus, but it is probably an anthroponym, not the theonym (Le Gall and Sénéchal 1974, 213). Figure 10.6 The stele of Dagolitous, found in 1953 at the sanctuary of the Seine Springs. (From Gallia (1954), 476. Reproduced with kind permission.) Figure 10.7 CIL XIII 2876. Dedication to Tiberius (?) with mention of Rosmerta (?). (Photograph and drawing by S. De Ricci, reproduced with kind permission. Reconstructed text by O. Hirschfeld and S. Reinach (Pro Alesia, I (1906–7), 25).) 222 Cazanove and Estarán Tolosa Figure 10.8 Gallo­Greek inscription G­268 on a ceramic vase (Claudius–Nero) from the temple area at Alesia. (Le Gall and Sénéchal 1974, 209; photograph by F. Creuzenet, reproduced with kind permission.) partially preserved under the basilica. Some fragments of entablature probably belong to it. They are carved with three severed heads in relief, pierced with arrows, of African type.62 It has also been proposed that one or two statues of gods or ‘heroes’ should be restored in the same place, one of them standing near a ram, and a monumental Gallo­Greek inscription using the same variety of limestone, studied by Michel Lejeune.63 The chronology of these sculptures is open to discussion; in any case, they cannot have been dismantled and buried before the middle of the first century ce, since they were discovered ‘au­dessus d’un sol’ dated from the time of Claudius and Nero.64 Therefore, the Gallo­Greek epigraphy, both monumental and on instrumentum, goes on until the Neronian period at least. In the sanctuary of Apollo Moritasgus, two thin fragments of lead were found during the previous excavations made by Emile Espérandieu in 1909.65 The place of the find is significant: next to the octagonal temple, built at the time of Claudius or Nero, and in the midst of offerings on bronze sheets (eyes, phallus, breasts, and so on) that date from the Flavian period.66 62 Olivier (1980). 63 RIG I, G­257, figs 311–14. 64 Le Gall (1990), 138; Toutain (1912); Provost (2009a), 361. 65 Espérandieu (1910), 273. One of them bears a name in Greek letters: karomaros (G­269), the other (G­270) presents a fragmentary and difficult to read text. Lejeune (RIG I) thought that they were two different inscriptions. In fact, they were found next to each other and are probably two fragments of the same text. It is possible to establish this fact by studying Esperandieu’s unpublished excavation journals. 66 Cazanove and Dondin­Payre (2016), 110, fig. 2. Religion, Language Maintenance, and Language Shift 223 Let us turn now to Gaulish epigraphy in Latin alphabet, which is represented uniquely in Alesia by the inscription of Martialis son of Dannotalos (Fig. 10.9):67 Martialis Dannôtali ieuru Ucuete sosîn celicnon etic gobedbi dugiìontiìo Ucuetin in Alisiìa. Martialis, son of Dannotalos made this building? for Ucuetis and for/with? the blacksmiths who worship Ucuetis in Alesia. The god Ucuetis appears twice on the inscription, which has given rise to many exegeses.68 There is consensus on the meaning of gobedbi, which refers to Figure 10.9 The inscription of Martialis, son of Dannotalos (Conseil Départemental de la Côte­d’Or, Musée Alésia, deposit of the Musée Municipal d’Alise­Sainte­Reine (D.2011.1.94)). (© Olivier Champagne, reproduced with kind permission.) 67 RIG II.1, L­13. 68 Most recently, Dupraz (2021), with bibiography. Some of the main translations previously put forward in a century of scholarly research are those of Thurneysen (1908): ‘Martialis, Sohn des Dannotalos, machte für Ucuetis dieses celicnon und für die Priester, die den Ucuetis bedienen in Alisia’); Lejeune (1979; 1996; RIG II.1): ‘Martialis fils de Dannotalos dédie à Ucuetis cet édifice avec les forgerons qui vénèrent [ou façonnent] Ucuetis en Alise’; Stifter (2011): ‘Martialis (son) of Dannotalos offered this edifice to Ucuetis and to/with the smiths who worship (?) Ucuetis in Alisia.’ 224 Cazanove and Estarán Tolosa blacksmiths,69 but not about what blacksmiths offer, though for a majority of scholars celicnon is a building.70 It is usually identified with the building furnished with a porticoed court and a facade overlooking the forum, in which was found in 1908 another dedication to Ucuetis and his consort Bergusia on a bronze vase, this time in Latin, dated to the second century ce.71 Before going further, we should note that the identification of the so­called Ucuetis monument is perhaps less solid than it seems. On the one hand, the Gaulish inscription of Ucuetis was found in 1839, in the first exploratory trenches on the site, about 75 m further south.72 The place of discovery seems very close to another small sanctuary, with a statue of Sucellus, dedication to Victory, and architectural elements.73 On the other hand, the vase with Latin dedication to Ucuetis and Bergusia was found in a basement, in the middle of an accumulation of various metal objects, iron and bronze: nails, keys, locks, rings, handles, and so on. The editors of the monument thought they were ‘ex­votos, offerings deposited . . . in a cultic room, as a testimony of grace or gratitude to the protective deities of the blacksmiths’.74 This is not likely. We might assume, instead, that we are dealing with a large foundry deposit. Therefore, the vase, like the rest of the metal, is in a secondary position. It is legit­ imate to raise these doubts, especially since the ‘Ucuetis monument’, interpreted as the schola of the corporation of blacksmiths, is considered the best provincial counterpart of the collegium headquarters of Italy, those of Ostia, for example. But it is better for the moment to leave the question open. Given the uncertainty, we should avoid using the dating of the monument of Ucuetis (late first century ce) to establish the chronology of the Gaulish inscription of Martialis. The evidence for intrinsic dating (hederae distinguentes, palaeography) is also not strong. Apart from the Alise inscription, the Gallo­Latin inscriptions on stone with the most careful layout all come from the territory of the Aedui: Autun,75 Auxey,76 and Nevers.77 All of them show, in fact, a good degree of famil­ iarity with the codes of Latin epigraphy (Fig. 10.10). However, these comparable inscribed objects were found in the nineteenth (Autun and Auxey) and fifteenth (Nevers) centuries and cannot be dated archaeologically. So Lejeune’s proposed dating for the Martialis inscription (Flavian period, last third of the first century ce) can be considered only an approximation, as archaeological confirmations are missing.78 The dating of this and much of the Gaulish epigraphy has 69 The ending of this word has been interpreted as a plural oblique case marker, possibly an instru­ mental (Dupraz 2021, 335–8). 70 Dupraz (2021), 334. 71 CIL XIII 11247; Provost (2009a), 401–9, with previous bibliography. 72 Maillard de Chambure (1841), 126, A 47. 73 Provost (2009a), 423–6. 74 Martin and Varène (1973), 159. 75 RIG II.1, L­10 = CIL XIII 2733. 76 RIG II.1, L­9 = CIL XIII 2630. 77 RIG II.1, L­11 = CIL XIII 2821. 78 It may just be noted, en passant, that there are some formal similarities with later inscriptions such as that of the Lingones fratres at Vertault, usually dated to the middle of the second century ce (CIL XIII 5661); Provost (2009b), 258–9. Religion, Language Maintenance, and Language Shift 225 Figure 10.10 Three Aeduan Gaulish inscriptions adopting the layout of Latin epigraphy: a. Autun (RIG II.1 p 129 = CIL XIII 2733); b. Nevers (RIG II.1 p 141 = CIL XIII 2821); c. Auxey (RIG II.2 p 123 = CIL XIII 2630). For these locations, see Fig. 10.1. 226 Cazanove and Estarán Tolosa traditionally been based, at least in part, on a hypothetical rapid Latinization of this territory. However, detailed analysis of the archaeological context of several non­lapidary Gaulish inscriptions has pointed, in some cases, to dates in the second and third centuries ce.79 In Alesia, the use of Latin for religious dedications perhaps really becomes prevalent only from the second century ce, as is demonstrated by two small series of inscriptions found in the Moritasgus sanctuary: handled pateras (or small pans) used to draw water,80 and anatomical ex­votos made of limestone (feet, trunk, knees). The inscribed handled pateras contain religious dedications that are valuable in considering the language choices of worshippers, as they may reflect the language used for worship. The inscribed pateras, in bronze and some­ times silver, are frequent in Gaul (Evaux­les­Bains,81 Châteaubleau,82 Mandeure,83 Berthouville,84 but are not exclusive to this territory: similar examples have been found in the sacred spring of Bath85 and in Celtiberia.86 The most interesting ones for our purpose are two almost identical pans, type Gödaker (Tassinari G.3100 and 3200), both of which carry dedications to the god Alisanos: that of Viévy (Visignot), in the territory of Aedui, is in Latin, whereas that of Couchey, in the territory of Lingones, is in Gaulish (Fig. 10.11).87 The latter object can probably be dated to 60–90, but may well still have been in use in the first half of the second century.88 The Viévy pan must be roughly contemporary, though its text deploys a votive formula that has been dated to the middle of the second century ce.89 There is, therefore, still in this period, for the same cult and the same support, apparently the possibility of linguistic choice on the part of the dedicant. 79 Mullen and Ruiz Darasse (e.g. 2020, 776) have argued that the dating of Gaulish needs to be properly reconsidered from the evidence itself rather than using circular argumentation based on approved histories. The project RIIG is working on addressing this. 80 Two inscribed bronze patera handles come from recent excavations in the sanctuary (2011 and 2013): the first one is a dedication ‘to the god Apollo Augustus Moritasgus’ made by a Roman citizen whose sole cognomen Iullinus is preserved—he fulfils a vow granted by his sister Fuscina; the second one is the dedication of a peregrine, Veio, son of Natalis, ‘to the god Apollo Moritasgus Augustus’. The contexts of discovery and the typology of the supports point to the second century ce, and more pre­ cisely for the first one to the first half of the century (Cazanove and Dondin Payre 2016, 112–16). A third dedication to Apollo (apparently without epiclesis) was found by Espérandieu in the same sanc­ tuary and was incised on a metallescent ceramic patera handle (CIL XIII 11239). It can be dated to the second half of the second century or the third century ce. 81 CIL XIII 1368. 82 AE 1998, 948. 83 CIL XIII 5408 and 5412. 84 CIL XIII 3183; Deniaux (2006). 85 Cunliffe (1988); Cousins (2014). 86 Hesperia SO.05.01–02. 87 RIG II.2, L­133, CIL XIII 5468: Doiros Segomari ieuru Alisanu. 88 The Couchey pan is similar (including in decoration) to examples found at Stittenham (Yorkshire) stamped by P. Cipius Polybius (RIB II. 2 2415.19 and 26), who owned the main bronze ware workshop in Capua in 65–85 (Kunow 1985). Its products were widely exported, in Gaul but especially in England (Bennett and Young 1981), Germany, and even beyond the Empire. A recent discovery in Grudna, Poland, gives an idea of the period of circulation of such objects: the Grudna pan was found in a tomb datable to sub­phase B 2b, thus between 110/120 and 160 ce (Kołoszuk 2015). 89 CIL XIII 2843: Deo Alisano Paullinus pro Contedio fil(io) suo u(otum) s(oluit) l(ibens) m(erito). According to Raepsaet­Charlier (1993), the formula deo + theonym appears at the time of Antoninus Pius. Religion, Language Maintenance, and Language Shift 227 Figure 10.11 Dedications to the god Alisanos: a. Couchey (RIG II.2, p. 353 = CIL XIII 5468) (photograph by kind permission of the Musée Archéologique de Dijon); b. Viévy (Visignot) (CIL XIII 2843) (photograph by O. de Cazanove, reproduced by kind permission of the Musée d’Archéologie Nationale, Saint­Germain­en­Laye). Another category of offerings is those depicting parts of the human body. In Gaul, the wooden anatomical ex­votos appear a generation after the conquest, at the Springs of the Seine, Chamalières, Magny­Cours, as shown by recent dendro­ chonological datings: they bring fresh evidence about the early adoption of Italian and Mediterranean practices.90 The first ex­votos on bronze sheets date to the 90 Cazanove (2017). 228 Cazanove and Estarán Tolosa Augustan period (we now have a well­dated set of them, in Orleans), but eyes on bronze sheets are especially common from the Flavian era. Finally, the stone ex­votos, which resemble those found at the same period in Greece,91 are mainly datable from the middle of the second century ce.92 In Alesia, three of them bear Latin dedications to Apollo Moritasgus.93 The only inscription of this set that can possibly date to the first century or the beginning of the second century ce is a Latin dedication to Moritasgus on an incomplete bronze eye­shaped sheet.94 If we take into consideration the dates of the survival of written Gaulish in Alesia and, more widely, in the Lingon and Aeduan territories (the most recent inscription dates probably from the last third of the first century or the early second century) and the second­century ce dating of all, or almost all, the Latin votive offerings of the sanctuary of Apollo Moritasgus (some of them, such as the anatomical limestone ex­votos, dating to the middle or second half of the century), we could consider a transition to a more Latinate epigraphic practice between the first and second centuries ce, or even during the course of the second century. During long periods of bilingualism there would be periods of perhaps faster or slower Latinization, differences between regions, communities, and individuals, and complicated domain­based choices. The great variety of rhythms would make up a completely heterogeneous and diverse process, to which we do not have access.95 It could be said, however, that it is likely that, by the end of the second century, Latinization had been successful, at least for dedications placed by individuals in a private capacity in public sanctuaries. In fact, it was in the second 91 Forsén (1996). 92 At the Apollo sanctuary in Alesia, two archaeological contexts are important in this respect: a dump pit (excavated 2011) filled with limestone ex­voto, sealed at the end of the second century/first half of the third century ce (Cazanove 2017, 63–7); and a manhole, set up in 186 ce (dendrochrono­ logical dating), put out of use and filled in in the middle of the third century (excavated 2016: first account in Cazanove 2016, 223–4). The filling included several anatomical limestone offerings. None of the ex­votos in either of these two contexts actually bore a dedication. But they belong to the same category as the inscribed ones. 93 CIL XIII 11240–11,242. 94 Cazanove and Dondin­Payre (2016), 109–12. It comes from the ancient excavations of Esperandieu, but only a recent restoration carried out at the Museum of Saint­Germain­en­Laye revealed the inscription. Other dedications on votive eyes come from Mirebeau, Pannes, Sources of the Seine, all in Latin. From the sanctuary at the Seine springs is also known a bronze sheet representing breasts, with Latin dedication: Cazanove (2009b); Dondin­Payre (2012). 95 One context where it seems likely that Gaulish has been used epigraphically for religious activity perhaps as late as 200 ce can be found in the small cult site at La Tannerie in Châteaubleau. It has as a focal point a double basin, surrounded by a porticoed courtyard. Its sculpted decoration is typical of the later Antonines or Severans (Revenu 2008). It has yielded c.1,400 coins, 186 fragments of statu­ ettes, some other ex­votos, and several inscribed tiles, including two alphabets and some Gaulish texts. One contains four lines of Gaulish, though there may be an opening phrase in Latin (Lambert 1998, 123). It mentions a Venerianum, perhaps a place dedicated to Venus or the name of a person (Lambert 2008, 145), and it may also contain a verbal form slanossiiet ‘may he/she heal’ (Lambert 1998, 123). It is possible that this text might have represented a sort of nuncupatio. The four­line tile was found in the south­west corner of the portico. Its stratigraphic position, at the base of the demolition, makes it possible to attribute it either to the collapsed roof or to a display that would have seen it suspended from the wall/beams (Zamboni 1999, 64). Religion, Language Maintenance, and Language Shift 229 century ce (and perhaps not before Antoninus Pius96 that a new Latin dedication formula was widely adopted in the civitas to which Alesia now belonged. For a long time, scholars have noticed evidence of the preferential diffusion of certain dedication formulae specific to a city, or a group of cities. Raepsaet­ Charlier has particularly highlighted what she calls ‘la formule éduenne’97—that is to say, the formula Augusto sacrum deo/deae and so on, with at least thirty­ seven occurrences in the territory of the Aedui and its immediate neighbour­ hood, including four times in Alesia. The well­delimited distribution area of the ‘formule éduenne’ allows us, according to Raepsaet­Charlier, to define the bound­ aries of the Aedui territory and in particular that of the Mandubian pagus, which is a part of it.98 In another recent paper, the authors—three specialists of eastern Gaul—are even more confident.99 Subsequently, however, during excavations in Mâlain (a minor settlement of the territory of the Lingones), a fragmentary statue of a female deity was found, with a dedication to Sirona on the base, which employs the ‘formule éduenne’ (Fig. 10.12):100 Aug(usto) sac(rum) dea(e) Sirona(e) Saturninu[s] Satair[—?] Admittedly, we are only 15 km from the boundaries of the former bailiwick of Auxois—that is to say, the old pagus Al(i)siensis, and the borders between the cit­ ies are purely administrative. But the distinction between the Aeduan and Mandubian pantheon, on the one hand, and the Lingonian pantheon, on the other, should not be too strictly drawn. Mars Cicolluis is the great Lingonian god, with his consort Litavis. The strong links between Mars Cicolluis and Lingonian identity are beyond doubt, as shown by the map of the dedications to the god inside the territory of the Lingones,101 the important sanctuary of the divine couple precisely in Mâlain102 and, outside the territory, a well­known inscription from Xanten: the cives Lingonum qui consistunt, who are resident aliens, make a dedication to Mars Cicolluis, pro salute Neronis.103 However, Sirona with the 96 If the syntagm deo (deae) + theonym actually appears at that time. See above, n. 89. 97 Raepsaet­Charlier (1993), 20, 55. 98 Raepsaet­Charlier (2013). On the date of the integration of the Mandubian territory into the Aeduan civitas, see above. 99 Kasprzyk, Nouvel, and Hostein (2012): ‘It is therefore absolutely obvious in our opinion that the very particular distribution of the Augusto sacrum deo formula indicates that these three territorial units (Aeduan territory in its traditional extension, region of Alesia and medieval diocese of Auxerre) form the ciuitas Aeduorum at the end of the Early Empire.’ Conversely, the ‘inscriptions found outside this area (Dijon, Lux, Hières­sur­Amby and Lezoux) [. . .] would reflect [. . .] the geographical mobility of Aeduan citizens, who adopt in the religious dedications they make outside their city a distinctive epigraphic element characteristic of their origin’ (our translations). 100 Widehen and Kasprzyk (2016). 101 Raepsaet­Charlier (2012a), 45–7. 102 Provost (2009b), 458–63. 103 AE 1981, 690. 230 Cazanove and Estarán Tolosa Figure 10.12 Inscription of Sirona. From Mâlain (Archéopages, 43 (2016), 28–33). (© Philippe Haut—Inrap. Reproduced with permission.) ‘formule éduenne’ is precisely present in the great place of worship of Mars Cicolluis in Mâlain. Conversely, at the northern end of the pagus Alisiensis, in Aignay­le­Duc, we find again the ‘formule éduenne’, but on a dedication to Cicolluis and Litavis.104 There are thus areas of contact, interpenetration between religious preferences, and linguistic/epigraphic interaction, which should not be rigidly compartmentalized. It is worth emphasizing the importance of large sanctuaries, such as the one dedicated to Mars Cicolluis and Litavis in Mâlain­Mediolanum, Apollo Moritasgus in Alesia, and many others, as meeting places where the linguistic exchange and change described in this chapter could take place. Not only are they frequented by the inhabitants of the neighbouring city, but they also attract visit­ ors more widely, from across the territory and even beyond the civitas. These hub sanctuaries must have played a role in the spread of literacy and in the process of Latinization at least with regard to the votive writing. Ton Derks drew attention, twenty years ago, to the seal boxes found in places of worship, and he supposed that they were used to close tablets on which were recorded conditional promises of vow, nuncupationes.105 The instrumentum of the sanctuaries also frequently 104 CIL XIII 2887 = ILingons 295. 105 Derks (1998), 225–31. The nuncupatio formulae are indeed attested—for instance, in Chateauneuf­ de­Savoie: Mermet (1993). See above, n. 30. Religion, Language Maintenance, and Language Shift 231 contains writing equipment, calami, little knives for sharpening them, and so on. The sanctuaries played important roles generally in the negotiation of local and Graeco­Latin cultures, in the sphere of religion, sculptural practices, and architecture among others. Their significance in linguistic and epigraphic exchange and change is also likely but unfortunately can be accessed only through snapshots. 4 The Pillar of the nautae: Interweaving or Juxtaposition of Language and Religion? We now turn to our final case study: a first­century ce monument that combines different cultural, religious, and linguistic spheres and that will illustrate the limits of our understanding (Fig. 10.13). Almost everything has been said on the pillar of the nautae discovered in 1711 under the choir of Notre­Dame de Paris, during restoration works of the vault of the archbishops.106 It immediately attracted the attention of learned Europeans, starting with Leibniz. Since then, the exegeses have proliferated, among others about the exact status, corporative or military, of the nautae Parisiaci,107 and their representation on the block that bears the inscription, about the Latin inscription itself and its atypical formula that places Tiberius before Jupiter108 and, of course, about the gods that make up this strange pantheon, all called by their names in the nominative,109 as if it were not obvious for the viewers to recognize them. It was assumed that some form of hierarchy existed between them. Van Andringa argues that the divinities proper to the nautae would have been subordinated to the great gods of Rome and, thereby, to the imperial authority.110 The presence and arrangement of local and Roman deities, as well as the Gaulish and Latin languages, in this extraordinary piece could be interpreted in many ways.111 On the one hand, if we look at the mixture of Latin and Gaulish in the same monument, with an exercise of extensive code­switching in which the 106 Most recently, with previous bibliography, Harl (2019). 107 Béal (2005); Scherrer (2013). 108 ‘an ambiguous and atypical formulary [. . .] at a time when linguistic and religious Romanisation was in its infancy in the civitas of the Parisii’ (Blanchard 2015, 25–7). 109 Iouis is rather an alternative form of nominative (like, e.g., Iouis Tifatinus) than a genitive. 110 Van Andringa (2006). 111 A long study on the nautes pillar has recently been published (Harl 2019), which has the great merit of being based on new photographic documentation carried out after the blocks were cleaned. It is too early to assess the impact of this new publication, which on many points makes proposals that are different from the generally accepted ideas about the pillar (e.g. assuming that the blocks were found almost in situ, and that the pillar was therefore erected at the upstream end of the Ile de la Cité). It may simply be noted that, on at least two points, the few remarks proposed here diverge from the conclusions of Harl’s study: on the one hand, Harl understands Eurises as ‘Kundschafter’ (‘scouts’), quoting rather surprisingly Greek terms in support of his interpretation (pp. 118–19). On the other hand, Harl places Pollux not on the main face of the pillar, above Jupiter, but on a side face, above Tarvos Trigaranus, because (he says) the right edge of Pollux’s face is irregular (pp. 89–98; 90, pl. 5). The argument is of only relative value, because the pillar was intended to be seen from all sides. 232 Cazanove and Estarán Tolosa Figure 10.13 A new proposed reconstruction of the Nautae pillar. The two ‘Roman’ sides are in the centre. (O. de Cazanove, based on the drawings of J.­P. Adam, modified with permission.) language adapts to the divinity, something without precise Palaeoeuropean com­ parandum, and the coexistence in the same piece of indigenous and Roman divinities, represented with their corresponding iconography (the local without apparent traces of interpretatio), we could argue that the Gauls were aware of the linguistic and religious transition they were experiencing and were able to prod­ uce a monument in which Gallic and Roman traditions were interwoven. On the other hand, we could look at this monument rather differently, and, rather than seeing a mix, we could instead witness the juxtaposition of cultures, distinct knowledge, and separate devotions, as shown by the distribution of images and writings on the four sides of the pillar. If the inscription in the car­ touche appears on the main face of the monument, as seems obvious, then Latin is put first. It seems logical to place on the same side the relief depicting Jupiter Religion, Language Maintenance, and Language Shift 233 himself (Iouis), as well as the image of Pollux, who, of the two dioscuri, is the son of Jupiter. The lower block, the widest we might expect for the base, bore four pairs of deities. The couple of gods, who are par excellence the protectors of Rome, Mars, and V(enus), seem particularly suitable to appear on the ‘Latin’ side of the pillar.112 On the reverse, above the group of three bearded men in arms, is written the word Eurises, ‘dedicants’.113 If this is so, the Latin and Gaulish dedications are located opposite each other, like, mutatis mutandis, the two inscriptions on the chest and the back of the colossal statue of Mercury in Lezoux.114 On the two upper blocks, there are the same number of Roman and Gallic gods. Then, one can perhaps restore two adjacent ‘Gallic’ faces and two Roman ones, as suggested by the fact that the same scene, that of Esus and Tarvos Trigaranus, is distributed on two contiguous faces, whereas it is treated as a single unit on a relief of Trier.115 Therefore, if the reconstruction of the pillar of the nautae suggested above is cor­ rect, one could argue that the gods of Graeco­Roman and Celtic origin are not really mixed, but remain juxtaposed, each group occupying two contiguous sides of the monument. The Latin is placed in front, the Gaulish at the back.116 This monument underscores the limits of our ability to interpret provincial Roman linguistic and religious contexts. It is hard to opine what the commission­ ers of the pillar monument intended to achieve with their creation, and even harder to grasp how the viewers of such a monument might have received it. 5 Domains for the Persistence of Gaulish? By focusing on a series of case studies, we have offered thoughts on religion, broadly conceived, as simultaneously a motor of Latinization, particularly in ‘offi­ cial’ and more public religious practices, but also as a domain in which the local language could be retained, especially when the context was local and more pri­ vate. According to our fragmentary textual record, which prevents us from draw­ ing precise conclusions, it could be hypothesized that in Gallia Comata the local language diminished owing to the advance of Latin throughout the second cen­ tury ce. At least the Gaulish epigraphy destined to be publicly exhibited and to attract a wider audience seems to have disappeared during the second century ce, 112 In the graphic restitution he suggests of the pillar (Fig. 10.13), Adam (1984), 299–306, places on the main face Mars and his female companion at the foot (but without recognizing Venus), then the Latin inscription above, above again Iouis, and finally Cernunnos on the upper block. But he acknow­ ledges that this option is ‘arbitrary’ and that the blocks ‘must be imagined to be susceptible to rotate’. 113 Lejeune RIG II.2 pp. 157–76, interpreted it as seniores, but they are rather the ‘dedicants’ accord­ ing to the more convincing analysis of Lambert (2018), 000, who sees in it a perfect participle on the verbal theme of ieuru. For a different interpretation, see Harl (2019), 118–19; see above, n. 111. 114 RIG II.1 pp. 112–18. 115 Espérandieu (1915), no. 4929; Schwinden (2003). 116 And on one of the lateral sides of the pillar, with a text difficult to interpret: RIG II.1 pp. 174–6. 234 Cazanove and Estarán Tolosa as we have seen. It could be said that the last testimonies of this nature are the famous calendars of Villards d’Héria and Coligny, the first dating to the second century ce and the second to between the first and second centuries ce, written in Gaulish,117 but whose formal aspects, for example, the palaeography, the choice of medium, and the layout, closely follow Roman official texts. From the second century ce, the publicly exposed religious texts of Gaul are written in Latin. It is during this century that Gaulish, in written form at least, seems to be relegated, at least in part, to the magical, or broader ritual, domain (although texts of other types such as the Châteaubleau tile are also late). Some of the literary testimonies that refer to the use of Gaulish in late antiquity refer precisely to religious contexts,118 and the interference of the Gaulish language in the Latin inscriptions of this period occurs precisely in terms or formulae of this type. These features can be found in Latin inscriptions with terms such as gutuater or vercobretos;119 in Gaulish medical and curative formulae in Latin texts such as those of Pliny and Marcellus of Bordeaux;120 in the defixio of Poitiers (fourth century ce);121 and in the fact that the expression of time is indicated in Gaulish mode, as shown not only by calendars but also by the inscription CIL XIII 2494, in Latin, where it is established that on the fourteenth days of the months of thirty days a certain funeral banquet must be held, using the Gaulish terms petrudecameto and tricontis.122 Therefore, two opposite vectors can be contrasted. On the one hand, Roman civic religion had the powerful visual element of writing and a capillary spread thanks to Augustus and his successors. These tools facilitated the implantation of Latin and its use in official and public linguistic domains and were adopted by the Gallic elites themselves. The appearance of Augustus inside the spring sanctuary of Nîmes (n. 21) and the very early dedication of a temple to the principes iuventutis are eloquent examples of the desire for Romanness on the part of certain Gallic elites. There is no doubt that Latin made its way through these channels into Gallo­Roman society. On the other hand, Gaulish continued to be spoken well into the imperial period. The roots of local religion and the magic and ritual practices linked to it could have played an important role in language maintenance,123 and the literary 117 RIG III. The Villards d’Héria calendar does not preserve any complete words, but its formal similarities to that of Coligny allow us to assume that it was written in Gaulish. 118 See, e.g., Historia Augusta, Vit. Alex. 60.6: mulier Druias eunti exclamauit Gallico sermone: ‘Vadas nec uictoriam speres nec militi tuo credas’, the medicinal/magical formulae in Marcellus of Bordeaux: in mon derco marcos anxatison, xi exucri cone xv criglion aisus scrisumio velor exucri cone xv grilau (Fleuriot 1974). 119 gutuater: CIL XIII 1577; 2585; 11,225; 11,226; vercobreto: AE 1980, 633 = AE 1981, 643; CIL XIII 1048 + 1074, AE 1980, 624; Bost and Perrier (1989). 120 Fleuriot (1974). 121 CIL XIII 10026, 86. 122 On this, see Marco Simón (2014). 123 According to Marco Simón (2008), the ‘ubiquity of the ritual’ was a characteristic of the Gallic societies of the Iron Age. Bell (1997), 173–209, expresses this idea as ‘ritual density’. Religion, Language Maintenance, and Language Shift 235 sources are suggestive, if not definitive, along these lines.124 These religious beliefs may have been an element that contributed to the preservation of Gaulish into late antiquity, linked to a strong component of orality and preservation of local practices.125 The fact that several texts on metal that can be related to the magical sphere are written in Gaulish reveals that the Gauls believed that their epichoric language was appropriate and effective to address their divinities.126 We know around a dozen examples written in Gaulish or at least in a language close to it (magical language is sometimes intentionally obscured in order to increase its effectiveness)127 and that are not in Latin.128 They seem to date up until the sec­ ond century ce, but if the lead sheet found at Saint­Marcel can be included among these texts, we could place this type of practice even in Constantinian times. Indeed, Baudecet’s gold foil, which could be an amulet or other prophylactic element and whose text contains elements not only of Gaulish but possibly also Germanic, dates to the end of the second century.129 It is not possible to know to what extent Gaulish was still alive in other domains or if it was used only in magical or other ritual contexts. One of the treasures of the Gaulish epigraphic set is the private documentation, where various phenom­ ena derived from the coexistence of both languages can be contemplated, such as the mixtures of languages in the set of firing lists of potters from La Graufesenque or the translingualism of the spindle whorls made in Autun.130 The private texts in Gaulish extend to the third, and possibly even the fourth, century ce, if we consider the Châteaubleau tile, and, although they are only a small sample of the linguistic situation of the imperial period, these documents allow us to hypothesize with relative security that the Latinization of Gallia Comata was still imperfect in the domains of private writing and, therefore, probably of oral communication in non­official environments, during at least the first two centuries ce, and possibly later in some contexts. It is in the purposes that interest us here, however, religious language in the private sphere, that it seems clearest that Gaulish may have maintained some degree of usage beyond the second century ce. In any case, the 124 See, e.g., Caesar, Gall. 6.16.1. Pliny may also be referring to this when he comments that ‘magic has ruled over Gaul’ (30.4). 125 One could possibly raise here the difficult question of the persistence of Druidism and its influence until the end of antiquity, well after the prohibitions of Augustus and then Claudius (Suetonius, Claudius 25.13), if the late texts that mention them (Historia Augusta, Carus 14.1–15.2; Ausonius, Commemoratio IV.7–14; X.22; Ammianus Marcellinus, Res Gestae XV.9.8) are not a mere revival, a purely literary phenomenon, as has been recently argued (Wiśniewski 2009). On the complex matter of Druidism, see, among others, in a vast bibliography and with partly dissimilar points of view: Aldhouse­Green (2010); Brunaux (2006); Clavel­Lévêque (1985); Marco Simón (2012b); Zecchini (1984). 126 On the particularities of the Celtic defixiones, see Dupraz (2018). 127 See n. 9. 128 Chartres, Chamalières, Les Martres­de­Veyre, Le Mas­Marcou (first century ce), Larzac (90–110 ce), Saint­Marcel (Constantinian), Le Mans?, Lezoux?, Eyguières?, Amélie­les­Bains? 129 Plumier et al. (1993). 130 For the language of the graffiti from La Graufesenque, see Adams (2003a), 687–724; Blom (2010; 2012); Marichal (1988); Mullen (2013b; in press), and Wilson and Wolff, both this volume. For translingualism and the spindle whorls, see Mullen (2022). 236 Cazanove and Estarán Tolosa coexistence of Latin and Gaulish in the public and private epigraphy of Gaul at the end of the first century ce and into the second century ce, if not later, marks an exception with respect to other Palaeoeuropean contexts. 6 Final Thoughts In this chapter we have considered the possibility that different aspects of religious practice in Gaul could be, on the one hand, a Latinizing factor and, on the other, a conservative factor, which could allow, in certain circumstances and situations, the maintenance of the local language. Certainly, the imperial cult in its various manifestations, the civic cults and non­public, but formal, group activities (let us recall the case of the temple of Vieille­Toulouse, probably commissioned by the members of a collegium), offered a context for Latinization. Conversely, certain aspects of the more local and private religion could encourage the ongoing use of the local language, though we will never be able to reconstruct the details of the undoubted chronological, regional, and social complexities. We have explored the nature of the Latinization of Gaul through religious language, investigating the early contexts for Latinization in the religious sphere in Narbonensis, and the nature of the spread of Latin and the use of epigraphic languages in religious contexts in Gallia Comata. Different regions of Gaul show different speeds of Latinization, but there is much more complexity than a simple Narbonensis–Comata divide. During the process of language spread, bilingualism, and shift, there would be periods of perhaps faster or slower Latinization, differences between communities and individuals, and complicated domain­based choices. The great variety of rhythms make up a heterogeneous process and diverse outcomes in the religious sphere, some of which we catch glimpses of, for example, through the first Latin dedication of a cult building in the highly multilingual oppidum of the Volcae, the localized religious formulae in publicly displayed Latin epigraphy (‘la formule éduenne’), the private mixed­language/ Gaulish magical texts, and the hard­to­interpret Gaulish–Latin religious creation of the pillar of the nautae.