SEEING NEW CONNECTIONS

In reading these volumes as a triptych, the reader truly appreciates the progress of the LatinNow project from its outset in 2017. The project’s database work was not focused on publishing new texts, except incidentally; rather it concentrated on collating, cleaning up and de-duplicating records that already existed, and then using GIS to map them. As anyone who has worked with epigraphic sources knows, this job was far from straightforward. The rewards for the team’s painstaking work are immense, as the LatinNow project allows researchers to visualise the data in new ways, making comparisons within and across provinces that were not possible before. The third volume, Mullen & Willi 2024, demonstrates the profound impact that the LatinNow database will have on the field of ancient epigraphy through the consideration of writing tools alongside inscriptions.

If the third volume is where the immense efforts of the digital project begin to pay dividends, the first two volumes lay the groundwork. Mullen 2023 builds on more than 20 years of existing work on ancient sociolinguistics, particularly the studies by J.N. Adams (Bilingualism and the Latin Language [2003] and The Regional Diversification of Latin 200 bc – 600 ad [2007]) and scholarly responses to them. Themes of literacy, bilingualism and language contact have been well-trodden in the past few decades, but they have not been studied evenly across the Western provinces; new evidence and methodologies in this book add fresh insights to ongoing debates.

Mullen & Woudhuysen 2023, focused on medieval language, does not have the same weight of scholarship behind it, as the editors note in the introduction (pp. 3–5). This volume is valuable in opening up areas of the field that have not yet been fully exploited. But its authors have had to do much more of the initial work to demonstrate the connections between linguistics, history and archaeology than the authors of Mullen 2023. There is more emphasis in this volume on the detailed work of tracing sound changes, dating loanwords and putting forward initial theories of who was in contact with whom when (particularly in the more linguistic chapters by I. Velázquez on Hispania, A.H. Blom on Gaul, W. Haubrichs on Germania, and D. Stifter & N. White on Britain and Ireland).

It is not possible in this review to summarise every chapter in these three books. Instead, I want to draw out some developments across the project’s work, especially where I think these will be influential. I also write from the perspective of an ancient linguist and historian, and not a medievalist or an archaeologist; those from other fields will doubtless find other points of interest.

WHOSE LITERACY?

Literacy is a prevailing theme throughout the three volumes. The authors take an expansive view of what research into ‘literacy’ might mean, and (joining the current direction of travel of the ‘material turn’ in ancient epigraphy) discuss the context, materiality, function, iconography, location, layout and lettering of epigraphic texts as well as the content and language of the texts. The result is a rich discussion encompassing a range of approaches.

As Woolf notes in the final chapter of volume 3, the landscape of evidence has changed unrecognisably since the first histories of Roman literacy were written. Documents produced by a broader cross-section of society, such as the Vindolanda tablets, the Bath curse tablets, the firing lists of La Graufesenque and the Bloomberg commercial records are now central parts of the conversation of ancient literacy, even though only a tiny fraction of the output of ephemeral writing has survived. Horster (in Mullen 2023) cautions against being too reliant on these kinds of texts – she rightly questions the representativeness of the dataset from Vindolanda or La Graufesenque in the context of the whole Roman world. But at the same time these texts are a healthy antidote to equating literacy with writing on stone. Ephemeral, everyday texts constituted most people’s main contact with writing (M. Feugère & Willi in Mullen & Willi 2024, p. 250). And there is still some way for this discussion to go: late antique everyday texts, such as the pizarra texts on slate discussed by Velázquez in volume 2, are only just starting to become part of this conversation in the way in which they deserve.

As readers of her previous work will know, precision of terminology is important to Mullen. In the introduction in Mullen & Willi 2024 she argues strongly for defining the term ‘epigraphic habit’ to include stone and monumental metal but not other media. ‘Epigraphic habit’ was introduced by R. MacMullen (‘The Epigraphic Habit in the Roman Empire’, AJPh 103 [1982], 233–46) to differentiate inscriptions on stone, with no specific addressee, from papyri and ostraca, with a specific recipient in mind. His aim at the time was to compare ‘epigraphic’ and ‘papyrological’ texts as indicators of literacy levels. Since his influential article scholars have variously used ‘epigraphic habit’ to mean stone inscriptions only (as MacMullen does; also E. Meyer, ‘Explaining the Epigraphic Habit in the Roman Empire: The Evidence of Epitaphs’, JRS 80 [1990], 74–96; C.W. Hedrick, ‘Democracy and the Athenian Epigraphical Habit’, Hesperia 68 [1999], 387–439) or monumental inscriptions on both stone and bronze (as suggested by Mullen; also Woolf, ‘Monumental Writing and the Expansion of Roman Society in the Early Empire’, JRS 86 [1996], 22–39). More recently it has been used to mean any inscriptions that are public-facing and without a specific addressee, including painted election notices, graffiti on external walls, inscriptions on trees and so on (see e.g. P. Kruschwitz, ‘Writing On Trees: Restoring a Lost Facet of the Graeco-Roman Epigraphic Habit’, ZPE 173 [2010], 45–62; also P. Kruschwitz, ‘Attitudes Towards Wall Inscriptions in the Roman Empire’, ZPE 174 [2010], 207–18), or in some cases the total epigraphic output of a community (e.g. Taylor’s definition of the epigraphic habit, in relation to Athens, as ‘the cultural impulse to record information on stone and other durable materials’, in: C. Taylor ‘Graffiti and the epigraphic habit’, in: J. Baird and C. Taylor [edd.], Ancient Graffiti in Context [2010]).

There are possible objections to the tighter definition of ‘epigraphic habit’ favoured by Mullen. It might be argued, for example, that the divide that MacMullen originally made between stone and papyri reflected a disciplinary divide rather than ancient categories, and that ‘public’ writing includes many other text types (Woolf [1996]). As epigraphers have worked more closely with papyrologists, and as texts such as curse tablets have come to be studied by both, the divide has come to appear somewhat artificial. Some scholars might also feel that ‘epigraphic habit’ has been superseded by the term ‘epigraphic culture’, a term not used much in these three volumes except by F. Beltrán Lloris (in Mullen 2023, p. 31); or that we might be better off speaking of ‘epigraphic habits’ rather than ‘the epigraphic habit’ (H. Mouritsen argues for the importance of the plural in ‘Freedmen and Decurions: Epitaphs and Social History in Imperial Italy’, JRS 95 [2005], 38–63). Not all authors in the volumes under review draw as strict a distinction as Mullen or Willi: J. de Bruin’s discussion of the archaeology of literacy in the Netherlands (in Mullen & Willi 2024, p. 326), for example, speaks instead of ‘monumental’ and ‘non-monumental’ evidence for literacy, where ‘monumental’ includes among other things a wooden tabula ansata from a military site in the Lower Rhine.

The interaction of bilingualism and literacy has long been discussed in scholarship, and is another key consideration across the three volumes – after all, it is nigh-on impossible to adapt a writing system to a new language without the involvement of at least one bilingual individual. It is rewarding to see ogam brought more fully into this conversation by Stifter & White (in Mullen & Woudhuysen 2023). They argue that the ogam script of Ireland (and Wales) required deep familiarity with the Latin grammatical tradition for its creation, although it is closely tailored to the phonology of Irish (pp. 215–16). They also note the interplay of materiality, form and function in the choice of language and script: ogam script is well suited to stone and wood, but not to manuscripts, which is part of the reason that Latin took over in the latter medium (p. 235).

Many of the authors also take pains to emphasise that there were, at least for a time, situations of stable bilingualism where two languages, and even two writing systems, were used side by side. Gaulish, for example, provides a rare but important example of a written local language being used well into the imperial period. N. Moncunill (in Mullen & Willi 2024) points out that the initial contact with the Latin epigraphic habit led to a period of expansion in the use of written forms of local languages – knowledge of Latin did not set in motion an immediate downward spiral for every language. These examples are helpful reminders that language death was not an immediate consequence of Latinisation, as Strabo might have believed; a few languages, like Basque, Irish and British Celtic did survive and ultimately gained written forms as a result of their contact with Latin and Romance.

LATINISATION: ‘ROMANISATION BY THE BACK DOOR’?

The use of the term ‘Latinisation’ here is inspired by Woolf’s definition of Latinisation as ‘the process by which the spread of a cluster of different Latin literacies and versions of spoken Latin interacted to create new ones all over the Roman West’ (Woolf, ‘Afterword: How the Latin West was won’, in: A. Cooley [ed.], Writing Latin, Becoming Roman? Literacy and Epigraphy in the Roman West [2002], 181–8, cited by Mullen in Mullen & Willi 2024, p. 5). The messiness is part of the point – Latinisation involves many Latins rather than one. Mullen explains that the use of ‘Latinisation’ is also intended to pull out the phenomenon of language change as a distinct process, rather than ‘as simply an unremarkable part of “Romanization’’’ (Mullen in Mullen 2023, p. 3).

Initially, it may seem surprising that there is not more pushback from the authors against the term ‘Latinisation’ – as Mullen says, it could be seen as ‘Romanization by the back door’ (Mullen in Mullen 2023, p. 22). In some ways the terminology does not escape this pitfall, particularly as the authors associate the arrival of the Latin language so closely with a distinctly Roman set of epigraphic practices (e.g. A. Wilson in Mullen 2023, p. 106, states that ‘Latin literacy was introduced to Londinium as a sophisticated pre-existing package’). Beltrán Lloris defends the older terminology of ‘Romanisation’, up to a point, arguing that attempts to replace it with ‘globalisation’, ‘creolisation’, ‘middle ground’ or other frameworks have not led very far (in Mullen 2023, p. 43). Other scholars may not share his view that new framings have been fruitless; the possibility of considering these processes of change through multiple lenses has been seen as a positive development, particularly for the integration of post-colonial approaches to Roman archaeology (A. Gardner, ‘Post-colonial Rome, and beyond’, Revista de Historiografía 36 [2021], 309–20). ‘Latinisation’, as Mullen defines it, fits well with the recent desire for diverse explanatory frameworks, but could also be integrated profitably into the object-led ‘Romanisation 2.0’ championed by M. Pitts among others (M. Pitts, ‘Towards Romanization 2.0: High-Definition Narratives in the Roman North-West’, Journal of Urban Archaeology 3 [2021], 117–30).

MILITARY, LEGAL AND RELIGIOUS LANGUAGE(S)

One of the most hotly debated themes of the volumes is the role of the military. In a richly evidenced chapter M.A. Speidel argues that numbers of Roman soldiers and veterans were proportionally so small that they could not have meaningfully affected the language of their communities. The number of veterans discharged each year was in the low thousands, he argues; they were unlikely to have had a major linguistic impact on the 70–100 million residents of the empire (Speidel in Mullen 2023, p. 135). Speidel draws on comparisons with Thrace, Iudaea and Syria Palaestina, where serving soldiers were commemorated in Latin but veterans had gravestones in Greek – was linguistic behaviour any different in the West when soldiers retired and went home?

This view is not taken up by many of the other authors in the volumes. One might argue (and indeed Horster does so) that sheer numbers are not the deciding factor when institutional power is in play. In Mullen & Willi 2024 De Bruin contends that correspondence between the Batavians stationed at Hadrian’s Wall and those who remained at home can be considered a key driver of Latin literacy in rural Batavia, where input from other written languages was rare, unlike in Speidel’s Thrace example (p. 323). In the same volume Willi (p. 318) argues that the military brought increased population diversity and density to the areas where they were active, but that their effect on language was indirect. The presence of the military created higher connectivity, but in her view it was the civilian element in military areas which was the catalyst for writing. Questions remain, then, about the role of the Roman army in the spread of Latinisation and Roman-style epigraphy, which further work may answer.

E.A. Meyer’s chapter on Roman law (in Mullen 2023) likewise turns assumptions on their head by arguing that access to Roman law did not profoundly influence non-Romans’ desire to learn Latin, because what little legal protection they could call upon as non-citizens could be accessed without using Latin (p. 193). Meyer argues that it was access to the informal personal justice of curses and ‘prayers for justice’ that might drive a provincial to learn Latin (pp. 204–5). ‘Here the push to learn Latin was strong, often urgent’, she says, ‘and the pull of Latin very direct, almost unavoidable’ (p. 183). This argument, while fascinating, relies on minimising the possibility of magical texts in local languages, which other contributors accept (such as O. de Cazanove & M.J. Estarán Tolosa, p. 207); they view magic as a particularly long-lasting text-type for local languages in which writers were slow to use Latin (p. 235). The exact nature of certain Gaulish texts on lead is a problem – if they are curses, that tips the debate in favour of de Cazanove and Estarán Tolosa’s interpretation. The role of Italians, not just Romans, in Hispania and Gaul in the Republican period is noted by several authors in these volumes, and Latin was not the only language of magic in Republican Italy: Greek, Oscan and Etruscan are all represented, and there are also mixed-language magical texts (K. McDonald, Italy Before Rome [2021], pp. 239ff.). Latin as a fast-track to the gods of revenge is a profoundly interesting idea, and this might have been the case in some communities and time periods, but this is unlikely to be the main driver of Latinisation. Not all the authors accept Meyer’s view of the law – indeed J.P. Conant’s chapter on late antique North Africa (in Mullen & Woudhuysen 2023) highlights access to the law as a key motivation for Latinisation.

The role of Christianity is also a subject of lively debate. R. Wiśniewski (in Mullen 2023) offers insights into the ways in which Christianity did not, in his view, have a major effect on Latinization (p. 237). Before the last quarter of the second century ce no Christian of Rome left evidence of being a Latin speaker (pp. 238–43). The language of Christianity, he argues, was Greek until Latinisation was well under way across the western provinces. In most cases Christianity took hold in areas where Latin was already spoken, rather than bringing Latin to new areas (p. 245). Stifter & White’s chapter in Mullen & Woudhuysen 2023 shows that there is something to be said for the idea that Christianity encouraged the use of Latin in Britain. In their view, it certainly sustained the use of Latin in Britain, even if it was not the first vector of the language onto the island (p. 212). In Ireland, as they note, Christianity arrived before Latin (p. 235).

And so, why did people in the West start to use Latin, if they were not motivated by the army or the law or Christianity? Trade and commerce are the reasons most commonly given by the authors. Latin as a facilitating ‘link’ or ‘vehicular’ language comes up many times. This is explored in most detail by Wilson, who imagines the immense initial effort involved in learning another language, but with the promise of reduced transaction costs (and costly mistakes) along the line (pp. 81–3 in Mullen 2023). Wilson also dwells on the experience of slaves, for whom language learning was a matter of survival (p. 98). These points are aptly illustrated by Wilson’s example, which was also discussed by Adams (2003, pp. 53–63), of a second-century ce receipt found in the Fayum, written by a slave trader named Aeschines Flavianus from Miletus, for the sale of a female slave to a soldier of the Ravenna fleet. The text, in grammatically imperfect Latin, is written in Greek characters and shows interference from Greek (Wilson in Mullen 2023, pp. 84–5). Although slightly outside the geographical scope of the volumes, such an example illustrates a language contact situation that was surely repeated daily across the Roman West, for both enslavers and enslaved people.

NAMING PRACTICES

M.-T. Raepsaet-Charlier’s chapter (in Mullen and Willi 2024) on the onomastics of the Batavians stands out as the only chapter devoted entirely to naming practices; it is a masterclass in how to use onomastic evidence sensitively. Her proposal is that, although the use of Latin names might not necessarily require the use of the Latin language, in practice these two things were closely linked, and that the choice of Latin names does require that a degree of Latinisation was present in the community. Alongside Celtic and Germanic cognomina and cover names, she also explores more unusual instances, such as elite peregrines with names such as Flavus Vihirmatis filius (a Germanic name with a Latin-style filiation), the rather curious Q. Phoebius Hilarus, whose gentilicium was a Latin derivation from Greek (p. 226), and Batavinia Romana, whose unique gentilicium may indicate that she was a public freedwoman of the city (p. 231).

On a more granular level the authors show that different families made different naming choices, such as the maintenance or non-maintenance of Palmyrene names as Roman cognomina (Speidel in Mullen 2023, p. 151). We can also see the changing naming practices of a single elite family Mausoleum of the Pompeys in Torreparedones (Estarán Tolosa & J. Herrera Rando in Mullen & Willi 2024, p. 107).

It is through onomastics that we observe some of the clearest exploration of linguistic gender differences in Latinisation. Women are sometimes assumed to be more traditional or local in their language choices: for example, it is possible that the relatively high number of women mentioned in later Gaulish texts indicates that they were keeping Gaulish alive in spheres where they had most influence (Mullen in Mullen & Willi 2024, p. 202). In some contexts they may also have had less access to transregional networks and therefore less reason to learn Latin; but this is not always born out in the epigraphy in a straightforward way. Speidel notes (in Mullen 2023, p. 138) that in the Cheffia region, in the hinterland of the coastal city of Hippo Regius, a local language known as ‘palaeo-Tamazight’ was used for many men’s funerary inscriptions, but the women all have Latin-style names and gravestones inscribed in Latin.

There are limits to the usefulness of names, as several of the contributors note. A name of a particular linguistic origin does not always indicate the use of that language or any identity connected to that language’s origin. An added complication is that it can be difficult to differentiate Celtic and Germanic names – many names can be analysed as either (Mullen in Mullen & Willi 2024, p. 28). There is still more to do, and Horster highlights onomastics as a particular area that could be taken further for future research (in Mullen 2023, p. 294).

POWER AND CHOICE

Power is not always Roman, nor is powerlessness always local. As Mullen (2023, p. 5) explains, an auxiliary soldier may have had high status in a specific local context, but low status in the army; local non-citizen elites may have higher status and greater social capital than citizen veteran soldiers. Nevertheless, the inscriptions used in these volumes were (for the most part) produced in a context of colonisation, where Latin was the language associated with military and political power. Although scholarship often concludes that there was no official language policy in operation in the Roman world, C. Wolff (in Mullen 2023) encourages us to consider the reaches of Rome’s soft power. She questions whether we can dismiss the idea of ‘language policy’ so easily when we know of instances of semi-official encouragement of Latin – for example, through the documented cases of Roman officials paying for the education of children of the local elite.

Challenges to – or accommodations of – Rome’s power appear in unexpected places. A. Kolb’s discussion of Roman milestones (in Mullen 2023), including those that use Celtic leugae rather than Roman miles as their base measurement, shows deftly that even the most formulaic, official inscriptions can show preferences for the local over the Roman. Another nice set of examples of local choice are the inscriptions put up when the community was not required to do so – such as the Tabula Siarensis (19 ce). Elites in somewhat out-of-the-way places, such as Siarum and Irni in Hispania, could communicate their participation in Roman networks, and compete in monumentality with elites of nearby cities (P. Houten in Mullen 2023, p. 72). We might not expect the Roman authorities to tolerate locally led variation, but several contributors note the broadly ‘pragmatic’ or ‘practical’ approach of the Roman imperial machine (e.g. Kolb in Mullen 2023, p. 118; B. Rochette in Mullen 2023, p. 290), and on the flip side the ‘pragmatic’ choices of locals in response to Roman power (Conant in Mullen & Woudhuysen 2023, p. 39).

Can the idea of personal choice be pushed too far, though? Meyer (in Mullen 2023) notes that the decision whether to use Roman law was a choice; but, undoubtedly, there were people on whom Roman law was imposed without consent. With too much emphasis on the individual, Rome’s power can end up feeling distant and abstract rather than immediate and violent. However, some authors are more sensitive to this than others: the role of death, enslavement and displacement in war is highlighted as a factor in language change in particular by Moncunill (in Mullen & Willi 2024, p. 50). Conant, in his work on language change in late antique North Africa (in Mullen & Woudhuysen 2023), reminds us that power was not always evenly felt, and what looks like individual choice from our perspective can be variation in how and when power was imposed (p. 39).

THE LATIN-NOW DATABASE

This review would not be complete without some mention of the LatinNow database, which forms the backbone of the LatinNow project (data downloadable from github.com/latinnow). The database includes 190,000 records of epigraphic and archaeological materials from Britain, Gaul, Germany and Hispania, plus a few from beyond the borders of the Roman Empire.

The team has taken care not to make false claims about the precision of the data. This is explained in detail by Mullen (in Mullen and Willi 2024, pp. 12–22), who justifies some of the decisions taken by the project in geographic presentation of the material. For example, they have instituted ‘buffer zones’ (5km for civilian settlements, 3km for military settlements and 1km for villas) in the metadata to compensate for the fact that inscriptions may have been either moved from their original findspots or assigned to the nearest known ancient or modern settlement when published (p. 21). An example of how this works in practice can be found in Houten’s chapter on the epigraphic habit of Spain (Mullen & Willi 2024).

Another step forward is the database of finds of writing equipment (styluses, tablets, spatulas and other tools). These finds can, for example, be suggestive of writing with ink and on non-durable supports, which may have been more common than previously thought in areas such as the Netherlands (de Bruin in Mullen & Willi 2024, p. 354). As explained in volume 3 in particular, the reporting of these finds is very variable from province to province. It is obvious simply from looking at the map that the documentation is much better for Britain than for Spain or Gaul – it is unlikely that Britain somehow had vastly more styluses than any other Western province. In general, it is the margins where these objects have been best recorded, because of scholarly interest in military literacy.

There are quirks to the system; an experienced user will adapt to these, but a student or casual user may find them frustrating at first. For example, at the time of writing, filtering for ‘Text type: Monumental’ only gives texts in Hispania; comparable text types are given other labels in other regions. In the ‘Search Places’ bar, if one types ‘London’, it returns a variety of possibilities: London (England), Londinium, Little London, London Bridge, London Docklands, Greater London, City of London and so on. ‘London (England)’ actually has two entries – one with 40 associated objects and the other with 1,008 objects. Searching ‘Londinium’, on the other hand, returns 700 associated objects. Searching ‘Alesia’ retrieves no suggested places; only the modern name ‘Alise’ or ‘Alise-Sainte-Reine’ works in the search. These quirks are the result of different databases, with different place naming practices, being combined together; future updates will hopefully improve the user experience.

There is also, currently, no way of searching the text of any inscription, although the full texts are included in the records. As the LatinNow database is aimed first and foremost not at editions of the texts, but at seeing patterns in materials, text types, writing materials and languages across space and time, it also privileges a map view rather than a list view or the ability to search within texts. Again, users will get used to the workflow needed to use the GIS to its fullest potential; the open-access data will help scholars who wish to build on LatinNow’s work in their own projects.

WHERE NEXT?

The province-level organisation of volumes 2 and 3 perhaps invites reading them together, to explore change over time within a region; but in some ways this would be an awkward task. As Russell notes in the response to volume 2, even reading two adjacent chapters is sometimes challenging and requires a great deal of work on the part of the reader (Mullen & Woudhuysen 2023, p. 269). Although links are made within and between volumes, in places these are under-exploited, particular in the pairs of chapters on Gaul (by Blom and I. Wood) and Iberia (by Velázquez and G. Barrett) in volume 2, which overlap in their source material without fully engaging with each other’s approaches. It is perhaps the job of a new round of scholars to explore the interconnections that these three books have uncovered.

One can always quibble about small aspects of any publication, Despite the project’s admirable emphasis on data visualisation, in places the maps and graphs are less than clear. Some require careful consultation not only of the map and map key, but also the surrounding text and the footnotes to decode what they depict. Others are not fully explicated and could have been rethought for clarity of presentation. In some cases, this is the result of reproduction of maps from other publications – for example, the map on Mullen & Willi p. 370 has a key that suggests a large number of surviving Roman funerary inscriptions at first sight (it is only clear with reference to the original publication that the markers show funerary remains, not inscriptions). Some maps showing variable-size circles to indicate numbers of inscriptions at different sites explain how many inscriptions are represented by each size of circle (e.g. on Mullen & Willi pp. 99 and 373) while others do not (Houten in Mullen 2023; Mullen & Willi, pp. 368 and 369). De Cazanove & Estarán Tolosa’s chapter in Mullen 2023 is rich in images; but some figures, such as site plans, warrant more explanation if their aim is to introduce archaeological evidence to non-archaeologists.

The epigraphic notation is not always adequately explained or consistent across chapters. For example, it would not be clear that the use of the circumflex on Mullen 2023, p. 223 indicates a ligature if we did not have the picture of the inscription directly below. The same texts (such as the slate letter P.Vis. 130) can appear in slightly different versions and translations in different chapters. In general, though, these are well-produced works, with remarkably few inconsistencies or typographical errors, given the diversity of languages and regions represented.