One cult, multiple problems: writing to the gods in north-western Roman Hispania

Introduction

The northwestern region of the Iberian Peninsula, stretching from the northern banks of the Tagus River to the Cantabrian Sea, remained largely illiterate until it was fully pacified at the end of the 1st c. BC. From that point onwards, a process described by J. d’Encarnação as epigrafização began, characterised by a marked increase in Latin inscriptions as part of the broader spread of epigraphic practices across the Roman Empire during the Early Imperial period.¹

A distinctive characteristic of the epigraphic culture in these areas – corresponding to what the Romans designated as Lusitania and Gallaecia – is the prevalence of dedications to indigenous deities. These theonyms have been linguistically and religiously linked to the Indo-European sphere and display notable features reminiscent of well-attested Celtic theonymy.² Given that the epigraphic habit was introduced as part of the region’s Latinisation and Romanisation, it is unsurprising that these dedications conform to Roman conventions in terms of epigraphic typology, formulae, and, naturally, language and script. Nevertheless, a small number of inscriptions have been identified. They are all religious in nature and use the Latin alphabet while incorporating various elements of an indigenous language.

Although limited in number, this epigraphic corpus presents a range of interpretative challenges, particularly with regard to the nature of the pre-Roman language conventionally referred to as “Lusitanian”. Issues such as the interplay between language, religion, and identity; the survival of indigenous cultural elements; Roman influence and its local adaptations; and the sociolinguistic dynamics in northwestern Hispania are all of particular relevance. In the following discussion, we will explore some of these themes in relation to the westernmost Hispanic territories that once constituted the Roman fines terrae.

The Epigraphic Record

Before proceeding further, a brief overview of the epigraphic material is necessary. The series of texts featuring indigenous linguistic elements can be classified into two categories, each one with distinct linguistic and epigraphic characteristics. The first group comprises five rock inscriptions containing complex texts written in Latin alphabet but recording the indigenous language.³

Although not fully comprehensible, the five inscriptions clearly exhibit a religious character, indicated by the presence of well-known indigenous deities such as Reve, Bandue, and Trebopala – frequently attested in Latin epigraphic documents. At least three of these texts – Arronches, Lamas de Moledo, and Cabeço das Fraguas – document ritual animal sacrifices, with names in the accusative case identified for sheep (oilam), pig (porcom), and a male bovine or bull (taurom ifadem).

It has often been noted that these three animals also feature in the Roman suovetaurilia ritual; however, in the Roman context, all three were sacrificed to a single deity, whereas in the Lusitanian context, they were offered to different deities.⁴ Moreover, all three types of animals appear together only in the Cabeço das Fraguas inscription, which cautions against drawing broad parallels. Lusitanian texts have traditionally been associated with open-air sanctuaries, although this is archaeologically confirmed only at Cabeço das Fraguas. The public nature of animal offerings has been interpreted as evidence of collective ritual sacrifices – with veamnicori in the Lamas de Moledo inscription possibly denoting members of a community – perhaps suggesting some form of sacrificial banquets. This interpretation aligns with Strabo’s account of such practices among the Gallaeci.⁵

Two of the inscriptions, Arroyo de la Luz I and Lamas de Moledo, are technically bilingual, each beginning with personal names as the subjects of the Latin verb scribo, although no further information is provided about these individuals. Their precise role in the creation of the inscriptions remains unclear, with interpretations ranging from commissioners and stonemasons to ritual officiants or potential witnesses tasked with recording a decision made by the community.⁶ Arronches also shows Latin linguistic interference through the use of the numeral X.

The second category comprises a somewhat larger group of inscriptions exhibiting linguistic interferences from the indigenous language, specifically in the form of code-switching. As mentioned in the opening chapter of the book, these interferences may be classified as instances of intra-sentential switching, as defined by A. Mullen, or as mixed inscriptions, following the classification proposed by M. J. Estarán Tolosa.⁷ This group includes approximately twenty altars or small votive stelae, with a few exceptions such as the wall graffiti from Fonte do Ídolo. Generally crafted from granite, these inscriptions contain votive dedications to indigenous deities, inscribed in Latin characters.

Their linguistic distinctiveness lies in the use of religious onomastics, particularly epithets, which feature endings not found in Latin but characteristic of the local language. These endings typically appear as singular datives in forms such as –bo(r), –ui, –oe, and –ico (with some graphic variation). Thus, these are essentially Latin texts that incorporate discrete features of the Lusitanian language, identifiable by their morphological particularities.⁸

The structure of these inscriptions presents the name of the deity, expressed wholly or partially in the indigenous language, while the name of the dedicant and the votive formula are in Latin. Although some of the deities occur in both groups,⁹ it is clear that they represent distinct types of expression, both religious and linguistic, each functioning within its own sphere.¹⁰ Linguistically, the altar of Viseu, which heads the list (fig. 2), occupies an intermediate position, featuring a somewhat more complex text. The first part is written in the indigenous language and is interpreted as a dedication to the goddesses and gods of Vissaium (Viseu), a variant of the formula diis deabusque commonly found in western Hispania. The second part of the text records the dedicant’s name, Albinus Chaereae f(ilius), together with the votive formula, both in Latin.

Linguistic Problems

Since the first linguistic studies of Lusitanian inscriptions in the mid-20th c., it has been widely accepted that their language belongs to the Indo-European family, showing some degree of connection to Celtiberian (the other epigraphically attested Indo-European Palaeohispanic language) and distinct from other linguistic areas of the Iberian Peninsula, such as Iberian in the east, Vasconic in the northeast, or Turdetanian in the south. However, two further questions have sparked considerable debate. The first concerns whether the language of the five canonical Lusitanian inscriptions is the same as that found in the code-switching texts across the Gallaeco-Lusitanian territory, or whether they represent distinct languages. The second involves the classification of this language (or these languages) within the branches of Indo-European, particularly with regard to its relationship to Celtic.¹¹

A map can help to visualise these issues (fig. 3). From a sociolinguistic perspective, the pre-Roman Iberian Peninsula has been divided into two regions based on the presence or absence of toponyms containing –brig(a)-, an element often found in Celtic-speaking areas. Celtiberian, Celtic, and Lusitanian attestations (in epigraphy, onomastics, or toponymy) are located to the west of this line. Gallaeco-Lusitanian theonymy, whether preserved in Latin or indigenous texts, extends along a corridor encompassing the north and centre of present-day Portugal, Galicia, and adjacent Spanish provinces. This region is distinct from the area where Celtiberian epigraphy is attested, which is centred in the northern Meseta and upper Ebro valley, as well as from the distribution zone of the “suprafamilial units” frequently found in Celtiberian and Latin inscriptions in the central peninsula. In the latter, a form of code-switching is also observed, notably in the use of the genitive plural ending –on/-om in place of the standard Latin declension.¹²

Within the Gallaeco-Lusitanian theonymic zone, the canonical Lusitanian inscriptions are concentrated along a band between the Douro and Guadiana Rivers, while dedications with code-switching extend further north into historic Gallaecia. In addition to these clear territorial distinctions, notable linguistic differences exist between the two types of indigenous inscriptions. The indigenous datives in -bo and -ui found in mixed inscriptions are absent from the Lusitanian rock inscriptions – an absence all the more striking given the religious character of both groups. In Arroyo de la Luz I and Cabeço das Fraguas, the word indi appears as a copulative conjunction, whereas in the bilingual text from Viseu, igo serves this function. Conversely, certain linguistic phenomena are shared by both groups, particularly the derivation of adjectives with the suffix -ko- and the use of -e as a dative singular ending in specific contexts.

Recent research increasingly supports the view that Hispano-Western indigenous epigraphy reflects a cohesive linguistic landscape, encompassing elements of personal onomastics, theonymy, and toponymy. A. Tovar Llorente originally designated this language as “Lusitanian,” a term still conventionally used, although “Hispano-Western” may offer a more accurate label, as proposed by C. Búa. The linguistic variations mentioned above likely stem from local dialectal differences and distinct linguistic developments; the prolonged persistence of indigenous epithets, for instance, may have facilitated specific linguistic evolutions.¹³ However, other scholars, such as B. Prósper Pérez and J. García Alonso, argue for the existence of multiple languages in the region: one, genuinely Lusitanian – Indo-European but non-Celtic – dominant in the area of the canonical inscriptions; and another (or possibly more) of Celtic type, covering the northern theonymic zone, including Gallaecia (where Celtic toponymy is documented), and partially overlapping with the former. This might be exemplified by the bilingual altar from Viseu.¹⁴

This brings us to the question of Lusitanian’s linguistic affiliation. The language preserved in indigenous epigraphy displays several features that clearly distinguish it from Celtic languages. Undoubtedly, the most significant distinction is the retention of the phoneme /p/ in initial or intervocalic positions (e.g. porcom or Trebopala), in contrast to its loss – a defining trait of Celtic languages and a recognised hallmark of “Celticity.” Additional phonetic differences include the evolution of the voiced aspirated Indo-European labial /bh/ into the voiceless fricative /f/, as well as the preservation of the diphthong -ei. In terms of vocabulary, Lusitanian shares numerous lexical similarities with Celtic, yet diverges in aspects of grammatical vocabulary. Take, for example, the presence of the copulative conjunction indi (and possibly igo, if we assume that the Viseu inscription belongs to the same language) – a term absent from the Celtic lexicon – and the present tense stem of the verb doenti (“to give”) attested at Lamas de Moledo.¹⁵

The non-Celtic nature of Lusitanian (regardless of its geographical extent) was largely accepted by scholars from the outset, though not without dissent. For instance, J. Untermann argued for a Celtic affiliation, citing lexical similarities between the complex Lusitanian texts and those written in Celtiberian, Gallic, and Insular Celtic. Additionally, he pointed to the abundance of local toponyms with Celtic roots, particularly those containing -brig(a)-.¹⁶

Although lexical similarities are evident, J. Gorrochategui Churruca has offered a more nuanced view of their relevance in linguistic classification. Common vocabulary, in general, is more easily borrowed between other languages, whereas grammatical lexicon tends to exhibit greater continuity. As noted, the few grammatical elements identified in Lusitanian, such as the copulative conjunctions, are absent from the known Celtic lexicon, suggesting the potential adoption of Celtic-derived vocabulary by a non-Celtic language.¹⁷

The diverse linguistic landscapes across the Iberian Peninsula suggest the presence of multiple Indo-European linguistic influences.¹⁸ Notably, the Gallaeco-Lusitanian theonymic zone corresponds closely to the extent of the Atlantic Bronze Age in the peninsula, encompassing nearly the entire western coast from Galicia to the Algarve. In recent years, J. Koch and B. Cunliffe have linked the spread of Celtic culture to the framework of intense cultural contact that occurred between these Atlantic regions, western France, and southern Britain between the 13th and 9th c. BC.¹⁹

This proposal has faced criticism, with both archaeological and linguistic evidence generally supporting a much later expansion of Celtic, no earlier than the first millennium BC and, in the Iberian Peninsula, likely progressing from east to west.²⁰ Given that Lusitanian retains the phoneme /p/ in both initial and intervocalic positions, its divergence from the Indo-European parent language must have occurred prior to the formation of Proto-Celtic, in which the loss of /p/ is a defining feature.²¹ It is therefore intriguing to consider the Indo-Europeanisation of the peninsula as unfolding in at least two major phases: an earlier phase associated with the Atlantic Bronze Age, to which Lusitanian would belong, and a later phase corresponding to the Celticisation of the central peninsula, to which Celtiberian would pertain.

While this model aligns with the existence of distinct theonymic and anthroponymic zones separate from Celtic Hispania, it proves less effective in accounting for toponymy. The northwestern region of Hispania exhibits the highest density of –brig(a)– toponyms across the Western world. It has been observed that this toponymic element frequently appears alongside non-Celtic features and that its introduction into the northernmost parts of the Iberian Peninsula likely occurred at a later date – suggesting it may overlay an earlier Indo-European stratum. Nonetheless, the greater prevalence of –brig(a)– toponyms in Lusitania and Gallaecia than in Gaul or Britain remains somewhat anomalous. Moreover, the Celtic elements within the Lusitanian lexicon are not fully accounted for, other than as the result of some vague cultural contacts.

The Epigraphic Context: Roman Ways for an Indigenous Language

As M. García Quintela has recently noted, the emergence of Lusitanian inscriptions is the result of a dual Roman influence: the introduction of writing and a heightened impetus to monumentalise connections between humans and the divine.²² This phenomenon must be understood within the broader context of Roman presence in the region, particularly within the framework of the aforementioned epigrafização of northern Lusitania and Gallaecia.

A distinctive aspect of the epigraphic habit in this area is the notable prominence of religious inscriptions, encompassing both indigenous deities and figures from the classical pantheon. The proportion of religious texts within the total number of inscriptions is significantly higher than that observed in other parts of Hispania; indeed, in the westernmost areas of Gallaecia, religious inscriptions even surpass funerary epigraphy, namely the predominant type across the Roman world. This pattern closely parallels that observed in Britannia, another province where the introduction of writing – and thus the beginning of an epigraphic tradition – occurred relatively late, emerging alongside Roman occupation.²³

Epigraphy became an additional element in the local monumentalisation of cult places. Two paradigmatic examples stand out due to their high concentration of inscriptions: the sanctuary of São Miguel de Mota (Alandroal, Évora), dedicated to Endovelicus, and Monte do Facho (Cangas, Pontevedra), associated with the worship of Berobreo. Hundreds of altars dedicated to these deities have been recovered from both sites, carved from local granite and inscribed with standardised formulae.²⁴

Wall inscriptions also contributed to this monumentalising effect. The open-air cult spaces typical of the Indo-European Iberian Peninsula, often situated on hills and rocky outcrops, provided an ideal setting for such practices. In some of these locations, inscriptions were carved onto rock faces, large stones, or within natural cavities and cup marks, closely associated with the rituals performed in these sacred spaces. Alongside the previously mentioned sites of Cabeço das Fraguas and Fonte do Ídolo, Latin wall inscriptions are also found at Panóias (Vila Real),²⁵ As Canles (Campo Lameiro, Pontevedra),²⁶ San Trocado (Punxín, Orense),²⁷ Fonte da Tigela (Sabugal),²⁸ Penedo das Ninfas (Lamoso, Paços de Ferreira, Porto),²⁹ Mogueira (São Martinho de Mouros, Resende, Viseu),³⁰ and the Castro of Três Ríos (Tondela, Viseu).³¹ In addition to wall inscriptions, some of these sites also preserve altars. Examples include an altar dedicated to Reve Anabarego from Penedo das Ninfas;³² a series of granite altars dedicated to Laepo from Cabeço das Fraguas;³³ and a small altar dedicated to Nabia at Fonte do Ídolo, near Bracara Augusta.³⁴

Despite the challenges posed by the legibility of some inscriptions, the prevalence of local anthroponyms and theonyms is notable. Particularly striking is the presence of Jupiter – a deity of the classical pantheon – at San Trocado and Penedo das Ninfas, where scholars such as J. C. Olivares Pedreño and A. Guerra have associated him with the indigenous god Reve.³⁵

Cabeço das Fraguas and, particularly, Fonte do Ídolo, illustrate the coexistence of Latin and vernacular texts and the choice of language according to the intended linguistic register. At Fonte do Ídolo, in addition to the small altar, the wall graffiti themselves display this alternation of languages. The rock sanctuary underwent renovations at some point in the 1st c. AD, during which a wall of ashlar blocks was added to the uppermost vertical façade. It was at this time that the four graffiti were inscribed, along with two reliefs: an anthropomorphic figure nearly a metre in height, and an aedicula with a triangular pediment, framing a male bust accompanied by a dove and a club in the tympanum.³⁶

The four texts appear to have been produced by different hands and convey different information, yet they display a consistent thematic and stylistic coherence.³⁷ The first and third inscriptions, located beside the anthropomorphic figure and within the aedicula, respectively, mention the dedicator’s name, Caelicus Fronto, a native of Arcobriga, likely the benefactor responsible for the sanctuary’s refurbishment. The third graffito also provides the deity’s name, Tongo Nabiago, in the dative case, featuring the indigenous suffixes –e and –ui. The fourth text, positioned above the aedicula relief, remains difficult to decipher and interpret. As M. J. Estarán Tolosa observes, Fonte do Ídolo offers an excellent example of the varied rhythms of cultural transformation. The Lusitanian commissioner adapts his name to conform to Roman naming conventions and incorporates distinctly Roman iconographic elements. However, the names of the deities are deliberately inflected in the local language.³⁸

In summary, the five canonical Lusitanian inscriptions can be situated within the broader context of the emergence of rock epigraphy and its use in sanctuaries. The vast majority of inscriptions associated with cultic spaces – many of which are decontextualised – document the fulfilment of vows, reflecting a more individualised form of worship distinct from the communal rituals documented in the Lusitanian texts. Recently, F. Marco Simón and M. García Quintela have drawn attention to three Latin inscriptions that reference animal sacrifices, potentially indicating both the persistence and Romanisation of rituals originally attested in the canonical Lusitanian texts.³⁹

The first example is a votive altar from San Vicente de Alcántara (Cáceres), which records the sacrifice of a bull to Jupiter for the health and return of Lupus Alboni f.⁴⁰ The second is a small granite fragment from Viana do Castelo (Minho), likewise recording the sacrifice of several bulls to Jupiter.⁴¹ The third inscription, from Marecos (Penafiel, Porto), is a sacrificial tariff in which Lucretius Vitullinus and Lucretius Sabinus Postumus Peregrinus offered cows, oxen, and lambs to Nabia and Jupiter, specifying 9 April 147 AD as the date.⁴² Although these pieces originate from the Gallaeco-Lusitanian theonymic region and are Roman in both language and iconography, they exhibit elements of indigenous anthroponymy or theonymy, albeit without morphological traces of the local language.

Language, Religion and Identity

Overall, we observe a local provincial population making use of Roman cultural instruments – specifically, epigraphic communication – to document religious acts, either in their own language or in Latin, displaying noticeable local linguistic interference. Two distinctive features of this Lusitanian epigraphic corpus stand out. First, the sustained use of the vernacular language, particularly at a time when Palaeohispanic writing had either disappeared or become marginal throughout the rest of the Iberian Peninsula. Second, and specifically with regard to the written record, the use of this language appears to be confined exclusively to the religious sphere, with no evidence of texts relating to other domains.

Given the evident influence of Latin, there is general consensus that a deliberate linguistic choice was made – intended to underscore the religious act and, in the case of the canonical Lusitanian inscriptions, the ritual act itself. Notably, the texts from Arroyo de la Luz I and Lamas de Moledo begin with the names of the individuals responsible for their production (Ambatus in the former, and Rufinus and Tiro in the latter), who serve as the subjects of the Latin verb scribo, followed directly by the text in the indigenous language. Although we do not know the precise roles these individuals played in the creation of the inscriptions, it seems likely that all three had some knowledge of Latin.

Two related issues continue to spark scholarly debate. The first concerns the ideological motivation behind this ritual use of language; the second, how it should be interpreted within the broader linguistic landscape of the area. With regard to the first issue and bearing in mind the significant gaps in the available evidence, an identity-centred interpretation appears compelling. In periods marked by the coexistence of diverse identities – particularly prior to Flavian municipalisation and the relative homogenisation of legal statuses – the use of the vernacular language may have served as a form of collective reaffirmation, especially within communal contexts such as the ritual banquets to which these texts refer.⁴³

This identity-based interpretation is not without its critics. As noted, the indigenous epigraphic record comprises 31 inscriptions, 26 of which exhibit code-switching, featuring texts primarily written in Latin, except for the divine names. This represents a mere fraction of the much larger body of regional epigraphy, which is overwhelmingly composed in Latin. Even when considering only epigraphic references to the local Gallaeco-Lusitanian theonymy, the Lusitanian texts account for a small proportion of approximately 270 inscriptions. Thus, in over 90 % of cases, religious communication with local deities occurred in Latin, apparently without issue.⁴⁴

It is important to recognise that these two types of inscriptions belong to distinct functional registers, although both serve a public, display-orientated purpose. The canonical Lusitanian inscriptions are explicitly public in a collective sense (though not necessarily institutional), documenting rituals that engaged the wider community.⁴⁵ In contrast, mixed dedications reflect the fulfilment of personal vows, with communication centred on the relationship between the individual (or individuals) and the deity. A visual example of these differences can be observed at Cabeço das Fraguas, where, alongside the Lusitanian rock inscription, six  and small altar stones have been recovered. On these, individuals with indigenous names fulfil their vows to Laepo, expressed entirely in Latin.⁴⁶

There are two possible interpretations of this coexistence. One possibility is that we are witnessing two parallel epigraphic registers: Latin, considered appropriate for individual religious communication, and Lusitanian, reserved for collective ritual – implying a clear identity-driven choice. Alternatively, this may reflect a form of “linguistic conservatism”, in which the retention of the indigenous language serves a purely functional purpose linked to ritual practice. The lack of evidence for private texts in the Lusitanian language certainly poses a challenge to the first interpretation.

This directly relates to the second issue: how should these texts be interpreted within the regional linguistic context? Several factors must be borne in mind. From the outset, Latin was predominantly the language of epigraphic expression and ultimately came to replace the vernacular language. Beyond its role as the language of authority, Latin was, from an early stage, associated in western Hispania with the introduction of writing and epigraphic communication.⁴⁷ Beyond the sociolinguistic implications, it is important to recognise that the surviving epigraphic record is likely biased in favour of Latin and offers little insight into the oral tradition. Furthermore, religion is one of the sociolinguistic domains most inclined towards linguistic conservatism, and it is common across the Roman Mediterranean to find instances in which indigenous languages and scripts persisted, even amid increasing linguistic Latinisation – as exemplified by some of the cases discussed in this volume.⁴⁸ The central challenge lies in the scarcity of written records, which prevents us from assessing the pace or nuances of this process of linguistic replacement. Moreover, given the lack of additional sources, were it not for the roughly thirty indigenous inscriptions that survive, one might assume that the linguistic shift was both immediate and complete within a relatively short timeframe.

Does indigenous epigraphy, then, reflect a bilingual reality, or rather a situation of advanced linguistic change – or even diglossia – as has sometimes been suggested in scholarship?⁴⁹ The first scenario aligns with the idea that linguistic change typically unfolds gradually, progressing at different rates depending on linguistic contexts and registers. The complex Lusitanian texts may have been intended to communicate with a local audience capable of understanding them (if not necessarily reading them), with the recording of ritual and communal acts constituting one of the domains in which the language could persist. Although of a different nature, the prolonged survival of local anthroponymic structures within Latin epigraphy likewise reflects these underlying continuities.⁵⁰

The second scenario aligns more closely with the argument for purely ritualistic linguistic conservatism. J. de Hoz Bravo has proposed that Lusitanian was an exclusively oral language, while Latin quickly became associated with written forms. In a truly bilingual context, one would expect bidirectional linguistic interference; however, since this can be observed only in inscriptions featuring code-switching, it may not reflect the broader sociolinguistic reality. According to J. de Hoz Bravo, Lusitanian had already begun to decline from an early period, becoming confined to a ritualistic language role – an explanation that would account for the absence of non-religious indigenous texts.⁵¹ The local language likely served to enhance the efficacy of collective rituals, emphasise their antiquity, and even confer a symbolic aura, regardless of how many individuals could actually read or fully understand the texts. This phenomenon is not unique, but rather common across both ancient and medieaval contexts; one need only consider the use of Latin in Catholic liturgy, which persisted well into the 20th c.⁵² As M. J. Estarán Tolosa observes, this perspective is consistent with the available historical data; however, it still leaves unresolved the question of why the vernacular language was committed to writing solely for ritual purposes.⁵³

Chronology introduces an additional variable. An earlier dating of the five Lusitanian inscriptions would increase the likelihood that they reflect a bilingual reality, whereas a later dating, closer to the period of full Latinisation, would lend stronger support to the hypothesis of a purely symbolic ritual use. The case of the mixed inscriptions is somewhat different, as they contain Latin texts incorporating indigenous morphological elements; however, their dating would significantly affect their interpretation. If, on the one hand, they are contemporaneous with the Lusitanian inscriptions, this would suggest the coexistence of two distinct linguistic registers. If, on the other hand, they are later, the preservation of indigenous dative forms would represent a regional peculiarity within the broader Latinisation process.⁵⁴

In the following sections, we will examine two topics that directly bear on these issues: the chronology of indigenous Lusitanian epigraphy and potential parallels for linguistic conservatism within the Roman world.

Problems of Chronology

The potential to establish a chronological sequence would greatly enhance the analysis of linguistic change. However, as is often the case, the available evidence remains limited. With regard to the context of the Lusitanian inscriptions, systematic archaeological excavations at Cabeço das Fraguas have revealed continuous occupation of the site from the 4th c. BC to the 1st c. AD. While indigenous materials dominate the assemblage, a few fragments of the late Republican black gloss ware from the 1st c. BC, along with a Drag. 27 vessels from the following century, are noteworthy.⁵⁵ In relation to the mixed texts, as previously noted, the inscription from Fonte do Ídolo is dated to the 1st c. AD, likely within the context of Braccara Augusta’s foundation and expansion. Two altars have been found in situ: one at Lucus Augusti, specifically from a structure near the entrance to the Roman city and dated to between the 2nd and 3rd c. AD; the other, a bilingual altar from Viseu, was located on the acropolis of the Roman city near the forum and is broadly dated to the post-Flavian period.⁵⁶ The specific archaeological contexts of the remaining pieces, however, remain unknown.

In the absence of archaeological context or internal evidence, palaeography may serve as a tentative dating criterion, though it must be applied with caution – particularly in the case of tituli from rural areas, where inscriptions were produced outside the more standardised urban officinae lapidariae. Furthermore, there are specific challenges associated with inscribing on granite, which is a prevalent epigraphic material in the region under study.

Nevertheless, I. Simón Cornago has suggested that the abundant use of ligatures and the presence of the small letter O (o minuta) in the Lamas de Moledo inscription may indicate a 1st c. AD date, possibly from the latter half of the century.⁵⁷ The palaeography of the Cabeço das Fraguas inscription likely also points to the 1st c. AD date; although the text contains fewer distinctive elements, the proposed dating aligns with the associated Roman materials from the site. The more southerly inscriptions from Arronches and Arroyo de la Luz II may be slightly earlier, as suggested by the presence of the open P – a feature common in Republican and Augustan epigraphy. However, this palaeographic characteristic shows considerable continuity across many regions and even appears in 2nd c. AD texts from Augusta Emerita.⁵⁸ Naturally, no conclusions can be drawn regarding Arroyo de la Luz I due to the loss of the original inscription. In any case, more extreme datings proposed for some of these documents (e.g. 2nd c. BC) can be dismissed as inconsistent with the regional sociolinguistic and epigraphic context. Conversely, a 3rd c. AD date is likely too late for the survival of indigenous languages in the Iberian Peninsula.⁵⁹

The palaeography of the mixed altars is even less conclusive for dating purposes. In general, the small size of some arulae and the difficulty of carving on local granite result in an irregular ductus, further complicated by issues of conservation. Nevertheless, examples such as those from Viseu and Guiães display a more regular and carefully executed ductus, consistent with the quality of the medium used. Despite these challenges, no features typically associated with early palaeography, such as the open P or the rectilinear S, are present.

Although it falls well short of providing anything approaching an absolute chronology, it may still be worthwhile to consider the evolution of personal onomastics in order to establish, at least, some relative temporal coordinates.⁶⁰ In locations with sufficiently substantial epigraphic corpora, it is possible to trace a sequence of onomastic change, as demonstrated in the case of Idanha-a-Velha, the ancient Civitas Igaeditanorum, where the anthroponymy has been studied by M. Alves Dias. M. Alves Dias identified four phases of change affecting both the structure and the linguistic origins of personal names:

1. single nomen,
2. nomen + filiation,
3. nomen + cognomen + filiation,
4. tria nomina, each with different subtypes depending on whether their components were of Latin or local origin.⁶¹

It is evident that the 15 personal names recorded in the mixed inscriptions represent a very limited sample. Two of these belong to women and display a bimember structure: Caburia Caturonia, from Proença-a-Velha, with an indigenous onomastic, and Albinia Albina, in this case Latin.⁶²

The only example of a single nomen is Arcuius, from the altar at Minhotâes – a hapax that refers to the Celtic root –arc- (e.g. Arcobriga). Most names follow the structure nomen + filiation and predominantly consist of Latin elements. The nomen + cognomen formula is observed only in the case of Flavus Victoris from the Portas inscription, with Flavus commonly used across northwestern Hispania as both a nomen and a cognomen.⁶³ Instances of the tria nomina are rare: [Cae]licus Fronto Ambimogidus, commissioner of the Fonte do Ídolo inscription, and C. Iulius Hispanus, from the altar at San Martín de Liñarán – the former displaying indigenous elements. There is no mention of a tribe, which would confirm Roman citizenship.

The persistence of regional indigenous onomastics and the continued use of the nomen + filiation formula call for caution when evaluating the existence of a standardised sequence of onomastic Latinisation. An inscription from the freguesia of São Romão (Armanar) in northern Portugal, dated to 217 AD, refers to the aedilis Vegetus Talabari f., namely a member of the local elite who retained indigenous forms both in the bimember structure and within his onomastic elements well into the 3rd c. AD.⁶⁴ Thus, despite the limited sample size, some degree of Latin influence can nonetheless be inferred from the onomastic analysis. The bimember structure remains the most prevalent, yet Latin onomastic elements dominate. Indigenous anthroponyms have largely disappeared or are restricted to patronymics – a phenomenon that J. M. Abascal Palazón has recently described as “lost generation”.⁶⁵

While it is impossible to provide precise dates, it is important to recognise that these mixed texts reflect a relatively advanced stage of contact between the indigenous populations and the Romans. As A. Redentor notes in this volume, in reference to the specific case of the conventus of Bracara Augusta, religious conservatism did not necessarily extend to onomastic practices.

In summary, efforts to date or establish a chronological sequence for the indigenous epigraphy of northwestern Hispania yield more uncertainties than certainties. Most scholars place the canonical Lusitanian inscriptions within the 1st to 2nd c. AD, although recent palaeographic studies tend to favour the earlier end of this range. While not definitive, this interpretation supports the view that these texts emerged during an initial phase of stable contact with Rome, following the pacification of the region. The possibility that Arroyo de la Luz II and Arronches may date slightly earlier, potentially to the Augustan period, would also be consistent with this historical trajectory.

The mixed inscriptions likely belong to a somewhat later period, when the indigenous language had largely disappeared from public use. Although the extent to which it survived in oral form remains uncertain, all indications suggest that it had declined to the point where it was no longer committed to writing. The exception lies in the indigenous datives preserved in the mixed inscriptions. If we propose a broad dating within the first three centuries of the Common Era, the inscriptions for which more precise data are available – together with general onomastic trends – point primarily to the 1st c. AD.

Religious Conservatism and Indigenous Languages in the Roman West

As previously noted, processes of linguistic change occur at varying speeds depending on the sociolinguistic register, with religion being one of the domains where this variation is most pronounced – either acting as a catalyst for change or serving as a bastion of conservatism. The western Roman world offers numerous examples of the latter, notably including Lusitanian epigraphy.

In her recent study of religious epigraphy in vernacular languages, G. de Tord Basterra highlights several distinctive aspects of the Lusitanian corpus in comparison with other Palaeoeuropean epigraphic cultures. Most of these cultures predate the Roman presence, although their periods of greatest epigraphic activity often coincide with their incorporation into the Roman sphere. G. de Tord Basterra emphasises that, in the Lusitanian case, only religious inscriptions survive, with no attested use of the indigenous language in other contexts. This results in a limited typology – rock inscriptions and altars – both of which have a public character and show a marked predominance of bilingual and mixed texts.⁶⁶

Italy, for example, is a region where local epigraphic cultures underwent substantial Roman influence. In its central area, a group of approximately half a dozen dedications stands out, featuring the Oscan dedicatory formula brateis datas alongside the names of dedicants and deities in Latin.⁶⁷. This intra-sentential code-switching once again illustrates linguistic retention associated with religious practices within the broader process of Latinisation. However, as M. J. Estarán Tolosa notes, the Lusitanian context does not precisely mirror that of Italic epigraphy, which reflects an earlier phase of linguistic change in which Oscan retained significantly greater vitality.⁶⁸

The persistence of the Gaulish language – attested in the written record at least until the 3rd c. AD and likely used orally for up to two centuries thereafter – offers further examples of linguistic contact and retention within the religious register.⁶⁹ To cite just a few well-known cases, the Pilier des Nautes in Paris features representations of both Gallic deities and figures from the classical pantheon. In the former, Celtic inflection is retained (e.g. Esus, [C]ernunnos, and Taruos Trigaranus), while the latter are declined in Latin (e.g. Iovis, Volcanus, Castor, and Fortuna). The ensemble also includes a dedication to Emperor Tiberius and Jupiter Optimus Maximus, which is particularly valuable for dating the monument and, as expected, is inscribed in Latin. The linguistic choices are both evident and deliberate, with Gaulish and Latin used in accordance with the cultural domain associated with each deity.⁷⁰

A similar phenomenon is observed in a stone statue found in Lezoux, which bears a Latin dedication to Mercurio et Augusto alongside a second inscription, in Gaulish, dedicated to the local deity Esus.⁷¹ A third instance of code-switching appears in a votive inscription, possibly originating from a portico in Sazeirat. Here, Latin appears to intrude upon an otherwise predominantly Gaulish text: the name of the dedicant, the verb, and the donated object are rendered in Gaulish – Sacer Peroco(s) ieuru duorico – while the inscription concludes with the Latin formula v(otum) s(olvit) l(ibens) m(erito).⁷²

To the south-west, in Aquitania and northeastern Spain, a non-Indo-European language related to Basque, known as Vasconic-Aquitanian, is primarily attested through onomastics and toponymy dating from the Roman period.⁷³ In a series of dedications to local deities, personal names and formulaic expressions are inflected in Latin, while the theonyms feature a vernacular dative ending in –e. Examples include Deo Larahe (Irujo, Navarra),⁷⁴ Selatse (Barbarín, Navarra),⁷⁵ Itsacurrine (Izcue, Navarra),⁷⁶ Helase (Miñano Mayor, Álava),⁷⁷ or Herauscorritsehe (Tardets-Sorholus, Pyrénées-Atlantiques).⁷⁸

Recently, M. J. Estarán Tolosa has drawn attention to instances of code-switching associated with linguistic retention in religious inscriptions from Germania Inferior – a phenomenon of considerable sociolinguistic interest, given the region’s late Latinisation and its clear parallels with the Lusitanian case.⁷⁹ In the cult of the Matres, certain epithets appear with the suffix –ims, while the remainder of the text is in Latin. Examples include references to Matronis Vatuims, found in inscriptions from Rodingen,⁸⁰ Lipp,⁸¹ and Morken;⁸² to Matronis Saitchamims in Hoven;⁸³ and to Matronis Aflims in Wesseling.⁸⁴

Conclusion

Turning to the bilingual altar from Viseu, we find that the dedicator, Albinus Chaereae f., employs both Latin and an indigenous language to consecrate the altar to the gods and goddesses of Vissaeum. The structure, ordinatio, and formula are consistent with hundreds of other votive altars throughout the Empire; however, the inclusion of the vernacular language in invoking local deities is noteworthy enough to have been preserved in writing. The personal onomastics also exemplify the coexistence of diverse cultural traditions: Albinus is a common Latin name; Chaerea is a Latinised Greek name, often found among freedmen, though not necessarily of Greek origin; and the two-part structure, combining personal name and patronymic, is a typical feature within the Hispanic context.

Roman imperial society allowed a significant scope for the negotiation, selection, and adaptation of cultural models, resulting in a rich amalgamation of local identities and cultures in which “Roman” and “indigenous”, although distinct, were neither opposed nor mutually exclusive. Northwestern Hispania provides compelling examples of this phenomenon. While language and religion clearly play pivotal roles in the construction of collective identities, this process should not be understood as inherently oppositional. Rather, it reflects adaptation and adjustment to the new reality shaped by Rome and the gradual transformation of local religious systems.

From this perspective, Hispano-Western indigenous epigraphy can be seen as the product of cultural contact, whether framed in terms of “Romanisation” or through alternative concepts such as “glocalisation” or “hybridisation”. The examples from Gaul and Germania demonstrate that northwestern Hispania is distinctive both for the relative abundance of preserved testimonies and for the absence of other types of epigraphic records outside the religious domain. At the same time, they show that linguistic retention within religious epigraphy is by no means unusual in the Roman West.

Of course, many questions remain unanswered. Aside from some indications suggesting that more complex texts predate the mixed inscriptions, we still lack a clear understanding of the sociolinguistic vitality of the Lusitanian language or of its evolution over time. The potential discovery of new epigraphic evidence, along with more detailed palaeographic studies and the identification of local workshops, could yield further insights. Likewise, a more comprehensive analysis of the dedicators’ anthroponyms in religious inscriptions might assist in identifying sociolinguistic profiles.

Over the past two decades, substantial progress has been made in compiling and analysing pre-Roman toponymy in Europe, supported by advances in computational methods – in which Hispanic materials have frequently played a pivotal role (e.g. Adopia or Hesperia). This work contributes to the delineation of onomastic regions and potential dialectal variations, as well as the identification of new isoglosses in the northwestern Hispanic area. Cross-referencing this linguistic data with archaeological findings, including palaeogenetics, will be crucial in resolving outstanding questions regarding the affiliation of the Lusitanian language and the Indo-Europeanisation of the Iberian Peninsula. In short, numerous avenues of inquiry remain open, and a multidisciplinary approach is essential for advancing our understanding.

Abbreviations

Universidade de Lisboa – UNIARQ. jherrerarando@gmail.com. ORCID: http://orcid.org/0000-0001-7524-0656. This work has been written within the framework of the research project The Birth of the Epigraphic Culture in Roman Lusitania, Fundação da Ciencia e Tecnologia (2022.03547.CEECIND / DOI: https://doi.org/10.54499/2022.03547.CEECIND/CP1762/CT0002).