Introduction

By inviting me to include this article in a volume entitled The Celts of the Iberian Peninsula, it is clear Manuel Alberro was already thinking of the relationship that has long been recognized between the large number of place-names in ancient Hispania that show the -briga element and the presence or movement of Celtic-speaking people in ancient times. The existence in Hispania of a very large number of place-names with the -briga ending has been considered of great importance for several decades in the reconstruction of the languages spoken in pre-Roman times in the Iberian Peninsula. The original map prepared by Jürgen Untermann (1961: 26-27 and Map 3) is probably the best example of this (Figure 1). He was able to show that Hispania could be divided into two main areas on the basis of these place-names. West and north of the dividing line he identified as the mainly Indo-European area. In his view, the presence of toponyms of this type was proof of the Indo-European (specifically Celtic) character of the peoples and languages of this area. To the east and south of this line other toponymic types (ili- /ilti-, etc.) would document the non-Indo-European character of these areas.

Figure 1. -Briga and ili-/ilti- place-names (after Untermann's 1961 map).
Figure 1. -Briga and ili-/ilti- place-names (after Untermann's 1961 map).

I think most scholars today1 would generally agree that things are not as clear-cut as this. We should expect (and I believe we find) a significant number of Indo-European (Celtic and non-Celtic) place-names on the 'wrong' side of the line, and the opposite is true for the non-Indo-European names.2 This does not mean, in general terms, that the whole hypothesis is completely wrong. It should be seen, I think, as one more example of the difficulties entailed by the reconstruction of the peopling of these different areas and the spread of languages that this involved. Only the detailed analysis of all place-names from all areas based on as few preconceived ideas as possible (or rather, without pre-conditioning the result of each individual case) can help us advance on the difficult road to a better understanding of what languages were spoken where and when.

In general terms, to summarize the above, we can expect to find a division of the Iberian Peninsula in antiquity into two regions: Indo-European and non-Indo-European. The first area would mainly correspond to the center, north and west, and the second to the rest, that is, Mediterranean Spain, together with part of the Basque region, the Pyrenees from Navarre to Catalonia, the northern and eastern part of the Ebro River valley and almost all of Andalusia (although its western part was subject to strong Indo-European influences). Recently, however, Francisco Villar (2000) has offered a new, somewhat revolutionary approach according to which there might have been a very old Indo-European layer that was particularly strong in the south.3 In my own work (see for instance 2003) I have also interpreted place-names as Indo-European that are found in regions generally considered non-Indo-European.

Before we proceed further, however, how do we know which languages were spoken in each area? Apart from the four main nuclei in the Iberian Peninsula with an indigenous epigraphy, consisting of Tartessos and the Iberian Mediterranean coast in the pre-Indo-European region, and Lusitania and Celtiberia4 in the Indo-European region, we have a long series of peoples in Hispania who have not left native texts. Names are our only direct testimony of the languages spoken in those areas and we have to give them all due attention, despite the problems entailed by their analysis.5 On this basis, Indo-European Hispania can be divided into areas according to the relative frequency of certain personal names in combination with what we know of those two areas just mentioned.

Proper names become very important when they constitute our only source of information. However, we need to bear in mind the danger of constructing hypotheses based solely on proper names, since they present inherent problems, including the fact that they may have spread far beyond the spatial and temporal boundaries of the language in which they originated.6

This problem is obviously even more acute when dealing with otherwise unknown languages. However, even when dealing with languages of which we know very little, it is indeed possible to draw general maps of the distribution of certain types of place-names and infer from this the expansion at a certain point in time of the languages from which they derive.7

-Briga and its significance as a place-name element

I have been working for over a decade now with ancient place-names in the Iberian Peninsula in an attempt to draw a hypothetical map of the languages spoken there in pre-Roman times.8 Against this background, it is high time to return to the special case represented by that special group of toponyms ending in -briga. We will review and analyse them here to provide, if possible, an answer to the question of what they may signify.

After9 the publication of Untermann's distribution map (1961: 26-27 and Map 310), María Lourdes Albertos Firmat (1990), Javier de Hoz (199411) and Francisco Villar (199512), among others13, have worked specifically on this subject. Albertos Firmat has gathered a very significant collection of names and made important efforts both on the linguistic front and in terms of the identification of the places so named. Villar's contribution has been more specifically linguistic, as the title of his 1995 chapter indicates ("La morfología de -briga y el tratamiento del grupo /ks/" [Villar 1995]). Based on de Hoz's (1994) and Villar's (1995) articles, it is clearer than ever that what we are still calling -briga names should perhaps more accurately be called -bri(g)s names, due to the declensional type apparently preferred by the Celts in Spain.14 Patrizia de Bernardo Stempel, in a recent and important article, also deals with this subject, suggesting a possible connection between ethnic names such as Artabri or Cantabri (2002: 107) and the -briga place-name series. Blanca María Prósper has contributed a significant chapter titled "Fonética y morfología de los compuestos toponímicos celtas en *-brig-"15 in her recent book (2002: 357-382). In my own work (see particularly 2003) I have always advocated the celticity of the different -briga names found in the different areas of the Peninsula. Javier de Hoz (1983, 1995) had already drawn our attention to the distribution of -briga names in the southwest, which coincides roughly with other, at first sight perhaps less visible, Celtic names, as well as with other non-Indo-European names. He also produced a very interesting map showing the repetition of -briga names in different areas of the Peninsula, which would support the idea of a secondary use not only of the -briga name, but more specifically of the very same name by groups of migrating Celts within Spain.16 These Celts, in some cases, may have been in a state of isolation among foreign languages and peoples, something that could be particularly connected with the use of -briga names. It would not be unthinkable that a -briga name may in some cases be taken as evidence that the Celts felt more at risk in such areas. That may have been the reason for choosing a name that alludes to the presence of a fortress. However, there are indications suggesting this type of name was used in the very heart of Celtiberia as well, where it could be supposed the Celts would have felt more at home. Nevertheless, it is true that the -briga names tend to exhibit a western distribution in Hispania, although it is indeed difficult to draw very detailed conclusions from this. I will return to this point later on.

In order to clarify the role that these names play in the identification of languages in ancient Hispania, we should begin by asking ourselves whether there is something from the point of view of language that actually connects -briga with Celtic-speaking people(s). Is there anything absolutely, undeniably Celtic in this place-name element?

The answer is, perhaps, not as clear-cut as one would expect. As one more Indo-European (IE) branch (to use a partially inaccurate but age-old metaphor), Celtic languages share a large repertoire of roots with languages of other IE groups, a sequence of (ideally) three or four phonemes, which form the basis of verbs, nouns, adjectives, etc. Pokorny's dictionary (1959-69) can be consulted for a large number of examples of these roots (and derivatives for each case in a very significant number of languages in this family).

-Briga is derived from one of these roots: *bhergh-. In Old Irish (OI) we have nom. Brí (< *brig-s), acc. brig, gen. breg, dat. brí. The word is also well attested in other Celtic languages.17 In fact, as I already mentioned above, there are strong indications that "la única forma autóctona del estereotipo toponímico celta -brig- en la Península Ibérica es un tema en oclusiva"18, -brig-s > briks > bris, in line with Old Irish. The Gaulish form would be -briga and this could have had something to do with the fact that we find many, if not most names in Spain in the form of the -briga variant, although Villar believes19 that "la modalidad -briga es fundamentalmente la forma latina de esta palabra". However, de Bernardo Stempel prefers to think that the form -briga "típica de galo y británico, no representa un latinismo, sino la forma normal para el celta del 4º período"20. The same Indo-European root is also the origin of the Germanic cognates berg or burg, Gothic baurgs. The first meaning is, apparently, something like 'hill', from whence comes 'fortress' or 'fortified town', until we may eventually get simply to the idea of an urban enclave that was not necessarily fortified. This is, by the way, the origin (through a Germanic mediation) of a very well known word in modern European languages: bourgeois, burgués, etc.

If, then, this is a derivative of a well-known Indo-European root well attested outside the Celtic languages, what is specifically Celtic about the toponymic element -briga? This brings us back to the question originally posed at the beginning of this essay.

According to Indo-European morphology, the roots whose structure we have just introduced can alternate the vowel that constitutes its nucleus following a set of rules that I will not discuss here. Theoretically speaking, we have three vocalic alternatives:

There are morphological reasons for choosing one or the other alternative. Evidently, this is not a free choice in a particular a language and even less so, of course, on the part of the speaker; -briga is a derivative of the zero-grade (< *bhrgh-a). It is in the phonetic results of the three alternatives ideally reconstructed for Indo-European that we can find the fingerprints of the individual languages, if we are lucky. What happens with the zero-grade form is that the most open phoneme of the sequence finds itself needing to assume the function of nucleus to that syllable. As a consequence, the -r- in our sequence (*bhrgh-) constituted / became the nucleus or 'vowel' of that syllable. But /r/ is a sound with certain limitations regarding its ability to function as a vowel, as is also the case with the /l/, the /m/, the /n/ and other sounds that, in Indo-European, found themselves in that position. Therefore, in those contexts, a vocalic 'reinforcement' of these somewhat 'special' vowels was surely perceived in the pronunciation (by the lips of Indo-European speakers) of those sequences, in some cases right before, in some cases right after, in some other cases both before and after, and, even, in some other cases, the new vowel completely replaced the liquid, nasal, etc. One may imagine that, in the first stage of this process, the exact quality of that vocalic 'reinforcement' could result from contextual constraints. For instance, in the context of a labial sound, it would be perceived as an /o/ or as an /u/, whereas in the context of a dental sound, it could be perceived as an /i/ or an /e/. It could be so. But, as far as we know, all branches of the major Indo-European family 'phonologized' those sounds, in all contexts, choosing one of these values more or less independently of one another. This is why the exact quality of the vowel that in historical languages appear relatively recently in the context of an old zero-grade has become an important clue, a kind of fingerprint for the classification of languages, with all the reservations that one must keep in mind in such cases. For instance, we know that from a vocalic /n/ in Greek we have -α-, in Latin -en- > -in-, in Germanic -un-, etc. Of course, when one or several languages share the same treatment, the identification of the result is much less evident. But in this respect we are lucky with the -briga example.

Celtic languages produced a vocalic phonetic result out of a vocalic -r- which is quite peculiar in the context of the whole set of languages of the Indo-European family, both in the order of appearance of the phonemes and in the exact quality of the new vowel. - Ri- from vocalic -r- is a quite isolated result. This is what gives us some basis for the argument that, phonetically speaking, it is quite unlikely21 that another Indo-European language from Western Europe in general or from Hispania in particular could have produced the same phonetic result. Lusitanian, which in this respect could have been a very plausible alternative candidate, shows /ur/ and then /or/ as the phonetic development of a vocalic /r/.22 The evolution of the two aspirates (-bh- and -gh- ) to the two voiced stops is much more common23 in Western Europe and therefore -briga from < *bhrgh-a is much less peculiar as far as the stops are concerned.

The answer, therefore, to our initial question is that the phonetics behind -briga, more specifically the evolution of the vocalic -r-, strongly suggest that this element was created by Celtic speakers. Therefore, the historical appearance of place-names with that element should, in principle, be attributable to the action, existence and presence of Celtic speakers. At this point we must define the limitations of this theory before proceeding further: although -briga, as a place-name component24, was generated by Celtic speakers, it could be (and was) appropriated by speakers of other languages as well. However, whether a particular name can be attributed to Celts or, for example, to Romans, is in some cases quite clear:

However, in other cases the distinction is not clear at all. As outlined above, -briga typically appears as the second element of a compound. When this is so, it is determined by the initial element of the compound, which, in most cases, linguistically pre-exists, i.e. precedes -briga: it was there before and can be ascribed to an older linguistic layer. The place is named by coining a compound that could be translated by a whole syntagma, including:

It is easily understood, I think, how, in the process of coining the new place-name, the Celtic speakers might incorporate pre-existing non-Celtic elements. This does not undermine the Celtic character of the toponym. The use of -briga can, in principle, be attributed to the presence of Celts: they were there and they created the new compound. It is true, however, that if they used a pre-existing personal or place-name that shows in its structure its non-Celtic character, that can be taken as a sign of the presence, in most cases difficult to date, of speakers of another language. When the place-name Cordovilla was created by speakers of Spanish, in the Middle Ages, from the pre-existing Córdoba, created by the pre-Roman inhabitants of the valley of the Guadalquivir, the newly coined name was a pale reflection of the fact that the old language that created the original name existed at some time in the past, but no more than that. In this case we know that the distance between both names and both languages is very large (several centuries). In many cases in ancient times, even if we are able to produce such refined analyses as this, it is impossible to assess the antiquity of the pre-existing element and therefore to know whether both languages were in fact at some point in direct contact.

To return to our -briga names, what we must bear in mind is that a -briga name with a non-Celtic initial component, is, paradoxically, not necessarily less Celtic than other 100% pure Celtic bi-partite -briga place-names. This is something to bear in mind when studying the construction of toponyms. We can consider a theoretical classification of these place-names into at least the following types:

Obviously, it is possible to have linguistic doubts about 1) the Celtic character of the initial element and 2) the fact that to classify a name such as Caesarobriga in Group 4 is much easier than doing the same thing with a name with a non-Celtic first element from an unidentified language. While it is possible that such cases existed, this is by far the least likely scenario.

-Briga names of ancient Hispania

Both Albertos Firmat and then Villar (like Untermann before them) made considerable efforts to gather significant collections of this type of place-name in Hispania. Villar even gathered a large repertoire of modern names, a good proportion of which may go back to ancient names of this type. I will not emulate their efforts here but will comment on a few examples, attempting to classify these place-names in one of the four categories just mentioned, followed by a few general comments on the possible implications of their distribution.

Type 1

Κοτταιόβριγα29
Name of a town of the Vettones. Near Almeida, by the Coa River, not far from the Spanish border (the province of Salamanca)?

*cott(o)-ai-o- + -briga: 'Hilly Fortress'. This place-name has a clear Celtic aspect, thanks to the suffix -briga. As for the initial element, it could also be Celtic. Rivet and Smith (1979: 259) relate this to Celtic *kotto- 'old' (Cornish coth and Breton coz 'old'). We would then have a Celtic equivalent to the Greek Παλίαα πόλις. See also Evans (1967: 186-7). There could be another explanation for the initial element, however. Ramón Menéndez Pidal (1968: 269-70) offers a wide selection of names from the Peninsula based on *cotto- and forms derived from it, relating them to modern Castilian cueto, 'hill'.30 I think it is clearly possible to relate it to the previously analyzed Celtic *cotto-, despite the initially apparent semantic gap.31

Λονδοβρίς32
Name of an island just off the Lusitanian coast. Berlengas Islands, near Cabo Carvoeiro.

*Lond-o-brig-s: "Londos' fortress". The first element of the compound is a perfect parallel to the British Lond-in-ium, London, ultimately, according to the traditional explanation, from the Celtic *londo-s, 'wild', related to OI lond, 'immitis, amarus, commotus', and Medieval Irish (MI) lonn, 'angry'. The details here are the problem. Londobris seems to be the only possible comparison. Even if the initial element is not so clearly Celtic, the compound is. We have here *brig-s > brix (> briss), apparently the most authentic Hispanic version of -briga.

Νεμετόβριγα33
Name of a city of the Astures (capital of the T(r)ibures). Trives Viejo, on the upper Sil River, near Puebla de Trives (TIR K-29, 101).

*Nemet-o-briga: 'Temple Fortress'. As for the etymology, it is rare to be confronted by such an obviously Celtic name, formed from the Celtic nemeton 'sacred grove', known in Gaulish, and Celtic -briga. There are forms with the same root attested in Britannia (Rivet and Smith 1979: 254-55 and 424). Holder (1896-1907, II: 708 ff.) prefers to translate neme-to-n as 'sanctuary', from the adjective *neme-to-s, 'sacred', 'noble', known as a personal name. Therefore, Nemetobriga may be 'Temple Town' or 'Nemetos' Town', this being a personal name. He gathers a long series of names based on this element: Nemetacon, Nemetavi (a people from Galicia, in whose territory included the town of Ούολόβριγα, [Ptolemy II, 6, 40]), Nemetes, Nemetiales, Nemeto-cena or -gena, *Nemeto-duro-s, *Nemeto-ialo-s, Nemet-ona, Nemeto-tacio, Nemeturicus.

Λακόβριγα34
Name of a city of the Vaccaei. Carrión de los Condes. It has variously been identified with: Aguilar de Campos, Villanueva del Campo, La Mudarra, Castroverde de Campos, Paredes de Nava and Villalpando. Lacobrigenses (Pliny III 3 26); Langobriga (It. Ant. 421)35.

*Lak-o-briga: 'Lake Fortress'. Here we have our old Celtic friend -briga, and an initial element Lac-o- that may also be Celtic. The variants Lango- or Lanco- take us into the sphere of names such as Segontia Lanca (a town of the Arevaci), Lancia (a town of the Astures and at least one of the towns of the Vettones) and various derivatives that also seem Celtic.

Αράβριγα36
Name of a city of the Lusitani, not far from the Tajo River, in central Portugal?

*ara + briga / *are + briga: 'the fort of the fields'? / 'the fort at the front'? This name seems Celtic, and is formed either with *are- 'ante, pro' (< ie. *p˚ri-) -- with the associated difficulty of explaining the a in Arabriga -- or with a Celtic ara, as in Gaulish Ara-Pennis, OI ar 'cultivated field', Welsh âr 'meadow', even Spanish serna (< sen-ara). This is again followed by our by now well known -briga.

Other37 Type 1 names38:
K-29: Adrobrica, Coeliobriga, (Flavia) Lambris, Mirobriga
K-30: Arcobriga, Deobriga, Deobrigula, Lacobriga, Meidubrigensis, Nertobriga, Segobriga
J-29: Arabriga, Arcobriga, Caetobrix, Eberobriga, Laccobriga, Londobris, Mirobriga, Nertobriga
J-30: Mirobriga, Segobriga
Prósper (2002) lists the following: Letiobris39, (B)laniobris, Agubri, Aviliobris, Ercoriobris, Miobri(s), Lubri

Type 2

Σαραβρίς40
Name of a city of the Vaccaei.41 Torre del Sabre, between the provinces of Salamanca and Zamora. As follows in Ptolemy's manuscripts XARVOKN: Σάραβρις Z: Σαλαβρίς C: Σαράβρις U: om. W: Σαραβίς in Müller's edition (1883-1901: II, 6, 49). Sabaris (It. Ant.), Sibaris, Sabaria (It. Ant.), Sebarium (Rav.).

*Sara-brig-s: 'The town by the Sara [River]'. This name is also found as Sabaris, Sibaris or Sabaria. If Sabaria is the right form, we could think of a radical Sav-, to be seen in a long series of names collected by Holder (1896-1907). These names could go back to the Old European (OE) root *sau-. Javier de Hoz (1963) lists several instances from Hispania. We could add to them a place-name of the Pelendones mentioned by Ptolemy, Savia, in addition to the Sabaria (or *Savaria) mentioned here, if this is the right form of this name. But if we take Σαραβρίς as the right form, this would take us into the sphere of the clearly Celtic -briga. Sarabris would have what seems to be the authentically Hispano-Celtic variant of the -briga element: -brix, for which we have several instances of development to -bris (cf. the name of a Lusitanian island, Londobris, clearly Celtic). As for Sara-, Holder has collected an interesting series of names with this element. Holder thinks of Irish suire, 'water, river', which would be convenient for a hydronym (Holder 1896-1907: 1365ff). But we should also consider OE *ser-, *sr-. Javier de Hoz (1963) has provided various examples from the Iberian Peninsula. Phonetically speaking, Sara- is exactly what we would expect in Old European. In Celtic we would expect *ser-, or *sor- instead. Therefore, I think that there was an OE Sara- to which Celtic speakers added their -brix (> -bris), creating their 'town by the Sara River'. A previous OE river name appears to have been appropriated by Celtic speakers to form this Celtic toponym.

Ταλάβριγα42
Name of a city of the Lusitani.

*tala + briga: 'the fort of the Tala [River?]'. This compound is Celtic even if the initial element is not. -Briga is Celtic and shows the handling of a previous name by Celtic speakers. Tala- could be classified as OE, to use our traditional label. Villar (1993) related it to *(s)tel- 'flow' (IEW 1018), the origin of Greek τέλμα 'standing water, pond, marsh, swamp' or OI talam 'earth'. This element is known elsewhere in the Peninsula (Talamine, a Galician town, Talabara, a place of the Carpetani, etc.).

Other Type 2 names:43
K-29: Avobrica, Calubriga, Lansbrica, Volobriga
K-30: Abulobriga-Amallobriga, Dessobriga-Tonobrica-Tenobrica
J-29: Conimbriga, Ierabriga-Perbrigam, Meribriga, Montobriga, Turobriga
J-30: Turobriga

Type 3

Βριγαικινω̑ν (Βριγαίκιον)44
An ethnic group and a city of the Astures. Dehesa de Morales, in Fuentes de Ropel (Zamora: TIR K-30, s.v.). Traditionally (Hübner, cited by Tovar, who seems to accept it [1989: 324]; Müller [1883-190 ]; Holder [1896-1907, I: 349]; Bosch Gimpera [1932: 523f.]), it has been considered the same town Ptolemy himself calls Βαργιακίς and locates in the neighbouring territory of the Vaccaei.45

*bhrgh-aik-ion (> *bhrgh-aik-in-oi): ' Fortress Town'. Brig-aik-ion looks linguistically transparent. Its consideration as a place-name created by Celtic speakers seems clear due to the presence of -briga. -Briga is more common as a second element in compounds (Segobriga, Nertobriga, Iuliobriga), but it also appears on its own and followed by suffixes. Rivet and Smith (1979: 278) state that -briga is rare as an initial element in compounds, except when, as in this case, it is accompanied by suffixes. They cite several examples, among them our Brigaecium, from *bherg-46 in zero-grade, with a clearly Celtic vocalization, followed by the well-known suffix -aik-.

Other Type 3 names:47
K-29: (Flavium) Brigantium

Type 4

Αὐγουστόβριγα48
Name of a city of the Vettones. Bascos, 93 km from Toledo?

August-o-briga: 'the fortress dedicated to Augustus'. -briga is Celtic, but the first element of the compound is Latin.

Ιουλιόβριγα
Name of a city of the Cantabri. Pliny III, 21. Near Reinosa, by Retortillo.

*Iuli-o-briga: 'Town/Fortress dedicated to Iulius or the Iulia family'. -Briga is Celtic, but the compound is Latin.

Other Type 4 names:49
K-30: Flaviobriga
J-29: Caesarobriga

Conclusions

Now that a few examples of each type have been presented, it might be useful to reflect a little on the implications of the presence of the toponymic element presented here for an understanding of the movement of Celtic speaking people(s) (Figure 2).

Figure 2. Celtic (red) and non-Celtic or obscure names (blue) in the Duero plateau.
Figure 2. Celtic (red) and non-Celtic or obscure names (blue) in the Duero plateau.

It is clear to me, according to the definitions of the different categories outlined above, that place-names of Types 1, 2 and 3 should, in principle, be taken as a strong indication of the presence of Celts in the area. If toponyms of this type have a Celtic identifier in their own phonetic structure (as we have demonstrated), the coining of such names is a linguistic activity that must be attributed to Celtic speakers.

Type 4 clearly consists only of names with a Latin initial element. However, there may also be some names with a non-Celtic (and non-Latin) initial element in Type 2 that could have been categorized as Type 4. There could not have been many of these, however, as the activity of place naming was certainly more straightforward among 'barbarians' than among the 'civilized', global-minded conquerors of the world. The Romans could easily have 'celticized' a name of their own50, without that necessarily being evidence of the presence of Celtic-speaking peoples. In any case, we must accept that at least one of the aims in many cases was probably for the new arrivals to adapt themselves into the linguistic atmosphere of the territory, some kind of concession to the alien surroundings. Of course, since the Romans cannot be taken as ethnological / linguistic experts (not their aim in any case), it is somewhat problematic to give credit to the Romans for the coining of names classified as Type 4 (such as Flaviobriga) as an indirect hint of the presence of Celts in the neighborhood.

The main conclusion that I think should be drawn from the preceding discussion is that the appearance of -briga/-bri(g)s names of Types 1, 2 and 3 should be considered direct evidence for the presence of Celts in an area, with a few exceptions that it would be difficult to confirm.51 By "Celts" I mean "speakers of a Celtic language"52, not necessarily Celtiberians, since we cannot and should not discount the possibility that other Celtic languages/dialects existed within Hispania (whether closely related to Celtiberian or not) as a methodological approach based on linguistic evidence53.

However, we cannot therefore conclude that:

In light of the preceding discussion, we can look again at the map presented at the beginning of this article and see something different in it (Figure 3):

Figure 1. -Briga and ili-/ilti- place-names (after Untermann's 1961 map).
Figure 1. -Briga and ili-/ilti- place-names (after Untermann's 1961 map).

In any case, I think this type of place-name has always been central to the identification of the spread of languages (Celtic in particular) in Hispania and will continue to serve this purpose. However, I feel we are still defining the way to evaluate such evidence appropriately.

In my view, all the place-names with -briga / -brig(s) should be classified as belonging to one of the four types described here (Figure 4). Types 1, 2 and 3 in particular are direct evidence of the spread of Celtic, although it is important to note that they show a higher concentration in western areas of Hispania. This does not have to be considered evidence for the celticity of Lusitanian55, which, among many other arguments, seems to show a different treatment of the vocalic vibrant56, of essential importance in the linguistic analysis of the -briga / -brig(s) element in particular. It is clear that Lusitanian was a language spoken in the western areas of the Peninsula, and that it was different in nature from Celtic. But nothing can confirm whether or not this language alone was spoken there. I believe the spread of -briga names, among other things, is proof of the presence of Celtic-speaking peoples in the west as well. Of course, this evidence also should not be considered in isolation. All conclusions based on the -briga names should be combined with the remaining name data (not only place-names) and all data from other sources (linguistic and non-linguistic). We are continuously gathering and refining the evidence for the ancient languages and peoples of the Iberian Peninsula. Clearly wide areas of Hispania were linguistically complex and the place-names presented here are just one more aspect of this evidence, one more color in the mosaic57, albeit a very important one.

Figure 4. The four different categories of -briga names in the Iberian Peninsula.
Figure 4. The four different categories of -briga names in the Iberian Peninsula.

Endnotes

1 See for instance Villar (2000: 441-2) for a very clear and thoughtful reflection on what Untermann's line actually implies: "... 'la Hispania de los -briga' podría ser equiparada con 'la Hispania indoeuropea' tan sólo en el caso de que estuviera probado que el pueblo de los -briga fue el único componente indoeuropeo en su suelo antes de la llegada de los romanos (…). En rigor ni siquiera sirve para deslindar lo celta de lo no celta. Podría haber en teoría territorios que hubieran ocupado los celtas sin que hubieran llamado -briga a ninguno de sus asentamientos, bien porque hubiera habido ya abundantes ciudades con nombres preceltas, bien porque los nuevos asentamientos no tuvieran las características de elevación y fortificación que tienen los castros."

2 Apart from the discouraging fact that the analysis of certain toponymic types as non-Indo-European itself has been recently challenged (Villar 2000).

3 See García Alonso (2001a: 214-17) for a more detailed introduction to the languages expected in the different areas.

4 See for instance Beltrán et al. (1996).

5 See M. Palomar Lapesa (1957), J. Untermann (1965), M.L. Albertos Firmat (1966) and J. Gorrochategui (1984), as well as D. Ellis Evans (1967), K.H. Schmidt (1957) or J.M. Abascal (1994), to cite just a few examples.

6 For details, see García Alonso (1995a: 25-28 or 2003: 26-28).

7 Last but not least, the fact that there is a temporal dimension in toponymy is a complicating factor. Place-names form layers or strata and we have to date them, if we can. We are not always able to. Using the same example as before, would it be correct to assume, based solely on the place-names, that southern California in the year 2006 was predominantly Spanish speaking?

8 1992, 1994a, 1994b, 1995b, 2000, 2001a, 2001b and particularly my global study La Geografía de Claudio Ptolomeo y la Península Ibérica, submitted as a Doctoral Dissertation at the University of Salamanca in 1993: J. Luis García Alonso (1995a), updated in my 2003 publication. The Greek geographer and astronomer Claudius Ptolemy worked in Alexandria between 127 and 141 AD, with the main objective of gathering all possible information and organizing it in an exhaustive synthesis that would become an essential tool for future scholars. He wrote, among other things, a world Geography, in eight volumes. He gathered all available information and ordered it in a systematic table of coordinates of the main towns and physical features. Ptolemy deals with Hispania in Book 2 of his work. He divides the Peninsula, in agreement with the Roman administrative division of his time, into three provinces: Baetica, Lusitania and Tarraconensis. One of the aims of the first comprehensive study was to provide a survey of the scholarship on the identification of every place mentioned (all 500 of them). Nevertheless, the publication of several volumes of the Spanish section of the TIR made suddenly obsolete a few of my suggested identifications. I have tried to solve this and other minor effects of time in my new updated version (2003). In the first part of my study I followed the text of the Geography (after carrying out a collation of the manuscripts that confirmed the general unreliability of Müller's standard edition). I commented on every single name (grouped in paragraphs after Ptolemy's own division in ethnic entities). Whenever possible, I tried to offer a linguistic affiliation of the name. Secondly, I presented the historic and linguistic data we have on each ethnic group, paying special attention to the information concerning the languages that may occur in their territory. Finally, together with a map of the territory of each group, I presented a tentative classification of the place-names, as well as some conclusions relative to the linguistic map of the area, referring to the general linguistic map whenever relevant.

9 Prior to this, several authors had dealt with these names, including Tovar (1949), Rix (1954: 99-107) and Hubschmid (1960).

10 He returned to this subject as recently as 2001 (Untermann 2001: 192-97). See also the series of maps in this article (2001: 213-8) and, particularly the map of the -briga names (2001: 214).

11 Also 1983 and 1995.

12 He also dealt more briefly with these names in 1996 and 2000.

13 Faust (1976: 183-4), Koch (1979: 406-8), and, more recently, Abascal (2002), de Bernardo Stempel (2002) and Prósper (2002).

14 According to Villar (1995: 164) the whole paradigm should be more or less like this:
N: briks > bris
Ab: brigez
G: *brigos
D: *brigei
Ac: *brigam¿-*em?
NP: *briges

15 She works with the idea that there might be several more names with this format (-brig-) that could be hidden behind phonetically modified forms (particularly the loss of the final voiced stop). As usual, she explains clearly the phonetic developments she is arguing for. She may be perfectly correct in many cases, although in other instances we cannot be certain, as she would no doubt agree. I suppose she accepts this format as Celtic, but, as far as I know, she does not offer an exact attribution for those phonetic developments mentioned. Should they be attributed to Celtic speakers? If not, are we to imagine that a group of Celtic names have been phonetically modified by speakers of a different language, such as Lusitanian? This would have consequences for how we imagine the succession of languages in the different areas of Western Hispania. A very interesting, if difficult, path.

16 De Hoz (1995: 606).

17 Welsh and Breton show bre, from briga, which apart from being the Gaulish form, is well attested in Hispania.

18 Villar (1995: 185).

19 Villar (1995: 186). He continues: "Con esa modalidad los romanos nombraban ciudades de cualquier región de Hispania, independientemente de cómo haya sido en cada momento la flexión nativa."

20 De Bernardo Stempel (2002: 107, n. 103).

21 Not absolutely impossible, however.

22 See Prósper (2002: 399).

23 However, there are some indications that Lusitanian could have a different treatment: ifadem from *eibh-. See J. L. García Alonso (2003: Introduction).

24 It is not a suffix but the second element of a nominal compound that is qualified by the initial element. This is the typical structure, although there are other cases: -briga can also be a base modified by one or more suffixes (Brigaecium).

25 However, even in cases like this, one could contemplate the possibility of the name being transported elsewhere by speakers of a different language (cf. the place-name Segovia in America, taken there not by speakers of Celtic, but by their descendants, speakers of Spanish).

26 An interesting approach is offered in Abascal's (2002) recent article in which he suggests a special relationship of -briga names with anthroponyms, not only with these very clear Latin examples, but also with native names (Coeliobriga). However, it is hard to be certain, for many of the examples he offers, that the origin of the place-name is specifically a personal name and that it cannot be anything else (e.g. another place-name, an adjective). In some cases both the personal and the local name may certainly share the same lexeme, but it is impossible to know which of the two came first; it is likely even that their origin is independent, i.e., they both originate from the same linguistic stock (the lexeme) available to the speaker that coined the name. I have the feeling that for many of the examples an alternative explanation is simpler and more likely (Adrobrica, Cottaeobriga, Dessobriga, Eberobriga, Langobriga, Longobriga, Nertobriga, Segobriga, Talabriga, etc.).

27 Pedrero (1996).

28 Sometimes perhaps the whole could be non-Celtic, but this is extremely difficult to detect and clearly less likely.

29 Ptolemy II, 5, 7. See L. García Alonso (2003: 119).

30 Albertos Firmat (1990) saw the more than likely relationship with the Aurelii Cottae.

31 A gap that has been discussed eloquently by de Bernardo Stempel (2000: 95).

32 Ptolemy II, 5, 7. See García Alonso (2003: 127).

33 Ptolemy II, 6, 32. See García Alonso (2003: 229).

34 Ptolemy II, 6, 49. See García Alonso (2003: 260).

35 Also It. Ant. 395; It. Ant. 449, 454; Rav. 318, 15; Plut. Sert. 13, FHA IV 173; Pliny III 3 26; It. Ant. 421; Rav. 307.

36 Ptolemy II, 5, 6. See García Alonso (2003: 109).

37 This is not an exhaustive list.

38 References in the TIR (K-29, K-30, K/J-31, J-29 and J-30).

39 For this group of names see Prósper (2002: 380-2).

40 Ptolemy II, 6, 49. Pliny (IV, 118) and inscriptions (CIL, II, 760 and 967). See García Alonso (2003: 273).

41 See as well It. Ant. 434, 5. See Holder ACS II, 1364.

42 Ptolemy II, 5, 6. See L. García Alonso (2003: 111). Appian (Iber. 75), It. Ant. (421, 6), Pliny (IV, 118), = Terebrica (Rav. 307, 2)?

43 References in the TIR (K-29, K-30, KJ-31, J-29 and J-30).

44 Ptol. II, 6, 31. Cited also in the 'vía de la Plata' (It. Ant.439, 8 and 440, 2 and Rav. IV, 45), and by Florus (II, 33, 55). See García Alonso (2003: 220).

45 Βαργιακίς is also the name of a city of the Vaccaei. Traditionally it has been considered to be the same place that Ptolemy (II, 6, 29) calls Brigaecium in the neighboring lands of the Astures. But if they were two different places, Bargiacis might have been today's Valderas (León), and Brigaecium Benavente. *Bhrgh- /*Bhorgh- + iac-is? > Bargiacis. *Bhrgh-aic-ion > Brigaecium (' Fortress Town'). If Bargiacis is a real form, there are some parallels in Holder, like Bargiacus (ancient name of Châteauneuf-Val-de-Bargis, dep. Nièvres), Bargos (an Irish river-name as well as a river name in Illyria) and Bargusii (Spanish ethnic name, from around Berga, in the province of Barcelona) (Holder 1896-1907). Could we postulate a development *bhrgh- > Barg-? That would not be Celtic. Celtic develops that to briga. But if we could correct it to *Bergiakís, there is a Celtic possibility. If the right form of the place-name is, after all, Brig-aec-ium, the celticity is clearer.

46 With an e-grade we also have Βέργιδου Φλαούιου, a city of the Astures, also mentioned in the It. Ant. (425, 4; 429, 2 and 431, 1), Rav. (320, 10) and an inscription (CIL II 4248; see Holder [1896-1907, I: 403]) mentioning a Bergidof(lauiensis), Villafranca del Bierzo or Cacabelos, in the Bierzo area, or *bhergh-i-d-on (Flavium) 'the town on the mountains dedicated to Flavius'. The name may survive in the modern Leonese area called 'Bierzo' (Schulten 1922ff, V: 195 and Tovar 1989: 324) (with some minor phonetic problems, though), which has the same Indo-European root (*bhergh) -origin of the Germanic berg or burg cognates (Gothic baurgs-) we see in the toponymic Celtic element -briga (<*bhrgh-a). But what is genuinely Celtic there is the vocalic phonetic result (-ri- < -r-) from -briga (< *bhrgh-a). So, with Bergidum (<*bhergh-), we cannot be sure that this place-name should be placed in a Celtic layer. We cannot discount the possibility that one or several pre-Celtic western Indo-European language(s) of Hispania used this toponymic element as well. The phonetics fit well with the Celtic hypothesis, although this is not the only possibility. However, if it is finally proven that Lusitanian treated the Indo-European voiced aspirates as voiceless fricatives, we could eliminate a candidate for Bergidum. But we still have the Old European language(s), and possibly other even more different groups. Nevertheless, if we eliminate Lusitanian for phonetic reasons (this being uncertain), I believe the Celtic possibility is more attractive than the Old European one. I prefer to place our Bergidum in the Celtic layer.

47 References in the TIR (K-29, K-30, KJ-31, J-29 and J-30).

48 Ptolemy II, 5, 7. Pliny (IV, 118), It. Ant. (438, 6) and Rav. (312, 12). See García Alonso (2003: 121).

49 References in the TIR (K-29, K-30, KJ-31, J-29 and J-30).

50 I feel it is much harder to imagine Lusitanians or Tartessians consciously doing something like this.

51 Theoretically speaking, for instance, an exception could be a name transported elsewhere by speakers of a different language ( Segovia in America), but this is an exception to the rule.

52 For a linguist at least this is a linguistic category and it has to be proven in linguistic terms.

53 See the excellent work by de Bernardo Stempel (2002) on this subject. The difficulties that some archaeologists feel concerning this are a problem that has to be dealt with, of course, but a linguist cannot close this door on himself just because it might look simpler. Who has the right to maintain without further proof that the linguistic mapping of pre-Roman Hispania was simple?

54 As early as 1906, H. d'Arbois de Jubainville had the following to say (cited by Delamarre (2003: 87): "…les Gaulois conquérants n'avaient pas complètement soumis une partie considerable de la population vaincue et, pour tenir tête à cette population toujours plus ou moins rebelle, il fallait multiplier les forteresses au milieu d'elle." See also Martín Sevilla (1980: 79, with bibliography): "-briga (...) debió aludir a lugares prominentes donde se establecían asentamientos por necesidades de defensa." An attractive theory is that this is the reason for the higher frequency of the -briga type of place-name in Hispania in comparison with other regions of the Celtic world, such as Gaul or Britain, where it is much more common to find place names with the element -dunum, perhaps implying a different type of settlement -- possibly not so heavily fortified? This would suggest a milieu where the Celts were able to live in greater security, regions where they did not feel as threatened as in (mainly western) Hispania. It is true, though, that the word dunum "désignait le fort, la citadelle circulaire, l'enceinte fortifiée et fermée, en général juchée sur une colline ou une hauteur, correspondant à l'oppidum latin" (Delamarre, 2003: 154). Both words, of different etymological origin, ultimately have almost exactly the same meaning (the Irish hill forts are called dún, whereas their Spanish counterparts are -briga). It would seem then that we are dealing with a dialectal question. Gaulish or insular Celtic preferentially make use of -dunum (although -briga place-names were also known, as can be seen in ethnic names such as Brigantes, etc), whereas Hispano-Celtic speakers almost always make use of -briga. Since both terms are almost mutually exclusive (to such an extent that almost all -dunum names in Spain are considered Gaulish and are found, significantly, in the Pyrenees), the possibility of this having to do with a dialectal split in vocabulary is very appealing.

55 This is the opinion still preferred by Untermann in 2001: "Hay que tomar briga como síntoma de una unidad lingüística que abarca y reúne todas las zonas de la Hispania indoeuropea" (2001: 195).

56 Lusitanian shows /ur/ and then /or/ as the phonetic development of a vocalic /r/. See Prósper (2002: 399).

57 For a very interesting map of the distribution of -briga names as well as of their internal echoes, see de Hoz (1995: 606).


Bibliography

Abascal Palazón , J.M.
1994. Los Nombres Personales en las Inscripciones Latinas de Hispania. Murcia: Universidad de Murcia.
2002. Téseras y monedas. Iconografía zoomorfa y formas jurídicas de la Celtiberia. Palaeohispania 2: 9-35.

ACTAS de los coloquios sobre lenguas y culturas prerromanas de la Península Ibérica (o lenguas y culturas paleohispánicas, o lenguas y culturas hispanas prerromanas) II, Tübingen 1976.
1979. Salamanca.
ACTAS de los coloquios sobre lenguas y culturas prerromanas de la Península Ibérica (o lenguas y culturas paleohispánicas, o lenguas y culturas hispanas prerromanas) V, Colonia 1989.
1993. Salamanca.
ACTAS de los coloquios sobre lenguas y culturas prerromanas de la Península Ibérica (o lenguas y culturas paleohispánicas, o lenguas y culturas hispanas prerromanas) VIII, Salamanca 1999.
2001. Salamanca.

Albertos Firmat, M.L.
1966. La Onomástica Personal Primitiva de Hispania Tarraconense y Bética. Salamanca: Consejo Superior de Investigaciones Científicas.
1990. Los topónimos en -briga en Hispania. Veleia 7: 131-46.

Appian
1939 (repr., 1st ed. corrected, 1962) Appiani Historia Romana, 2 vols. Iberica Vol. I, 62-140. P. Viereck, A. G. Roos and E. Gabba (eds), Leipzig: B.G. Teubner; 1905 (repr. 1986) Vol. II Leipzig: B.G. Teubner.

Beltrán, F., J. De Hoz and J. Untermann
1996. El Tercer Bronce de Botorrita (Contrebia Belaisca). Zaragoza: Gobierno de Aragón, Departamento de Educación y Cultura.

Bosch-Gimpera, P.
1932. Etnología de la Península Ibérica. Barcelona: Editorial Alpha.

CIL Corpus Inscriptionum Latinarum.
1862. Berlin: Deutsche Akademie der Wissenschaften.
CIL Corpus Inscriptionum Latinarum II 2/5: CIL II ed., Part V.
1998. Berlin and New York: Conventus Astigitanus.
CIL Corpus Inscriptionum Latinarum II 2/7: CIL II ed., Part VII.
1995. Berlin and New York: Conventus Cordubensis.
CIL Corpus Inscriptionum Latinarum II 2/14.1: CIL II ed., Part XIV, Fascicle I. Pars meridionalis conventus
1995. Berlin and New York: Tarraconensis.

De Bernardo Stempel, P.
2000. Ptolemy's Celtic Italy and Ireland: a linguistic analysis. In D.N. Parsons and P. Sims-Williams (eds), Ptolemy: Towards a Linguistic Atlas of the Earliest Place-names of Europe, University of Wales, April 1999, pp. 83-112. Aberystwyth: CMCS Publications.
2002. Centro y áreas laterales: formación del celtibérico sobre el fondo del celta peninsular hispano. Palaeohispanica 2: 89-132.

Delamarre, X.
2003. Dictionnaire de la langue gauloise. Paris: Éditions Errance.

Evans, D. Ellis
1967. Gaulish Personal Names: A Study of Some Continental Formations. Oxford: Clarendon.

Faust, M.
1976. Cuestiones generales de toponimia prerromana. ACTAS I, pp. 165-79. Salamanca.

Florus
1967. Florus. Oeuvres, 2. vols. P. Jal (ed.). Paris: Les Belles Lettres.

Florus
1886. L. Annaei Flori Epitome Libri II and P. Annii Flori Fragmentum de Vergilio Oratore an poeta. O. Rossbach (ed.). Leipzig: B.G. Teubner.

FHA: Fontes Hispaniae Antiquae
1925-1987. A. Schulten, P. Bosch Gimpera and L. Pericot (eds). 9 vols. Barcelona: A. Bosch.

García Alonso, J.L.
1992. On the celticity of some Hispanic place-names. Proceedings of the 9th International Congress of Celtic Studies, Paris, July 1991. Études Celtiques 29: 191-201.
1994a. Sobre algunos topónimos hispánicos en Claudio Ptolomeo: los nombres autrigones. In F.R. Adrados (ed.), Actas del VIII Congreso Español de Estudios Clásicos, Madrid, September 1991, Vol. I, pp. 503-510. Madrid: Ediciones Clásicas.
1994b. La toponimia de los antiguos pelendones en la Geografía de Claudio Ptolomeo. In A.A. Tejedor and H. Perdiguero (eds), Actas de la Reunión Científica sobre Toponimia de Castilla y León, Burgos, November 1992, pp. 117-25. Burgos.
1995a. La Geografía de Claudio Ptolomeo y la Península Ibérica, Ph.D. thesis University of Salamanca. (Microfiche).
1995b. La contribución de la toponimia prerromana de la P. Ibérica al conocimiento de la lenguas paleohispánicas. Los nombres de los cántabros. In A. Agud, J. A. Fdez. Delgado and A. Ramos (eds), Las Lenguas de Corpus y sus Problemas Lingüísticos, pp. 55-67, Madrid and Salamanca: Ediciones Clásicas.
2000. On the celticity of the Duero Plateau: place-names in Ptolemy. In D.N. Parsons and P. Sims-Williams (eds), Ptolemy: Towards a Linguistic Atlas of the Earliest Place-Names of Europe. University of Wales, April 1999, pp. 29-54. Aberystwyth: CMCS Publications.
2001a. The place-names of ancient Hispania and its linguistic layers. Studia Celtica 35: 213-244.
2001b. Lenguas prerromanas en el territorio de los vetones a partir de la toponimia. ACTAS VIII, pp. 389-406.
2003. La Península Ibérica en la Geografía de Claudio Ptolomeo. Salamanca and Vitorio: Universidad del País Vasco, Servicio Editorial.

Gorrochategui, J.
1984. Estudio sobre la Onomástica Indígena de Aquitania. Bilbao and Salamanca: Servicio Editorial Universidad del País Vasco y la colaboración de Universidad de Salamanca.

Holder, A.
1896-1907 (1961-62). Alt-celtischer Sprachschatz, 3 vols. Leipzig: B.G. Teubner.

Hoz, J. De
1963. Hidronimia antigua europea en la Península Ibérica. Emerita 31: 227-242.
1983. Algunas consideraciones sobre los orígenes del alfabeto griego. In J. A. Fernández Delgado (ed.), Estudios Metodológicos sobre la Lengua Griega. Cáceres: Universidad de Extremadura.
1994. Castellum Aviliobris. Los celtas del extremo occidente continental. In K.H. Schmidt, R. Bielmeier, R. Stempel and R. Landsweert (eds), Indogermanica et Caucasica: Festschrift für Karl Horst Schmidt, pp. 348-62. Berlin and New York: W. De Gruyter.
1995. Tartesio, fenicio y céltico, 25 años después. Tartessos 25 años después: Actas del Congreso Conmemorativo del V Symposium Internacional de Prehistoria Peninsular, pp. 591-607. Jerez de la Frontera: Universidad de Cádiz.

Hubschmid, J.

1960. Toponimia prerromana. In M. Alvar et al. (eds). Enciclopedia Lingüística Hispánica I (M. Alvar et al. eds), pp. 447-93. Madrid: Consejo Superior de Investigaciones Científicas .

IEW = See Pokorny.

It. Ant.: Itinerarium Antonini.
Itineraria Romana. Römische Reisewege an Hand der Tabula Peutingeriana dargestellt. K. Müller (ed.). 1964 edition. Rome: Bretschneider.

Koch, M.
1979. Die Keltiberer und ihr historischer Kontext. Actas II, pp. 387-419. Salamanca.

Marstrander, C.J.S. (ed.)
1913-1976. Dictionary of the Irish Language Based Mainly on Old and Middle Irish Materials. Dublin: Royal Irish Academy.

Menéndez Pidal, R.
1968. Toponimia Prerrománica Hispana. Madrid: Gredos.

Palomar Lapesa, M.
1957. La Onomástica Personal Pre-Latina de la Antigua Lusitania: Estudio Lingüistico. Salamanca: Consejo Superior de Investigaciones Científicas.

Pedrero Sancho, R. M.
1996. El hidrónimo prerromano Mira. Emerita 64: 361-74.

Pliny
1947-61. Pline l'Ancien. Histoire Naturelle, 15 vols. A. Ernout et al. (eds). Paris: Les Belles Lettres.

Plutarch
1964-1971. Plutarchi vitae paralelae. K. Ziegler (ed.). 1969 Vol. 1 (4th edition). 1964 Vol. 2 (2nd edition). 1971 Vol. 3 (2nd edition). Leipzig: B.G. Teubner.

Plutarch
1914-1926 (1967 ed.) Plutarch's Lives. B. Perrin (ed.). Vols. 1, 4, 5 and 11. 1914, 16, 17 and 26 (repr. 1967, 68 and 62). Cambridge: Loeb Classical Library.

Pokorn y, J.
1959-69. Indogermanisches etymologisches Wörterbuch. 2 vols. Bern and Munich: IEW.

Prósper, B.
2002. Len guas y Religiones Prerromanas del Occidente de la Península Ibérica . Salamanca: Ediciones Universidad de Salamanca.

Ptolemy
1883-1901 (1962). Claudii Ptolemaei Geographia . K. Müller (ed.), 2 vols. Paris: Alfredo Firmin Didot .

Rav. (Ravenna [Anon.])
1860 (1962). Ravennatis Anonymi Cosmographia et Guidonis Geographica. M. Pinder and G. Parthey (eds). Aalen-Meisenheim: O. Zeller.

Rivet, A.L.F. and C. Smith
1979. The Place-Names of Roman Britain. Princeton: Princeton University Press.

Rix, H.
1954. Zur Verbreitung und Chronologie einiger keltischer Ortsnamentypen. Festschrift für Peter Goessler, pp. 99-107. Stuttgart: Kohlhammer.

Schmidt, K.H.
1957. Die Komposition in gallischen Personennamen. Zeitschrift für Celtische Philologie 26: 33-301.

Schulten, A. and P. Bosch Gimpera
1922. Fontes Hispaniae Antiquae, Vol. I. Barcelona: A. Bosch and Berlin: Weidmann.

Sevilla Rodríguez, M.
1980. Toponimia de origen indoeuropeo prelatino en Asturias. Oviedo: I nstituto de Estudios Asturianos.

TIR K-29: 1991 = Tabula Imperii Romani. Hoja K-29: Porto, Madrid.
----- K-30: 1993 = Tabula Imperii Romani. Hoja K-30: Madrid, Madrid.
----- J-29: 1995 = Tabula Imperii Romani. Hoja J-29: Lisboa, Madrid.
----- K/J-31: 1997 = Tabula Imperii Romani. Hoja K/J-31: Pyrénées Orientales-Baleares, Madrid.
----- J-30: 2002 = Tabula Imperii Romani. Hoja J-30: Valencia, Madrid.

Tovar, A.
1949. Estudios sobre las primitivas lenguas hispánicas. Buenos Aires: Coni.
1989. Iberische Landeskunde. Segunda Parte. Las tribus y las ciudades de la antigua Hispania. Tomo 3: Tarraconensis. Baden-Baden: V. Koerner.

Untermann, J.
1961. Sprachräume und Sprachbewegungen im vorrömischen Hispanien. Wiesbaden: O. Harrassowitz. Spanish translation in APL 10 (1963).
1965. Elementos de un Atlas Antroponímico de la Hispania Antigua. Madrid: Bibliotheca praehistorica Hispana.
2001. La toponimia antigua como fuente de las lenguas hispano-celtas. Palaeohispanica 1: 187-218.

Villar, F.
1993. Las silbantes en Celtibérico. In J. Untermann and F. Villar (eds), Actas del V Coloquio sobre Lenguas y Culturas Prerromanas de la Península Ibérica, Colonia, 1989, pp. 773-812. Salamanca: Ediciones Universidad de Salamanca.
1995. Estudios de Celtibérico y de Toponimia Prerromana. Salamanca: Ediciones Universidad Salamanca.
1996. Los indoeuropeos y los orígenes de Europa. Lenguaje e Historia (2nd ed.). Madrid: Gredos.
2000. Indoeuropeos y no Indoeuropeos en la Hispania Prerromana: las Poblaciones y las Lenguas Prerromanas de Andalucía, Cataluña y Aragón según la Información que nos Proporciona la Toponimia. Salamanca: Ediciones Universidad de Salamanca.