Ancient writing systems in the Mediterranean

A critical guide to electronic resources


edited by: Coline Ruiz Darasse (translation revised by Melanie Rockenhaus)

  • Introduction
  • Index
  • Ancient Writing Systems
  • Further information

There were three ancient Continental Celtic languages :

  1. Lepontic, which was spoken in Northern Italia between the 6th and the 4th c. BC.
  2. Gaulish, a Celtic language spoken in Gaul, first found ca. 4th c. BC. and known until the 5th c. AD. There were most probably different dialects, but at the moment their specific characteristics are unknown.
  3. Celtiberian, a language spoken in Spain by Celtiberians whose presence is attested as early as 4th c. BC but whose language is epigraphically attested only from the 2nd c. BC. It was spoken and written continuously but only in some areas of the Meseta norte area of Spain.

These three Indo-European languages derive from Common Celtic, with evolutions particularly in the treatment of the labiovelar *kw. The opposition P-Celtic /Q-Celtic is not the most important way to distinguish these languages: according to Schmidt, the treatment of *ṃ / *ṇ must also be considered. Celtiberian treats it with -a : am/an.

This scheme (sometimes criticized because of its simplification) has been proposed by Michel Lejeune, in order to differentiate the languages attested :

Gaulish :         p + an

Celtiberian :       kan

Brittonic :        p + en

Irish :                kw + en



Lepontic is a Continental Celtic language, spoken between 6th and 4th c. BC. Inscriptions in this language have been found near Lugano, between Lake Como and Lake Maggiore in Northern Italy.



Lepontic has only two typical characteristics: the weakening of n before the stop and the assimilation -nd- > -nn. 

Phonetically Lepontic does not distinguish between voiced and voiceless stops as Celtiberian does.

The Ornavasso vase shows that the language in which it is engraved treats the labiovelar *kw as /p/. This would make it similar to Gaulish.


Vocalic characteristics are those known elsewhere for Celtic languages:

•   passage from *ē > ī

•   tendency to open from -ĭ to -ě

•   in the final syllable, passage from *ō to *ū.


According to Michel Lejeune, there is no flexional characteristic in Lepontic which is not present in Gaulish, in Celtiberian, or (by reconstruction) in Gaelic (Lepontica, 1971, p. 113).

It is probable that from the 4th c. BC the Gaulish occupation of Northern Italy strengthened the relationship between Lepontic and Gaulish even more.

For a complete review of these issues, see Michel Lejeune, Lepontica, 1971, p. 111 to 123; the author reviews the linguistic data of the Lepontic epigraphic inventory until 1970. 

Because of the small epigraphic corpus, it is difficult to fully understand the linguistic characteristics of the Lepontic language, and specialists still doubt its affiliation to Celtic languages.


Gaulish was a Continental Celtic language belonging to the P-Celtic family. Labiovelar *kw , as in the other Brittonic languages, evolved to p.

Another analogy with Brittonic was the same treatment of the -nm- > -nu.



The loss of aspiration in aspirated stops (which are then assimilated to simple stops) was inherited from Common Celtic (an Indo-European evolution).

The stops are thus as follows:

voiced: /b, d, g /

voiceless: /p, t, k/

two nasals: /m, n/

two liquids: /l, r/

one sibilant: /s/

one affricated: /ts/ written with théta in Gallo-Greek script and with đ in Gallo-Latin script

one aspirate: /x/, written χ in Gallo-Greek script and X in Gallo-Latin script.


•    Passage from *ē > ī and from *ei > ē , which is a Celtic characteristic. *rēx in the Gallo-Greek alphabet is written sometimes ρειξ sometimes ριγς.

•    In the final syllable, passage from *ō to ū. There were many orthographic variants, in particular in Gallo-Greek writing, for which the reader is referred to the grammatical and orthographical appendices in the RIG I.

There are also two semi-vowels: /w, y/.


Laryngeals in intervocalic position were treated as ă : -gnātus (Cintugnatus, "born from", > *genHe).


Gaulish, as other Indo-European languages, is a flexional language. Nevertheless, at this time it is difficult to tell how many cases there were in Gaulish.

Documentation is fragmentary. Most data is anthroponymic, and this does not allow the establishment of a full declension table.

The same is true for verbal morphology and syntax.

For a complete recapitulation of these issues, see P.-Y. Lambert (chap. IV: Morphologie and chap. V: Notes de syntaxe).


Celtiberian is a Q-Celtic language with archaic characteristics. As in Gaelic, the labiovelar *kw  has not followed its development into p as it has in the other Celtic languages: *kwetṷor > Welsh  petguar ; Cornish peswar ; Breton pevar and Gaulish petorritum (“four-wheel chariot”).

For the historiography and bibliography of Celtiberian Grammar, see Villar, 1995, chapter V, p. 84-107.


Celtiberian phonetics were very conservative and rather close to what is known about Common Celtic.


From Common Celtic (an Indo-European evolution), it inherited the loss of aspiration of aspirated stops (which were then assimilated to simple stops) and the loss of the voiceless labial (p). Stops were therefore very few:

t k kw

b d g (gw)

•    loss of aspiration of aspirated and voiced stops and fusion with voiced stops

•    *gw > b : labialisation of voiced labiovelar only (strictly Celtic characteristic)

•    p > *f > ø : initial and intervocalic. *pro > ro ex. robiseti (BBI, A8).


•    passage from *ē > ī and from *ei > ē , which was a Celtic characteristic. Ex. Bronze plaque from Luzaga: teiuoreikis *dẹ̄ṷo-rīks (here in the final syllable)

•    in the final syllable, passage from *ō to ū. About this mark, Kim McKone proposes the hypothesis of a change from*ō to *ā in the second-to-last syllable, even if this mark is not attested

•    passage from *ā to *ō in non-stressed syllables can be illustrated in  stoteroi (BBIV, B7).


Laryngeals in intervocalic position were featured in ă tuateros (g. sg. BB III, III-24) and tuateres (BB III, II-40) < *dhugh2ter cf. θυγάτηρ.


•    ei was written for ẹ̄ / ī coming from an old diphthong *ei , as well as for ī > *ē. ex: ueizos (K.0.11) *ṷeid- (in initial position); sekeiza (A.8).

•    *oṷ was stable in Celtiberian (boustom, BBI, A4) and also *ai and *oi : cf. belaiskom (A.80) and tokoitos (BBI, A1, A10); *eṷ did not exist.

*ṛ and *ḷ before stops became ri and li. It may have been a Celtic specificity.

In other cases, * and * became ar and al just as *ṃ / *ṇ : am/an.

F. Villar has investigated sibilants extensively. Some points mentioned are:

•    *s original: stayed in the initial position, before stops and in the absolute final position. In these cases, it was written M, transcribed by s.

•    *s > z in intervocalic position. In this case, writing was s and transcription is z.


A study of the whole morphology and syntax of Celtiberian can be found in the two volumes by Carlos Jordán Colera.

Further information

  1. Bibliography
  2. Online resources