Celtic Languages - Classifications

Classifications

Proto-Celtic divided into four sub-families:

• Gaulish and its close relatives Galatian, Lepontic, and Noric. Lepontic, the oldest attested Celtic language (from the 6th century BC), is treated as a primary branch by some researchers, including Schumacher, perhaps even the first language to diverge from Proto-Celtic. These languages were once spoken in a wide arc from France to Turkey and from Belgium to northern Italy. They are now all extinct.
• Hispano-Celtic; also extinct:
  • Celtiberian, anciently spoken in the Iberian peninsula, in parts of modern Aragón, Old Castile, and New Castile in Spain.
  • Gallaecian, anciently spoken in north-western Iberia (north-west Spain and northern Portugal).

• • Tartessian, attested from inscriptions found in south-western Iberia (southern Portugal and south-west Spain) dated to the 7th-5th centuries BC. Prior to 2011, Tartessian was treated as unclassified, with no obvious external relationship. It has since been tentatively classified as Celtic by American linguist John T. Koch, which would make Tartessian the oldest attested Celtic language.

• • Possibly Lusitanian, from central interior Portugal and western Spain; while its precise classification is unclear, the Indo-European affinity of Lusitanian is not in doubt.
• Brythonic, including the living languages Breton, Cornish, and Welsh, and the extinct languages Cumbric and Pictish though Pictish may be a sister language rather than a daughter of British (Common Brythonic). Before the arrival of Scotti on the Isle of Man in the 9th century, there may have been a Brythonic language in the Isle of Man.
• Goidelic, including the living languages Irish, Manx, and Scottish Gaelic.

Scholarly handling of the Celtic languages has been rather argumentative owing to lack of much primary source data. Some scholars (such as Cowgill 1975; McCone 1991, 1992; and Schrijver 1995) distinguish Continental Celtic and Insular Celtic, arguing that the differences between the Goidelic and Brythonic languages arose after these split off from the Continental Celtic languages. Other scholars (such as Schmidt 1988) distinguish between P-Celtic and Q-Celtic, putting most of the Gaulish and Brythonic languages in the former group and the Goidelic and Celtiberian languages in the latter. The P-Celtic languages (also called Gallo-Brittonic) are sometimes seen (for example by Koch 1992) as a central innovating area as opposed to the more conservative peripheral Q-Celtic languages.

The Breton language is Brythonic, not Gaulish, though there may be some input from the latter. When the Anglo-Saxons moved into Great Britain, several waves of the native Britons crossed the English Channel and landed in Brittany. They brought with them their Brythonic language, which evolved into Breton – still partially intelligible by modern Welsh and Cornish speakers.

In the P/Q classification schema, the first language to split off from Proto-Celtic was Gaelic. It has characteristics that some scholars see as archaic, but others see as also being in the Brythonic languages (see Schmidt). In the Insular/Continental classification schema, the split of the former into Gaelic and Brythonic is seen as being late.

The distinction of Celtic into these four sub-families most likely occurred about 900 BC according to Gray and Atkinson but, because of estimation uncertainty, it could be any time between 1200 and 800 BC. However, they only considered Gaelic and Brythonic. The controversial paper by Forster and Toth included Gaulish and put the break-up much earlier at 3200 BC ± 1500 years. They support the Insular Celtic hypothesis. The early Celts were commonly associated with the archaeological Urnfield culture, the Hallstatt culture, and the La Tène culture, though the earlier assumption of association between language and culture is now considered to be less strong.

There are two main competing schemata of categorization. The older schema, argued for by Schmidt (1988) among others, links Gaulish with Brythonic in a P-Celtic node, originally leaving just Goidelic as Q-Celtic. The difference between P and Q languages is the treatment of Proto-Celtic *kʷ, which became *p in the P-Celtic languages but *k in Goidelic. An example is the Proto-Celtic verb root *kʷrin- "to buy", which became pryn- in Welsh but cren- in Old Irish. However, a classification based on a single feature is seen as risky by its critics, particularly as the sound change occurs in other language groups (especially the Osco-Umbrian and Greek).

The other schema, defended for example by McCone (1996), links Goidelic and Brythonic together as an Insular Celtic branch, while Gaulish and Celtiberian are referred to as Continental Celtic. According to this theory, the "P-Celtic" sound change of to occurred independently or areally. The proponents of the Insular Celtic hypothesis point to other shared innovations among Insular Celtic languages, including inflected prepositions, VSO word order, and the lenition of intervocalic to, a nasalised voiced bilabial fricative (an extremely rare sound). There is, however, no assumption that the Continental Celtic languages descend from a common "Proto-Continental Celtic" ancestor. Rather, in the Insular/Continental schema, Celtiberian is usually considered to be the first branch to split from Proto-Celtic, and the remaining group would later have split into Gaulish and Insular Celtic.

There are legitimate scholarly arguments in favour of both the Insular Celtic hypothesis and the P-Celtic/Q-Celtic hypothesis. Proponents of each schema dispute the accuracy and usefulness of the other's categories. However, since the 1970s the division into Insular and Continental Celtic has become the more widely held view (Cowgill 1975; McCone 1991, 1992; Schrijver 1995), but in the middle of the 1980s, the P-Celtic/Q-Celtic hypothesis found new supporters (Lambert 1994), because of the inscription on the Larzac piece of lead (1983), the analysis of which reveals another common phonetical innovation -nm- > -nu (Gaelic ainm / Gaulish anuana, Old Welsh enuein "names"), that is less accidental than only one. The discovery of a third common innovation, would allow the specialists to come to the conclusion of a Gallo-Brythonic dialect (Schmidt 1986; Fleuriot 1986).

The interpretation of this and further evidence is still quite contested, and the main argument in favour of Insular Celtic is connected with the development of the verbal morphology and the syntax in Irish and British Celtic, which Schumacher regards as convincing, while he considers the P-Celtic/Q-Celtic division unimportant and treats Gallo-Brittonic as an outdated hypothesis.

When referring only to the modern Celtic languages, since no Continental Celtic language has living descendants, "Q-Celtic" is equivalent to "Goidelic" and "P-Celtic" is equivalent to "Brythonic".

Within the Indo-European family, the Celtic languages have sometimes been placed with the Italic languages in a common Italo-Celtic subfamily, a hypothesis that is now largely discarded, in favour of the assumption of language contact between pre-Celtic and pre-Italic communities.

How the family tree of the Celtic languages is ordered depends on which hypothesis is used: