Celtic languages

• 1. Introduction

  ›

  Celtic languages survive today as six living tongues clinging to the northwestern edge of Europe, spoken by communities from Wales to Brittany to the Isle of Man. Once, they ranged across much of the continent and reached as far as central Anatolia. That collapse from dominance to the margins is one of the great reversals in the history of human speech. Who first gave these languages their name? What holds them together as a family, and what nearly erased them? The answers involve Alpine inscriptions older than the Roman Republic, a Welsh scholar writing in 1707, and arguments among linguists that remain unsettled today.

• 2. Naming and Origins

  ›

  Edward Lhuyd was the scholar who first applied the label "Celtic" to this group of languages, in 1707. He was building on the work of Paul-Yves Pezron, who had drawn the explicit connection between the Celts described by ancient writers and the speakers of Welsh and Breton. That act of naming anchored a scholarly tradition that continues to the present.

  The languages descend from a hypothetical ancestor called Proto-Celtic, itself a branch of the broader Indo-European family. The question of when Proto-Celtic fractured into distinct branches is contested. Gray and Atkinson put the division of the family into its main sub-families at around 900 BC, though they acknowledge uncertainty spanning the range of 1200 to 800 BC. A paper by Forster and Toth reached a more startling conclusion, placing the break-up at roughly 3200 BC, plus or minus 1500 years, a figure that has remained controversial.

  The earliest physical evidence of written Celtic comes from Lepontic inscriptions found in the Alps, dated to the 6th century BC. Lepontic, spoken in what is now Switzerland and parts of northern and central Italy, is the oldest attested Celtic language by that measure. It belongs to the Continental branch, a group that also includes Gaulish, Celtiberian, and Gallaecian, all of which are now extinct. Lhuyd's own 1707 theory about the origins of the Q-Celtic languages, placing them in western Iberia, would resurface centuries later in debates about Lusitanian and its possible kinship with Celtic.

• 3. The Continental World

  ›

  Gaulish was spoken in a wide arc stretching from what is now Belgium all the way to central Turkey. Galatian, the variety spoken in Anatolia, may have persisted as late as the 6th century AD. The Continental Celtic languages collectively went extinct by the 7th or 8th centuries AD, leaving behind inscriptions and place names but no living descendants.

  The Hispano-Celtic languages occupied the Iberian Peninsula in two distinct zones. Celtiberian was spoken in the eastern part of Old Castile and the south of Aragon, in the area of modern provinces including Soria, Guadalajara, and Zaragoza. Gallaecian, also called Western or Northwestern Hispano-Celtic, covered what is now northern Portugal and the Spanish regions of Galicia and Asturias. Whether Celtiberian and Gallaecian were closely related or merely neighbors is still uncertain.

  Lepontic left behind coins carrying its inscriptions; examples have been found as far from the Alps as Noricum and Gallia Narbonensis. The Cisalpine variety of Gaulish was spoken in northern Italy until the 1st century BC, and scholars including Eska see it as closer to Lepontic than to Transalpine Gaulish. Beyond the main branches, a cluster of languages may have been Celtic without certainty. Tartessian, known from 95 inscriptions in the southwest of the Iberian Peninsula, has been argued by John T. Koch to belong to the Celtic family. Camunic, spoken in the Val Camonica and Valtellina valleys of the Central Alps in the first millennium BC, has more recently attracted similar proposals.

• 4. The Insular Survivors

  ›

  Irish and Scottish Gaelic are both descended from Middle Irish; Welsh and Breton share a descent from Common Brittonic. These four languages have kept substantial communities of native speakers into the present day, forming the core of the living Insular Celtic world.

  Welsh holds the strongest demographic position among them. About 538,000 people in Wales, or 17.8 percent of the population, claimed they could speak Welsh in the 2021 census. A 2022 survey counted 788,927 people in Wales, roughly 26.7 percent, as speakers. Unusually for Celtic languages, Welsh is not classified as endangered by UNESCO, and it carries official status in Wales. A community of around 5,000 Welsh speakers exists in the Chubut Province of Argentina, a trace of 19th-century migration.

  Irish holds official status across the island of Ireland and within the European Union. Daily use outside the educational system is concentrated in the Gaeltacht regions; in the Republic of Ireland, 73,803 people use the language daily outside school. Total speakers across the island and diaspora are counted in the millions in survey data, though the number of people who use Irish as a first, everyday language is far smaller. Scottish Gaelic had 57,375 native speakers recorded in Scotland in 2011, with a further 2,500 counted in Canada, mainly in Nova Scotia.

  Breton occupies a different political context: it is spoken in continental Europe, in Brittany, making it the only living Celtic language on the European mainland. Its origin is British, not Gaulish. Settlers from Britain carried it to Brittany in the post-Roman period, which is why Breton sits within the Brittonic branch alongside Welsh and Cornish rather than descending from the Gaulish that once filled the same territory.

• 5. Revival From the Brink

  ›

  Cornish died out with its presumed last native speaker in 1777. Manx reached the same point in 1974. Both languages were gone in all but memory when revitalisation movements gained force in the 2000s. Adults and children took up both languages deliberately, creating a new generation of native speakers where none had existed for decades.

  Cornish now has 563 native speakers recorded by SIL Ethnologue and a skilled-speaker community estimated at between 2,500 and 5,000. The language is overseen by the Akademi Kernewek and the Cornish Language Partnership. Manx has more than 100 native speakers, including a small number of children who have grown up with the language from birth. The 2021 census on the Isle of Man recorded 2,023 people who could speak Manx. Both languages are still classified as minority languages, but their existence as living speech rather than archive materials is a recent and documented fact.

  The Welsh Language Commissioner and the Bwrdd yr Iaith Gymraeg before it represent the formal institutional backbone that has supported Welsh through decades of pressure. Irish is regulated by Foras na Gaeilge. Scottish Gaelic is overseen by Bòrd na Gàidhlig. Breton has the Ofis Publik ar Brezhoneg. The existence of named regulatory bodies for all six languages reflects a deliberate political choice to treat each one as worth preserving, a development without precedent in earlier centuries.

• 6. Structure and Sound

  ›

  Consonant mutations are among the most striking features shared across the Insular Celtic languages. A word changes its initial consonant depending on its grammatical context, so the same root can begin with different sounds in different sentences. This feature does not appear in Continental Celtic, making it one of the distinguishing marks of the Insular branch.

  All the Insular Celtic languages place the verb first in the sentence, giving them a verb-subject-object word order. This contrasts with English and French, which put the subject first. An Irish example from the source illustrates it directly: Ná bac le mac an bhacaigh is ní bhacfaidh mac an bhacaigh leat, where the second half of the sentence opens with the verb. The counting systems in some Celtic languages use a base of twenty rather than ten. Welsh expresses the number 99 through a structure meaning "four twenties and nineteen"; the Cornish example of 56 breaks down as "sixteen and two twenty."

  Celtic languages also lack a simple verb meaning "to have." Instead, possession is expressed through a composite structure using the verb "to be" together with a preposition. Cornish says "there is a cat to me," Welsh says "a cat is with me," and Irish says "there is a cat at me." The inflected preposition, where the preposition itself changes form to incorporate a pronoun, is another feature restricted to Insular Celtic. These shared innovations across Irish, Scottish Gaelic, Manx, Welsh, Cornish, and Breton are part of what makes the Insular Celtic hypothesis, now the more widely accepted scholarly view, compelling to its proponents.

• 7. The Classification Debate

  ›

  Scholars have disagreed for decades about how to arrange Celtic languages on a family tree. Two rival hypotheses have dominated the argument. The Insular Celtic hypothesis, championed by scholars including Cowgill in 1975 and McCone in 1991 and 1992, holds that the primary division is between Continental Celtic and the languages spoken in the British Isles and Brittany. The P/Q-Celtic hypothesis draws the main line between languages that replaced an initial Q sound with P and those that kept Q.

  The P/Q distinction shows up in core vocabulary: Goidelic and Hispano-Celtic languages are Q-Celtic, while most Gallic and Brittonic languages are P-Celtic. Ranko Matasović wrote in the introduction to his 2009 Etymological Dictionary of Proto-Celtic that Celtiberian is almost certainly an independent branch that separated from the others very early. That position, if correct, complicates both major hypotheses.

  A 1983 inscription on a piece of lead found at Larzac brought new energy to the P/Q debate in the mid-1980s. Analysis of the inscription revealed a shared phonetic innovation between Gaelic and Gaulish, expressed as the change from -nm- to -nu, visible in Gaelic ainm and Gaulish anuana, both meaning "names." If a third such shared innovation were identified, it could support the existence of a Gallo-Brittonic dialect group. As of 2008, Stifter described the Gallo-Brittonic view as "out of favour" and the Insular Celtic hypothesis as "widely accepted," though the interpretation of the Larzac evidence remains part of an ongoing conversation among specialists. The hypothesis proposed by Calvert Watkins in 1966 that Celtic and Italic might form a joint subfamily within Indo-European fell somewhat out of favour after his reexamination, though scholars including Ringe, Warnow, and Taylor have argued for an Italo-Celtic grouping in 21st-century work.