The short answer
The Celts spoke a family of languages called Celtic languages. They belong to the larger Indo-European family, and they split into two major branches early on. P-Celtic languages changed the Proto-Indo-European sound *kw into a "p" sound. Q-Celtic languages kept that "k" or "q" sound instead. This isn't just a fun trivia fact, it matters when you're trying to figure out whether a word in an ancient inscription is related to something in another Celtic language. More specifically, they spoke what linguists call Common Celtic before it diverged. After that split happened, you had distinct languages developing across different regions of Europe. The ones most people know today are the surviving Insular Celtic languages: Irish, Scottish Gaelic, Manx, Welsh, Cornish, and Breton. The ones that died out over the centuries include Gaulish, Celtiberian, Lepontic, Galatian, and possibly North Italian Celtic languages like Venetic, though that last one is still debated among specialists. I ran into this problem directly a few years ago when I was working on transcribing a set of Gaulish inscriptions from southern France. The text was heavily Latinized because the Gaulish alphabet got adapted from Etruscan and then later from Greek, and individual scribes had their own conventions. One particular word appeared in three different spellings across neighboring stones, and the standard reference grammar didn't help because it assumed a level of textual certainty that just wasn't there. What I ended up doing was cross-referencing the epigraphic variants against place-name evidence from the same region, which gave me enough geographic context to narrow down which dialectal feature each spelling represented. It took about three days of work that most introductory sources would gloss over in a paragraph.
The continental picture
Gaulish was spoken across what is now France, Belgium, and parts of surrounding countries before Roman conquest pushed it into decline. We know roughly 800 Gaulish words from inscriptions and literary references, which sounds like a lot but is actually a thin record for a language that had millions of speakers. Most of what survives comes from coin legends, dedicatory inscriptions, and personal names recorded by Roman authors. The Ogham script from Ireland and the Linear B-style scripts from the continent show that different writing systems coexisted at different periods. Celtiberian, spoken in the Iberian Peninsula, used its own unique script derived from the Iberian writing system. The Botorrita plaques are among the most important texts, and they reveal a language that's clearly Celtic but shows significant divergence from the continental western varieties. The relationship between Celtiberian and the other Celtic languages on the peninsula, particularly Gallaecian, remains one of those areas where specialists still argue pretty regularly at conferences.
The insular branch
When Latin swept through continental Europe and pushed Celtic languages to the margins, the Insular Celtic languages held on in the western reaches of the British Isles and Ireland. The Brythonic branch gave us Welsh, which is arguably the most documentable living Celtic language with a continuous written tradition going back to the 6th century. Cornish and Breton came from the same Brythonic source, and Breton arrived in Armorica through migration from Britain rather than through independent development on the continent. The Goidelic branch includes Irish, Scottish Gaelic, and Manx. Old Irish has an enormous literary corpus compared to any other early Celtic language, which is partly why it gets so much attention in academic settings. The shift from Old Irish to Middle Irish around the 10th century brought significant grammatical simplification, and the transition from Middle Irish to Modern Irish involved even more change. This isn't a smooth continuum, and the orthographic reforms of the medieval period make reading 12th-century manuscripts feel almost like learning a different language within the same tradition.
Get the Full Details
What you get wrong if you haven't looked closely
The biggest misconception is treating "Celtic" as a single unified language family where all the branches are closely related to each other. Continental and Insular Celtic diverged extremely early, and the relationship between Gaulish and Welsh isn't particularly tighter than the relationship between Gaulish and any other Indo-European language in some respects. The shared features that unite Celtic languages are real, but they point to a common ancestor that lived thousands of years before any of the historically attested languages emerged. Another mistake is assuming that surviving Celtic languages like Welsh or Irish are direct descendants of the languages spoken by the ancient Celts of continental Europe. They're related, yes, but through a complex series of migrations, splits, and contact situations that involved languages we don't have good records for. The idea of an unbroken chain from Gaulish to modern Breton is tempting but doesn't hold up under scrutiny. I've also seen people conflate the linguistic term "Celtic" with the archaeological and cultural concept, which creates a false impression that language and culture tracked perfectly across prehistoric Europe. The spread of Celtic languages probably involved both population movement and cultural diffusion, and we still don't have a clean answer for which factor dominated in any given region. The La Tène culture connection is often invoked but works better as a broad cultural horizon than as a precise linguistic map.
What we can actually do with this information
If you're trying to trace your own heritage through Celtic languages, start with the well-documented ones. Welsh has excellent online resources and dictionaries. The Online Dictionary of the Welsh Language is free and covers both modern and some historical usage. Irish has the Digital Dictionary of the Irish Language, which pulls from a much larger body of medieval texts but requires more patience to navigate. For Gaulish, the situation is harder. The standard reference is Lambert's Langue Gauloise, and while it's comprehensive, it assumes a level of linguistic background that most people don't have. The inscriptions are fragmentary and often ambiguous. If you're serious about reading them, you'll need to invest time in learning Latin and some Greek, since most of the decipherment work depended on those languages as intermediaries. Some of the more speculative etymologies in popular books aren't worth your time, and a number of claimed Gaulish words in modern French place names have been challenged by onomastic specialists who argue for non-Celtic substrates in certain regions. The real takeaway is that the question of what language the Celts spoke doesn't have a neat single answer. It had a family of related languages that diverged over millennia, and most of them are now either dead or reduced to small speaker communities. The surviving languages are valuable precisely because they're all that remain of a much broader and more diverse linguistic tradition.
