What did British Latin sound like?

Reading time: 15 minutes

Available again in audio format! Click here to hear this article read in my slightly rough, post-cold voice, with my sincere apologies to the Welsh language.

THE TRANSITION from Roman Britain to Medieval Britain is a fascinating historical, archaeological and linguistic puzzle. The fifth, sixth and seventh centuries AD in Britain are like a black box, into which we put a well-integrated region of the Roman Empire, and out of which emerges a patchwork of new kingdoms, cultures and languages. Explaining the workings of this change on the basis of the available evidence is a challenge that continues to keep historians very busy, and keeps me up at night.

Roman Britain and its very Roman roads. Image from here.

One opinion I, as a linguist, hold is that by the end of the official Roman administration of Britain (c. 410 AD), Latin had become a common language of the population of Britain. This is to say, at least in the south of what is now England, Latin had become the majority mother tongue of the population, just as it had on the Continent. I disagree with the alternative view that the Romans brought Latin to Britain and then took it all home with them, leaving the barbarian Britons none the wiser. Elsewhere, Latin would over time produce the Romance languages (French, Spanish, Catalan, Italian and so on). However, in Britain, popular Latin was not to endure, since the incoming Angles and Saxons would upset the linguistic lie of the land.

I don’t hold this opinion because of lots of direct evidence; our sources for Latin of this era in Britain are minimal. Furthermore, texts like the Vindolanda tablets and the Bath curse tablets are amazing and insightful, but they are not unproblematic sources for the analysis of a distinctly British Latin. Writing has its own rules, and may not accurately reflect everyday language, sticking instead to an expected formal, perhaps archaic register. Besides, sources like the Vindolanda letters come from a military context, and the Roman army was a multinational and multilingual organisation, conscripting and moving soldiers around the empire. Distinctive features in this Latin may not necessarily be British-isms.

A letter from Octavius to Candidus, from Vindolanda. Image from here.

Where then does our evidence for British Latin come from? It comes indirectly, from seeing its influences on other languages and from trying to fill in the gaps in the story of language in Britain. For an article that exemplifies this, I recommend Schrijver’s The Rise and Fall of British Latin (2002). One way in which British Latin left a big impression was in the development of Brythonic. This was the main Celtic language in Britain in those ancient days, and it would become Breton, Cornish and Welsh.

I’ve written previously about the great legacy of Latin within Welsh words, but only briefly touched on what kind of Latin the Pre-Welsh language was in contact with. From its lasting silhouette, we may identify some specific features of the Latin spoken by ordinary people based in Roman Britain, since presumably it was these people who were transferring words from one language into the other. This route back to British Latin helpfully avoids the complications of written sources.

Discussing what we can learn about British Latin from the languages it had influences on, namely Welsh, is my task today. In particular, my target is British Latin immediately after the end of imperial governance, when it was perhaps at its geographical peak. This piece looks at seven features that will together provide my answer to the following question:

What can Welsh tell us about what British Latin, spoken in southern Britain c. 450 AD, sounded like?

1: Yvowel Yprothesis

Prothesis is a sound change, in which a new sound over time is added to the beginning of a word or set of words. A well-known case of this comes from the Romance languages, in which we see how a vowel was added to words in Latin that began with /s/ (as in English sun) followed by another consonant.


Since we find this phenomenon across French, Catalan and Spanish (and Italian used to have it too), we can assume that the rule has early origins in the Late Latin period, when the different Romance dialects were yet to emerge. The thing is, Welsh loanwords from Latin show it too. The first three of the words above have the Welsh counterparts ysgol, ystad and ystafell.

The Main Arts Building of Bangor’s prifysgol – university, or prīma schola in Latin. Image from here.

In fact, Welsh displays this addition of y- even for words that once started /sC/ but are not of Latin origin, such as ysgwydd ‘shoulder’. Did Latin therefore inspire the Welsh rule, as well as donating words that it affected? Granted, we cannot rule out the possibility of two separate developments of prothesis, but I think that from the evidence of Latin loanwords like ysgol, we may assume that British Latin too had this rule of vowel addition that we see in Romance.

2: H-Less

Another feature of Latin loanwords in Welsh is that they don’t seem to have come with their H’s still attached. We know from the written record that the consonant /h/ (as in English hat) was lost from Latin words at an early point. This is how hōra ‘hour’ has become Italian ora.

Yet in Welsh too, Latin hōra developed into the H-less awr. Likewise, habēna ‘rein’ and historia are behind Welsh afwyn ‘rein’ and ystyr ‘meaning’, while hospes ‘guest’ is the Latin root of ysbyty ‘hospital’. The consistent absence of H from the Latin loanwords where we might expect it leads me to think that its loss had occurred in British Latin too.

3: O’s Becoming A’s

Another apparent change is that we find the vowel /a/ in Welsh words for which the presumed Latin origin has an /o/. For example, the word achos ’cause, case’ goes back to Latin occāsiō. The archaic Welsh verb achludd ‘to hide’ derives from occlūdere, while Latin monachus ‘monk’ produced Welsh mynach but also manach in Welsh and Irish. There is evidence for this change happening too within the development of French (Jackson 1953: 82), so it looks like an innovation within the Latin of northwest Europe on both sides of the English Channel.

4: Missing Syllables and Nasals

Welsh words from Latin also regularly show the effects of syncope, through which vowels within a word are lost. Latin populus ‘people’, discipulus ‘student’ and viridis ‘green’ turn up in Welsh as pobl, disgybl and gwyrdd, having lost the short middle vowels along the way. We know for a fact that this exact syncope was a feature of popular Late Latin, since syncopated versions of words like viridis are condemned in the Appendix Probi document. The most plausible account for what we see therefore is that British Latin had this syncope too, and passed on the shortened forms to the Brythonic languages.

We also see that Latin words that once had the sequence of sounds /ns/ lost the nasal consonant /n/ over time; thus Latin nsīs ‘month’ has become French mois and Italian mese. Something similar seems to have happened in some Welsh words, since the Latin for ‘table’, nsa, appears in Welsh as mwys ‘basket’. The great Roman name nstantīnus likewise comes to be borne by figures in Welsh history and mythology called Custennin. Once again, it seems fair to conclude that British Latin’s words got denasaled too.

The statue of emperor Constantine in York. Having been proclaimed emperor in York, he left quite a legacy in Britain; Constantine is the name of many leaders in Welsh and Scottish medieval history. Image from here.

5: Wine, Not Vino

One point of departure between British Latin and some of the new dialects of Romance is how it treated the sounds /b/ and /w/ (as in bit and wit). In brief, it seems to have kept them separate and unchanged. The evidence from Welsh does not point to Latin’s change of /w/ into /v/, which turned Classical Latin vīnum ‘wine’ (once ‘wee-num’) into Italian vino and French vin, and vitrum ‘glass’ into vetro and verre. Nor do we have British evidence for the merger of the sounds /b/ and /w/, as has happened in Spanish.

We have pretty early evidence for changes to these two sounds from Latin closer to the heartland of the empire, such as in Italy (Clackson and Horrocks 2011: 242), so it seems strange that we don’t see similar shifts and mergers happening within the words that Welsh gets from Latin. British Latin seems to have preserved the /w/ consonant that was changing elsewhere in the empire, and this /w/ in Latin words would fit in with the /w/ that already existed in native Celtic vocabulary. A Latin word like vīnum would undergo the same changes as a Celtic word like *windos ‘white’, becoming Welsh gwin and gwyn. Perhaps British Latin was in this regard a bit old-fashioned, a product of Britain’s geographical location.

6: Hard Consonants Only

Another widespread sound change in Latin that left no mark on Welsh words is the palatalisation of C and G. The sounds /k/ and /g/ (as in cut and gut), which C and G originally represented, would later change according to what vowel followed them. This led to the difference between ‘soft’ and ‘hard’ versions of C and G in various Romance languages, as in French cent and collège or gentil and gâteau.

However, the words that Latin bequeathed to Welsh show no sign of this change. It seems that C remained /k/ in Latin words like cella ‘room’ and certus ‘certain’. In Welsh, these have become cell and certh, retaining a hard C sound to this day. Even the name of the letter C testifies to this plosive pronunciation, since Welsh gwyddor ‘alphabet’ is a remnant of Latin abecēdārium (‘the A-B-C-D-thing’), in which the initial /g/ of gwyddor developed out of an original hard /k/ pronunciation for the letter C.

On the basis of Welsh words, Latin in Britain does not seem to have undergone this big shift.

7: Voice to the Voiceless

In the fifth century AD, something big happened to the consonants of both Brythonic in Britain and in what would become the Western Romance languages – that is, French, Spanish, Catalan, Portuguese, but not Italian. One part of this shift was that the consonants /p, t, k/, when positioned between two vowels, became /b, d, g/. In technical terms, they underwent lenition and became their voiced equivalents. For example, Latin words like lupus, rota and amīcus (‘wolf, wheel, friend’) have become lobo, rueda and amigo in Spanish and lobo, roda and amigo in Portuguese, with the ‘softening’ of the consonants between the two vowels.

Curiously, the same change occurred in Brythonic at around the same time (Jackson 1953: 92). The same three sounds (and others) in the same contexts softened. The Celtic words *stratos ‘valley’ and *brātīr ‘brother’ have become Welsh ystrad and brawd. It affected Latin loanwords too, turning populus ‘people’ into Welsh pobl and catēna ‘chain’ into cadwyn. It’s believed that Welsh did not borrow such words with the consonants already softened, but rather that the lenition happened later. We can believe this because of St Patrick!¹

Some scholars consider these two cases of lenition (one in Late Latin, one in Brythonic) to be just a coincidence. Others think there was some sort of causal connection, as do I. Regardless, in 450 AD, if both Late Latin on the Continent and Brythonic in Britain were showing this change, I reckon that British Latin would have it too.

My Answer

Welsh words allow us to paint a complex picture of the Latin that was once spoken in southern Britain. Through its considerable contact with that Latin, Welsh has preserved something of its shape for us today, like an echo. What we see in that shape is a variety of Late Latin in Britain that was very much integrated into the wider Roman linguistic world in its sounds, its vocabulary, even its morphology².

British Latin in 450 AD would have sounded barely distinguishable from the Latin over the water. Perhaps it may have seemed somewhat phonetically old-fashioned to someone from Italy, since certain sound changes (like 5 and 6) don’t appear to have reached Britain’s shores. All of the seven features considered here are up for debate, especially the sources of the general rules behind features 1 and 7.

Yet, considered altogether, I believe these seven features lead us to think that “spoken British Latin was to all intents and purposes identical with the type of Romance underlying Old French” (Schrijver 2007: 7). The English Channel was no barrier to innovations in language, and the people of Britain swam in the great flow of life and language in the later Roman Empire. This Roman and post-Roman status quo was not to last though – but that’s another story.


An enchanting and lonely stretch of Sarn Helen, a route of Roman origin that runs through Wales from north to south, and a symbol of the lasting legacy of Roman Britain.

Bonus Word Nerdery:

1: St Patrick and Historical Linguistics

Patrick’s name (Latin: Patricius) got borrowed twice into the Irish language. The first borrowing occurred during his lifetime (died c. 461) and his mission to Ireland from Britain. This produced Old Irish Cothraige, a rare alternative name for him. Then it was borrowed again from either Latin or Brythonic in Britain, producing the usual Irish name draig, which shows us the change from /t/ to /d/. Thanks to the saint, we can argue that it was sometime between these borrowings, likely over course of the fifth century, that Brythonic lenition occurred.

2: The Morphology of British Latin

In a futile effort to save space, I’ve concentrated on sounds in this piece, but I should add that we can tell things from Welsh about other aspects of British Latin, such as its morphology – the shape and composition of its words. For instance, the planet Venus appears in Welsh as Gwener. The shape of this word indicates that Brythonic did not borrow the classical nominative form Venus from the Latin in Britain, but rather the Vener- stem that is behind case forms like genitive Veneris and accusative Venerem. From such words, we may argue that British Latin too participated in the breakdown of the old case system, the legacy of which is not only to be seen in Welsh Gwener but also in Italian’s word, Venere.


  • Barahona, O. (2016). The chronology of Romance lenition: the testimony of Gothic loanwords. MA Thesis. Leiden University.
  • Clackson, J., & Horrocks, G. (2011). The Blackwell History of the Latin Language. Wiley-Blackwell.
  • Jackson, K. H. (1953). Language and history in early Britain; a chronological survey of the Brittonic languages, first to twelfth century A.D. Edinburgh University Press.
  • Lloyd-Jones, J. (1910). Some Latin Loan-words in Welsh. Zeitschrift für celtische Philologie 7(1), 462-474.
  • Matasovic, R. (2008). Etymological Dictionary of Proto-Celtic. Brill.
  • Schrijver, P. (2002). The rise and fall of British Latin: evidence from English and Brittonic. The Celtic roots of English: Studies in Languages 37. University of Joensuu.
  • Schrijver, P. (2007). What Britons Spoke around 400 AD. Britons in Anglo-Saxon England. 165-171. Boydell & Brewer.
  • Geiriadur Prifysgol Cymru online.

Cover image of Constantine’s statue from here.

4 thoughts on “What did British Latin sound like?

  1. Thanks for the article. Could the similarities in changes in sounds in the Latin in Britain and France have something to do with the fact that Gaulish was in the same group of Celtic Languages as those spoken on Great Britain?

    P.S. Prothesis doesn’t seem to have occurred in Rumantsch: schola > scola; status > stadi; stabulum > stalla (this for some reason has become a feminine noun); stēlla > staila. Basque appears to have prothesis, too, in loanwords from Latin, e.g. status > estatu. Interestingly, it also appears before r- in loanwords, e.g. rota > errota.

    Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: