Digging Old Irish

Reading time: 15-20 minuftes

Old Irish has a bit of a bad reputation. This historical form of the Irish language, dated to between c. 600 AD and c. 900 AD, is a challenge for learners, even for those who already know some Modern Irish. Its systems of spelling, sounds, word order and grammar are very distinct. It requires its students to get acquainted with such linguistic delights as ‘initial mutations’, ‘broad and slender consonants’ and ‘the verbal complex’. From the perspective of the larger Indo-European family to which the language belongs, Old Irish bucks many trends and can feel very unfamiliar.

Yet scholars of the language quite rightly go to great pains to demonstrate just how much method there is in the madness, and how the poor reputation is pretty unfair.

Old Irish is a great example of how a small number of very reasonable linguistic changes can together have massive, language-wide consequences. In doing so, their cumulative effect drastically altered the shape of the language from how we can see it used to be. These changes have obscured many connections back to the wider Indo-European family. Undo the changes, and the ghost of Irish past will emerge…

This is all very abstract, I know. So, the goal of this month’s blog post is to offer a concrete example of how familiar a little analysis can make Old Irish, and how much fun it can be to dig into. I call this ‘language archaeology’, as archaeology is what it feels like. With the right conceptual tools, you get to unearth the different linguistic layers of history and grammar within the very words you observe on the surface.

For me, excavating Old Irish is an exciting and rewarding process, as well as one with very practical benefits, since it makes this ancient language more easily learnable. To illustrate what I mean, allow me to guide you through the wonderful world of the Old Irish noun.

A ninth-century manuscript copy of Priscian’s Institutiones Grammaticae. Written in Latin, but with tiny Old Irish glosses added too! From here.

A Bit of Background

Within the Indo-European family of languages, Old Irish’s cousins include Latin, Ancient Greek, Sanskrit, Gothic, Old Church Slavonic and Classical Armenian. The noun as a word category shows some common features across these languages. Crucially, nouns make use of grammatical endings. That is to say, the end of the noun expresses some kind of grammatical information.

One part of this information is case, which tells you what role the noun plays in the overall sentence. The nominative case, for instance, is usually reserved for the subject, the thing that ‘does’ the action. If nominative, these languages will convey that in how they end, often making use of a dedicated nominative suffix:

LanguageNominative wordTranslation
Ancient Greekhuiósson
Old Church Slavonic gradŭcastle
Classical Armenianhogispirit

Because case is a feature with multiple options, these endings change according to what is needed. The Latin noun dominus has a set of twelve forms that combine its six cases with singular and plural number. In each form, the domin- root remains stable, while the endings alternate.


This may all be elementary stuff, but it needs saying that it doesn’t have to be this way. Case can be expressed differently, or not at all. In some Indo-European languages, a particular case may not have a specific expression. Others may have almost no system of case at all, such as in English today.

The Old Irish Situation

So, how does Old Irish fit into all this? Well, in short, a bit awkwardly.

Old Irish does also have grammatical cases, five in fact. These are the nominative, the vocative, the accusative, the genitive and the dative. These perform much the same roles as their counterpart cases in related languages, and they likewise combine with number to produce multiple possible variants for each noun.

Part of a diagram of the Indo-European family, including Irish as part of the Celtic family. This diagram does not include the long-standing theory that Celtic forms a distinct sub-branch with Italic, but does position them near each other. From here.


Special endings are only one way in which Old Irish expresses this grammatical info, and they’re by no means the most common way. Old Irish instead often conveys case not through the ending of a word, but through the beginning of the word that follows it. Bear with me here.

An Old Irish noun can have a specific influence on the first sound of the following word. It is through this knock-on effect that the case and number features of the noun itself are made known. This is called mutation, something also utilised by Welsh and other modern Celtic languages.

Let’s take the noun fer, which means ‘man’. The nominative singular of fer is fer. The accusative singular, used for when a man is the object of the verb, is also fer. So, when we meet a form of this word, how can we tell its case and number?

Crucially, it’s the accusative singular form of fer that can affect the start of the word that immediately follows it. Such a word is usually an adjective, which contributes to the meaning of the noun. Specifically, accusative singular fer will be able to ‘nasalise’ the first sound of that adjective. That is to say, it may add a nasal quality, turning an adjective like óc ‘young’ into n-óc. Nominative singular fer has no such effect.

It is therefore this nasalisation of the next word that tells us the case and number of fer we are dealing with here.

  • Nominative singular: fer óc (‘a young man’)
  • Accusative singular: fer n-óc (‘a young man’)

Or we can do with this with ech ‘horse’. It’s ech when nominative and singular, and ech when accusative and singular, but only accusative ech has the nasalising effect.

  • Nominative singular: ech álaind (‘a beautiful horse’)
  • Accusative singular: ech n-álaind (‘a beautiful horse’)

We can create a short sentence that includes fer and ech like

  • ad·cí fer óc ech n-álaind (‘a young man sees a beautiful horse’).

It’s thanks to the nasalising of álaind that we can tell which one is accusative and therefore the object, the thing being seen.

Three Marvellous Mutations

Including nasalisation, there are three canonical mutations that can happen to the following word in Old Irish.

  1. Lenition (L)
    • This involves the ‘softening’ of the following consonant. The specific sound changes that can come under lenition are numerous, but they are very systematic. An original voiceless /p/ sound (as in English pat) might soften to a voiced /b/ (as in bat) and this would be considered lenition. An original /b/ sound will itself soften to a /v/ (as in vat).
  2. Nasalisation (N)
    • This originally involved the addition of a nasal consonant to the start of the next word, though it isn’t always written. So, if nasalised by the preceding word, ben ‘woman’ becomes mben and othar ‘work’ becomes n-othar.
  3. Aspiration/gemination (H)
    • If the next word begins with a vowel, a /h/ (as in hat) is added. If it begins with a consonant, that consonant is geminated or ‘doubled’. This third mutation is not regularly shown in our written sources, but we can be confident it was there.

These three combine with actual endings and with changes to the root of fer to give us our different forms for the noun, as Latin dominus has above. Following scholarly practice, the three mutation effects are indicated with a single letter written up high after the word (L, N or H).

Nominativeferfir L
Vocativefir Lfiru H
Accusativefer Nfiru H
Genitivefir Lfer N
Dativefiur Lferaib
Ten forms of fer ‘man’. The five dual forms have been removed for simplicity.

So, a ‘thin man’ in the nominative singular would be fer cáel, with no mutation. ‘Thin men’ in the nominative plural would be fir chaíl, with a change to the vowel in fer and the lenition of cáel. ‘To a thin man’, using the dative singular, would similarly be do ḟiur cháel.

Digging Deeper

So, that’s what we’re dealing with. The example of fer should serve to illustrate the complexities of the Old Irish noun. It uses a combination of processes to convey the right case and number – processes that may feel very alien to someone familiar with other early European languages, like Latin.

What’s gone on here then? How did this combination of processes come about? It all in fact has a fairly unified origin: the phenomenon of sandhi.

This term, taken from Sanskrit, refers to the way adjacent sounds can influence and alter each other within a word or in a string of words. I dare say it’s a universal phenomenon across all spoken languages. It’s the reason why English gets the words incorrect, impossible, illogical and irregular from Latin, in which the negative prefix in- has adapted to be more like the sound that immediately follows.

It’s the reason why English alternates between a and an as its indefinite article (e.g. a banana vs. an apple). We can also hear it in the ‘linking R’ that some dialects of English put at the end of a word. If you, like me, speak the English of Southeast England, you can try out these phrases:

  • better bread, better fish, better milk
  • better eggs, better apples, better oil

Can you hear the R in the second set, but not the first? In casual speech, we would say better eggs and better apples, because the following word begins with a vowel. This way, the flow of sounds goes vowel > consonant > vowel. These are useful examples of sandhi working across the boundary between two words.

Sandhi is the theoretical key that we can use to unpick the Old Irish noun.

Since the time of Old Irish, mutations have an integral part of Irish grammar, but it wasn’t always so. The three aforementioned mutations originally arose from the sandhi interactions between the end of one word and the start of another. It was a process of sounds that ended up a process of grammar.

Just like its Indo-European cousins, Irish nouns used to have a whole set of endings! They operated happily at the end of nouns as they did in Latin and Ancient Greek. We can even see this in our rare sources for the earliest stage of Irish (‘Primitive Irish‘) and in related Celtic languages like Gaulish. The sounds of these endings could naturally have an effect on the words that followed, with nouns’ endings modifying the initial part of their dependent adjectives.

A Gaulish inscription, from Alise-Sainte-Reine, Franche, dating to the first century AD. A Celtic language with a rich array of grammatical endings! From here.

Then something happened. Many original endings were lost in early Irish, through a widespread change to the sounds and stress position of words. This was no big bother for nouns though; their influence on following words’ sounds could step in to pick up the slack. Thus sandhi effects, once inconsequential, gained a new grammatical importance.

Sandhi is also the reason why mutations are limited in their use. A noun can’t mutate any old word that follows it. Instead, they usually operate between the noun and its dependent adjectives. This is to be expected. Nouns and adjectives have an especially close association, and together form a phrase and a distinct constituent of the overall sentence. Thanks to this association, sandhi could occur and mutations could get started.

The Old Irish mutations are therefore like ghosts of grammar, shadows of where sounds and endings used to be. We may take comfort in this; Old Irish isn’t so different after all. But we can do better. Let’s work through the mutations one by one.


Lenition developed when the ending of the noun was a vowel. Sandwiched between the vowel ending and the vowel of the first syllable of the following word, an initial consonant would soften. This is itself a reasonable and well-documented outcome. Lenition between two vowels can be seen for instance in some Spanish words, in which what was spelled T in Latin is now spelled a D.

  • Latin prātum ‘meadow’ > Spanish prado ‘meadow’
  • Latin cīvitātem ‘city’ > Spanish ciudad

So, when we see forms of a noun like fir L that trigger lenition, we can presume that they once had an ending that was just a vowel.


Nasalisation is similarly an indicator that the word once ended in a nasal consonant, something like *n or *m. That nasal can no longer be seen at the end of the noun, but its effect on the following word endures.

Here’s my rough attempt to show how a bit of grammar like nasalisation could develop, creating Old Irish accusative singular phrases like fer n-uile ‘a whole man’.

Aspiration and Gemination

Aspiration and gemination are really two processes with two origins that get lumped together as one Old Irish mutation. The source for the aspiration effect is that the noun used to end in a vowel and then the consonant *s. If the following word began with a vowel, the *s found itself caught between two vowels and softened to a /h/ sound. So, in the case of firu H, we can presume that such nouns once ended with an *s.

Bringing It All Together

With all that in mind, let’s revisit the full table for fer.

Nominativeferfir L
(Once ended with a vowel)
Vocativefir L
(Once ended with a vowel)
firu H
(Once ended with an *s)
Accusativefer N
(Once ended with a nasal)
firu H
(Once ended with an *s)
Genitivefir L
(Once ended with a vowel)
fer N
(Once ended with a nasal)
Dativefiur L
(Once ended with a vowel)

Now, does any of this look familiar?

What if we bring back the table for the Latin word dominus? Some striking similarities emerge Old Irish fer and Latin dominus, with its set of endings.

(Ends with a vowel)
(Ends with a vowel)
(Ends with a nasal)
(Ends with a /s/)
(Ends with a vowel)
(Ends with a nasal)

We can see now for instance that the nasalising accusative singular fer N isn’t so different from Latin’s nasal-ending accusative singular dominum. Likewise, the leniting genitive singular and nominative plural fir L matchs the vowel-final dominī.

What does all this mean? That Latin is the origin of Old Irish? No, not that, but they are related languages, and the two once had the same or very similar grammatical endings for their nouns.

These endings go back their common ancestor. In Classical Latin, they’re still going strong, but by the Old Irish period, they’re fading fast. These ancient elements are mostly realised in Old Irish not by their presence, but by the phonetic impression that they left behind.

We can hammer the point home by setting Old Irish fer side by side with its Latin cognate: the Latin word vir, which is related to fer and likewise means ‘man’. It’s from vir that English gets virtue and virtuoso. Through these two, we can match up Old Irish’s mutation effects with Latin’s endings, given here in bold.

Old Irish ferLatin vir
Nominative Singularfervir
Vocative Singularfir Lvir
Accusative Singular fer Nvirum
Genitive Singularfir Lvirī
Nominative Pluralfir Lvirī
Vocative Pluralfir Hvirī
Accusative Pluralfiru Hvirōs
Genitive Pluralfer Nvirōrum

It’s not an exact match, true, as things like vocative plural fir H leads us to predict that the Latin will end with an -s, and yet we have virī. This for me though makes the puzzle all the more thrilling, as it points to the prehistoric swapping and merging of different endings, even when we don’t have direct evidence for it.

This is what I mean by language archaeology! With a little digging, past stages of the Irish language emerge, which reveal a time when endings were more prevalent and mutations were as yet no more than the interactions of sounds between words.

Deeper Still…

Feminine nouns in Latin are in general characterised by an A-vowel, and the nominative singular ending is often just -a. We can see this in the words fēmina, toga and porta (‘woman, toga, door’). As a cousin of Latin, we might predict that this final A-vowel will have been lost from the end of words in Old Irish, but as a vowel, it may have left behind a leniting effect. Sure enough, that’s what we find. The feminine nouns fáilte L, túath L and frac L (‘joy, people, woman’) all cause lenition.

The third gender, the neuter, likewise shows similarities across Latin and Old Irish through the mutations that neuter words cause. One common class of neuter nouns in Latin end in -um when nominative and singular. These include for example bellum ‘war’ and templum ‘temple’. Meanwhile, its Old Irish counterpart class of neuter nouns triggers the nasal mutation, as we can see in the words cenn N ‘head’, dún N ‘fortress’ and gein N ‘birth’. The old consonant continues its nasalising in Old Irish.

Mutations are not limited to nouns either. Prepositions can also trigger them, which affects the first sounds of a noun that follows. Simple i N is an Old Irish preposition meaning ‘in’ or ‘into’. It triggers nasalisation, so ‘in Ireland’ would be i nÉrinn. Its nasalisation is hardly surprisingly when we know that it points to an old nasal consonant, one still seen in English in, Latin in and Welsh yn.

To Conclude

If you’ve made it to the end of this long and meandering post, thank you! I hope at least it has made sense, and that it has achieved its principal aims: to walk you through the connections between nouns in Old Irish and nouns in related languages, and to lay out the key changes that have obscured those connections.

There is a lot more that could be said, and no doubt a real scholar of Old Irish would do it differently, but I hope at least that this introductory piece is clear enough and serves as a good base for your own linguistic archaeology. Needless to say, Old Irish deserves praise and admiration for its sake, rather than just for its connections to related languages. However, those connections can be very helpful, and great fun to dig into too!



  • Stüber, K. (2017). 69: The morphology of Celtic. Handbook of Comparative and Historical Indo-European Linguistics: An International Handbook. Vol. 2. 1203-1218. De Gruyter Mouton.
  • Thurneysen, R. (1975). A Grammar of Old Irish. Translated by Binchy & Bergin. School of Celtic Studies, Dublin Institute for Advanced Studies.
  • Old Irish: Glottothèque: Ancient Indo-European Grammars online (electronic resource). University of Göttingen.

Featured image: the stunning first page of the Gospel of John in the Book of Kells. From here.

One thought on “Digging Old Irish

  1. I’ve only dabbled in (modern) Irish, so it’s fascinating to see the source of the n- etc. It reminds me of Finnish, where if an imperative is followed by a word beginning with a consonant, that consonant is pronounced (though not spelled) double: Tule tänne! [tulet tänne] = Come here! According to materials from the University of Minnesota (can’t find credits beyond that), “This phenomenon is due to the fact that the imperative used to end in a consonant (-k) at an earlier stage of the language. Although the consonant is no longer present in the modern language, in a sense it appears (although not as
    a -k, but as a copy of the following consonant), but only when another word follows.”

    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: