Silent letters and consonant pairs in Irish

I’ve had a fascination with Gaeilge, the Irish language, for a long time. Its long words and complicated writing, which together allow for such fun things as fheicfeadh [ɛcətʲ]. Oh! And its consonant mutation, one of the coolest features I think a language can have. Initial consonants changing based on prepositions, adverbs, gender, tenses, and so on. It’s so amazing and interconnected! Add to that, Irish’s long literary history and the modern attempts to save the language from extinction, and I just can’t help but love the language.

But I’m not here to write about any of that. I’m here to write about the Irish consonants, more precisely, the leathana and caola pairs, the two categories that most Irish consonants fall into. In English, they are called broad and slender, and every Irish consonant has one version that falls into each category (except /h/).

A consonant being broad or slender often surfaces as a secondary articulation, broad consonants being velarised, and slender consonants being palatalised. This can be seen in the minimal pair of buí /bˠiː/ (yellow), and bí /bʲiː/ (be), where the only difference is velarisation/palatalisation. But if you look at it phonetically, a lot more than just a secondary articulation happens for some phonemes. For example, the coronal plosive pairs /tˠ/ /tʲ/ and /dˠ/ /dʲ/, despite using the same symbol, differ in their place of articulation, with the broad /tˠ/ /dˠ/ being dental or denti-alveolar [t̪ˠ] [d̪ˠ], and the slender ones being alveolar in some accents, but alveolo-palatal and even slightly affricate in the accents in and around county Mayo and the southern parts of county Donegal, becoming [tɕ] and [dʑ].

Another good example of this is between /vˠ/ and /vʲ/. Because, in the Ulster dialects, the broad version is not realised as [vˠ] as one might expect, but as [w], shifting from velarised labio-dental to labio-velar. This total shift of [vˠ] to [w] does not occur everywhere in Ireland though, as the Munster dialects keep /vˠ/ as labio-dental, and the central Connacht dialects have both [vˠ] and [w] in complementary distribution, with [w] initially and [vˠ] elsewhere, as seen in this example: bhfuil [wɪ̈lʲ] (blood) and naomh [n̪ˠɰïːə̯vˠ] (saint).

The extra characters (ɰ, ə̯, ¨(Velar approximant, non-syllabic mid central vowel, and the centralisation diacritic)) are part of another process of vowel allophones and on-offglides, where glides are inserted between front vowels and broad consonants, and between back vowels and slender consonants, and where vowels slightly assimilate to the place of articulation of the surrounding consonants. Of course, there are more details as to how that works with all the different vowels and glides and so on, but that’s boring, and instead I want to talk about Irish’s spelling, and how the broad/slender pairs come into play there.

Most Irish dialects hold these consonant phonemes, at the very least
Most Irish dialects hold these consonant phonemes, at the very least

When it comes to writing Irish with the phonetic alphabet, the broad/slender distinction is written with diacritics and such, ˠ, ʲ, t̪, and that’s pretty easy. But for the medieval monks who first had to write down Irish with the Latin alphabet around the 10th century, it was much more of a challenge. Old Irish had around 46~ consonants and 11~, maybe 12~, vowels, however, the version of the Latin alphabet that the monks used only had 18 letters. But then the monks discovered something peculiar. They could split all the consonants into two groups (even /h/ back then!), and these groups often correlated with specific vowels. This meant that the first group, the broad consonants, were often surrounded by the vowels written a/áo/ó and u/ú, and the second group of consonants, the slender ones, were often surrounded by vowels written e/é and i/í. So, the monks came up with a system where if a consonant was preceded or succeeded by a/á, o/ó, or u/ú, it would be broad, and likewise, if it was preceded or succeeded by e/é, or i/í, it would be slender (this is a generalisation, there were of course variations from monastery to monastery). While this did mean that it became much easier for the monks to write Irish with their few letters, it also meant that Irish suddenly had a lot of silent vowel letters. For example, the word which in primitive through to old Irish was pronounced /kʲeNː(the exact quality of /Nː/ is unknown)/, was written ceann, with the “a” not being pronounced, but instead showing that the nn and was broad /Nː/ and not slender /Nʲː/. This system is still in use today, and other examples include the modern name for the Irish language, Gaeilge, where the a, again, is not there to be pronounced, but to show that the G is broad, and that the name is [ˈɡeːlʲɟə] and not [ˈɟeːlʲɟə].

Looking outside of Ireland, I’ve noticed lots of other languages with similar systems. There are the related Goidelic languages, Scottish Gaelic and Manx, with systems that developed alongside Irish from their shared ancestor, Primitive Irish, through a series of consonant assimilations, where consonants near front vowels would become palatalised, and consonants near back vowels would become velarised, and then later, the vowels would be dropped, and all which was left was a palatalised/velarised distinction.

Looking further out, many of the Slavic languages of eastern and central Europe have similar systems, with soft (palatalised) and hard (plain) consonants. All these languages are Indo-European though, and while that might lead one to think that this is some shared Indo-European feature, eroded away in all the other PIE-descendants, but retained in these scattered languages, it is not. All these languages developed their enigmatic systems by themselves at later dates, with no relation to each other.

On a final note, I would like to mention that, in the middle of the Pacific, there is a language with a very similar palatalised/velarised system, called Kajin M̧ajeļ. In English it is called Marshallese, and, like Irish, it has a nearly complete set of velarised/palatalised consonants, the only exception (because of course there has to be an exception) is in the Marshallese velar consonants, where there is instead a contrast between plain and labialised consonants instead.


Tristan is 16 years old and he attends Eisbjerghus Boarding School. He is very fond of music and linguistics, especially phonology and phonetics. In the future he aspires to be a musician.


Primarily used:

Also used:

2 thoughts on “Silent letters and consonant pairs in Irish”

  1. Oh, this is a good opportunity to ask: are [c ɟ] real, or are they just abstract transcriptions for [kʲ gʲ]? I’ve seen so many languages that have [kʲ gʲ] transcribed with [c ɟ] that I’m a bit allergic to these symbols now (actual [c ɟ] are found in Hungarian and Latvian in any case, and they’d be rather hard to keep apart from [tʲ dʲ]).

    The non-palatalized consonants of Russian are mostly velarized, too.


Leave a Comment