Irish initial mutations
For people who start learning Irish (and other Celtic languages), initial consonant mutations often become stumbling blocks. They were certainly confusing for me, and still are: knowing how they work and how to use them is one thing, but to be able to understand the language and speak fluently, you need to train yourself to mutate and un-mutate words without thinking. There is only one way to do that — constant practice and repetition. From that point of view just memorizing mutation tables and practicing them (e.g., with help from spaced repetition software) is a perfectly good approach.
If you have a student memorize the mutation tables up front, you can tell them that the past tense of many verbs is formed by leniting them, for example. It is just that memorizing mysterious tables without any context can drive any student insane. But not telling students that the processes that are used to form the past tense, possessives, relative clauses, and many other constructs are the same processes and that they are predictable also deprives them of an important learning tool.
Most books and courses use a mixed approach: show that the patterns exist and then gradually teach them by example (“táim i mo chonaí i gCorcaigh, tá siad ina gconaí i nGallimh…”). However, the phonetic nature of mutations is very rarely highlighted. The main reason is that it may take a while to explain to people without any prior training in phonology. But it may in fact be an excellent opportunity for students to learn something about human speech that they did not realize before, and for some people it may be easier to memorize mutations when they see that those mutations are not random at all. So, I made an attempt to present information about mutations in a way that emphasizes the phonetic nature of what is going on.
For people who are not phonology nerds yet, I tried to annotate all terms and IPA symbols with Wikipedia links for further reading.
Initial consonant mutations in Goidelic languages tend to preserve the place of articulation but change its manner. For example, the voiceless velar plosive [k] becomes [x] (voiceless velar fricative) after lenition and [g] (voiced velar plosive) after eclipsis. Notice how its place of articulation — velar 1 — remains unchanged.
It is not always so simple. The extinction of dental fricative sounds in Modern Irish made the situation with lenited [d] and [t] quite complicated, for example: in Old Irish, their lenited versions were always [ð] and [θ], but in Modern Irish, [t] is lenited to [h], and [d] can be either [ɣ] (broad) or [ʝ] (slender). And that is not counting “dh”, “th”, and “gh” in non-initial positions where they can be either completely silent or just cause compensatory vowel lengthening.
Still, the phenomenon has clear patterns:
| Unmutated | Séimhiú (lenition) | Urú (eclipsis) |
|---|---|---|
| voiceless fricative | silent or glottal fricative | voiced fricative or unchanged |
| voiced fricative — never occurs in initial positions in native stems | ||
| voiceless stop | voiceless fricative | voiced stop |
| voiced stop | voiced fricative | nasal occlusive |
| nasal occlusive | voiced fricative (when possible) | unchanged |
More specifically, lenition turns every occlusive sound into a fricative, when it is possible. Eclipsis makes voiceless fricatives and plosives into their voiced counterparts, and voiced plosives into nasal occlusives.
We can present all mutations in a table, grouped by the place of articulation.
| Unmutated | Séimhiú (lenition) | Urú (eclipsis) | ||
|---|---|---|---|---|
| Spelling | Pronunciations | Spelling | Pronunciations | |
| Bilabial and labiodental consonants2 | ||||
| f3 | fh | Silent | bhf | [v] (only slender), [β], [w] (broad) |
| p | ph | [ɸ], [f] | bp | [b] |
| b | bh | [v] (only slender), [β], [w] | mb | [m] |
| m | mh | [v] (only slender), [β], [w] | Cannot change — already nasal | |
| Dental and alveolar consonants | ||||
| t | th |
Old Irish: [θ] Modern Irish: [h] |
dt | [d] |
| d | dh |
Old Irish: [ð] Modern Irish: [ɣ] (broad), [ʝ] (slender) |
nd | [n] |
| n | Does not change | Cannot change — already nasal | ||
| Velar consonants | ||||
| c | ch | [x] | gc | [g] |
| g | gh | [ɣ] (broad), [j] (slender) | ng | [ŋ] |
| Sibilants, approximants, and taps | ||||
| s | sh | [h] | Does not change4 | |
| l | Does not change5 | Cannot change6 | ||
| r7 | Does not change8 | Does not change | ||
Acknowledgements
Many thanks to Emma Nic Cárthaigh for beta reading, suggestions, and corrections.
1Produced with the back of the tongue against the soft palate. The soft palate is called velum in Latin, hence the term.
2In Modern Irish, lenited versions of bilabial occlusives (bh/mh) are often realized as labiodental rather than bilabial fricatives. Generally, [v]/[β]/[w] and [f]/[ɸ] are allophones in Modern Irish.
3F is usually labiodental in Irish but in some words and in some speakers it may be bilabial — one common example where it often happens is faoi (under).
4Could become [z] if Irish had voiced sibilants. Reportedly, that was the case in the Clare Island dialect and it was spelled “ds” there, e.g., “ár dsagart” /a:r zagart/.
5L could become [ɬ] in an alternative universe. Welsh has that sound and there are pairs of cognates that have [l] in Irish but [ɬ] in Welsh: lámh vs llaw (hand), leabhar vs llyfr... The likely reason is that initial [l] tended to be realized as [w] in many Irish dialects for a long time and was lenited to [l] in phrases like “mo lámh”.
6A “nasalized L” sound is technically possible to articulate but nasalized approximants do not seem to occur in any human language, for better or worse.
7R in Modern Irish can be either a trill or an approximant, speakers usually consider them allophones.
8Welsh has a voiceless counterpart of the alveolar trill ([r̥]), but that sound does not occur in Irish.