When you hear the word ‘mutation’, your mind may jump to the dramatic result of a scientific experiment gone wrong, or maybe to the occasional blip in the amazing unravelling and duplication of our genetic code. But it’s not surprising that this term has a well-known meaning in linguistics too: after all, languages are themselves products of nature which can unravel, evolve and transform in an incredible number of ways.
If you’ve ever started learning a Celtic language, you may have come across a particularly interesting grammatical phenomenon in which the beginning of words can change or ‘mutate’ in certain contexts. For example, let’s take the Welsh word Cymru, which means ‘Wales’. Normally this word starts with a c, but if you have ever visited Wales, you most likely would have passed a sign with the words Croeso i Gymru, or ‘Welcome to Wales’, and you can truly say that you are now yng Nghymru, that is, ‘in Wales’. As we can see, the same Welsh word appears in three different linguistic contexts in three different forms, all differing in their first letters.
While many language learners may be familiar with grammars that require the ends of words to change, grammatical rules requiring the modification of the beginning of a word are much less common. Mutations are a type of grammatical function that linguists call phonological marking, a way of conveying grammatical information about a word and its role in the sentence by changing the way it sounds. As we shall see below, even though mutation is a clearly phonological variation, the exact environments in which this process occurs suggest that there are various different underlying causes. Amazingly, these mutations are some of the last remaining fossilised remnants of extremely old sound rules dating back to the early stages of the Celtic language family, and still remain a topic of great interest for linguists of all academic fields.
Living transformations Once spoken across Western Europe and the British Isles, the Celtic languages are now hard to find in spoken form, and are limited mostly to small communities in coastal areas. The two surviving branches of this language family are Goidelic, comprising Irish Gaelic and its two descendants Scottish Gaelic and Manx, and Brythonic, which contains Welsh, Breton (spoken in Brittany, France) and the recently revived Cornish language. Understanding how these languages are related is important for making sense of the striking similarities between some of the mutation patterns we see in all six of these languages, as well as the clear differences in how they are structured.
For example, Manx and Irish have only two main mutation patterns for initial consonants, whereas a Brythonic language like Breton can have up to four. Mutations even vary according to dialect, region and speaker, and show interesting variation between communities, though for the sake of simplicity here we will use the standard mutation forms usually given in Celtic grammars.
So how exactly do mutations work? Simply put, the initial sound of a word undergoes a transformation of some sort, which is represented both in a change in pronunciation and in a change in spelling. Let’s take a common example of mutation in which the beginning of a noun changes after a possessive adjective, namely a word such as ‘my’, ‘your’ or ‘her’. Comparing the unmutated or ‘radical’ form of the word for ‘house’ in Irish, Welsh and Breton we can see that the following changes are possible:
|Lenition||a theach||his house|
|Eclipsis||a dteach||their house|
|Soft||ei dŷ||his house|
|Aspirate||ei thŷ||her house|
|Nasal||fy nhŷ||my house|
|Soft||e di||his house|
|Spirant||va zi||my house|
As can be seen from the table, the word for ‘house’ begins with the same sound [t] in each language. What happens to this sound depends on the rules of each language and on the word that precedes it, processes that are (rather confusingly) given different names depending on which language we’re talking about. Starting with Irish, we see that there are two possibilities: [t] in the word teach may undergo what is called lenition to become [θ] in theach (a sound like English ‘th’ in ‘thing’) or it may be eclipsed to become the sound [d] in dteach. Clearly, this depends on what precedes the word: mo ‘my’ causes the mutation called lenition in the following word, whereas a ‘their’ causes eclipsis.
Comparing this to Welsh and Breton, we see that Irish lenition has the same effect as the Welsh aspirate mutation and that eclipsis has the same effect as Welsh and Breton soft mutations, namely that the consonant [t] is voiced to [d]. However, there are also some clear differences. The word va, ‘my’ in Breton, triggers the so-called spirant mutation, transforming [t] to [z], whereas the corresponding word fy in Welsh triggers a nasal mutation, changing [t] to a voiceless nasal sound [ṇ] (pronounced like ‘n’ but with an accompanying puff of air through the nose).
At first glance it seems that mutations are already extremely complicated, even after fairly common words! However, despite the technical terminology, it has been observed that these changes are actually quite intuitive to native speakers, and once it is learnt exactly how letters transform, it’s simply a matter of applying the phonological rules correctly.
Another observation worth noting is that these sound changes are often quite predictable with a bit of knowledge about phonetics (the study of sound production). To illustrate, let’s take the complete table of possible Welsh mutations:
|Original letter||Soft mutation||Nasal mutation||Aspirate mutation|
For each letter there are some interesting phonetic patterns. For soft mutation, voiceless consonants are voiced, whereas voiced consonants are made fricatives (hissing sounds). For the nasal mutation, sounds simply keep all their phonetic features (voiced or voiceless, pronounced with the teeth or the lips) but are pronounced through the nose, that is, made nasal! The aspirate mutation simply involves making the original letters more like fricatives and pronounced with slightly more aspiration, and so for native speakers there is a clearly learnable link between the original consonants and their mutated forms. While the exact nature of these links differ from language to language, a fascinating feature of Celtic mutation systems is that the same phonetic patterns occur more or less consistently.
Another question that arises when faced with understanding mutations is when they actually occur. When talking about sound changes, linguists try to isolate the environments or triggers associated with the change in order to determine the conditions for its occurrence. In Celtic languages, the environment of a mutation categorises it into one of two broad groups: contact mutations and grammatical mutations.
Contact mutations are so-called since they occur directly after or in ‘contact’ with common mutating words such as ‘my’, ‘your’, and ‘their’, question words, verbal particles and adverbs, the exact members of which vary from language to language. Less common (and more elusive) are grammatical mutations, in which words are affected if they come before or after a word or phrase with a particular grammatical feature. An illustrative example of this is the common mutation of an adjective that follows any feminine singular noun:
|‘big door’||doras mór||dorrys mooar||dŵr mawr||daras meur|
|‘big river’||abhainn mhór||awin vooar||afon fawr||avon veur|
Here the word for ‘door’, which is masculine in all of these languages, induces no change in the following adjective, whereas for ‘river’, which is feminine, the following adjective mustbe mutated. Intriguingly, when several different mutation-triggering environments occur in a sentence, these may lead to a whole chain of mutations as in this Irish sentence. Shown below is a sentence as it would appear grammatically, with mutations, and then ungrammatically, without mutations (here an asterisk before a phrase means ‘ungrammatical’ or ‘not attested’):
Chaill mé mo fháinne i lár na habhann i ngar don fheirm bheag.
I lost my ring in the middle of the river by the small farm.
*Caill mé mo fáinne i lár na abhann i gar don feirm beag.
Here there are a total of six mutations, with four contact mutations after the words mo, na, i and don, and two grammatical mutations triggered by the past tense (chaill mé) and a preceding feminine singular noun (fheirm bheag). As we can see, features such as tense, gender and definiteness (whether or not we are talking about the thing or a thing) are often marked by mutations in Irish and its sister languages, alongside the expected inflectional endings we see in many other languages. While it would be possible to write this sentence without mutations, as above, this would be highly ungrammatical, and in some cases ambiguous or even incomprehensible.
Indeed, as much as mutations have been studied by linguists, it is still incredibly difficult to ascertain why certain grammatical features require sounds to be changed at the beginnings of words, and even native speakers have to learn individual cases off by heart or simply do not use them at all. In order to understand why these linguistic patterns occur, it is necessary to dig deeper into the history of the Celtic language family.
Making sense of mutations: the historical picture
Mutations are not a recent phenomenon: they are recorded in the earliest Celtic texts written in the Roman alphabet, and so studying these can give one a clue about their origins. More information is available in the observable nature of mutations in the living languages themselves. Firstly, in certain cases, mutations are surprisingly similar between languages, even between the Goidelic and Brythonic branches. Usually, this suggests that some common factor is at work. Secondly, many of the phonological processes involved in mutations are very similar, such as the transformation of the stop [m] to the fricative [v] in the words for ‘big’ after a feminine singular adjective.
Thirdly, as we have seen, mutations are usually triggered by similar grammatical environments or by the proximity of a small word like a preposition or possessive adjective, even though the exact words involved differ from language to language. Since this evidence applies to all remaining Celtic languages and has even been recorded in the extinct Continental Celtic branch (Eska 2008), historical linguists argue that mutations must have a common historical origin, which would most likely have originated in their mother language, Proto-Celtic.
By recognising that mutations are processes with solely phonological effects, historical linguists argue that mutations originated as a regular phonological process, triggered by the presence of certain sounds in the surrounding environment which are no longer present in the modern language. It is a well-observed linguistic phenomenon that when adjacent in speech, sounds which are articulated differently will change their features in some way so as to become more like one another—a process called assimilation. For example, in English the plural of cat is pronounced as [kats], with a voiceless fricative [s] since it is preceded by another voiceless sound [t].
But when a word ends with a voiced consonant as in log, the plural ending assimilates to this consonant by also becoming voiced, so we actually pronounce the plural as [logz] and not as [logs].Coupled with historical information about previous stages of the Celtic languages, this well-observed tendency of sounds to become similar to each other gives us a clue as to why many of the observable mutations are so similar between different Celtic groups, even in the case of seemingly strange mutations such as [b] to [m] (e.g. Welsh brawd ‘brother’ vs. fy mrawd ‘my brother’).
As an example, let’s take the word for ‘two’ which causes mutation of the following word in all of the Celtic languages:
|English||Irish Gaelic||Scottish Gaelic||Manx||Welsh||Breton||Cornish|
|‘two cows’||dhá bhó||dhà bhò||daa vooa||dwy fuwch||div vuoc’h||diw vuch|
Despite the differences in spelling, there is a remarkable consistency here in the nature of the resulting contact mutation after ‘two’. In each case the word for ‘cow’ begins with the sound [b], but after ‘two’ this mutates to the sound [v] (note that in Welsh this sound is written as ‘f’). To figure out why this is so, we need to be able to take a look at what Proto-Celtic words may have looked like so as to suggest some motivation for mutations at work.
Since there are no actual records of Proto-Celtic, linguists compare evidence from all the Celtic languages and from other Indo-European language families to come up with an approximate reconstruction of individual words. Using these methods, and noting that the Celtic languages all have quite similar words for ‘two’ and ‘cow’, the Proto-Celtic forms are reconstructed as *dwi and *bows respectively (Geiriadur Prifysgol Cymru). As in the example above, here an asterisk preceding a word means it has never actually been attested in real language, but in this context it is nevertheless a good working reconstruction of an older form.
Putting this together, we can see that in the expression *dwi bows ‘two cows’ the initial [b] of *bows is surrounded by a vowel on either side, which is the phonological environment for the expected mutation. Appealing to the study of sound production in language, linguists can tell that a vowel sound is simply a continuous vibration of the vocal chords as we breathe out, whereas the stop [b] involves a vibration of the vocal chords with closed lips, before suddenly releasing the tension in a burst of air.
Although such a sequence of sounds is fully possible (English speakers do this every day) in fast speech, it is a little tricky to vibrate the vocal chords while exhaling, stop the flow of air, and then release it again in a short space of time. Historically, it seems that Proto-Celtic speakers dealt with this by assimilating the initial stop [b] of *bows into the fricative [v], which is a little easier to say between two vowels since it involves both vibration of the vocal chords and a continuous outgoing stream of air from the mouth.
Consequently, after the word *dwi nouns beginning with consonants changed phonetically to accommodate the preceding vowel, i.e. *dwi [b]ows became *dwi [v]ows. Over time, as Proto-Celtic changed and became the Celtic languages we know today, such mutating environments either disappeared or were no longer understood as phonological, allowing the effects on following words to be fossilised as a seemingly arbitrary mutation.
The evidence for these mutations is widespread, and doesn’t just explain mutations after words that are quite similar to their suggested historical forms. In Irish Gaelic, consonants always undergo eclipsis after the preposition i ‘in’, though this seems to include a set of unrelated sound changes: i mbaile ‘in town’, i gcarraig ‘in rock’, i nÉirinn ‘in Ireland’. Historically, this can be explained by noting that i is originally descended from the Proto-Celtic word *in, which ends in a voiced nasal consonant [n]. Consequently, the following sound assimilated to this final [n] by becoming voiced if it was voiceless (as in gcarraig) or becoming nasal if it was already voiced (as in mbaile) or transferring the [n] to the affected word (as in nÉirinn). When *in lost its final [n] the rule was preserved, thus destroying any phonological motivation for the change but retaining the mutation.
This methodology has turned out to be relatively successful in explaining why certain mutations appear where they do, and in charting the development of the notoriously tricky spelling system in Irish as a way of representing these sound changes with ingenuity. However, the fully successful way of modelling all mutations is still being debated in linguistic scholarship, and there are still unanswered questions about certain types of grammatical mutation and how they operate in the mind of a speaker.
Despite these difficulties, a historical approach to Celtic mutations provides linguists with useful knowledge of change and transformations from two different perspectives. Firstly, what appear at first to be arbitrary and confusing alternations in the beginnings of words are shown to have underlying patterns that can be uncovered by concentrating on exactly how mutation changes the sounds of a word. On the other hand, the historical approach enables linguists to look back into the past and learn how the previous forms of a language can structure and explain its present.
Language change is an inherent part of human communication, and it is the task of linguists to look between the lines and find the hidden regularities underneath seemingly irregular effects in the hope of understanding phenomena as unique and fascinating as mutations.
Eska, Joseph F. (2008).Continental Celtic in ed. Roger D. Woodward, Ancient Languages of Europe. Cambridge University Press: Cambridge.