Y Treiglad Meddal • The Soft Mutation

First, let's take a look at exactly what the Soft Mutation is, what the process is. Most books present the Soft Mutation as a whole series of changes to various consonants, as if it were a dozen independent changes that take place randomly. These books often obscure the mutation still further by presenting the letters as they are affected in alphabetical order or some such awful thing. In reality, the Soft Mutation is one process, and not a complex one. But in order to step back and appreciate this, we need to understand a bit more about the way that consonants are made in the mouth.

Consonants can be located in several places in the mouth. Consider the English letters "p," "t," and "k." If you pronounce them in order, you can appreciate that they are similar in some ways. (And I don't mean to pronounce the names of the letters, "pee," "tee," and "kay." I mean the sound that they make in a word, just the puff of air part.) They are all "voiceless," meaning you aren't using your larynx to say them; there's no "hum" in your throat. They are interrupt the airflow through your mouth completely; they are "stop" consonants. The only way in which they differ is the location in the mouth where they are made. The /p/ is made in the very front, with the lips pinching together and interrupting the airflow. The /t/ is made by pinching off the airflow through raising the tip of the tongue behind the front teeth. And the /k/ is made by raising the back of the tongue to pinch off the airflow at the back of the roof of the mouth.

But these aren't the only kinds of stops in a language. Let's add voicing to these consonants -- let's move them one "step" over in terms of making them more resonant, or "softer." Add voicing to a /p/, and you get a /b/. Same consonant, but this time you can feel a "hum" in your throat when you say it. /t/ becomes /d/, and /k/ begomes /g/. They are all pronounced in the same locations in the mouth, and they all interrupt the airflow completely, but one is voiced, and one is not.

Okay, let's move them yet one more "step" over in terms of adding sonorance. Let's not pinch the airflow off entirely, but just narrow it a bit. A /b/, for example, becomes something like a /v/ sound, where the lips (or teeth and lips, in the case of the English /v/ sound) restrict the airflow, but don't close it off entirely. The /d/ sound can become a voiced "th," as in the beginning of words like "there" and "this." Even a /g/ could, strictly speaking, soften to a voiced version of the voiceless "ch" that you get in Scots words like "loch." (This doesn't happen in Welsh for another reason.)

So, let's lay out a table of consonants, where the first column lists the voiceless stops, the second column adds voicing, and the third column adds airflow:

Each column is one step over in terms of adding a more fluid, resonant, or sonorant quality to the consonant. The voiceless stops are the "hardest" sounds, and the sounds like /v/ and the voiced "th" are very soft and flowing; they have both voicing, and they permit freer airflow through the mouth.

Viewed this way, the Soft Mutation is one process -- each consonant in the table is moved one column over in terms of its sonorance. Q. E. D. There you go. The last, rightmost column can't be softened any further, so those consonants are unaffected. If the consonant in question can't be moved over one "click" without falling off the table entirely, or if it's not even on the table, like the /s/, then the Soft Mutation won't touch it.

The mutations that I've marked in light red are the "tweaks" to the table that I'll explain below.

Tweaks to the Soft Mutation

There are of course some little tweaks. The /ll/ and the /rh/ consonants, which are not stops, also are affected by the Soft Mutation in a lot of places (but not all). These voiceless consonants do not exist in English, and when they are softened, voicing is added to them, turning them into the more familiar voiced /l/ and /r/ sounds that English speakers use all the time.

Also, the other tweak is that the /g/, when affected by a Soft Mutation, actually vanishes. The reason for this is the following. In all cases, when a consonant is softened, it "turns into" another consonant. A /p/ turns into the already existing /b/, or the /k/ turns into the already existing /g/. Each consonant when mutated "jumps into" another consonant. However, there is no counterpart to a /g/-with-airflow-added-to-it in Welsh; there is no consonant for a softened /g/ to "jump into." As a consequence, it jumps away entirely and disappears.

The third and last tweak is a little odd. The letter /m/ mutates, just like the letter /b/. Just as a /b/ is the same as a /p/ with voicing added, an /m/ is the same as a /b/ with nasality added. Both of them are voiced, they are both made by pinching the lips closed, and they both interrupt the airflow through the mouth completely. The difference is that the /m/ allows airflow through the nose. Other than that, they are the same consonant, and they mutate the same way.

Of course by this logic, the /n/ should mutate the same as the /d/, but it doesn't. Why not? I haven't the damnedest, which is why this is the only aspect of the Soft Mutation that really annoys me. /n/'s often tend to be a bit "stickier" in a lot of languages, so it's not entirely strange, but it still bugs me.

Why does Welsh even do this? Where did this mutation come from?

Frequently, the mutations (the Soft Mutation in particular since it is so ubiquitous) are used as evidence as to why Welsh takes the Gold Medal Bull-Moose Grand Prize for Weirdest Language on Earth. Welsh and the other Celtic languages (Breton, Cornish, Manx, Irish, and Scots-Gaelic) are all specially known for having complex and intricate systems of mutations, but honestly, they aren't really that strange. Other languages often do something roughly similar under the radar, as it were. Celtic languages are distinctive in that this slightly moldable quality of some of their consonants carries grammatical value.

What in God's name does that mean? Read on!

Some of you may know a bit of Spanish. (For those of you who, like me, live in southern California, you may well have inadvertently learned a good amount of it.) Okay -- remember the classic pronunciation of a word like "dedo," meaning "finger?" There's two "d"'s in that word, but if you're a native speaker or particular about your accent, you don't say them both the same way. The second one is more of a voiced /th/ sound, like in "this" or "that."

Another good example of consonants in Spanish getting a little mushy in some circumstances is the word for lawyer: abogado. Those consonants aren't really hit very hard. They become very soft and fluid. Why?

Simple: they are sandwiched between two vowels. In the hierarchy of sonorance, fluid, resonant sound quality in language, nothing but nothing is more sonorant than a vowel. They are the most sonorant sounds in any language. And sometimes, if a consonant is flanked by two vowels, the sonorance of the two vowels will bleed into the consonant, making it more fluid.

As a consequence, you don't say dedo in Spanish with two hard "d"'s. You say something that is more like /deddo/, to use the spelling conventions that I've set here where a voiced /th/ is written as "dd." The main thing in Spanish is that this doesn't carry any grammatical information; it's just an artifact of the accent. If you want to sound like a native, you will soften your voiced stops, the /b/, /d/, and /g/, between vowels -- but your teacher won't mark you wrong if you don't do it. British English speakers often drop /r/'s after vowels, as in the English pronunciation of a word like "harmony" or "art." But it doesn't affect the grammar to say it the American way.

In Welsh though, the mutations carry grammatical value, so a phrase or sentence that doesn't contain the correct mutation is actually ungrammatical, and often spelled wrong. (The native speakers will understand you but in the same way that a native English speaker will understand something like, "You go store?" spoken by someone who is still learning English.)

The prevailing belief, in fact, is that a process something like the one at work in Spanish was once at work in Celtic languages thousands of years ago, and that this is the origin of the intricate and beautiful mutation system now at play in those languages. For an idea of how this could have taken place, consider the following two words in Spanish:

bal: "ball," masculine
boca: "mouth," feminine

Unlike Welsh, Spanish does have an indefinite article: "un" for masculine nouns, and "una" for feminine. "A ball" and "a mouth" are:

un bal
una boca

Take a look at the /b/ sound that starts both words. The first one has a consonant on one side, and a vowel on the other. The second one, however, is sandwiched between two vowels. The /b/ at the beginning of the word "boca" in the second phrase, therefore, will soften for native speakers or people who want to have a good native accent.

(Again, this is not a matter of grammar, and if you don't do this, it won't make the sentence wrong. It'll just mean that you have a thick foreign accent.)

Now suppose that, over the next couple hundred years, the -a on the end of the feminine indefinite article drops off, making "un" the proper indefinite article for all nouns. There's no reason to suppose this will happen, but it's not unheard of; languages often misplace things over time. If this were to happen, perhaps the softening effect that was prompted by the presence of that -a might still remain, an echo of the former existence of the -a on the end of "una." In that case, the softening at the beginning of the word "boca" might be required to signal that the word is feminine. Now, instead of being an artifact of the accent, it would actually carry grammatical meaning.

The prevailing belief among linguists is that a process similar to this one occurred thousands of years ago in the Celtic languages -- and that's where the mutations originated. An unintentional example that illustrates this is given in Gwynfor Evans' "Land of My Fathers," where he mentions the Welsh historical figure Cynfelin. Originally, this name was the Latin "Cynobelinos." The /b/ in the middle of the word was initially sandwiched between two vowels, and it softened as a consequence, much like the /b/ in abogado. Then over time, the triggering /o/ vanished, leaving the first syllable "Cyn-" and a now softened v-sound after it, an echo of the /o/ that was once there. (Remember that "f" in Welsh is pronounced as a /v/ sound.) Over the intervening millennia, these mutations grew and branched off and became more and more intricate until now, they are a complex and engaging hallmark of all Celtic languages.

So honestly, it's not so implausible that a language might employ a system of altering its initial consonants based on a variety of triggers. That's not to say that Welsh doesn't have some pretty weird stuff going on; their plural suffixes alone would make it a shoe-in for semi-finalist in the Weirdest Language on Earth contest at least. But the mutations are more reasonable that you'd think at first glance.

And really -- as a native English speaker, I don't think we have any business calling out other languages for weirdness. Ask yourself why a word like "knight" fails to pronounce 50% of its letters, and then talk to me about Welsh being weird.

Where does the Soft Mutation occur?

All right, so there's the simplified way of looking at the Soft Mutation -- if the consonant is on the table, move it one column over unless it would fall off the edge. Poof, there you go. Now, where do you do it?

The answer is Damned. Near. Everywhere. Or, to a learner, it can sure seem that way. In reality, there are five general places where the Soft Mutation occurs, and they are the following:

1. After the subject of a sentence,
2. In connection with feminine nouns,
3. To mark some inflected verb forms,
4. After almost every preposition in the language, with some exceptions, and
5. After some of the possessive pronouns.

Believe me, it's not that bad. In general, just keep in mind that if you're anywhere near feminine gender, prepositions, inflections, or possessives, or you're following a subject, you'll probably have a Soft Mutation to keep track of. Let's take a look at examples of each.

Examples of the Soft Mutation

After the subject of a sentence

This spot in a Welsh sentence is a very special one. In all cases, if a word that follows a sentence subject can Soft Mutate, it will do so.

Take as an example, the sentence "The dog bit a horse."

Fe frathodd y ci geffyl.

(There's actually another mutation going on there, but we'll address that later.) For the moment, let's look at the dictionary forms of the words in this sentence.

"bit" -- fe frathodd
"the dog" -- y ci
"a horse" -- ceffyl

(There's no Welsh word for "a" or "an" -- "a horse" is just the word ceffyl without any indefinite article in front.)

The first thing you notice is that the word for "horse" in that sentence is not ceffyl, despite that being the version that's in the dictionary. Instead, since the word follows the subject of the sentence, the /k/ sound of ceffyl is softened to geffyl.

Suppose, the horse, being a little miffed at this, turns around and bites a cat:

Fe frathodd y ceffyl gath.

Just as you suspect, gath isn't actually the Welsh word for "cat," but a softened version of it. The dictionary form of the word is cath.

And then, the cat gets understandably annoyed and lets the mouse have it:

Fe frathodd y gath lygoden.

Again, there's some more mutations going on there, but they'll come in later. For now, just notice that lygoden isn't, as you would guess, the actual Welsh word for "mouse." The dictionary form of that word is llygoden.

In all cases, the word immediately following the subject of the sentence softens. This affects all words that are there, not just if they happen to be the object of the sentence. Consider the following sentences:

Dw i'n mynd i'r siop. -- "I'm going to the shop."
Dw i ddim yn mynd i'r siop. -- "I'm not going to the shop."

The word that is used to negate the verb isn't actually ddim. Its strict form is dim. But since it always pops into the sentence immediately after the subject, it always mutates.

After the subject of a sentence = Soft Mutation. Q. E. D.

Surrounding feminine nouns

One of the mutations that occurred in the last sentence above, the cat biting the mouse, had an extra mutation in it if you noticed:

Fe frathodd y gath lygoden.

Shouldn't "the cat" be y cath? Turns out, nope. After the definite article "the" -- y in Welsh -- feminine singular nouns undergo Soft Mutation. "Cat" is feminine, so we get gath. Horses and dogs are, for some reason utterly opaque to someone like me who speaks a language without any grammatical gender, masculine and hence: y ci and y ceffyl.

Now, suppose the cats are just feeling bad-tempered today. Suppose both the white cat and the black cat bite a mouse. Now the Welsh words for "white" and "black" are:

"white" -- gwen
"black" -- du

Okay, let's add those words into the sentences and see what happens:

Fe frathodd y gath wen lygoden.
Fe frathodd y gath ddu lygoden.

Huh? Hey, the adjectives that come after feminine nouns also undergo Soft Mutation!

So -- singular, definite feminine noun? Soft Mutate that sucker. If it has an adjective after it, then that mutates, too.

This one's a bit easier to buy into if you are familiar with some other Romance languages, French in particular. Adjectives that come after feminine nouns in French often change the pronunciation of their final consonant by adding an "e" to the word. The "vert" used to describe a book, for example suddenly regains the /t/ at the end when it is used to describe a teacup: "verte." In the French example, the final consonant on an adjective following a masculine noun frequently softens out of existence, where the final "e" that is placed at the end of the feminine version prevents the consonant from disappearing.

It's not the same process in both languages, but the general idea of a consonant at the beginning or end of a word that gets a bit mushy or even fades out of existence depending on the gender of the noun it modifies isn't entirely strange, at least.

Beginning some inflected verb forms

This is the last mutation that was popping up in those sentences, but that I didn't mention until now. Fe frathodd means "bit," right? Third person, past tense. And don't forget that the f's are actually pronounced as /v/.

But the dictionary form of "to bite" is brathu. Okay, we can see where the -odd might be a verb ending to mark third person past tense, but why is it frathodd and not brathodd?

Well, because you Soft Mutate inflected verb forms when they aren't negative. If the dog/cat/horse didn't bite the horse/cat/mouse, then fine. Brathodd it is, with an extra little bit added on:

Brathodd y ci ddim ceffyl.

But if it's not negated, then the verb Soft Mutates. (And sometimes, just because Welsh loves its Soft Mutations, it still does in some dialects. There is also an interesting hybrid of this and the Aspirate Mutation that sometimes shows up beginning a negated inflected verb, where the initial consonant will undergo AM if possible, and if not, then SM. I think if you stick with a SM, you'll be understood. These things vary even among different speakers of Welsh.)

(Also, you'll notice in the above example that since the dim falls into the slot after the subject, it takes the Soft Mutation and leaves ceffyl to revert to its dictionary form.)

After every damned preposition under the sun (almost)

Yep. Get used to it. Prepositions mutate the living daylights out of everything they precede, just about. (They do some other weird things as well, especially before pronouns.) But for now, just keep in mind that if you are going to someplace, coming from there, going over or on or talking about, then you've got a Soft Mutation following.

Even the ones that don't Soft Mutate things still often induce mutations of one or the other sort. Gyda, meaning "with," will cause an Aspirate Mutation in the following word. Yn, meaning "in," will cause a Nasal Mutation in the following word.

A few prepositions don't cause mutations, like mewn, the indefinite version of "in." But for the most part, you're pretty safe in assuming that they all do, and just working backward from there.