Where do sound changes fit in?

[Akubra]

In conlanging, I find it difficult to grasp what I should do with sound changes. I have the beginnings of a conlang called K'aach. My main goal is to create a relatively naturalistic conlang with a mixture of sound features from Maya languages, Hindi, and Australian languages. The vocabulary (currently about 100 words) is divided according to their semantic domain. For example, words related to "universe/creation" are influenced by Australian languages, "person" by Hindi, and "language/thought" by Maya languages. I don't have a conculture, but the language is supposed to be spoken by humans somewhere on Earth.

Now I have put the vocabulary through some sound changes and I have a result. But what should I do with that result? Should I use the language before it has undergone sound changes (if so, why were the sound changes necessary?), or should I use the resulting words (which I like less than the original ones)? I have read some other topics about sound changes, but they don't seem to answer my questions. I'm feeling stuck and need some help.

[Creyeditor]

I think you kight be confusing diachronic sound changes with synchronic phonological processes.
A sound change changes (roughly) some sounds in an older stage of the language into some other sound in a newer stage of the language. If you do naturalistic diachronic conlanging, i.e. you derive your conlang from an older stage of the same language, you need sound changes.
Synchronic phonological processes (often) result in phonological alternations. In English for example the letter <c> in <electric> has a different pronounciation from <c> in <electricity>. This is due to some phonological process that happens synchronically.
Similarly, the <t> in <tea> and the <t> in <writer> are not pronounced the same. Patterns that interact with morphology are often called morphophonology and other patterns allophony. Naturalistic conlanging needs phonological processes because natural languages have phonological processes.

[Salmoneus]

I don't think Crey is wrong in what they've said (though I don't think I agree with their reading of what you meant), but I thought it would be helpful to break things down more explicitly.

A "sound change" is a slightly loose way of describing a process in a language whereby one sound 'changes' into another. A process involves an input state and an output state.

A "process" can be diachronic or synchronic. These are just fancy Greek-derived words for whether the input and output state exist at different times and hence the process operates 'through time' (dia-chronic) or the two states in some sense exist at the same time (syn-chronic).

---------------------------------------

A diachronic process is a real change between how things were and how they are now (or how they are now and how they will be). Sound change normally refers to diachronic sound change.

Diachronic sound change therefore inherently relates two different languages, a parent and a daughter. [the borders of a 'language' are vague, so sometimes we might instead talk of earlier and later 'stages' of the language or some similar terminology, but it's ultimately the same thing]

-----------------------------

Synchronic change is instead change between two things that exist at once. Some people would prefer to say that there is "change" between an "underlying" linguistic reality and a "surface" realisation, while others would prefer to say that there is a "change" between the sounds that appear in one place and those that appear in another. These amount to the same thing in practice. It basically means "the sound used in this word isn't what we might immediately expect".

For example, the first vowel in "nation" doesn't sound the same as the first vowel in "national", even though both the spelling and the clear connection in meaning would lead someone to assume that they would be the same. So we talk about a "change" between these two forms (or between both forms and some underlying abstract form that produces both these surface forms).

Synchronic changes would really better be considered "alternations", because "change" implies time. However, alternations are often laxly called changes.

----------------------

Synchronic change can be divided into three kinds (at least):

- allophonic variation, where the "same" sound in different words is actually pronounced differently because of the neighbouring sounds or position in the word (etc). For instance, the "p" sounds in "pat" and "spat" are actually different, even though most English speakers don't notice this. This is because the presence of the "s" predictably changes the "p" in a certain way in English.

- productive phonemic variation, where the sound in two seemingly related word is totally different, but in a way that can be learnt as a general pattern. For instance, adding -al to a two-syllable word with a long first vowel (usually, except with some weird or new words) causes the first vowel to shorten - hence NATION, with the same vowel as FATE, becomes NATIONAL, with the same vowel as FAT.

- non-productive phonemic variation, where the sounds in seemingly related words are different in a way you just have to be learnt as a seemingly-random pair.

Some changes are more or less productive; it's a continuum, not an absolute distinction.

Productive alternations can be conditioned phonologically (the 'change' is produced when adjacent to certain other sounds), morphologically (the change is produced when certain morphemes (meaning-bearing word units) are present) or in very rare cases syntactically (the change is produce when certain words are nearby). Importantly, some morphologically-conditioned change uses morphemes that are superficially invisible, and hence are ONLY seen in the alternation they cause. For instance, foot > feet - the plural morpheme is only visible in the alternation between the two vowel sounds.

----------------------------------------------------

Synchronic change is a part of "naturalistic" conlanging - that is, conlanging that aims to produce something that looks like a real language. Because synchronic change of some sort is a part of almost any real language (probably all of them but I don't know for sure).


---------------------------

Diachronic change links two languages, so if you are only interested in one language then you don't need to worry about diachronic change.

However, most naturalistic conlangers DO worry about diachronic sound changes, almost obsessively!

Why?

Firstly, because working out the diachronics is how you create sets of languages - not just a mother and daughter pair, but also sister languages and cousin languages. If you create your equivalent of Latin, you can then use diachronic to create your equivalents of French, Spanish, Romanian, etc, and create an entire family of languages.

Secondly, because having a parent language is a part of being a natural language, and so people trying to emulate a natural language may want to emulate the parent-daughter diachronic process simply as a matter of principle.

Thirdly, because it's fun.

Fourthly, because diachronics lets you create a posteriori languages - conlangs derived naturalistically from real languages. This is fun and a helpful shortcut for those who don't want to have to invent everything from scratch.

And fifthly, because diachronic change is where most synchronic changes come from. (Well, it's where phonemic synchronic alternations come from, at least).

You don't actually need to think about diachronic change to produce synchronic changes for your language. However, since "good" naturalistic changes are those that look like they were created by diachronic changes, it's generally easier to just make up the diachronic changes. This also lets you 'discover' new synchronic changes you might not have imagined.

[Akubra]

Thanks for your answer, Creyeditor.
And thanks a lot for your comprehensive explanation, Salmoneus. I have read it and it has become quite clearer now, but I will need to reread it more than once to completely grasp it.
In the meantime, here is my first try at diachronic sound changes, as mentioned in my first post. I've used the searchable Index Diachronica for examples, and I've put them through the Lexurgy Sound Changer. It would be interesting to know how "natural" they are?

assimilation-nasal-plosive:
{n, ŋ} => {m, m} / _ p
{m, ŋ} => {n, n} / _ t
{m, n} => {ŋ, ŋ} / _ k

depalatalization-of-affricates-and-deaffrication:
{t͡ʃ, d͡ʒ} => {t͡s, d͡z} / _ {e, i}
{t͡s, d͡z} => {s, z} / _ {e, i}

devoicing-and-deaspiration:
{bʱ, dʱ, ɡʱ} => {pʰ, tʰ, kʰ}
{pʰ, tʰ, kʰ} => {p, t, k}

sibilantization:
# Index Diachronica: 9.1.2.1 Middle Vietnamese to Hanoi Vietnamese (north)
# <r> /z ɹ/ ‘z’ as in zoo (north), ‘r’ as in ring (south)
ɹ => z

lenition:
{p, t, k} => {b, d, g} / @vowel _ @vowel

[Creyeditor]

Thanks @Sal for veing much more explicit and well-ordered.
Regarding the original question about the vocabulary. The simple answer is that you can include both forms but you don't need to include any (except for an ortographic form). For diachronic sound changes, the form before the sound change gives etymological information and the form after it gives the actual pronounciation. For synchronic phonological processes, the form after the process gives more detailed information about pronounciation and the form before the change gives more general information on it.

Edit: The sound changes look fine to me.

[Nel Fie]

To add my own grain of salt, and in addition to what's already been said:

From the point of view of naturalistic conlanging, something to know is that we have no idea where languages originated, and what shape they took. We can only track long trails of sound changes to earlier forms of languages, which were themselves only the product of further changes from even earlier languages, neither of which we know about. And once you go back far enough (in terms of sound changes, not necessarily time), the languages become so different as to be incomparable to their contemporary forms.

So, what does this mean for conlanging? Well, my own perspective is that one could effectively say that sound changes ARE the origin of language, in so far that the distinction is meaningless in the context of conlanging, because once you have enough sound changes stacked up, you can basically turn any proto-language into anything you want. In that sense, diachronic sound changes are the generator from which a language is made, rather than the proto-lang or its phonology. I've tried to apply this attitude to my own conlanging, and it's helped a lot.

As an example, an approach that's helped me to build inventories is to create them from the outputs of sound changes rather than a selection of phonemes. For example, imagine you want /d/ in your conlang. I wouldn't say "I'll just add /d/ to the inventory". Instead, I think: "I'll add the change < t → d / V_V > to my diachronics". And I find this approach particularly helpful since like you, I want my final result to sound a certain way. You said you didn't like how your conlang sounded as much after applying sound changes - well, this should fix it since you're essentially building a list of diachronic (or even synchronic) changes that will produce the sounds you actually like, instead of forcing yourself to add changes that destroy your beautiful, pre-existent phonology.

Another detail that I think hasn't been pointed out:

Speakers can to some degree be aware of sound changes across time and space, and they can intentionally be referred to as part of regular discourse. In that sense, sound changes can essentially be part of your conlang's arsenal to express ideas. To give a real world example, think about how someone might pronounce things differently to illustrate the speech of an American English speaker, or of a British English speaker - and might yet again give a third pronunciation if they try to intentionally enunciate the word clearly and carefully.

Or simply put, sound changes can also have as simple a use as giving your theoretical speakers something to talk about.

[Akubra]

Thank you so much for your contribution, Nel Fie. Thanks to your combined efforts, I now understand the significance of diachronic sound changes and how to effectively apply them to conlanging. Previously, I was approaching it incorrectly; I had a sound inventory I liked, but the changes I applied took me further away from it, which was frustrating. Now, I realize that I can work towards it instead of away from it.

There’s something both of you, Salmoneus and Nel Fie, mentioned that particularly resonated with me:

...because diachronics lets you create a posteriori languages - conlangs derived naturalistically from real languages.

...once you have enough sound changes stacked up, you can basically turn any proto-language into anything you want.

This means that, theoretically (and perhaps practically), I could take the sound inventory of Dutch (my mother tongue) and, by applying a series of sound changes, achieve the sounds I want and like. That "sounds" like fun!

I think I'll give it a try and see what results I get!

Thanks again to all three of you for your help. I truly feel I can move forward now!

[Nel Fie]

Happy to help, if it did!

Most of all, conlanging is supposed to be fun, so feel free to pursue whatever sounds fun to you! People can get very caught up in various "do's" and "don't's", especially in regard to naturalism, but at the end of the day they only matter if you are happy with the result and enjoyed your work. And I'd hope even the most prescriptive naturalist wouldn't want you to waste your time creating something you thought was a chore to work on.

[Arayaz]

depalatalization-of-affricates-and-deaffrication:
{t͡ʃ, d͡ʒ} => {t͡s, d͡z} / _ {e, i}
{t͡s, d͡z} => {s, z} / _ {e, i}

This change confuses me. What motivates it? It'd be normal on its own, but why would it only occur before those front vowels? If anything, the presence of the front vowel would make depalatalization far less likely ─ I'd expect a shift from *t͡s to t͡ʃ before /e i/, not the other direction!

[Akubra]

depalatalization-of-affricates-and-deaffrication:
{t͡ʃ, d͡ʒ} => {t͡s, d͡z} / _ {e, i}
{t͡s, d͡z} => {s, z} / _ {e, i}

This change confuses me. What motivates it? It'd be normal on its own, but why would it only occur before those front vowels? If anything, the presence of the front vowel would make depalatalization far less likely ─ I'd expect a shift from *t͡s to t͡ʃ before /e i/, not the other direction!

Well, it turns out I didn't know what I was doing. I thought I had copied it from the Index Diachronica, but now I can't find it. I must have seen it somewhere else, but I didn't take notes and can't remember where. It's not important anymore, though, because I was doing it wrong, as you can see in my previous post, so I discarded it. Please note that I'm a beginner at conlanging and applying sound changes. Now, I'm using the Index Diachronica as my guide and recording the sources in the sound changer.

[Ælfwine]

Don't use Index Diachronica as a guide. Often it omits things or there is more to meets the eye with some sound changes.

Rather learn the environments in which some sound changes are likely to happen. Some sound changes are unconditional or motivated by a chain (i.e. d > t so ð > d because there is a gap in the phonetic inventory.) Others are influenced by what consonants or vowels are adjacent to them (i.e. the combination kj is very suspectible to becoming something like t͡ʃ).

As for yours t͡ʃ > ts is likely to happen unconditionally, but the environment you posit for it tends to have the opposite effect.

[Arayaz]

Yeah. The Index Diachronica may tell you, say, that d → t in some language, without letting on that this was accompanied by b → p and ɡ → k. Also, many of the sources it uses are questionable at best. A much better way to come up with sound changes is simply to learn more about phonetics.