One of my works-in-progress is a language called Proto-Oreadin. As the name suggests, it is a proto-language, which eventually I am hoping to expand into a family of languages. Currently I have a completed phonology, most of the grammar, and I'm beginning the lexicon. My plan to create these daughter languages is by using the zompist.com sound change applier app to evolve the words of my language into multiple daughter languages, and then apply semantic shifts, borrowings, and other pieces of worldbuilding context—but that's a separate issue.

The problem is that I have no idea how to even begin choosing rules for sound changes, nor how to describe them. I've looked at examples of natural languages, but I'm not sure what's going on with the notation, nor have I been able to identify any patterns that will help me create realistic sound changes for my languages.

  1. How are sound changes notated?
  2. What are the general categories of sound changes?
  • 1
    I don't see how this can be made less broad. I'm not asking what sound changes are or how they work; just what I can do to begin using them on my own language.
    – CHEESE
    Feb 9, 2018 at 2:13
  • @curiousdannii I'm simply asking people to inform me on correct notation of sound changes and common patterns in real examples. That doesn't seem too broad to me. If that was unclear from the question please tell me and I will edit it to bring this out more. I think both answers have done a good job answering my question without leaving too much out.
    – CHEESE
    Feb 9, 2018 at 4:22
  • Hope this edit helps :)
    – curiousdannii
    Feb 9, 2018 at 7:13
  • Two different questions only mildly related = too broad.
    – Jan
    Feb 9, 2018 at 7:55
  • David Salo's book A Gateway to Sindarin contains a reconstructed sequence of 248 changes from proto-Eldarin to late Sindarin, which might be useful as a collection of examples of such shifts as well as illustrating notation. Jan 24, 2023 at 23:19

3 Answers 3


To answer part a), the basic syntax of the notation goes like this:

[before] > [after] / [context]

The part after the slash gives the situations in which the sound changes occur. For example, the hypothetical spirantization of [b] to [v] intervocalically could be notated as follows:

b > v / V__V

This rule indicates that b becomes v when surrounded by a vowel (V) on each side. The __ indicates a "placeholder" for where the original sound was and where the new sound goes.

Some other common notations are # for word boundaries and [+feature] or [-feature] to indicate the presence or lack of a certain feature respectively. For example, final-obstruent devoicing in e.g. Russian could be

[+stop][+voiced] > [-voiced] / __#

where the __# means that the stop in question (__) is at the end of a word (#).

As for b), the best way to get an idea for what kinds of things generally happen in languages is honestly probably just by reading about lots of different languages and seeing what tends to happen. Some of the most common general processes are:

  • Lenition, probably one of the most common sound changes, in which a "softer" sound becomes a "louder" sound. (There are some general rules that determine what this actually means, although it can vary between languages). A few examples are intervocalic flapping in many English dialects (/bɛtər/ → [bɛɾɚ]) and the pronunciation of /b, d, g/ as fricatives or approximants, roughly [β, ð, ɣ], in most contexts in Spanish.

  • Assimilation is another frequent sound change. This occurs when features of a given phoneme are influenced by the phonemes around it. For example, in English and Spanish, /n/ becomes [m] before bilabials and [ŋ] before velars (e.g. <rainbow> [ɹ̠ʷejmbow], <concordar> [koŋkoɾðaɾ]).

  • Similar ideas are umlaut and vowel harmony, in which vowels change depending on other vowels in the word. Remnants of Germanic umlaut can be seen in English, e.g. foot/feet, and vowel harmony is prominent in e.g. Turkish, where all of the vowels in a word have to harmonize (for the most part) so that affixes have several different forms (for example, the genitive suffix is [in] after front unrounded vowels, [ɯn] after back unrounded vowels, [yn] after front rounded vowels, and [un] after back rounded vowels).

While there are general patterns, it's also important to note that some sound changes seem completely illogical and random, and just... sort of happen. The most infamous examples are the ruki rule, where [s] became [ʃ] but only after [r, w, k, j] (?), and strangest of all, Proto-Indo-European *dw to Armenian erk (???). So you really have a lot of room for creativity here, while still being able to be naturalistic.

  • 1
    Just as a comment, umlaut and vowel harmony are really just subsets of assimilation. Feb 9, 2018 at 12:48
  • *dw to erk was not just one change, of course. I think I've seen a reconstruction of the sequence of shifts, most of them not out of the ordinary. (Wish I knew where I'd seen it!) Oct 9, 2018 at 20:32
  • @AntonSherwood I don't think the intermediate steps have ever been reconstructed, but as far as I'm aware, *d became r (which isn't unusual), the labiovelar glide *w turned into velar k, and e was an epenthetic vowel inserted for phonotactic reasons.
    – Doorknob
    Oct 9, 2018 at 20:59
  • @AntonSherwood this article has one such proposal: academia.edu/6375253/…
    – Tristan
    Jan 24, 2023 at 13:45

I guess this forum really isn't set up for the kind of intense assistance & interaction you really need. Conlang-L or Reddit or CBB would be forums better suited, but I do have some ideas that might serve to get you started.

Since it seems like you've got a handle on the basics and are really asking for a directional nudge, I'd suggest the following:

  • Make a short list (maybe a dozen) of Proto-Language words and sort them by various characteristics: words that begin with a voiceless stop; words that contain a voiced stop followed by a syllabic liquid; words with an accented long vowel root.

  • Pronounce one of the words over and over and try "permutations in the phonetic neighbourhood" like CWADRUT CWADRUT CHWADRUT CHWADHRUT CHWAAHRUTH HWAARRUS HWAARUH HAWARƏH HAWWAR

  • Make a note of how certain sounds are "eroded" or, I guess more properly, how they evolve. [kw] relaxes into [hw]; [dr] > [ðr] > [r]; as the final syllable weakens, the accent shifts towards the word head; as the accent shifts towards the word head, medial long vowels become shortened.

  • Sit back in astoundment how you've just discovered several sound change rules!

If you don't like that progression, try another one.

Kind of a simplistic exercise, but it's a good method to start. Eventually, other rules you discover are going to intersect and sometimes abrogate another law. Sometimes two parallel dialects run along slightly different tracks.

Sometimes dialects are "behind the times" when compared to others, and this will launch you into the wonder world of diachronics. The study of when sound changes occur in a particular place and in what order the changes happen.

And of course, you're considering a whole family tree, so you're going to need to do this exercise multiple times. You might find that two or three daughter lineages kind of align in certain ways. For example, *cwadrut yields hawar in language A and qewarro in language B and kawrs in language C but shwuntz in language D and shhwandaras in language E.

As you examine other words, you notice that ancestral [k-] remains [k] or becomes [h] (both palatal sounds) in several languages, all of which are at the eastern end of this proto language's continuum; while most of the languages where ancestral [k-] has become [ʃ] are all in the west. A nice geographic split (that may or may not actually mean much), much like that which exists within Indo-European languages.


Stolen from this answer by sumelic on linguistics: There is searchable version of the Index Diachronica that allows to search for some sounds and explore postulated and observed changes of that sound. The notation used is the same as in @Doorknob's answer above.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.