As time progresses and a language is often used, sounds start to change; different phonemes are used. The words get "smoothed" like gravel at a beach or in a desert. After centuries the stones will be smooth. Sound changes, as they're called, are a major driving force of language change.
Sound changes are born every time we speak. People rarely say any word perfectly; perhaps your <t> was a little too far forward, or your <u> wasn't rounded as much as it normally would be. Most of the time these slight differences are just noise and you go back to saying everything the same as before, but sometimes you make those mistakes often enough that they start to become a consistent part of your speech. Anyone who respects or admires you — even if it's just your group of friends — will start to subconsciously copy the way that you speak and that sound change will begin to spread.
Sound change is nigh unstoppable. If you write two novels in the same setting in different periods of time using the same conlang, it's quite likely some sound changes will have happened, so you'll want to implement them.
Sound change also has no memory. If at a certain point in time there are some sounds X in words, they all will change to Y even if some of them were W a few centuries ago while some have been X since the beginning of the language.
Although it seems like sound change happens regardless of grammar, this is not necessarily true. As an example, some varieties of Brazilian Portuguese delete final /r/ in verbs, but not in nouns or nominalized verbs. The infinitive "poder" (can) is usually pronounced /poˈde/, but as a noun, "poder" (power) is pronounced /poˈder/, even colloquially. Another possible inconsistency for sound changes is that more frequent words are more subject to changes.
In the case that two words would be pronounced the same if a certain sound change happens, one of the following things can happen:
- Nothing, because the words can be differentiated by context; so they will become homophones.
- The words will merge, beginning to mean the same grammatically — synonyms.
- The word to be changed will refuse to change.
- People will stop using one or the other word, replacing it with another construct.
Common Sound Changes edit
Some kinds of sound changes are more common than others. Let's take a look at the most common ones.
Assimilation is by far the most important sound change. Assimilation is when a sound changes to become more similar to the surrounding sounds. A consonant may change to match the place or type of articulation of an adjoining consonant.
- In English, /n/ often becomes [ɱ] (labiodental) before /f/, like in <infant>.
- The word <assimilation> is an example. It comes from adding <ad-> onto <similate>. The /d/ assimilates to /s/.
Lenition is the "weakening" of sounds. It usually refers to consonants becoming voiced and moving down the type of articulation table closer to being a semivowel. It also refers to sounds that disappear altogether. Lenition is especially common intervocalically (between two vowels).
- /t/ » [d]
- /d/ » [s]
- /k/ » [∅]
- In some varieties of English, /t/ and /d/ can become [ɾ] word-internally.
Palatalization is the shifting of a consonant towards the palate. This is a common type of assimilation. Consonants can palatalize before or after a front vowel ([i], [e]) or a palatal consonant ([j]), perhaps ending up as an affricate or fricative.
- /tj/ » [t͡ʃ]
- /dj/ » [d͡ʒ]
- /sj/ » [ʃ]
- /zj/ » [ʒ]
- /hj/ » [ç]
- In English, pronunciation of the -tion suffix as /ʃən/ is the result of palatalization of a former /siən/.
- In Japanese, the pronunciation of <si> as /ɕi/, <ti> as /t͡ɕi/, <di> as /d͡ʑi/, and <hi> as /çi/ is the result of palatalization.
Velarization is a secondary articulation of a consonant where the back of the tongue is raised towards the velum. In some languages, such as Russian and Irish, velarized consonants often contrast with palatalized consonants.
- In some varieties of English, when /l/ is found in the syllable coda position, it is velarized (e.g. <wall> as /wɔːlˠ/). When /l/ is found in the syllable onset position it is non-velaried (e.g. <lawn> as /lɔːn/).
Monophthongization is the simplification of a diphthong (or triphthong) down to a single vowel. This feature was very common in Old French and Ancient Greek, leading some the diphthongs of these languages to be monophthongized. For instance, the French <ai> and <eau> are now pronounced [ɛ] and [o]; in Modern Greek, the combinations <ει> and <οι> are pronounced [i].
- In Australian English, the diphthong /eə/ (as in <air>) is often pronounced [eː].
- In some American English dialects, the diphthong /aɪ/ (as in <eye>) is pronounced [aː].
Vowels next to nasal consonants very often become nasal themselves. This is a type of assimilation. If a nasal consonant disappears, the mark it left on the vowel may remain, causing nasal vowels to become phonemic.
- Again, this is a feature very common in the French language. Consider the word <maison>, house pronounced [mɛ.ˈzɔ̃]. The <o> was nasalised because a nasal sound (namely /n/) was following.
This is the change from /z/ to a trilled /r/, which has occurred in various European languages. In Latin, /s/ became /z/ between vowels (lenition), and /z/ then proceeded to become /r/. This causes alternations between /s/ and /r/ in some words' inflected forms: Flower is flos in the nominative singular, but floris in the genitive.
You can regularise these sounds over time. Latin did this, so that original flos and honos became flor and honor, to match their genitives floris and honoris.
Sound Changes can happen both unilaterally (in every possible location) or only in certain environments. For instance, a language may lenite a particular sound, only if it follows a particular consonant. When logging sound changes, a standardized notation is used, which looks something like this:
[x] > [y] / [z]_
In this formula, the underscore indicates where our phoneme in question would be, and it can be read as "when [x] follows [z], it becomes [y]". This basic structure can be expanded for more tricky rules. For example
[x] > [h] / #[V]_
Here we are indicating that [x] becomes [h] when following [V], where [V] is "any vowel". We have also added the hash to the second half of the equation, which indicates a word boundary (either initial or final). This means we can read this as "post-vocalic [x] becomes [h] in initial syllables only."
Choosing sound changes edit
The basic idea here is that when you're making your conlang you should have in your mind a parent language (or proto-language) and a child language. The proto-language is going to be a conlang just as we have been making up until this point and should not have any history to it. The child language is going to contain all the history. We will evolve the child language by applying sound changes to the parent. The child language is the result; the language that you will present to other people, or put in your novel, or whatever other reason you conlang for.
Before you begin, you may want to have some idea of the kinds of sounds that you want your child language to have. Create a rough draft of the phonology of the child language. Once you have that, you can start trying to change the phonology of the proto-language into this child draft by selecting sound changes and adding them to a list. It takes some practice to be able to do this well, so don't worry too much if the final product isn't exactly the same as your draft.
Alternatively, you can decide not to worry too much about the final product and simply select sound changes randomly. You won't have much control over what you get, but you may get something interesting.
Once you have a list of sound changes, you will want to go through the dictionary of the proto-language and apply those sound changes to every word there.
Sound change appliers edit
You may have noticed that applying sound changes to words is quite a tedious process. To help with this, some conlangers have written computer programs called Sound change appliers that automate much of this work for you.
The original and most famous sound change applier is the SCA by Zompist.
Sound change appliers are powerful and useful tools, but they can have trouble with certain kinds of changes. They can get confused by any change that needs to happen in particular syllables, such as syllable-based syncope, or any change where the environment spans multiple syllables, such as umlaut.