Languages are never set in stone, and your conlang shouldn’t really be either. This is to be the first part of a recurring series on the basics of historical linguistics and how to evolve your conlang. In this first part, I will discuss the various kinds of sound changes that can occur and how to apply them to conlanging.
The obvious way to use historical linguistics in conlanging is to first create a proto-language and then derive a daughter language from it. While this approach will produce rich, in-depth conlangs; it also very time consuming. Even if you don’t want to commit to that, I still think it is worth learning about historical linguistics, as knowing and applying the basic principles can help make your conlang more naturalistic.
Method
First you must have a proto-language ready. This language can be as in depth as you like, but you don’t need to go too in-depth. About 250-500 words plus a basic grammar will do.
Next you need to plan out the series of sound changes. You could either plan the phonology of the daughter language and then work out what sound changes are needed to get there; or you could apply sound changes that seem interesting and then see where it takes you. I personally prefer a mix of the two.
Sound changes can be written as a series of rules. A rule applies across all instances where a certain condition is met. It is important to note that some words may be unaffected by a particular rule; and some sound changes such as metathesis (see below) are inherently irregular. Some sound changes are applied to all occurances of the target sound, which are called unconditional. Other rules are conditional, which means they depend on a condition based on adjacent sounds or relative position in a word or syllable.
You will also need a way of applying those sound changes. I’ve heard that some people do this manually, which I think is crazy and incredibly time consuming. The more sensible approach is to use some kind of algorithm. Some people make their own, but if you’re not into coding like me there are some free programmes you can use. I use Mark Rosenfelder’s Sound Change Applier, which is a free application that can be used in a browser. You do have to learn how to write sound changes for the Sound Change Applier, but I think it’s quite intuitive to learn.
Applications
The most obvious application for sound change is to create a language family. At first this may seem incredibly daunting to make an entire family, but it is worth the effort since languages don’t exist in a vacuum. Languages can influence each other over time in various ways. This is mainly loan words, but grammatical and phonological features may also be shared.
The other obvious application is to create dialects. This is basically the same as creating a language family but on a smaller scale. A language is a dialect plus an army and a navy, so don’t worry too much about your ‘dialects’ being too different (or your ‘sister languages’ being too similar). I would recommend focusing on changing the vowels, since vowels are more susceptible to sound change, they are more likely to vary across dialects.
Another application is to create allophones, which I will explain more about later.
Kinds of Sound Change
In this section, I will go through the various kinds of sound change you can apply to your conlang. For demonstrative purposes, I will apply some of these sound changes to my own conlang Nìmpyèshìu.
Assimilation/Dissimilation
The most important kind of sound change to understand is assimilation. This is when two close together sounds become more similar to each other. This makes a sequence easier to articulate since there are fewer phonetic differences between each sound. Assimilation is the reason why we say impossible and not *inpossible.
A sound can either obtain only some of the features of the other sound (partial), or it can become identical (total). A sound may also assimilate to an adjacent sound (contact), or to a sound that occurs later in the word (distal). Assimilation can also occur in different directions: when a sound assimilates to a sound that occurs later in the word it is call anticipatory assimilation; while assimilation to a sound that occurs earlier is called preservative assimilation. There is also mutual assimilation which occurs in both directions.
| Type of Assimilation | [hwɐmbɐu] hwambau ‘coin’ |
|---|---|
| Partial-Contact-Anticipatory | [xwɐmbɐu] |
| Partial-Contact-Preservative | [hw̥ɐmbɐu] |
| Partial-Contact-Mutual | [xw̥ɐmbɐu] |
| Partial-Distal-Anticipatory | [fwɐmbɐu] |
| Partial-Distal-Preservative | [hwɐmpɐu] |
| Partial-Distal-Mutual | [fwɐmpɐu] |
| Total-Contact-Anticipatory | [hwɐbbɐu] |
| Total-Contact-Preservative | [hwɐmmɐu] |
| Total-Distal-Anticipatory | [bwɐmbɐu] |
| Total-Distal-Preservative | [hwɐmwɐu] |
While there is a tendency for sounds to assimilate, sequences with too many similar sounds (such as a tongue-twister) may become more difficult to pronounce. In these situations dissimilation, the opposite of assimilation, may occur. In Nìmpyèshìu the word emnìm [ɛmˈnɪ̂ːm] ‘telepath’ has a lot of nasal consonants, so one of those nasals could become a stop resulting in something like [ɛbˈnɪ̂ːm].
Weakening/Strengthening
Weakening (also known as lentition) refers to a set of sound changes that can be broadly defined as a decrease in obstruction of the airstream.
| Strong | Weak |
|---|---|
| Geminate consonants | Single consonants |
| Stop | Affricate |
| Afficate | Fricative |
| Fricative | Approximant |
| Stop | Taps, flaps and trills |
| Taps, flaps and trills | Approximants |
| Oral consonants | Glottal consonants |
| Non-nasals | Nasals |
| Voiceless consonants | Voiced consonants |
| Lateral Approximant | Semi-vowel/Closed vowel |
Weakening can occur almost anywhere, but it most commonly occurs in-between vowels.
As the table above shows, there can be multiple possibilities for weakening a consonant. For example the Nìmpyèshìu word adi [ɑdɪ] ‘destruction’ could become either [ɐd͡zɪ], [ɐrɪ], [ɐɾɪ] or [ɐnɪ].
The sound changes listed in the table above may also occur in reverse. This is called strengthening or fortition. Strengthening is less common than weakening but is still rather frequent. This is most likely to happen in stressed syllables, but can also occur in other situations. Another common example of strengthening is tendency for consonants at the end of a word to lose voicing.
Deletion/Epenthesis
Deletion is obviously the complete loss of sounds. It can be considered the ultimate form of weakening. Certain sounds such as the glottal fricative [h] and the schwa [ə] are particularly likely to be lost. For example, if an intervocalic [ɦ] is deleted, then hlàiha̤sle̤ [ˈhlɐ̂ɪɦɐ̤́slɛ̤] ‘to use telekinesis’ will become [ˈhlɐ̂ɪ.ɐ̤́slɛ̤́].
A particularly interesting form of deletion is when a vowel (usually short and unstressed) is lost in the middle of a word. This is called syncope. For example, a word like anarain [ɐnɐɹɐɪn] ‘brother’ may become [ɐnrɐɪn].
The opposite of deletion is epenthesis, which is when a new sound is added. This can happen for various reasons: including breaking up an unpleasant cluster; or making the transition between two sounds easier. For example, a [t] could be inserted between the cluster [ns] in linsan [lɪnsɐn] ‘headquarters’, becoming [lɪntsɐn]. This basically makes the transition between [n] and [s] easier. For [n] to transition to [s], the gap between the tongue and the alveolar ridge must close, the vocal folds need to start vibrating and the velum needs to be lowered in order to redirect air through the nose. The inserted [t] acts as an intermediate state, where the tongue touches the alveolar ridge but the velum has not been lowered yet.
Fusion/Unpacking
Fusion is when a sequence merge together resulting in a sound with phonological features from each sound of the former sequence. This could be interpreted as a contact assimilation followed by deletion. If I apply fusion to hwambau [hwɐmbɐu] ‘coin’, then I could get [xʷɐmbɐu], the [xʷ] being a fricative like the [h] while having the same place of articulation as [w].
The opposite of fusion is unpacking, where a sound breaks up into a sequence of sounds. This usually happens to more complex sounds, such as those with secondary articulation.
Chain Shift
Vowels are more likely to undergo sound change than consonants. Vowel change is also more likely to occur unconditionally. A common phenomena that occurs with vowels is called a chain shift. When a vowel undergoes an unconditional change, it leave a gap behind. Another vowel will then unconditionally change to fill the void. This then leads to a chain of vowel shifts.
Hiatus Avoidance
Vowels are particularly unstable next to a hiatus, which is syllable break that occurs between two vowels. Fittingly, the word hiatus /haɪˈeɪtəs/ and also the alternative name diaeresis /daɪˈɛrɪsɪs/ both contain an example of this. Various kinds of sound change may occur in order to remove a hiatus:
- Glide formation is when the more closed vowel becomes a glide (also called a semivowel).
- Glide insertion is similar to glide formation, but a glide is inserted instead of replacing one of the vowels.
- Glide strengthening is a further sound change that can be added to the above
- Fusion, where the two vowels combine in to a single vowel with the properties of both.
- Deletion of the weaker vowel.
In Nìmpyèshìu there is already a rule for avoiding hiatus, which is to add a /ɹ/ in between the two vowels. So to demonstrate various kinds of hiatus avoidance, I will use the result of the example for deletion:
| Sound change | /hlɐ̂ɪ.ɐ̤slɛ̤/ ‘to use telekinesis’ |
|---|---|
| Glide formation | /hlɐ̂jɐ̤slɛ̤/ |
| Glide insertion | /hlɐ̂ɪjɐ̤slɛ̤/ |
| Glide strengthening | /hlɐ̂ɕɐ̤slɛ̤/ |
| Fusion | /hlɐ̂ɛslɛ̤/ |
| Deletion | /hlɐ̂ɪslɛ̤/ |
Metathesis
Metathesis is when two sounds (or sometimes sequences) switch places. For example, hónmi [ˈhǒːnmɪ] ‘to greet’ could become [ˈhǒːmnɪ]. Unlike previous sound changes I have mentioned, metathesis is inherently irregular.
Hapology
Hapology is the deletion of a repetition of a sequence. It hard to find potential examples in Nìmpyèshìu, but gàgà [ˈgɐ̂ːgɐ̂ː] ‘inside’ could become just [ˈgɐ̂ː]. Hapology is the reversal of reduplication, which is a morphological process that carries some meaning. So hapology may only happen if the information expressed by reduplication is no longer significant.
Allophones Vs. Phonemes
A language’s phonological inventory consists of phonemes, which technically refers to a group of sounds which are considered to be the same. A phoneme may have multiple realisations, each variant is called an allophone. You may have noticed that in English (at least in some accents) the l in cool, [ɫ], is different from the l in light, [l]. The former being velarised and the latter having no secondary articulation. Since these two sounds are similar and are never contrasted, they are both considered to be allophones of the phoneme /l/.
The rules in which allophones occur are determined like conditional sound changes. In English, [ɫ] only occurs in the coda and voiceless stops are only aspirated if they are not part of a cluster.
It also important to bear in mind whether a sound change produces a new phoneme or just a new allophone of a phoneme. If you want to make a new phoneme, then you have to make sure that it can be contrasted with all other phonemes in the language.
The most basic way to create new phonemes is to take an allophone of a phoneme and then remove its condition. In my conlang for example, [ŋ] is an allophone of /n/ when it occurs before a velar consonant [k], [g] or [w]. But if a consonant was lost when preceded by a nasal, then [ŋ] would become phonemic. For example: slingen ‘hole’ [slɪŋgɛn] would become [slɪŋɛn]. Since /slɪŋɛn/ can now be contrasted with /slɪnɛn/ (although it isn’t listed in my dictionary, it is phonotactically valid so it could exist in theory), /ŋ/ is now a phoneme.
Also, there is way that phonemes can change without the condition disappearing. If a conditional sound change causes a phoneme to become another pre-existing phoneme, then the result would be considered the latter phoneme rather than allophone of the former phoneme. Let’s say a front vowel such as /ɪ/ causes /t͡s/ to become /t͡ɕ/, then tsi̤m ’round’ /t͡sɪ̤m/ would become /t͡ɕɪ̤m/. This is considered to be a change in phonemes since /tɕ/ is a pre-existing phoneme.
What is interesting about that previous process is that it can used to produce morphophonological rules. For example in Nìmpyèshìu, there is a rule that states when two words are compounded together if the first word ends in /n/ while the second begins in a /n/ then the second /n/ will become a /d/. If I had rule where the plural marker nèm was suffixed to the end of a word, then that previous rule would may be applied to words ending in /n/ but not to others. So nau /nɐu/ ‘ear’ would become naunèm /nɐunɛ̂ːm/ ‘ears’ while naun /nɐun/ ‘snake’ would become naundèm /nɐundɛ̂ːm/. There are now two forms of the plural suffix depending on the structure of the stem. I could go on all day about this, so I might go in-depth with this in a future blog post.
Afterword
If you’re interested in reading more about sound change, then I highly recommend Trask’s Historical Linguistics.
I would like to thank those who voted for the topic of this post on twitter. For my next post I plan on doing another conlang review, so there will be a poll on my twitter account for that.
Also I have a Ko-fi account now. So if you like my blog posts and want to support me, you can do so there.
Stephen Escher 