So far in this series, I’ve been explaining the various kinds of sounds that you can include in your conlang. But in addition to having a limited inventory of sounds, languages also have restrictions on how those sounds can be arranged in a syllable. These restrictions are called phonotactics.
The Structure of a Syllable
First, it is important to understand the structure of a syllable. A syllable can be divided into three parts: onset, nucleus and coda. Think of these as the ‘beginning’, ‘middle’ and ‘end’ of a syllable.
The only necessary part of a syllable is the nucleus. This usually consists of a vowel, but can sometimes be a syllabic consonant. English words can sometimes be pronounced with a syllabic consonant, for example, rhythm [ˈɹɪ.ðm̩] and bottle [ˈbɒtᵊɫ̩]. However, they can also be pronounced with a schwa /ə/ before that consonant. Chinese Mandarin allows its dental and retroflex consonants to be followed by the syllabic consonants /z̩/ and /ɻ/ respectively. Any continuous sound can be a syllabic consonant since they can be extended without changing articulation. However, I don’t think it’s possible to have a syllabic stop since there has to be some kind of release to be audible. Another thing to keep in mind is that a syllabic semivowel would actually be the equivalent close vowel.
In some languages, such as Chinese Mandarin, syllable structure is normally analysed a bit differently: the onset is broken into an initial consonant, a medial semivowel, while the nucleus and coda are grouped together and called a final.
Chinese languages, in general, are very limited in which consonants can occur in the coda. This is probably due to how tone developed which I discussed in my post about tones and pitch accent.
Hawaiian has incredibly minimalist syllable structure: no constant clusters are allowed and all syllables must end in a vowel.
A good first step for creating phonotactic constraints is to define a general syllable structure. This can be expressed using a formula like this: (C)V(C). This basically means that each syllable must contain a vowel and may contain a consonant in the onset and/or coda. This should only be used as a summary and I recommend you should list phonotactic rules for clarity.
Example: English’s phonotactic constraints
English has the following basic syllable structure:
(s)(C)(l, r, w, y)V(C)(C)(C)(C)
Here is a general list of English’s phonotactic rules:
- No geminate consonants
- No /ŋ/ in onsets
- If the onset consists of a consonant cluster then:
- Neither consonant can be Affricates or /h/
- The first consonant must be a stop or a fricative
- A consonant must not be followed by a voiced stop or voiced fricative
- If a consonant is not /s/, then the following consonant must be an approximant
- No semivowels in codas
- No /h/ in codas
- If the coda consists of a consonant cluster then:
- A consonant must not be followed by /ɹ/, /ŋ/, /ð/ or /ʒ/
- A stop or fricative following /m/ or /ŋ/ must be voiced and share the same place of articulation with the nasal.
- Two stops or fricatives must share the same voicing
There are also some rules that apply at the word-level:
- /ə/ cannot occur in stressed syllables
- /ʒ/ cannot be word-initial, but sometimes borrowed words can be pronounced with an initial /ʒ/ such as genre /(d)ʒɒn.ɹə/.
There are more rules than this, but they’re a lot more specific and I want to keep it brief.
Even if your phonology is very similar to English, you can still make your conlang sound exotic by breaking the phonotactic rules defined above.
The Sonority Sequencing Principle
When choosing phonotactic rules, it helps to follow the Sonority hierarchy. This is basically a way of ordering sounds by their general loudness (sonority). There are many variations of the hierarchy, but it is generally ordered like this:
- Open vowels
- Mid vowels
- Close vowels and semivowels
- Laterals and rhotics
Also, voiced sounds would be higher on the hierarchy than unvoiced sounds of the same manner of articulation.
Usually, the more sonorous sounds (at the top of the list above) will occur closer to the nucleus of a syllable. But remember, this is a general rule not an absolute, so exceptions can occur.
When segments become more sonorous as they get further away from the nucleus, it is called a reversal. Although this sounds strange and exotic, these actually happen quite frequently in English: /s/ can precede voiceless stops in onsets; and in the coda, /s/ can follow voiceless stops.
Another kind of exception is called a plateau. This is when adjacent segments share the same place on the hierarchy. These also can occur in English: for example words such as sphere /sfiə/ or act /ækt/ which both contain two adjacent consonants of the same voicing and manner of articulation.
Affricates Vs Consonant Clusters
An affricate is simply a stop followed by a fricative of the same place of articulation. So you might think: why are they considered to be phonemes and not consonant clusters? The difference is in how they are analysed. An affricate is considered to be a single phonological unit while a consonant cluster is a sequence that can be divided into segments. The /ts/ in the word cats is identical to a voiceless alveolar affricate /t͡s/, but if you look at a chart of English’s consonant inventory (see my post on consonants) /t͡s/ is not considered to be a phoneme. One reason for this is that the /t/ can be substituted with other consonants. As /kæps/ and /kæks/ are both considered valid syllables under English’s phonotactic rules; therefore, /ts/ is considered to be a consonant cluster rather than a phoneme. On the other hand, the /t͡ʃ/ in catch /kæt͡ʃ/ is considered to be phonemic since the /t/ can’t be substituted; /kæpʃ/ and /kækʃ/ are not valid syllables in English, so /t͡ʃ/ is considered a unique phonological unit.
Another reason for this could be that /s/ in English can function as a subsyllabic morpheme, ie. it’s a segment of a syllable that carries its own meaning. Cats /kæts/ is the plural of cat /kæt/, therefore /ts/ is not an individual unit as the /s/ is removable. On the other hand, /ʃ/ does not function as a subsyllabic morpheme in English, so the /tʃ/ in catch /kæt͡ʃ/ is a phoneme since the /ʃ/ cannot be removed without completely changing the word.
I recommend that you consider what affricates there are in your phonemic inventory before writing your phonotactic constraints. If you want /t͡s/ to be a phoneme, then you shouldn’t have a rule stating that /s/ can follow any other consonant or /t/ preceding any consonant. Basically, you would want to /ts/ to be a unique combination that wouldn’t normally be allowed under your conlang’s phonotactic constraints.
I recommend taking a look at J.C. Catford’s A Practical Introduction to phonetics, which has a chapter on co-articulation and sequences.
For my conlanging post, I’m not sure where to go from here. They may be about sound change or romanisation systems. Also, I might do a post about allophones at some point. Feel free to suggest any ideas for posts to me.