Aleks wrote:I only know of some I don't want. It would be tedious to figure out what it is not allowed.
If you don't allow onset-clusters, that's a restriction against pairs of consonants appearing in the same syllable-onset;
if you don't allow coda-clusters, that's a restriction against pairs of consonants appearing in the same syllable-coda;
if you don't allow diphthongs or longer polyphthongs, that's a restriction against pairs of vowels appearing in the same syllable-nucleus.
Do you allow hiatuses? (An open (vowel-final) syllable followed immediately by an onsetless (vowel-initial) syllable in the same word.)
One naturalistic and realistic thing to do is rank the phonemes by sonority. (How many ranks there are, and what they are exactly, is language-specific in detail, though there's some general consistency from language to language.)
Many natlangs have a rule that in a syllable-onset sonority must not decrease, or must increase, or even must increase by some minimum number of ranks (for instance, the next consonant must be at least two ranks more sonorous than the previous consonant in the same onset).
And many natlangs have the corresponding rule for syllable-codas; sonority must not increase, or must decrease, or even must decrease by some minimum number of ranks (for instance, the next consonant must be at least two ranks less sonorous than the previous consonant in the same coda).
If you have either onset-clusters or coda-clusters or both, and you have both voiced and voiceless consonants, you might have a rule that if both a voiced consonant and an unvoiced consonant occur in the same syllable-margin, the voiced consonant has to be closer to the nucleus than the unvoiced consonant.
Do you have, or want, any rules that a certain consonant cannot be immediately followed, or cannot be immediately preceded, by a certain vowel, in the same syllable, or in the same word?
TTBOMK that would be rarish in natlangs, but not completely unknown.
For pairs of pulmonic egressive consonants, try going by rows and/or columns of the IPA chart.
Which {manners-of-articulation | places-of-articulation | phonemes} cannot be immediately followed by which {manners-of-articulation | places-of-articulation | phonemes} in the same {syllable-onset | syllable-coda | word} ?
If you don't choose "phonemes" both times, you'll eliminate several pairs at once.
Similarly, for consecutive pairs of vowels, you might go by closeness/openness (height) or frontness/backness or roundedness/unroundedness or peripheral/interior.
Do you want to disallow certain {heights | frontnesses | roundednesses | peripheralnesses | vowels} to be immediately followed by certain {heights | frontnesses | roundednesses | peripheralnesses | vowels} in the same {syllable | word}?
Do you have non-pulmonic (e.g. glottalic or velaric) or ingressive airstream consonants? (Ejectives are glottalic egressive; clicks are velaric ingressive.)
If so, you might require that they appear only in syllable onsets, or only as the first phoneme in a syllable, or even only as the first phoneme in a word.
Those are naturalistic and realistic IIANM.
If you have certain pairs of consonants that you won't allow in the same word, but the first of them can be the last sound in a syllable-coda and the second can be the first sound in a syllable onset, you have to figure out what to do if such a pair would come up in otherwise-normal lexicogeny/rhematopoeia/word-coining or morphology.
Choices include:
* insert some epenthetic sound between them
* metathesize them (switch their order), if that results in a legal pair
* mutate one of them (word-internal sandhi)
* drop/elide/omit one of them.
You'll probably have different restrictions for different locations; and you'll probably have different restrictions depending on what type(s) of sound(s) they are.
I hope that helps to at least start you off.