Shembaba Phonology
Originally, I had wanted to make my first post Lexember post about Shembaba's classifiers. But because I don't wanna keep transcribing all the examples in IPA I thought it might be good to give you Shembaba's phonology first.
I'll give you a more detailed version first but feel free to jump to 'Summary' if you only want the super short version to be able to read Shembaba examples in future posts.
Consonants
Consonants are mostly written with the easiest possible form of their respective IPA character (i.e. no diacritics or special characters).
[ʃʲ] and [ʒʲ] are mostly allophones of /s/ and /z/ before front vowels and are transcribed as <sh> and <zh>. Sequences like /siV[+back]/ and /ziV[+back]/ as well as loan words make [ʃʲ] and [ʒʲ] at least marginally phonemic. Also, in consonant clusters, [s, z] will only palatalise if the preceding consonant also is coronal.
[l̪] and [n̪] can become alveolar [l] and [n] before front vowels for most speakers and they some is true for [d̪, t̪] > [d, t] for some speakers
Geminated consonants are phonemic and simply get doubled in the romanisation.
[ɾ] is pretty much in free variation with [r] (the latter being especially common for emphasis and pre-tonic syllables)
Vowels
Vowels have a phonemic length distinction. A macron is used on long vowels in the romanisation – except [øː] which, somewhat unsatisfyingly to me, is <ő> but if you read my Lexember 4 post you know that I've made my peace.
/e/ is pronounced [ɛ] in unstressed syllables for most speakers and has even fully become [ɛ] for a few speakers. /eː/ is always [eː].
before vowels is simply to write [j] - I just felt whimsical and wanted to do it the old-fashioned way to make it more archaic-looking – this will not happen when preceded by sibilants /s/ and /z/ which will absorb the [i] into [ʃʲ] and [ʒʲ].
Stress and Prosody
Shembaba is mora-sensitive. What counts as a mora? One vowel is one mora. A diphthong or a long vowels constist of two morae. Syllable codas don't count as moraic except for geminate codas. This has an effect on stress as well as the distribution of long vowels and geminates.
Stress usually falls on the penultimate syllable. However, the antepenultimate will be stressed instead if it is heavy (= has two morae) and the penultimate isn't. The only time the final syllable is stressed is for imperatives and then it’s marked with an acute accent.
Stress the final syllable if it is a verb in the imperative mood (odārió! 'run!')
Stress the antepenultimate if it contains a long vowel or ends in a geminated consonant and the penultimate doesn't (uzhȳlidme 'filth')
Stress the penultimate (uzhikas 'coal')
Furthermore, there can’t be two syllables with a long vowel next to each other unless they’re the last two syllables of a word. In all other circumstances the less stressed syllable gets a short vowel:
/as.tyː.liː.ɾeː.ŋaːn/ > [as.tyː.liˈɾeː.ŋaːn] <astȳlirēngān> x - xx - xx - xx - xx > x - xx - x - xx - xx (the x's count morae) climb-ABIL-skill-GEN 'of the climbing skill'
/as.tyː.liː.reː.ŋaːn.saːn/ > [as.ty.liː.ɾɛ.ŋaːn.saːn] <astylīrengānsān> x - xx - xx - xx - xx - xx > x - x - xx - x - xx - xx climb-ABIL-skill-GEN-AGN-GEN 'of the professional climber'
If two syllables with a long vowel appear next to each other only keep them when they are the final two syllables of the word
Otherwise, counting from right to left in disyllabic feet, keep only every other syllable long (iambic style), with the last syllable being extrametrical, so that there’s never two long syllables next to each other except the final two.
There’s a weaker constraint that there can’t be two syllables with a geminated consonant next to each other. To resolve, the syllable with higher stress ordering (same as above) wins out and gets to keep its geminated consonant.
→ If two syllables ending in a geminated consonant appear next to each, counting from right to left in feet, keep only the second syllable’s geminated coda (iambic), with the last syllable being extrametrical.
This rule, however, can be broken by inflection (but not derivation).
Summary
To read Shembaba's romanisation you only need to follow these short rules:
Letters correspond to their IPA values with the following exceptions: [ŋ] is <ng>, [ʃʲ] is <sh>, [ʒʲ] is <zh>, [ø] is <ö>, [øː] is <ő>, [β] is <v> and [j] is pre-vocalic <i>.
Coronal consonants usually are dental.
Long vowels are written with a mācron exept [øː] which is <ő>.
Double consonants transcribe gemination. In digraphs only the first element is doubled – e.g. enngēvro 'autumn' is [ˈɛŋˈŋeːβ.ɾo] but whekönggu 'peach' is [ʍɛˈkøŋ.gu]
Stress always falls on the penultimate syllable except when the penultimate syllable is light and the antepenultimate syllable is heavy (= has a long vowel or geminate coda) or in imperatives when the final syllable is stressed with an ácute accent (odārió! 'run!')











