Matthew McVeagh | my conlangs
A language with trivocalic roots, like the triconsonantal roots of Semitic but using vowels.
A language idea that occurred to quite a few conlangers – many languages of the Afro-Asiatic family, especially the Semitic branch, structure words with roots made up of several consonants differentiated in inflection and derivation by interspersed vowels; would it be possible to switch the roles of consonants and vowels in such a structure?
This general style of morphology is best called 'transfixation' (other often used terms like non-concatenative or templatic refer to wider phenomena which include this one but also others). In Semitic languages the transfixes are the vowels (plus zero phoneme elements and affixes), the roots being expressed in multi-consonantal sequences (generally three, with some two or four). In an equivalent trivocalic language a sequence of vowels would form the roots and the transfixes would be made up of consonants (plus the same zero phoneme elements and affixes, probably).
In order to make this work at least as well as Semitic languages Trivocalic would need about as many vowels as Semitic has consonants. This is in order to get enough possible phoneme combinations to make a decent root vocabulary size. The only two alternatives are (1) to use a vowel sequence longer than three to multiply to even more combinations, and (2) to structure the lexicon on the basis of few basic roots and express greater complexity by means of dependents and qualifiers.
It's an open question whether as many consonants would be needed in this language as in a 'normal' language (including Semitic ones). Since the consonants would be performing the same role as vowels in Semitic-style consonant-root languages, you could have as few vowel phonemes as they do. And in the case of Arabic, that's just three.