I am very confident my implied premise is valid. There is no sense in trying to construct a language without a complete list illustrating the semantic relationship of the anticipated words.
I have not invested any time into understanding the modular construction of sound that I seem to be seeing as the focus of the questions posted here, but it would seem that the number of modular sounds might equal to the number of items in the collation format for the ontology.
Seem reasonable???
Constructing an ontology first is one approach to language construction, and it leads to so-called philosophical languages or a priori conlangs. This approach was especially popular in the 19th century, before more natural conlangs like Volapük and Esperanto appeared.
Naturalistic conlangs are most often constructed without an explicit ontology and have semantic irregularities of all kind.
We can read ontology a little more broadly, but it's hard to do so without getting into worldbuilding. Generally the division between meanings (e.g. bend vs. fold vs. break, in French your knees don't bend, they fold, also the love-like distinction of English) can be considered of an ontological nature, and quite often these are strongly informed by the specific cultural history that lead to them
The distinction between "cooking" and "baking" derives from how historically in Europe these were the remits of different people working in different places. It is quite artificial when you really look at it, and other languages use the same work for the heating of bread (or bread equivalent) and meat.
An ontology is useful to figure out where concepts are classified by the speakers (e.g. what counts as a flower, what counts as art, what counts as crime, as a bird, as a bug?). Even on a grammatical level (what is dubious or unreal enough that it qualifies for the subjunctive?) Otherwise, you're really just relexifying your mother tongue.
