Has anyone developed a complete, hierarchical ontology of “language”?

Question

If we consider whatever the phenomenon of “language” is, what might be the most immediate way of subdividing it into types?

For example, it could be broken into specific instances of languages, vs. general categories of “language”.

The former could be pretty easy to break down into language families. The latter could just be seen pretty much as the field of linguistics.

I am more interested in a hierarchical way of studying a particular language. I am especially interested in very small numbers of “children” for each concept - like binary branching, or trinary, max, to force oneself to find as many categories as possible to envelop each lower level of the hierarchy.

So instead of an obvious way of subdividing “Azerbaijani” into topics like “grammar”, “vocabulary”, etc, I am trying to consider a more strict requirement on how to subdivide it. For example, what could be a single most specifying feature of Azerbaijani - almost like the principle or rule that delimits, for a person with no prior knowledge of Azerbaijani, what Azerbaijani is like, with the most effect, from the beginning?

As of now I can only thing of something like S-V-O order, but I haven’t really proven that there isn’t something even more determining.

It would be important at this point to not just jump ahead and start listing grammatical parameters like “it has 7 cases and it’s agglutinative and it has these inflectional particles, etc.” Instead, the goal would be to feel that the next assertion about the language is somehow justified, somehow follows logically from the previous one, rather than just listing off various topics you read about.

As mentioned, if you just jump ahead to saying something like, “and here is a list of endings on pronouns that convey ‘person’”, it feels too limited - why did you choose to explain person-related suffixes next, rather than verb tense suffixes, or something else? You have to justify that the next thing you say is the next available, most general specification on that language that bears on the language as a whole. For some reason my intuition feels like we would still be on the level of sentences, and perhaps we would continue with a rule about word order, for example, subordination and coordination.

Maybe the best next topic could be any inherent property of clauses - not yet a detailing of each clause type, however.

Maybe next, some attempt at justifying what fundamentally separates different clause types - instead of saying “there are five clause types”, you are required to group things ideally in twos - so maybe you could say that there are clauses that prompt responses on the discourse level (questions), and those that do not mandate it (declarative, imperative, exclamative, exclamations).

You would still be a long, long way off from detailing various noun and verb inflection patterns. Hopefully you could then subdivide, conceptually, between declaratives (sentences which add information) vs “communicative” utterances (those that relay an attitude or intention or wish). It would not be satisfying to divide “declaratives” into something like “simple” and “complex”. Maybe I’m not able to articulate it at this moment but I think it is because it feels too arbitrary, like classifying types of ice cream into those that are colored vs. black and white, it doesn’t seem to be a meaningful way to classify in order to understand the order of principles that give something its interesting or significant characteristics. Maybe this paradigm can be made a bit more identifiable just by asking the right question, before providing the answer. Something like, “what is the fundamental characteristic of a declarative sentence - what function does it fundamentally provide? And: does that aspect have any high-level groupings or ‘types’?” And from that perspective we could say, perhaps at least illocutionary force is a good next step, because while declarative sentences “inform”, in some way, there may be a few different fundamental ways, in which they inform.

And so on.

I wish to pursue the same paradigm for other dimensions of the language, beside grammar. For example, for vocabulary, it could be interesting to consider if there is a single word or concept which is perhaps the single most general, all-encompassing concept, with no differentiability, it is so featureless - for this, I nominate the first word of the language you should learn to be “thing”. And so on. From there you seek a single way of dividing the concept into halfs; and continue specifying, on and on.

Has such a way of taxonomizing the elements of language been done, moving in a strictly, totally ordered, binary branching sequence, from the most general to the most hyper-specific?

Answer 1

This may be of interest to you. However, as far as the question you asked is concerned, no. Here's why. You say "As of now I can only thing of something like S-V-O order. It would be important at this point to not just jump ahead and start listing grammatical parameters". You jumped way ahead – what is this "S" and "V" stuff? If you want to build up knowledge of a specific language, you have to start by identifying – and justifying – some really elementary concepts. You are allowed to have some unproven a priori concepts that inspire you to make specific claims about Azerbaijani, but you have to devise the framework for proving claims.

Before you can prove that there is some pattern of word order, you have to prove that there are words. I invite you to prove that there are words in Azerbaijani. Prove that there are words in that language that translate into English as "apple, horse, blue, name, fire".

In a different forum, we would have a back and forth discussion where you would claim to have an answer, and I would deny that your claim is true – the issue is, how do you justify that claim. Before you can justify the claim, you have to understand what the claim implies. One of my lines of denial would be "That's not Azerbaijani". Another line of denial would be to deny the specific claim for a word (but I'd have to see what you presented as the word for "name", for example).

It seems that you are only interested in parameters of Azerbaijani from the perspective of typological syntax, so you can presuppose certain fundamentals. For example, you could take it to be sufficient to point to the written form of the language and say that "blue" is abı. With suitable disclaimers, you could ward off my objection that that is not "real Azerbaijani", or that it doesn't exactly translate "blue". You can entirely ignore all of the analysis that is involved in getting from speech to standard orthography, and instead empirically base your analysis on the distribution of blank-delimited letter sequences in the corpus of Azerbaijani writing. This will fail if you wanted to know about an unwritten language, but Azerbaijani has a reasonable corpus of written forms.

Therefore, you might try to devise a statistical table that tells you the probability of encountering the string at immediately before the string pişiyi, and the probability of encountering atı immediately before pişik. More generally, you would look at all permutations of three words selecting one from the two-member sets {atı, at}, {pişik, pişiyi}, {qorxutdu, qorxutdu}. Now the problem is, suppose we stipulate (since I do not know the language) that the sequences at pişiyi qorxutdu and pişik atı qorxutdu exist, and no other permutations exist – what claim about SVO order does this support? We have no proof that this sequence of letters contains any verbs or nouns, much less subjects or objects. The attempt to reach a conclusion about an extremely advanced grammatical topic fails because we didn't attend to much less interesting prior analysis. How do you establish "meaning", how do you determine that there is some shared meaning behind atı and at, likewise pişik and pişiyi, how do you justify the claim that a certain string of letters in a set of sentences is "the subject"?

Rather than first learning "thing" (which does not coherently translate into all languages), I suggest learning a ubiquitous concrete noun, like "water". Your choice seems to be based on counting up the number of "things" subsumed by a word, and finding the term whose set is the largest. So "existent" or "universe" would be the obvious starting points since "existent" includes not just entities, but actions and attributes.

And that's more or less why linguists don't engage in the exercise that you are advocating – we're more interested in concrete and testable claims.