Attempting to make a language where each letter has a meaning and so that each word is actually a description of itself...

for every word you need to define a single symbol.
the combination of these symbols will then form a sentence such that each sequence of symbols is a description of itself.

trivial

>what is Loglan

>what is Lojban

Attempting to have an idea composed of smaller ideas so that each idea is actually an idea about ideas. What should each idea mean? Pic related.

I'm constraining it to the English alphabet.

^

What about pictograms?
blissym.com/lesson1.pdf
If every letter has a meaning you might as well make them visual.

>Lojban and Logban use the english alphabet

Constraining it to the English phonemes as well but without digraphs

Take a look at this:
towardsdatascience.com/sentence-embedding-3053db22ea77

Basically it's a data-driven approach to map ideas, from single words to whole documents, to a vector space in some way that preserves meaning (i.e. so that the distance between the vector mapped to by "king" and the vector mapped to by "queen" will be closer together than the former and, say, the vector mapped to by "car").

Maybe you could try to partition these spaces into 52 sets and then assign to each partition a letter (uppercase and lowercase have different meanings), and define letters grouped together as acting on each other in some way (if "a" is in the general vicinity of "monarch" and "b" in the vicinity of "automobile", "ab" could mean "a monarch's automobile", or something).

This sounds really cool
thanks user

citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.157.1636&rep=rep1&type=pdf

I don't really have the qualifications to say this, but shouldn't you start with the syntax?

Assume we knew the perfect letters to use and not just in meaning, but for rhythm and pedagogy and for any other good thing that can be said about it. It would be incredibly lucky if that result could be randomly achieved.

I am assuming you want a person to use this so the primary element is the English phonemes. You then want your letters to represent these phonemes. We can set the letters aside because they are secondary.

Now, there are different ways for phonemes to be staged (even if we ignore body/kino/facial complements to language). Not every sound can exactly come right after another, but this still is not what I am trying to get at. These are more like physical boundaries for the phoneme.

I am trying to think of conceptual boundaries so the phoneme can accurately and definitively map to a meaning, but it should also be able to direct to new meanings which have not been discovered.

We'd have to be careful on classification or our language will drone like bagpipes. The same sound will repeat between more illustrative utterance. Suppose some classification is weighted too much, like if used a single sound at the start of any noun (/dʒ/) (think the j in judge), then that sound is going to happen a whole lot.

>syntax
I was thinking we could have a syntax similar to maths Reverse Polish notation where the meaning would be modified at the end

It's not phonetic tho

I'm pretty sure Hebrew does what you describe, maybe Google some examples of how words are constructed in hebrew.

I tried that
you will soon run out of letters and are forced to make more and more exact and abstract connections
at the end you petty much end up with loglan/lojban or basically written chinese

Letters should represent syllables and ka, ma and ba should be reserved for numbers in trinary (in preparation for quantum computing), ended with d. So counting from 1 to 10 would be pronounced mad bad makad mamad mabad bakad bamad babad makakad makamad. Simple, efficient, effective, practical, clear and legible. Perfect even.

Yeah it does but it isn't as logic based as would be good