Principles:
-Should be easily interpretable with computers using camera, mircophone or directly from character input
-humans should be able to read this fast and with less errors
-in case written with altering surface texture, it should be also easily readable (for blind, or those want to have their eyes a rest)
-Some level of cognitive optimisation. For example, the "sentence" would be built up, so that it is gradually becomeing clear with every word what kind of mental functions should activate. This is like: there is a saparate part responsible for denoting meant validity period of information, so in this way one knows right away if it is necessary to concentrate somewhere on the surrounding, or use imagination, or use snythesis, higher amount of analisis after this period..
it facilitates logical thinking, expression, using three state logic (true, false, undefined)
-Its easier to sign then latin letters (with your hand or body parts)
Structure of a relation
(specifiers: sp resp, sp val, sp loc; relation body: relbdesc -> relcenter -> relend)
----------------
sp resp {specifier of responsiveness}:
would determine if the transmitter intends respons from receivers (completion [questions], completion or action [!], omitted means general information
----------------
sp val {specifier of meant validity}:
made up of sub parts: first specific words describe time past, or future, or present. thes time identifiers have options to denote finished or unfinished expression, and continious or general, and single occurance, numbered occurance (just put a numerical determiner), or cyclical undefined occurance. then you can give a relation body, so that you define time as the occurance of something (as mostly time is defined this way).
You can put multiple such blocks connecting them with "and" ("and" is implied, so no words for it, just put many).
ommiting all together just means previous time specification holds
----------------
sp loc {specifier of location}:
wells, here just comes a location specifier word, then a relation body. You can ommit both, but you must have relation body for location specification if you have location specifier word.
###################
Main relation body is not ommited except if you answer questions with true, false, or undefined question answering words, or with numbers.
--------------
relbdesc {relationbody descriptor}:
never ommited when there is relation body. There are options, only one can be picked, mainly it desribes the compositon of relcenter (*listed in relcenter section). optionally negation (not the same word as false) can preced this sign
-------------
relcenter {center part of relation body}:
it contains a focus part and optionally after this (starting with a cause identifier word, another "node" which was the cause of relation).
focus part depends on reldesc, if it has mutliple parts some are the main focus part.
options:
1) nodesup relationword nodeinf [relationword nodeinf]: main focus is on all of the relation (so that is mainly on all the relationwords), relation word defines if the first(previeous) is taken as superior over the second(next) inferior, or they could be equal. at least two nodes separated by a rleationword must be present
2) nodeinf relationword nodesup [relationword nodesup]: main focus is on downermost nodinf being equally connected (or first node if equal relation). "superior inferior means mainly direction of relation". also, there is another version, where the order is also this way, but the main focus is instead on specified number of nodes counting left from the leftmost directed/hierarchical relationword, not just on the leftmost (this version requires another word after reldesc, or a numerical descriptor). Both supcategoris are similar to previous, so at least it have to contain two nodes separated by relationword
3) relationword: there is only relation word, it is meant to be an answer to a question where someone asked you to complete a relation where the relationword was unknown
3) nodesup: you answer a question giving the asker the unknown superior node (or in case of expressing one way that something has a propertie)
4) nodeinf: you give the asker the inferior node (or in case of expressing one way that something has a propertie)
5) blank: this just marks that you are talking about relations of any sort
*nodes usually are either an object (or numerical descriptor alone, means its is the object, but only non relative numerical descriptors in this case) descriptor word, or a relation, plus followed by a property descriptor word (which can be derived from object words, and also object words can be derived from property words, just changing the first sound) so one way to express you have fixed the car well, then you would open a relation which only has a node then the node would be a relation so you open another relation by describing finished past, then you give (with reldesc that) the relation has three parts and the focus is on relationword, then you put you as the nodesup, "to fix" as the relationword, and car as nodeinf then you put a statment anding kind relation ending word, then you jus put another porperty word (good). So here you see why the focus is important, mainly in case you would give relations as nodes of other relations (that is subsentences to some degree).
nodes can also have numerical descriptors following them (digits would use the same char, as it would be used in math equetions). further description of numerical descriptors ojbects and property descriptors come later.
also, Nodes and relation words can be listed by logical connective marks ("and" is not marked, its implicit, others are inclusive or, exclusive or, uncertain or) Imlication (uncertain implication only) and equivalance is a relationword. Nodes can be preceded by negation (which leads to uncertain nodesup or uncertain nodeinf) so bascily you can easier ask in this language that something is not done by this but by whome is it done? (see details on relend)
relationwords can also be modified by those logical connective words.
-----------
relend {relation ending}:
always given, it denotes you intend action (relation be implemented) or answer from the person hearing it(this way what to answer is given in relend, its either evaluation, adding something, or adding something which is undefined there but noted[in realbody.. see description later on objects property words]), or you just give information.
reldesc and relend is used as a bracket for logical statments basicly.
so only node (containing object) or only relationword reldescs can be used to implement bracketing in large listing of logically connected objects for example (when objects are used) such as if i would like to express: personX and personY and not (personZ and personV and personT) (here the brackets are replaced first by a reldesc word, describing onlynode relbody, and the ending with a general statement ending. if this bracketing is not implemented, you would have to write personX and personY and not personZ or(inclusive) not personV or(inclusive) not personT.
##################
Numerical descriptors, property-words and objcet words.
All the word types begin with a special character/sound, relations begin with a vowel sound, everything else begins with a consonant sound.
numerical descriptors:
This is a separate wordgroup, it stands after object word, and either before or after property-words. A numerical-descriptor begins optionally with the sign specifing the numbersystem up till base 20 (the default is base 20), then goes a specifier, which would tell if it is a serial number, just a number. In case of serial or simple number a sign would follow, and then goes the number descriptor. The number descriptor can be given two ways, it is another sign, telling if you give it in a normalised form then you would use this same sign before the exponent. The serial number is a relative "property" so its specifiercharacter is similar in sound and shape to those which would be the second character of all relative property-words. It makes one remember to give (and grammatically have to give) point of reference. (which is one character denoting previous, or more the same way as it will be with relative properties [serial number is a one dimensionally relative "property"])
There are also uncertain numerical descriptors, but denoting the term "any/all" or "some" or "many"
for giving undefined number values, basicly labels, a couple of characters (possibly no more then 1 times number of vowels) would be reserved for this purpose.
------------------
Propperty-words:
Propperty words describe the property of something, a color a shape or soemthing like these (better, is a different property word then good, however it might have some more similarity). Property words and objectwords just by swaping the first consonant (the first character would be swaped this way, a character is a cons+vow), so this way its easier to describe somethig as "box like" or things like that. Some properties are relative they have dimension, they contain a character which is the first char of joining word for listing "points" of reference (this is to make you remember, that you have started to write those joining words, so specifing points of reference), than after this goes just a simple number character specifier (inside this word, without numerical-descriptor begining character), which specifies the dimension of relativity (above, below is three dimensionaly relative, towards away is one dimensionaly relative and things like this), then comes every other character of the property-word, after which must have nodes listed connected with the referenceppoint connective words (the first char of which is the second char of relative propert words).
-------------------
Structure of property-words and objectwords:
They begin with a consonant which would tell if its a property or an object word. The second is a vowel, which would describe cognitive cathegory (so far 10 categories exist): absolute concept (color word, etc), relative concept, unanimated/continious object, many state non automated object, semi automatic animated object, automated animating object, automated quick-adaptive object, undefinedmaincategory, name for automated objects (like name for people), name for other objects (including places, cars or such).
If no more characters given, then it would mean uncertain object/property within the main category (such as someone, something, "some multi-state object", "some whatever" so there are more possible undefined then in any other well spread human languages [i would guess]).
After this characters for words should be picked, by writing a program which reminds the operator to check similar words if they describe something similar, and if words are relative or not (check dimensionality of every relative words). So similar terms would have similar writing.
#################
Script/speak
General: A consonant is always followed by a vowel, vowels are on the begining of the relation-words. spaces are not necessary, has no meaning, since every category of word begins with a special sound. So this way the utilised syllables would also not cause any merging of sounds.
Voiced and unvoiced consonants mean the same (maybe this featur later would be droped if not enough characters).
Here are the syllables with international notation ( some link to listen http://www.paulmeier.com/ipa/consonants.html):
consonants:
p/b, s/z, t/d, ʙ, r, f, k/g, v (not this but in the flap or tap row the similar looking), c/ɟ, m, n, ɲ, h, l, tʃʼ (maybe not this but, it sound like "T" in the line "how was i supposed TO know" in they song Baby one more time by Ten Masked Men), tsh [not ipa, it sounds like t in picture, could be also replaced by j as in job]
vowels (ipa):
i, ʌ, ɤ, y (ü), ɯ, ɛ, o, a, e (é), u
u is never used before a consonant or alone.
Two scripts will exist, for now there is only one done, which writes consonants on top, vowels bottom, and writes chars from left to right always and lines from top to bottom always.
Another one charset would be ambitextrous, and it writes continiously, with a mark denoting begining of sentence (write order), then next line begins where this one ended. the pages would act the same. odd pages (page numbering starting from one) begin on top.
later i post the charsets, if needed.
None.
Currently Working On: 
