In order to develop a system of speech, it is necessary to understand what kind of sounds a larynx is capable of making in the first place. This would take you deep into the study of phonetics along with tons of experimentation, but it is good practice to consider critically before performing any hard work behind it. The thing with speech is, that irrespective of how you define it, you aren’t going to be around to teach every single speaker how something is supposed to be pronounced. They will have their own personalities and unique traits, and will majorly be concerned with whether the other person understands them and vice versa, rather than bother to perfect their pronunciation. That is the entire reason why dialects and accents exist to begin with. So, what should be done is to define general rules about how something is supposed to sound and let the people do the rest. What monstrosity it evolves into later is not your botheration. When assigning sounds, we cannot forget that these sounds must be approximated in writing later, and randomly making up sounds for different concepts will lead to disaster. Symbolic writing also helps the brain to assign a sound(intangible) to a character(tangible) and encode them into memory in clusters, instead of trying to remember complex sounds by themselves. Ideally, these clusters should be simple, since it’s easier and more accurate to remember small clusters of information. This sound-letter association also allows you to use the power of permutation and combination to use some basic sounds denoted by concrete letters, and chain them later to make complex ones. Another advantage is that you can write the letter associated with a tiny sound unit, and learners will automatically translate the symbol into sound when they see it. These associations are the foundations of your language. An important consideration for these clusters is thereby the simplicity: in both speech and writing. Phonetically, most languages have two categories of letters : vowels and consonants, of which the vowels are used frequently. Observing the vowels yields an interesting result : their sounds are mostly present in languages throughout the world with either a short or long pronunciation. (English) a/e/i/o/u or (German) a/e/o/u/i/ü/ä or (Indian languages) अ/इ/उ/ए/ओ or (Japanese) あ/い/う/お. It is not all encompassing by any means, but the bottomline is that all languages have commonalities in their vowels. It is not difficult to see why : Phonetically, it is not possible to accurately chain sounds without passing through one of these phases, since vowels are pronounced in the simplest way possible. Just generate a sound and slightly modify your mouth from its natural resting position, and there you have a vowel. As nobody stops generating sound while speaking, and merely modifies their oral structure to create various sounds, vowel sounds are inevitable, even though the vowels themselves may not appear in words (English is a major offender-rhythm). The structure of all the consonants hence normally pass through vowels while chaining them, unless consonants are similar in oral shape, and all these computations decide which sound may be possible to chain after a particular one. Needless to say, one does not need to calculate so many things when speaking, since they become second nature, yet it becomes necessary to recognize these trivialities when constructing a language. Thus, we come to the phase of actually selecting what vowel and consonant sounds we want in our medium. It’s time to get creative… or so I say but it’s quite a tedious process to create a script, assign sounds to them and then create words by chaining those sounds. Not only is it a particularly irrelevant thing to spend time upon, but it will also take unnecessary effort to type in this new language, so we will just use one that is scarcely known, but exists for our purposes. For efficiency, we should logically have unique clusters that link a sound and letter, with easy methods to chain letters(and hence sounds) as a word progresses. Doing this means that the generated clusters will be phonetically consistent(like an invertible mathematical function), with one sound having only one letter and vice versa. In fact, the oldest known languages of the world, from whence most other languages are said to have originated share this feature of phonetic consistency : Deutsch and संस्कृतम् . For our purposes, we will use the देवनागरी(Devnagri) script of the Sanskrit language. ///Script photo It is therefore imperative that readers look up the pronunciation of the Devnagri script(whose pronunciations I clearly cannot write here), which is also used in Hindi, Marathi or most other Indian languages. Learning them perfectly is unnecessary, just the basic idea is enough. Eventually, as they are used and you keep revising them, the letter-sound associations will become engraved into your bloodstream. Another prudent thing to be mentioned is that… this is supposed to be the development of a new language: nothing that hasn’t been discussed will be used. दएवनआगरई अशई दइसतए…