spiega tech
shared documentation across projects
1. Phonosymbolic language
Phonosymbolic words are composed of sounds that give symbolic meaning through cultural and linguistic conventions. More on omniglot.
Each phonosymbolic word is designed to evoke certain emotions, ideas, or cultural associations rather than directly translating into physical descriptions. This approach allows for a more abstract and flexible form of communication.
During high school I was fascinated about creating a language where sounds represent basic concepts and words are just the literal composition of those. The idea was to divide the sounds into consonant and vowels and diacritics as in natural languages. Vowels, consonants and diacritics should belong to different realms of meaning where each word is the composition of those. Basically if `t` and `d` are both plosive but `d` is the alveolar version. Let's imagine now we have a sign for t which will represent a subject and a diacritic which will represent an attribute. In phonosymbolic languages, these distinctions might be used to convey different meanings or nuances. For instance, in some hypothetical language system, "t" and "d" could represent different symbolic concepts based on their subtle differences in articulation.
The language will then have a sign for t and a diacritic to move to d. Back then I had no idea on how to assign sounds to meaning but after 10y of physics and 15y in ML I guess LLM, clustering and embeddings can help a lot.
2. Alphabet
We now need to decide on an approximate basis (that would be natural language specific) what are the basic sound and diacritic to use. The phonetic alphabet is not precise because a single natural language won't need the complexity of phonetical extensiveness. We start with Italian. We start picking Italian as natural language and we have then selected the basic sounds: t, p, ci, n, f… and created the softer version
| hard | soft |
|---|---|
| t | d |
| s | s |
| f | v |
| p | b |
| ci | gi |
| ch | gh |
| n | m |
| sc | j |
We will then use a diacritic sound to move between the two, same for the vowels.
| closed | open | soft |
|---|---|---|
| à | á | |
| è | é | |
| ò | ó | ö |
| ì | ì | |
| ù | ü |
We need then to find 8 consonant signs, 5 vowels and 1 diacritic and we start with simple signs like \ or /.
3. Font
We firstly use fontforge to create the font to visualize the main features and I will decide a fixed mapping to ease typing on the keyboard like:
| key | sound |
|---|---|
| k | ch |
| c | ci |
| g | gh |
| j | gi |
| ś | sc |
| ź | j |