TL;DR: In 2021 Martin and I created this Project. We started with the idea to create fonts which respond to the context of each writing. We used some machine learning algorithms to analyse the text. It ended up in a small tool where our font but also any other variable font can be tested.
The Inspiration
In »Sprache und Schrift« (Spoken and written language), published in 1920, Dr. Walter Porstmann propagates a written language which is closer to the spoken language.
In 1927 Kurt Schwitters published his take on Porstmann’s idea, the »Neue Plastische Systemschrift« (new vivid letter system).
Neue Plastische Systemschrift by Kurt Schwitters, 1927
In 1928/1929 Jan Tschichold published the »phonetisches minuskelalphabet« (phonetic lower case alphabet), a close visual of Porstmann’s ideas.
Phonetisches Minuskelalphabet by Jan Tschichold, 1928/29
Schwitters and Tschichold developed font systems which only worked for the German language. Herbert Bayer, who moved to the States in 1938, developed the »Fonetik Alphabet«(phonetic alphabet) in 1959, which represents a version for the English language.
Fonetik Alphabet by Herbert Bayer, 1959
While phonetic fonts try to bring the written language closer to the spoken language we try to bring the mood a text provokes closer to the aesthetics that a font implies. We call these fonts optosentiment fonts.
The Tool
The Font we used was created with Open Type Face. OTF is a project I have created in 2019. It is a tool where anyone can design a font. The design process is based on facetracking. We used it to create a happy font and a negative font!
The tools we used to realise the analysis are:
Keras.js — Bidirectional LSTM
ThisAndAgain — Sentiment