An industry worth USD 24 Billion annually is failing to deliver its core service.
Largely unchanged for over 100 years, language is taught using a failing, book driven approach. This, unsurprisingly, does not create competent English speakers.
“Pupils with years of classroom instruction often cannot hold a conversation.”
Technology now allows us to simulate the natural way we are designed to learn a language.
Below are the features of the Grammarfone method
Comprehensible Input
To be able to speak students must first be provided information that connects the concept to the relevant sounds.
In this example grasping the concepts and related sounds of ‘green’, ‘blue’, , ‘lizard’ and ‘I see’.
With enough information to guide the speaker but also to challenge them.
Productive Output
Generating speech requires knowledge and skill.
Knowledge of what you want to say based on what you are seeing. (Not reading out loud.)
Skill to perform the complicated sequence of mouth shapes and air expiration required to make a specific set of sounds.
This requires constant practice
Accurate Analysis
The data collected from students practising to speak in this way is a highly valuable and rare speech corpus.
This data will be annotated and put through a machine learning process to build an analysis system capable of providing feedback hitherto only possible by an experienced native language teacher.
A simulation of the software is provided here.