Oxford Language Primer
Oxford Language Primer is a new language learning platform
which incorporates the latest advances in AI technology.
Invited Beta Starting early 2024...
Launch Oxford Language Primer
Learning a new language is challenging.
Native speakers produce words at around 2.5 words a second, drawing from a
lexicon of around 30,000 words as well as a network of grammatical knowledge.
New language learners must acquire
sufficient vocabulary and grammar to be able to converse with such proficient speakers, recognizing and producing words
and grammar at speed.
This requires robust and automatized language knowledge, and this can only be achieved through
many and frequent opportunities to practice understanding and producing the language.
Finding sufficient opportunities outside of an immersion context is difficult.
In recent years, technology has emerged which aims to support learners in meeting this challenge.
For example, flashcard
apps support vocabulary learning by offering extensive retrieval practice over systematic
intervals. However, these generally use written rather than spoken language, support learning of words but not grammar,
and require learners to self-correct without additional feedback.
Other language learning apps have more comprehensive goals, but have serious limitations for serious learners;
For example: slow learning pace in the
early stages (due to the use of “easy” slot filling and matching tasks, which can be done without fully retrieving and
processing the language) and sharp difficulty jumps at later stages (because earlier learned language has not been fully
integrated and automated before new structures and vocabulary are introduced). In addition, while these programs now
incorporate some speaking, writing remains primary, particularly in the early stages.
Our
goal is to create language programmes which take learners from complete beginner to independent language user as
effectively and efficiently as possible.
Our philosophy is that spoken language is primary, and we build in speaking and
(natural pace) listening from the very first lesson - drawing on the latest in AI Speech Recognition and Synthesis technology
- whilst also teaching the written language.
Critically, every task the learner is asked to do provides an active
opportunity to practice understanding and producing the language (spoken or written), and we provide instant correction
and personalized feedback using large language models (LLMs) as well as the opportunity to ask follow-up questions.
Through this process, the learner continuously accesses and updates a developing mental lexicon and grammatical system.
Some explicit instruction in grammar and vocabulary is also provided (since this is known to be useful for adult
learners) and learners progress through a series of structured lessons, moving forwards when their responses indicate
sufficient mastery of the previous material.
Old material is continuously reviewed using spaced repetition algorithms.
Once learners have progressed sufficiently, the AI technology can allow them to practice conversing with an AI “partner”
directly.
The Team
Founder and CTO:
Edward Longhurst
Previously