Language Learning #1 — Stuart Barrass from Kaizen Languages
Biography of the speaker
Stuart Barrass studied Business and Economics and completed a postgraduate diploma in Japanese, Finance and Economics at the University of Waikato, New Zealand. Before the end of this degree, he worked for a Fintech company in New Zealand to then move to London to start his job as a management consultant and product manager for EY. He is now the CEO and co-founder of Kaizen Languages, a Zanichelli Venture’s portfolio company that uses conversational AI to teach languages.
The interview was recorded virtually and transcribed by Zanichelli Venture.
Question 1: Kaizen was only developed a year ago and you and your team are now doing really well, with a 4.8/5 star rating on US Apple Store. Can you tell us more about your venture?
SB: Kaizen Languages is an AI platform for teaching languages which uses AI tutors instead of traditional ones. The platform is designed to reflect the types of conversations that you would have in your day to day; restaurant conversations, casual chats with friends and more. A key advantage of Kaizen is that it gets you to speak from day one; pronunciation, grammar and context are all taken into account. The AI tutor corrects you and challenges you back to accelerate the learning. Our new approach to learning language finds its roots on the basis that the most effective way to learn a language is through conversations. Part of our success is probably attributable to the variety of AI tutors we showcase; these allow the user to have a conversation based on what they are mostly interested in.
ZV: It sounds like a really interesting approach on how to learn languages leveraging the new technologies of the Fourth Industrial Revolution. Can you tell us a bit more about your background and how that helped you develop Kaizen?
SB: I guess we start from my time at university. I studied Business and Economics in New Zealand, but I have always had a passion for languages, studying Japanese and Mandarin Chinese. As part of my degree, I spent a year abroad at Kansai Gaidai University, Osaka, improving my Japanese language skills; here I experienced first-hand what it takes to learn a foreign language in a foreign country. After going back to New Zealand, I initially started my career in Fintech and I struggled to maintain my Japanese language skills. I then moved to the UK to broaden my career and started working as a management consultant at EY, where I sat on the Architecture and Advanced Technology Teams, focusing on AI, Blockchain and other emerging technologies. I had the chance to explore the technology space from multitude angles and, more importantly, to meet Patrick, the co-founder and CTO of Kaizen Languages. Patrick shared a similar story: grew up in Hong Kong and moved to the UK for university, he found maintaining his Mandarin language skills challenging. We bonded over a similar struggle and we started to think about how to solve this problem on a larger scale. Fundamentally, it really boils down to one factor: the speaking practice. How often are you in a conversation in the foreign language? How often do you practice it? The other solutions to learning foreign languages weren’t giving us what we need; online platforms and applications were engaging but did not help you to confidently speak a language; traditional classes didn’t let us practice our speaking nearly as much as you would need to become proficient in one language. In fact, research shows that a student might need around 700+ hours of speaking to become fluent in Spanish; that would be very expensive and time consuming for many students. So this is how the idea behind Kaizen came about and the rest… is currently being written!
Question 2: On a recent HolonIQ Executive Brief, we notice that there has been a growing interest in the use of voice-based AI for language learning. In what way do you think that the current advancement of tech can help people learn languages better?
SB: In the language learning space, we have seen quite a big leap when it comes to AI. We can see that there are many more interesting ways to create engaging content for learning languages. Ten years ago the most innovative transformation that technology brought was textbook digitalisation; things have now changed. I find the rise of AR/VR really interesting: one can build completely new worlds where language learners can immerse themselves in, similar to what happens in the game industry. There has also been growing attention directed towards gamification, especially in early stages. Games can help keep learners motivated and engaged, which is fundamental to achieve the consistency required to become proficient in a foreign language. However, we should never lose the key focus of our journey, which is to help people speak a foreign language comfortably. Some platforms/apps can sometimes put considerable emphasis on the game design, and less on the language learning process, which steers the user away from the main goal.
ZV: This analysis is somewhat similar to formal language assessments: are you actually learning the language or are you simply mastering the exam format and style? Very interesting. Do you think there will ever be a time where in-person tutors are replaced by AI ones?
SB: Well, there are tasks that humans can do much better than AI, such as answering cultural questions or explaining why certain words are used in specific contexts. The AI tutor has its own advantages: you can feed it with as much information as you like, whether that is the types of words you use, the interests you have, the language level you are at and more. The AI tutor has a holistic view of your preparation and it will always do; if you suddenly change your in-person language tutor, you might have to start from an initial assessment again. The goal is not to eliminate tutors; we want to help people learn language faster. Ultimately, you will use the language to speak to humans, not to robots.
ZV: As you mentioned earlier, the digital space is providing new ways of creating content. We touched on AI and VR/AR; do you see any other trends in this space?
SB: I would say that AI shows the greatest use for language learning: from pronunciation to grammar or context, AI algorithms can help you master each learning area. Recently there has been an increased focus on optimisation. For example, given 100 words, how do you best order them to speed up your learning? However, we are now moving from a set of words to a set of sentences, which is adding an extra level of complexity both for the user and the developer. I guess this trend can fall under the broader theme of hyper-personalisation; education has been very prescriptive with little user input up until now. Now the trend is reversing: users have unique learning experiences specifically tailored to their needs. This is what happens with Kaizen: the user will have different conversations depending on what words they use and what response they give. The AI tutor gives you that extra level of variation that will help diversify the conversation and practice multiple answers. It also removes a degree of social awkwardness that is created when you are practicing a new language with a native; they perhaps don’t feel comfortable correcting you any time you make a mistake. Additionally, being a native speaker does not necessarily translate to being a good teacher either.
Question 3: Kaizen Languages is mainly focused on teaching Japanese for now. Why did you choose to focus on this language? What do you think will be the languages that are most suitable to a teaching model that deploys innovative technologies?
SB: Aside from Japanese being a passion of mine, I feel the language has been neglected by our competitors. Learning Japanese is quite frankly challenging, given the three writing styles and the amount of characters that one needs to know to become fluent in the language. Characters might also have different pronunciations simply based on their context or their order in the sentence. Nonetheless, the language can be found in many scenarios, such as while watching anime or reading manga, and it represents the 3rd largest economy in the world, ahead of the UK and France. For these reasons, positioning ourselves with Japanese seemed more appropriate. Identifying what languages can be taught through AI tutors really depends on the language itself, as each one presents its own challenges. For example, in Spanish the word order is much more important than in Japanese, whereas in the latter you may find two words that are spelt the same but pronounced differently depending on context (this added an extra level of complexity to the Japanese AI tutors!). The learners’ experience is also heavily influenced by the starting language: for a Spanish native speaker, it won’t be as difficult to learn French as it would be for a Korean native speaker.
Question 4: Many eager learners now turn to language learning applications to improve their language skills. Thinking about the future of this education space, do you see language learning applications incorporated into the more traditional learning paths in a B2B model (e.g. early school, high school, etc.) or will they be more suitable for a direct B2C consumer use?
SB: There are interesting trends in the language learning sector. If you look at the number of students enrolling at university in the UK or New Zealand, they are often declining. This might be due to the rise of new platforms which are giving more opportunities to learn languages outside the university space. That doesn’t mean that universities and learning platforms are mutually exclusive; you can easily do both. However, when you are learning a language through a university degree, you are often tested on your knowledge at a specific time (e.g. when you have an assessment); with platforms/applications students can get a better idea of their learning development. Additionally, university language assessments are often multiple choices or essay based, which show a degree of language mastery, but not to the extent that students would be comfortable speaking the language to natives. For the B2B and B2C models, I find that reaching schools and universities is particularly challenging: the sales cycle is very long and there are several stakeholders you need to convince before reaching the end user. I believe there will soon be an influx of EdTech tools in schools, though, as more and more institutions are adopting the use of laptops and smartphones in their curricula; definitely a space not miss out on.
ZV: Really interesting thoughts, especially those revolving around the B2B sales cycle. I am curious to elaborate more on why you think university enrolments in language courses are decreasing – do you have any thoughts on this?
SB: Well, the cost of degrees is increasing massively year on year; students can spend that money living in the country of interest while learning the language and working part-time. The opportunity cost of reading a language at university is too big to ignore. There are definitely specific jobs which you would need an academic preparation for, such as professional translator or foreign language teacher; however, for more common uses of foreign languages, you probably don’t need a university degree.
Key takeaways
- AI language tutors can help students improve their pronunciation, grammar and vocabulary skills;
- Technology creates new opportunities to engage and motivate students in their language learning journey;
- We are now moving from a prescriptive education system to an hyper-personalisation of learning content based on students’ needs.
Closing statement
We thank Stuart Barrass for taking the time to speak with us; if you want to read more about Kaizen Languages, you can find more information on their website. Have a look at our other interviews on the Language Learning space: