An Analysis of Language Learning Startups

The urge to learn foreign languages is as old as human history itself. With the increasing prevalence of language applications usage, we reviewed the most common business models that provide unlimited language learning to the user with a single click anywhere around the globe. We have examined four commonly used business models, presenting a brief summary listed with similarities and differences among them. We are hoping such inception will be enriching for the field of work and the passionate interests in the language learning.

1. SOFTWARE-BASED LANGUAGE LEARNING

This business model utilized by the following cases of language learning applications is considered the very most popular and common-used model for teaching languages regarding daily life scenarios, using VR Techs, speech recognition software, and chatbots. It provides mini lessons with a narrow focus on daily life routine, engaging useful topics such as introducing one’s self, ordering food, or travelling arrangements.

Application mechanisms ensure that learners stay motivated to achieve their goals as each lesson contains a chief lesson followed by additional units of practice for pronunciation, reading and listening. It also provides a small set of vocabulary and short phrases by the image-word match on the screen, allows users to write some exercises or repeat the word aloud using the microphone. Still, it varies in number of languages provided; whilst some of them contain 4 languages, some of them offer up to 30 languages.

Primarily, depending on the language case, there may be placement tests assessing the user in order to place them in the correct level and then new units become active once the previous unit is completed.

Lessons in this type of applications do not approximately exceed 5 minutes and repetition is somewhat necessary to achieve the goals of the learning process in acquiring a basic base of vocabulary. Most of these applications are based upon the Freemium Model, while Pro users get other features ranging from: (ad blocking, unlimited languages and lessons, offline access, chatbots, videos for native speakers of selected languages, some statistics showing the number of lessons passed, etc.).

PROS

· Simple useful content, good learning app combination for another tutorial or classroom assistance.

· Contains dozens of languages.

· Useful in teaching phonetics, letters and basic concepts.

· Affordable, because content’s marginal cost for an additional student is low (approximately zero), which is why users can be offered reasonable prices.

CONS

One of the main drawbacks of these applications as a whole is that they do not provide much ‘original input’ or ‘real output’.

What does that mean?

Humans acquire languages by being exposed to the context of language molded by practicing communicating with others in real-world, everyday contexts. This is how we all learned our first language.

While these applications provide some input (listening and reading exercises) and a small portion of output (writing and pronunciation exercises), this stays far from the original input (such as reading a textbook or watching videos) and the real output (such as speaking with a native speaker). These two factors — authentic input and real output — are key to achieving language fluency, not just tapping the screen. Best of all, real input and output have the added benefit of making language learning more fun, exciting, effective and efficient.

In most of these applications, speaking and listening are significantly minor. Speaking exercises are optional. It can be enabled or disabled in settings, as well as the debugger is the “software” and this does not guarantee validation considering the tone of voice or dialect spoken by.

· Some language learning programs use the same images for basic vocabulary in all languages — whether you are learning Arabic, Chinese or any other language, and this of course arouses boredom those aiming to learn more than one language, as well as using the same approach and design reduces brain stimulation to remember and associate images with new words.

· Variation in the ‘size and quality’ of content according to the language chosen. For example, In the German language course we find six levels, while in the Arabic language we find only four levels.

· Exercises can get rabidly boring, especially for learner who fend off the learning approach with pictures and games.

· The intermediate and advanced levels are not much different from the beginner compared to the obvious difference we see in real classrooms.

· The speech recognition system is not accurate, as well word /sentence voice detection by the “software”, especially in some of the languages I have tried, such as Arabic, it happens to have many errors in the “pronunciation” in several applications.

· While these apps focus on words and sentences, they lack grammar lessons for each language, and so it seems impossible to know the conjugations of verbs in Arabic, or whether the word “sky” is masculine or feminine. These apps have dropped that.

· Due to the similarity in the content of different languages, we will lose some cultural context in between. For example, the words rice, bread, and milk may sound universal, but there are languages and cultures where kabsa, shawarma are often useful.

The Rosetta Stone has addressed this somewhat with additional material covering various topics on new videos on demand. These videos are not available in all languages yet, but they are great material and I hope to see more of them.

· One of the cons of Memrise app is that there is no guarantee of the content quality , because part of the content is user-designed. Also in the Busuu app, which is functioned on a network of learners who correct each other.

Briefly: these apps will not make a native tongue out of a foreign speaker, still excellent study aid. They are a good choice for beginners learning vocabulary, sentences and some simple verb conjugations, but on the contrary, not useful when it comes to strengthening conversation skill or learning complex grammar or cultural context.

In software-based language learning applications, we attribute the success of one of them to several reasons, including:

· First: the number of languages provided by the application. Therefore, the reason for the popularity of Drops and Duolingo is due to the multiplicity of languages available, the more languages there are, the more users will have, reputation and fame.

· Secondly: a well-supported educational structure. This is determined by the quality, diversity and gradation of the content. Once a skill is mastered, the user shifts to a higher level, providing more challenging content and an entertaining method that does not trigger boredom in repetition.

This is shown in Duolingo as the exercises become progressively more difficult, each unit contains a topic, of which contains multiple lessons. Same goes for Rosetta Stone. I like that Rosetta Stone lessons take about half an hour to finalize, so the user invests time in paying attention and keeping new words in the Long-term memory. Babbel has a reasonably clear structure but does not have the versatility of Rosetta Stone or Duolingo, but it is simple, straightforward, and easy to use. As for Drops, it is distinguished by its reliance on games in learning the language, which adds a lot of fun and breaks the boredom.

· Thirdly: Other resources provided by the application, such as podcasts or videos. A fulfilling content with a challenging method for anyone who is not up to the conversation level yet. Duolingo, Rosetta Stone, and Babbel offer this great feature. Duolingo podcasts tell real stories (short radio documentaries) and each episode uses a bilingual pair of English and user-learning language. Duolingo also has a feature called Stories (available for certain languages) that allows for extensive reading and listening practice as well. Rosetta Stone contains additional content including beginner videos that help teach how to use the language down to the minutiae of everyday situation.

· Finally: the price. Mondly’s annual price is way low compared to other language learning apps, followed by Drops, Busuu and Duolingo. However, the free version of Duolingo is a solid option for unlimited language lessons to keep you on track.

In short: Drops and Duolingo are solid options for those interested in learning a new language (and they’re more fun than Rosetta Stone’s traditional approach). If game-based learning is something the user prefers, we recommend trying Drops first. In addition to essential lessons and additional content, Rosetta Stone has optional 1:1 teaching sessions and a new live streaming feature. Busuu is still very good and the reward systems, comments, and social elements have been improved to increase students’ sense of interaction. However, the problem with educational feedback exchange from another user increases the lack of reliability in the quality of the content. Same problem with Memrise. Whereas in Babbel and Mondly there is an ample room for improvement.

This table evidently shows details of major players using the previous business model, ranked in order of highest funds.

2. Live language learning (ONE-ON-ONE)

In the previous model, we got to know the most important applications based on learning by software. There is another option; it is joining live streaming private classes, where e-learning platforms connect language learners with teachers via online video chat. Teachers are native speakers of different languages, available around the clock all over the world.

These platforms enable the teacher to set the fees per hour and allow the user to choose the teacher based on factors including (their country of origin, specialization, rating, availability, and the rate they charge). The duration of the conversation ranges from 30 to 60 minutes and prices vary from one application to another and from one teacher to another, over average $24 per hour.

This business model does not offer structured lessons as in the previous model, but rather allows language practice with a real human tutor.

These platforms provide an environment in which one can improve skill of speaking with a native speaker on different topics without committing to a specific educational curriculum. The focus is personalized on one-to-one instruction classrooms via one-way video conferencing, targeting the simplicity of practicing the language for people in their spare time.

Briefly, the user does not learn a language using this model, as nothing is formally taught. Instead, they are applications to practice conversation through live interaction with other people. It is a suitable option for intermediate level where people practically push themselves into situations where they approach the language through conversations with native speakers.

PROS

The most divisive advantage of this model exceeding the previous one lays within interacting with real person in language learning. This feature cannot be neglected — from the learner’s point of view — because direct interaction with a real person pays several advantages, including:

· Overcoming Glossophobia and breaking the physical barriers of language, and the fear of speaking out loud.

· Instant correction of pronunciation by the teacher.

· Live interactive learning experience.

Applications using this business model also have other advantages:

· Speaking and listening skill gets a lot of attention due to the instructions of a real teacher

· Flexibility: It is one of the most attractive features, especially for learners who may not have vacancy in their schedules to study in regular classes. The learner can log in anytime and anywhere with one click.

· The payment plan is also flexible, allowing users to pay for each lesson instead of subscribing to a full course or monthly plan.

· In (one-to-one) Classes, the user learns at his own pace, and the teacher is bound to cope with him, until the learner fully absorbs the lesson.

· Voluntary interactions: the learner prefers learning in a free driving environment and it is rather interesting, as it allows the user to broach several topics based on his/her own interests, and enables the teacher to formulate a lesson plan that meets the individual needs of the learner.

· Integrating language learning with culture knowledge: This is because the user will learn from several native speakers, holding conversations with them about local culture and customs.

· Multiple options availability: User can browse teacher profiles, search by language, price, availability, or other important factors. They can also see comments made by previous students. Some apps provide an option to book a trial lesson as well.

CONS

· There is no structured educational content or study plan curriculum. This type of app has no exercises, quizzes, or grammar lessons, just a chat!

· The educational value is highly dependent on the teacher. Teachers are not required to have specialized certificates but are simply native speakers of the language, informally teach and hold conversations with the learner. Who knows if they are right? Moreover, what if they don’t bother correcting you when you’re wrong? Therefore, the learner either finds a good and loyal teacher or have a bad experience with another naive teacher.

· This type of application does not rely on setting goals compared with traditional classes, the learner can know how many new words, grammar and exercises he should do each week and how many hours he spends.

In language learning applications that provide one-to-one sessions, it does not appear to be a significant difference in the features between them, but we attribute the success of one of them to several reasons, including:

· First: the number of languages provided by the application: the multiplicity of languages available, the more languages there are, the more users will have, reputation and fame.

· Secondly: Teacher quality. Since these applications rely heavily on language delivery by independent teachers, the quality of learning is highly dependent on the teacher himself. Although after a quick browsing of these applications, I found the same teachers of the “Arabic language” with profiles in different applications, but the number of hours of instruction they provided in one application may be twice times the hours in another application. Perhaps the popularity of the application in a particular country in learning a particular language is a huge factor.

· Thirdly: Other features provided by the application. Verbling for example allows teachers to share files such as worksheets and text examples that can support learning. There is even an option to create custom flashcards to use between live lessons. Also in Cambly it offers an option of structured study content that allows the teacher and the learner to walk on it in addition to the open conversation. While in HelloTalk there is a feature to send voice messages to the teacher according to the privacy settings.

HelloTalk differs in some ways. It operates as a social platform that also allows users to exchange messages with each other. The number of its users exceeds 20 million where they are encouraged to post a short update or a picture that enables others to correct and reply directly. I don’t like this part of the app that much because it’s so open and free, it seems distracting. When I compare this property with Busuu, it is clear that Busuu architecture goes a long way in facilitating useful interaction. HelloTalk seems more random. In Busuu, the learner is required to complete a short language learning exercise and once the exercise is completed it is transmitted to the community, then anyone can offer to correct it. This seems more useful because it is clear what each person has to do to make the interaction work. However, I can see how some people may prefer having fewer rules because it allows for more personal interaction. Chatterbug also offers a unique and wonderful feature that allows direct and quick conversation (whether text, audio or video) with the teacher using the WhatsApp application.

I appreciate the succession of this type of application for gathering speakers of different languages with learners. It is not easy. As I mentioned earlier, there are no big differences between these applications in terms of quality, because the success is due to the availability of the language the learner wants to learn, in addition to the tutor himself and the experience he provides, as well as the price determined by the teacher and not the application. Sure, the user experience of the app plays an important role but I have not put that under investigation yet.

This table evidently shows details of major players using the previous business model, ranked in order of highest funds.

3. STRUCTURED LIVE LANGUAGE LEARNING

This approach differs from the previous two models in that it tends to be more like an online language school.

Technically, they are platforms that use an online classroom delivery system that is more organized than just an open conversation. These platforms provide structured education and a study plan for different levels designed by language education experts. These courses are delivered by dedicated teachers experienced in language teaching, so learners can rest assured that they are being taught not only by native speakers, but also by trained tutors.

Such as language schools; these platforms provide the ability to download studying materials and provide periodic exercises and tests after each unit. The platforms we will show below have group classes (often around three to four students per class) as well as one-on-one tutoring.

This approach is suitable for people who are looking for a learning opportunity equal to what’s already is being offered by institutes and universities, containing placement tests, minimum number of classes to complete, exercises and live lessons to move to the next level.

PROS

· A systemic structured curriculum with specific goals and multiple language levels designed by language experts and proper to all levels of learning.

· Tutors are language-teaching specialists, trained to deliver courses. In addition to obtaining the certificates required for teaching languages, candidates undergo a selection process that includes personal interviews and previous experience, and the process of preparation and continuous training of teachers on training programs takes place before they engage in teaching.

· Teaching all language skills; listening, reading, speaking and writing.

· Instant correction and feedback directly by tutors.

· Culture is part of the learning journey. Thus, specialized curricula do not usually neglect such aspect in learning any language, and native-speaking teachers have a distinctive role in integrating culture during lessons.

· These platforms provide other specialized courses such as, corporate language training and preparation for international exams such as TOEFL, TOEIC and IELTS.

CONS

· This business model requires the company to exert great efforts in selecting and training teachers and constantly assessing their performance.

As there are only a few good teachers available on demand, it seems like a big challenge when we have many students, we will need a huge number of teachers and it is very difficult to control the quality, and this will have a negative impact on the customer experience and on efficiency and effectiveness.

· Teacher costs are high comparing to other classes charged by a freelancer teacher in the previous business model.

· Lessons cost much higher than previous models.

In this business model, the quality of the educational experience is often measured by the competence of the teachers and the quality of the educational content. There are also some different features distinguish each platform.

Here we point out some of the differences between these platforms.

1. OPEN ENGLISH:

Open English is specially designed for native speakers of Spanish and Portuguese. This platform is characterized by the availability of live group learning sessions every 30 minutes.

The learner can join the learning session appropriate to his level at any time, if he has completed the requirements of the previous lesson, and every week he must complete the entire unit to move to the next.

This platform provides assessment exercises after each unit includes “writing” that the teacher marks directly and sends feedback to the learner. The platform offers one-on-one individual classes for an additional cost, as well as business English classes and preparation for TOEFL and IELTS. The platform is available on the web IOS and Android

2. LINGODA:

Lingoda is an online language school best known for its promotion of Marathon Courses, which offers partial refunds for completing a certain number of classes within a certain period. After scheduling a class on the platform, the learner will be able to join the class with (2–4) students from all over the world, and the learning materials can be downloaded at the end of the class. Lingoda is based on the monthly subscription for the number of sessions. Lingoda does not have an application on mobile phones, but rather exists on the web only.

3. FLEUNTIFY:

Fluentify is a platform offers self-study courses to enrich the learning process. It has 3 subscription plans by number of sessions 3, 10 and 25 sessions.

For me, I believe nothing can replace real interaction and direct feedback from a real person regarding language learning, with clear goals, a study plan, and qualified and trained teachers.

However, there are significant challenges facing this business model that limit its expansion, and undermine its service for a small number of users compared to the first and second models, because it mainly does not depend only on the human teacher, but on the qualified and trained human teacher.

This table evidently shows details of major players using the previous business model, ranked in order of highest funds.

4. AI-BASED LANGUAGE LEARNING APPs

The following companies are using artificial intelligence techniques to rely on a virtual teacher to teach “English” specifically. These techniques mimic the entire process of a human teacher: listening, understanding, thinking, and providing feedback. Some of these techniques include:

· ASR — (Automatic Speech Recognition) for Listening.

· NLP — (Natural Language Processing) for Comprehension.

· Adaptive learning and TTS — (Text To Speech) for method thinking and providing notes

The technology, the product, and the content that is built into the system, all have their stakes in the success of this business model. Since the system is able to respond to the learner with content appropriate to his language level, this is a dream for many teachers simply because, it is “adaptive” personal learning that saves plenty of efforts.

PROS

· Affordable, because content’s marginal cost for an additional student is low (approximately zero), which is why users can be offered reasonable prices.

· Scalability. Where this model does not rely on real teachers — as was one of the challenges of the previous model — this model based on artificial intelligence techniques can be predicted to serve more students.

· Efficiency and effectiveness. Although many potential customers are still skeptical about adopting technology in language teaching, their behaviour may slowly change as these technologies will evolve and provide a learning experience with consistent standards of efficiency and effectiveness.

· A user experience that gives more confidence. Because the learner will speak with the machine, he will not feel insecure to speak a second language.

CONS

· Like the negatives in the first business model, the learner does not learn all the language skills, there is no plan for the content, no exercises, and no knowledge of the culture, as well as the voice of the machine is not like a real human voice and may make pronunciation mistakes.

· AI Techniques are still in their infancy with regard to language teaching applications, and they do not appear to be widespread in multiple languages, as it is currently available to teach “English”.

· The technical effort required to build this type of application reduces competition in this field

· The delay in adopting these technologies by the user reduces their success at the present time. Learning with a human tutor remains a better option for many (personal point of view).

This table evidently shows details of major clients using the previous business model, ranked in order of highest funds.

SUMMARY

In this review, we presented a brief summary of the most common business models in language learning techniques, a brief overview of each of them, pros and cons, as well as information about the most important major players in each business model.

Through this reading, we concluded several factors for the diffusion of language learning applications, according to the business model that they adopt, and we can summarize them with these factors:

1. Scalability: The software-based model (the first model) appears to serve more users who are literate in different languages ​​for less effort in creating consistent content that is visible to all users regardless of their individual differences.

The second model, based on instant conversation with a real tutor, also appears to be highly scalable as it relies on connecting native speakers “freelancers” to have an open conversation with the language learner.

While in the third model, which is based on a real tutor, a study plan and study content, it shows a challenge in its expansion due to its heavy reliance on a qualified and trained teacher working for the company within strict standards and conditions.
The fourth model is similar in its expansion to the first model due to its reliance on virtual learning using artificial intelligence techniques, but it remains limited to teaching English and needs time to mature and prove its success.

2. Cost: The third model seems to be the highest in this regard, both from the point of view of the customer or the company, as it requires the development of customized content for all levels, teacher training, and a commitment to pay a fixed commission for the teacher. Whereas in the second model the teacher is the one who sets the price and the company is just an intermediary that takes a percentage of the amount charged by the teacher. The first and fourth models seem easier to calculate as they depend on monthly, annual or lifetime subscriptions.

3. The human factor and the importance of immediate correction: This factor seems more serious from the user’s point of view, as people who prefer learning with a real teacher for the reasons we mentioned earlier will prefer the second or third model, and vice versa. In the third model, all language skills are taught because it is more like a real classroom, while in the second model the focus will be only on improving the speaking skill, which is the most suitable option for the intermediate level or higher.

4. Efficiency and effectiveness: Efficiency and effectiveness seem more challenging in these models, because in the second and third models it is highly dependent on the real teacher. Who can guarantee the competence of all teachers? Even though there is an evaluation feature, efficiency is a challenge in these two models in terms of ensuring the efficiency and commitment of teachers and their continued development and follow-up. While in the first and fourth models, the technology replaces the human element, thus increasing the efficiency of the system and reducing the error rate.

5. Multilingualism: All previous business models are capable of building applications in multiple languages; the fourth model seems the most difficult because artificial intelligence techniques related to voice recognition and natural language processing seem to be advanced in favor of the English language.

However, applications that include multiple languages in any of the above models will attract a larger audience and therefore greater expansion and greater opportunities.
In the first model, we saw how languages vary in the difficulty of building their own content using the software. For example, teaching English is not the same as teaching Arabic.

6. Targeted Audience: In the success of any business model, we cannot turn a blind eye to the target group, which is, in my opinion, the most important factor that shall be taking for granted before starting to develop a business model.

The needs of a teenage language learner are different from those of, say, a diplomat. By looking at the characteristics of each model, as we presented previously, we can arrive at the most appropriate models for the target group based on individual differences, such as age, preferred teaching approach, desired language skills, time flexibility, educational goals, and others.

7. Providing a competitive yet better education: The presented business models differ in their approach to presenting the language to the learner, and since each model has its pros and cons, each application that uses one of these models seeks to compete either in providing a better user experience, more enjoyable, more valuable, ease and flexibility, and others.

We’ve seen how some apps add additional educational resources like podcasts, translation, free hours, and comprehensive quizzes. All this increases the chances of success of the application in particular.