Introduction

Problems in education are such a well known problem to be almost a cliché in public discourse: Education has seen little to no innovation, is underfunded, teachers are underpaid, have no time, need more material, the classes are too big and students have outdated textbooks. These problems, of course, also extend to the language classroom, which has its unique problems as would any other subject such as math or history.

One of the main solutions to improve education by increasing quality, access and make all that scalable, is to digitalize it. While this trend has, mainly in the private sector, caught on, a common observation is that attempts to digitalize educational content is to transfer it and so tends to merely recycle the status quo of conventional mediums (Heil, Wu, Lee, Schmidt & Edu, 2016).

One such example that tackles all those classic problems is the digitalization of the transfer of knowledge. Traditionally this has been done via classrooms. However, with the help of new technologies, classes can be supplemented and even replaced by online video software such as the video platform Youtube (Youtube, 2019).

Students can use such a platform and also software in general any time they want, but they need to attend a class at a specific time. The student also needs to share the teacher with other students – the more students in the class, the less personalized the teaching becomes. The benefit a class has is, that a teacher, due to his profession, might be good at explaining things and also can interpret the feedback of the class (confused looks, etc.) to repeat or emphasize learning material. However, even personal verbal and nonverbal instruction might be improved, if supplemented by online video content.

A teacher has to repeat his or her material over and over, for every class. A video only needs to be produced once and may be updated sometimes. Also, with a likely advancement of production techniques, its quality could be greatly improved. Videos can also be designed with limited user attention in mind. A two hour class containing many different topics can be broken up into its different sub-topics. Moreover, a user can decide which video he wants to watch, videos can be recommended to him, he can watch them when and as often as he wants. He can pause them and repeat certain parts. There can even be a question and answer section (the math learning website Khan Academy offers this), in which the user can, in addition to reading the existing ones, ask his own questions even when he might have been too shy to ask in a real classroom (Khanacademy, 2019). While this is already widespread as there are numerous videos on Youtube that follow this style, such as (Khanacademy, 2019; 3Blue1Brown, 2019), ironically, when the cutting edge forefront of education, universities, take on this topic (often in the form of "massive open online courses" due to their size, short MOOCs), they are slow to adapt and often don't implement any of those advantages. Frequently they insist on low quality video media players, which tend to be less technically stable and performant than Youtube. They also tend to keep the original lecture length of about 1.5 hours.

While this example of online videos is not the focus of this thesis, it shows two contrasting approaches to digitalization. The comparatively novel approach educational content on Youtube follows, incorporates most of the advantages while universities simply try to transfer their existing approach of lecture (or class) centered teaching.

A less clear cut example, which relates closely to the focus of this thesis, is the approach the well known and successful language learning app (Duolingo, 2019) has. While Duolingo could be viewed as one of the most innovative language apps, it still primarily only uses traditional language learning methods (Munday, 2016, p. 88). It is unclear if this is because traditional methods are of high quality and so can hardly be improved or if this is a failure to create new learning methods based on increased possibilities through current technologies and reduced limitations originally imposed by the lack of said possibilities. Sufficient research in this regard has also been lacking (Heil, Wu, Lee, Schmidt & Edu, 2016, p. 33), which might be caused by the pace of technological progress. Studies such as (Lu, 2008), who used 'short message service' (SMS) as a vocabulary trainer laid out groundwork but became partly outdated as their technological methods (such as SMS) are limited compared to today's software. We thus ask the question: How should language education be digitalized optimally?

To answer this, the goal of this thesis is to build a framework that explains how a second or foreign language should be learned. The method of construction for our framework is to critically review the appropriate scientific literature and extend our framework with the acquired insights. As a consequence, in the first part of this thesis, we will slowly extend our framework by continuously raising questions and seeking their answers in the literature. Our framework thus will be a combined collection of existing theories, models and frameworks. Due to the complex nature of the task, the final construct of our framework is laid out at the beginning to aid the understanding of the process which is in the first part of this thesis.

In the second part of this thesis, the application of the framework will be shown via the learning steps of a hypothetical user. This example will show the limitations of our framework. Those limitations can be reduced by utilizing software, as will be shown. Thus the second part will discuss the role of software and the requirements and features the ideal software should have based on our framework. Finally, the implementation in the form of a built prototype will illustrate the potential that software can have as a means of improving traditional language education.

A key insight of our framework is that it is critical to put the individual nature and variance of learning needs and requirements into consideration for sustained long term learning success when time, motivation and cognitive energy are limited. This requires a tailored learning experience which can be provided by software. Software supports the user by collecting data about his learning activities. This user learning data provides insight into the individual learning needs and requirements and an immediate recognition and application of them by the system is desired. With this, a constant adaptation of the learning system of the software occurs that is based on the user generated learning data. The adaptiveness of software thus fits the individuality of learning processes.

Our framework is targeted towards adult learners, ideally university students or any group of people that has the time and means to study something in the long term. Conditions that apply to adolescents, elderly people, immigrants, or any other group that falls outside the norm (of the ideal student) and thus represents special cases with adjustments, are not of primary concern. An example for such a non-primary concern would be a lack of access to the internet or lack of access to educational institutions in general. These groups and use cases have their validity, but they are out of the scope of our general framework even though it might still apply to them.

We conceptualize the 'learning system' as any sort of learning material, teaching practices or learning activities, and we will use the term 'user' for the person, student or learner who uses the learning system. For the sake of simplicity, we will refer to the user with the male pronouns, but it refers to both genders. We call words learned by the user 'vocabulary' but mainly use the abbreviation 'vocab' and its plural 'vocabs'. While we refer to the language a user is learning as the 'second language' or L2, it could also be called the 'foreign language'. The difference is that a user generally comes into contact and has access to the second language within the context of his first language (a Hispanic user in the USA might have Spanish as his first language and English as his second language, or the other way around). A foreign language refers to a language, where no such contact exists, such as a German learning Japanese in Germany (Ellis, 1994). For our thesis this difference will not play a role and 'second' or 'foreign' are interchangeable, even though 'second' is used.