13.07.2020Artificial Intelligence in language learning

Artificial intelligence is in vogue. However, certain key criteria must still be met before being able to replace a real language teacher: spontaneity, creativity and shared knowledge.

Artificial intelligence (AI) with a written or oral interface is supposed to make our lives easier in the future.

It therefore seems natural for some to take advantage of innovative technologies in the teaching of foreign languages ​​as well. These technologies would be able to make learning possible anywhere and anytime. Many commercial suppliers and non-commercial institutions are already developing applications with often very different technical approaches, such as:

- Classic graphical user interface

Language learners work their way through digital exercises that strongly recall those found in traditional textbooks. The drag and drop function is used to match words with images or to fill in gaps in masking texts.

From the point of view of learning psychology, the problem here is that it is mainly about moving predefined blocks of text, so that the learners receive almost no training on how to formulate creatively and spontaneous their own contributions to the discussion.

- Language interface with dialogue function:

It is an attempt to simulate a natural verbal interaction with a virtual tutor that can be described in a narrower sense as an AI.

These dialog systems follow the principle of a simple chatbot which helps the learner to communicate intuitively in natural language. Learners are free to make oral contributions that technology analyses to detect the presence of predefined keywords. If the correct keyword is used, an appropriate predefined answer from the artificial tutor is selected and output. The input and output language can be written or verbal.

The problem is that most systems simply answer simple keywords and have great difficulty assessing whether the entries are grammatically correct - let alone appropriate to the situation. Such interactive artificial tutor systems operate in clearly defined scenarios with predictable dialogues and the corresponding sources of error.

- Virtual learning environments with pedagogical agent systems:

Dialog systems with complex avatars capable of even gestures and facial expressions represent the most advanced development.

The problem with teaching foreign languages ​​is that interaction with the agent will only work smoothly if learners understand the questions and answers that the system designers were able to predict. Human reactions are, however, only predictable to a limited extent. Outside scripted application areas, dialogs with chatbots and agent systems are irregular, inconsistent, and prone to error. They cannot serve as a model for learners of foreign languages.

- Big data analysis:

The analysis of large amounts of data via algorithms and statistical models can also be used for foreign language courses - for translation exercises, for example, and in the form of dictionaries.

The advantage of working with large data collections containing empirical remarks made by native speakers is that the foreign language is learned not as an abstract system, but as it is actually used. On the other hand, one of the problems is that the analysis of information is based on a predefined algorithm which does not always search for and provide what it should.

So, what is the potential of AI and what are its limits in learning a foreign language?

In terms of interaction, it is problematic that systems are designed on a purely deterministic basis, that is, they follow a program and have access only to limited knowledge resources such as knowledge social or cultural.

Human communication works exactly in the opposite direction. We assume that we share a lot of knowledge and are extremely effective in how we communicate only what we recognize to be relevant to a specific interaction situation. At the same time, we can act spontaneously and flexibly. Artificial intelligence systems cannot do this because they do not have a fundamental prerequisite: self-reflective consciousness.

AI will not be able, in the foreseeable future, to replace teachers in real situations. Online learning scenarios with artificial tutors do not replace classroom instruction.

However, e-learning applications with graphical user interfaces can perfectly replace workbooks and allow learners to complete or prepare classroom lessons with a human teacher. This is simply called blended-learning, an effective language learning system offered by VOXEA by making available to its learners, alongside its face-to-face courses, its digital platform e-VOX.