Learning by Doing: Space-Associate Language Learning Using a Sensorized Environment

This report describes a Learning-by-Doing (LBD) system for assisting learners of English as a foreign language, by providing educational content generated based on real-world behaviors in a sensorized environment. The LBD system consists of an ultrasonic location sensing system developed by the authors and an educational content server. The location of the student is monitored by ultrasonic sensors embedded around the learning room. The system voices English verbs related to objects around the student so that the student can obtain the language skills for expressing various activities of everyday life in English by performing real-world behaviors. The system was implemented experimentally in an English language school. Six subjects participated in an experiment evaluating the educational efficacy of the proposed system. Recall rates of learned verbs were compared among three methodologies, "Learning by text," "Learning by pictures and text," and "Learning by doing". Recall rates were measured at the following intervals: soon after the learning task, 2 days after, 4 days after, 7 days after, and 14 days after the learning task. The results indicate that the proposed LBD system increased the recall rate of learned words by 113% 14 days after the learning task. Overall, the system significantly increased the average recall rate by 102% in comparison to the other learning methodologies.

The authors have developed a three-dimensional ultrasonic tagging system that can be used as a location sensor. The system calculates the position of a tag by trilateration or multilateration. Examples of the tags used in the system are shown below. The LBD system was constructed based on this technology.

The Learning-by-Doing system consists of both hardware and software components. The former is the ultrasonic 3D tag system for measuring the position of a tag, and the latter is the Learning-by-Doing content system for monitoring the foreign language content. By sending an ultrasonic wave to receivers, the ultrasonic tag system calculates the 3D position of an object or a person by an attached transmitter. Macromedia Flash® content can be changed according to this position information. Specifically, the context, in relation to the position of an object or a person, is monitored and the associated contents are displayed.

The position of objects and users are tracked by the system, while the educational content is generated based on the detected position and is presented instantaneously. For example, when the user picks up a wine bottle, the system vocalizes "spill" or "stain". Or, when the user picks up a model of ice cream, the system voices "melt". Similarly, when the user approaches the fireplace, the system voices "burn" or "you are burned" while showing helpful visuals on a projector screen, as shown below.

Experiments investigating the effectiveness of the LBD system were conducted with the participation of 6 Japanese students aged 22 to 24. In the first task, each subject was required to learn 50 English verbs within 20 min using texts only, then a 10 min free recall test was administered. In the second task, each subject was required to learn 50 English verbs (different from those in the first test) using visuals and text, followed by a free recall test. In the final task, each subject was required to learn 50 verbs (again different from the previous test) using the LBD method, followed by a free recall test. The free recall test evaluated the number of verbs that the students could recall.

The following figure shows the relationship between the methodology and the change of recall rates over time. For all subjects, words learned during the LBD method were clearly more easily recalled during the free recall tests. In total the LBD increased the average recall rate by 102% in comparison to the other methodologies. The forgetting rates were 52% using text alone, 48% for visuals and text, and 24% for LBD. Specifically, the proposed system presented a harder-to-forget studying method than the other methodologies. The t-value between forgetting rates in the 1st and 2nd group = 1.38, and p = 0.23 (p > 0.05). Conversely, the t-value between forgetting rates in the 2nd and 3rd group = 28.55, and p = 8.94x10E-6 (p < 0.01), and the t-value between forgetting rates in the 1st and 3rd group = 17.12, and p = 6.82x10E-5 (p < 0.01).

This report presented a Learning-by-Doing method for teaching a foreign language. The LBD system consists of an ultrasonic location sensing system that controls the presentation of context-specific material to students. The effectiveness of the system was evaluated using an experimental LBD system installed in an English language school. Following the experiments conducted with the participation of 6 Japanese students, it was shown that the proposed system increases the recall rate of learned words by 113% 14 days after the learning task. In total, the system increased the average recall rate by 102% in comparison to the other methodologies. Therefore, the proposed system was more effective for learning verbs. The forgetting rates were 52% for using text alone, 48% for visuals and text, and 24% for LBD. Specifically, the proposed system presented a harder-to-forget studying method than the other methodologies.

Our future work also will also focus on improving the learning process using knowledge of databases and learning strategies, such as CYC or WordNet.