Existing research has established immense importance in contextual learning for second language development. However, to provide real-life context in the learning of a foreign language is challenging. To help create such a learning environment, our previous research developed an innovative system called the R&T System, utilizing robots (R) and tangible objects (T) with Internet of Things (IoT sensors). As the R&T system is still new, there is no sufficient knowledge about what learning obstacles learners would encounter in such a context-rich environment and how best to support them. To explore these issues, in the current study, we designed an English vocabulary game to engage learners to interact with the R&T System. Adopting an action research methodology, we first conducted a pilot study with 12, fourth grade students to identify learning obstacles in a game setting. They were randomly assigned to two groups, one with learning support and the other without learning support. The experiment lasted for four weeks. Both pre-test and post-test on English vocabulary were used to measure the effectiveness of these learning support mechanisms. From the video recordings of the participants' engagement with the R&T System, the results reveal that the learning support mechanisms had positive impact on the participants' learning performance and significantly reduced the occurrences of learning obstacles. This research could serve as a useful reference for future research when designing learning support for contextual vocabulary learning supported by a robotic system.