This paper argues that studying expressions of semantic concepts may be enhanced by using at least two tasks with the same participants. It demonstrates how the study of boundedness is enhanced by employing visually modified stimuli to increase salience of motion-paths in a production and a recognition task. In the personalized narrative retell task motion-paths were elongated to encourage the linguistic encoding of events as I flew up [bounded] vs. I flew up up up [unbounded]. L2 production data revealed that learners (n=35) systematically encoded bounded events in simple predicates (particles, verbs, or particle-verb structures) and unbounded events in reduplications of simplex predicates (unexpected forms). An original, computerized, timed audio-visual judgment task complemented the production task. It assessed the mental representation of bounded and unbounded descriptions by manipulating path-boundedness in motion-videos. Participants watched live-motion videos, then a sentence appeared on the screen describing either a bounded or unbounded event. Participants determined whether the sentence described the motion-video. Statistical analysis confirmed that video-boundedness significantly predicted native speakers' (n=115) and one learner's performance. The remaining learners developed toward native-like contrast in the bounded context, lagging behind in unbounded. The two tasks in combination provide a reliable and valid measure of L2 knowledge.