This manuscript reports on two studies. Of interest is at what point in development toddlers recognize paths of motion embedded across event contexts when they are named with a novel spatial particle. In the first study, data were collected from 40 toddlers aged 18 and 24 months in a participant-controlled habituation procedure, followed by a preferential looking task and behavioral reenactment. Participants were habituated to videos of a figure illustrating a single path (up, down, over, or under) while hearing either "Look, he's going" or "Look, he's going keet" in the no word and novel word conditions respectively. They then viewed a three-action event in a preferential looking paradigm with the familiar path embedded within the event on one screen, and an entirely novel event on the other. Following this, they participated in a reenactment procedure. Toddlers who heard a novel label at habituation spent significantly more time looking at videos containing the target spatial relation. Despite this effect, few toddlers reenacted the spatial relation. Study 2 explored whether providing a novel spatial particle at both habituation and test changes looking behavior and increases reenactment rates. Data were collected from 52 infants aged 18 and 24 months. Procedures followed the first study, with one important change. During the preferential looking task, participants heard additional prompting to locate the target spatial relation, such as "When does he go keet?" or "When does he go there?" in the novel and no word conditions respectively. With this manipulation, toddlers hearing a novel word continued to show a preference for familiar videos, but interestingly reenactment rates remained low, highlighting the difficulty of learning these word types.