This study used an online second language (L2) vocabulary lesson to evaluate whether the physical body (i.e., embodiment) of a robot tutor has an impact on how the learner learns from the robot. In addition, we tested how individual differences in attitudes toward robots, first impressions of the robot, anxiety in learning L2, and personality traits may be related to L2 vocabulary learning. One hundred Turkish-speaking young adults were taught eight English words in a one-on-one Zoom session either with a NAO robot tutor (N = 50) or with a voice-only tutor (N = 50). The findings showed that participants learned the vocabulary equally well from the robot and voice tutors, indicating that the physical embodiment of the robot did not change learning gains in a short vocabulary lesson. Further, negative attitudes toward robots had negative effects on learning for participants in the robot tutor condition, but first impressions did not predict vocabulary learning in either of the two conditions. L2 anxiety, on the other hand, negatively predicted learning outcomes in both conditions. We also report that attitudes toward robots and the impressions of the robot tutor remained unchanged before and after the lesson. As one of the first to examine the effectiveness of robots as an online lecturer, this study presents an example of comparable learning outcomes regardless of physical embodiment.