Motivated by mixed evidence on the effectiveness of reading companions on children's reading performance, as well as the clear and present need for libraries to conduct literacy education, this study sought to investigate the feasibility of using social robots in library literacy activities and to extract the essential functions of effective reading companions by comparing human and robot co-readers. A humanoid robot, Robot Julia, was developed as a social robot to provide oral conversation and offer children tactful and stimulating support for their reading. An experimental study was conducted to examine child patrons' perceptions and performance in reading activities with the robot companion as compared to human companions. A total of 36 elementary school children participated in the study. The results positively supported that the participants perceived the robot companion as more favourable and desirable to read with than a human co-reader. The children favoured robotic verbalisation over human verbalisation. According to the results of the comparison, cognitively it was found that human and robot companions facilitated children's reading comprehension in different ways and that the children performed similarly well with both kinds of reading companions. Affectively, the robot co-reader induced more social interaction during the reading sessions. Despite all the positive aspects, it is also necessary to consider the limitations including the novelty effect of the approach. This study contributes empirical evidence in the pragmatic field of library science and expands upon social robot research by exploring one-on-one child--robot interactions in reading, as opposed to the group contexts in previous studies.