In a first study, we considered whether modelling and the type of an adult's request influenced children's ability at 1;8 and 2;2 to comprehend gestures and replica objects as symbols for familiar objects. In a second study, we evaluated whether modelling and type of request influenced children's ability at 1;8 (N=24) to understand unfamiliar (i.e. unconventional) objects as symbols. Results of Study 1 indicated that children at 2;2 comprehended a gesture as a symbol in the absence of any previous model demonstration. All children comprehended a replica as standing for another object, although modelling marginally improved children's performance at 2;2 and decreased performance at 1;8. In general, the type of request did not influence children's comprehension of gestures or replicas as symbols. Results of Study 2 showed that modelling and request type did not influence children's understanding of objects that are unconventional and novel, as symbols. The studies converge to suggest that symbolic comprehension is a highly context-dependent ability that continues to develop over the second year.