The present study addresses two questions: On a non-visual multimodal interface for textual information browsing, (1) how prevalent is input modality switching as an error handling strategy, and (2) how much does an input modality need to fail before input modality switching occurs. The results indicate that although switching input modalities to correct errors is an expected practice on multimodal GUIs, it is not the prevalent strategy for non-visual multimodal interfaces. We believe that users are more likely to diversify their error handling strategies within a modality, if different strategies are possible, but we have not found conclusive evidence for this belief. However, our analysis suggests that the failure to switch modalities when errors occur may, in part, be due to the prevalence of alternative error handling strategies in a particular input modality, that is, the user prefers to stay in the same modality rather than assume the cognitive load of a switch.