We examine the problem of message recognition by reviewing the definitions and the security model in the literature. In particular, we examine the Jane Doe protocol, which was proposed by Lucks et al., more closely and note its inability to recover in case of a certain adversarial disruption. Our paper saves this well-studied protocol from its unrecoverable state when such adversarial disruption occurs. We propose a new message recognition protocol, which is based on the Jane Doe protocol, and incorporate the resynchronization technique within the protocol itself. That is, without having to provide a separate resynchronization procedure, we overcome the recoverability problem of the Jane Doe protocol. Moreover, we enumerate all possible attacks against the new protocol and show that none of the attacks can occur. We further prove the security of the new protocol and its ability to self-recover once the disruption has stopped.