First, remember that you actually have some sort of DCE device between the two DTE's that are trying to communicate. So, a generic example:
Router --> CSU/DSU --> Network Cloud --> Remote CSU/DSU --> Remote Router
In the "real world", you would never just place two random DTE devices in synchronous communication with each other without configuring an agreed clock source prior to activation. In other words, there must be a source driving a clock. Both DCE devices (assuming a point-to-point connection) must agree on what will be driving the clock.
There are three generally accepted methods of clocking: internal, external (also called network or master/slave), and derived (also called recovered).
By far, the prevalent source of clocking is external clocking. This permits your ISP/carrier/telco to set the clock rate and your device follows.
Thus, with all respect, your question is slightly misworded/misguided. When a new line is brought up, if (and only if) external (or master/slave) clocking is configured, the slave clock will be "forced" to a certain clock rate by the master clock. It's possible they two clocks had different speeds originally, but they will be "forced" into alignment by the nature of the master/slave relationship.
Now, let's take your question and assume an internal clocking configuration. Since each CSU (or DCE) runs their own clock, they will never synchronize. Remember this is at layer 1: there is no inherent "negotiation" ability. Recovery and fallback requires on the transmitting and receiving hardware.
FWIW: Troubleshooting clocking issues takes up a large percentage of my WAN troubleshooting. Add scenarios when jitter (short-term) and wander (long-term) deviations occur, and clock issues can be very difficult to troubleshoot.