I have never worked on a SCO box. And I have never neard of "ditty". But I will describe a similiar sounding problem and solutions involving more mainstream versions of unix and the "stty" program.

Tty ports on systems like hp-ux have a default set of characteristics. You can vary them by opening the device file and issueing an ioctl() call. The
program stty does this. But here is the key...when the last process closes the file, the characteristics go back to the defaults.

So a command like:
stty 9600 < /dev/tty01
changes the baud rate for only as long as the stty command is running.

An ugly but common solution is to run a script like:
( stty 9600 ; sleep 1000000 ) < /dev/tty01
at startup.

A more correct solution is to insure that any program or script that needs to use the port shoulders the responsibility as setting it up the way it need to be. Thus the stty statement should be placed in the interface script used by the lp subsystem to actually do the printing.

Maybe your ditty problem is something like that. But the mere existence of a system command called ditty shows a disregard for unix standards...so maybe not.