LCDs are traditionally "parallel" interface - they have 7 data pins (although there are techniques for using only 4 of them), and you need to put data on each pin and then "clock" the data into the LCD all at once, which means you need 7 (or 4) GPIO pins free on the device writing to them. One classic way around that is to use a shift register, which turns serial data into parallel - basically, you clock serial data one bit at a time into the shift register which holds it in a buffer, and then tell the shift register to move the data to the LCD pins all at once. That means you only need 2 pins to talk to the shift register. These days, instead of shift registers, it's becoming common to use micro controllers dedicated to LCD (or 7segment LED) displays. The principle remains the same - you talk serial over 2 pins to the micro controller, send it the data and then say "display", and it takes care of talking directly to the LCD.
The short answer for the Raspberry Pi is that you will be able to talk to any "Serial LCD" via either the serial RX/TX pins, or the SPI pins, depending on the protocol. That means you only need those 2 pins, and most modern programming languages have a serial comms capability. In Python (which I believe is available on most linux ARM ports), there's PySerial, which means you can talk to the LCD basically with one line of setup.
It also means you can use something like the FTDI USB->232 converters to put the LCD on the USB bus - in fact, except for the device ID, the code for talking to an LCD via RX/TX and via USB would be exactly the same. The converters are another significant increase in cost, though, but it says a lot about how cheap the Raspberry Pi is that a $20 cable is a "significant increase"