A lot of good shields will give you the option to use hardware or software serial, and that is what's happening here.
While you think you see four pins on the arduino side and six on the jumper side, what you're actually seeing is (yes) the four pins on the arduino, but those four pins go to four pins on the jumpers.
The other two pins on the jumpers (the ones in the middle) go to the TX and RX of the shield. The shield itself only has one TX (TXD0 pin 2 on JP4) and one RX (RXD0 pin 2 on JP6). So the jumpers form the connection between your arduino and the shield. This is why they are the middle pins, because you put the jumper on the first two, or the last two, depending on what it is you want to achieve.
You will note the net names of pin 2 of JP4 and JP6 relate directly to (and are connected to) pin 19 and 20 respectively of the chip on the shield.
The whole benefit of the jumpers is you connect JP4 and JP6 pins 1 and 2 to which closes the circuit and uses the inbuilt hardware serial (pin 1 of the jumper being arduino hardware and pin 2 being the TXD0 source from the shield). Connecting pins 2 and 3 means you close the circuit between the shields TX and RX to the arduino GPIO pins D2 and D3 which you can assign when initialising your software serial.