Let's perform a theoretical thought-experiment.
We start with two brass terminals attached to some suitable insulating material, say
ebonite (as used by all the pioneering electrical engineers in the 19th and early 20th centuries).
To those two terminals we attach a "measuring device" which, depending upon its configuration, can accurately measure both very small and very large resistances and capacitances.
We have available a box of "perfect" quality resistors, ranging from 1 Ohm to 10000 MOhms. Each resistor is connected, in turn, to those brass terminals and the reading on the "measuring device" is noted.
We have available a box of "perfect" quality capacitors, ranging from 0.001 pFarad to 1000 mFarad. Each capacitor is connected, in turn, to those brass terminals and the reading on the "measuring device" is noted.
At some point in the experiment with the resistors we will be seeing readings (with the very high resistances, tending towards infinity) that when plotted out against some appropriate ordinate match that of the experiment using the capacitors (with the very small capacitances).
So in our "thought experiment" we see that very large resistances can be approximated by very small capacitances.
Now coming back to the real world of joints in a metallic pathway over which both telephony and xDSL signals are carried. I'll choose the fifth joint counted from which ever end of the circuit that is more convenient. That joint is perfect. To all intents and purposes its resistance is zero Ohms, its capacitance is zero Farads and it is perfectly symmetric in its signal carrying ability. Now, as we are in the real world, that joint can go faulty. It might develop high resistance tendencies, it might become more capacitive than resistive, it might become asymmetric in those tendencies (i.e. become semiconducting). Of those defects the most common is HR, which is then followed (in the order of appearance) by semiconducting.
A HR joint will predominately attenuate across the entire spectrum of interest. A semiconducting joint will predominately attenuate across the entire spectrum of interest and add intermodulation products as a result of combining, asymmetrically, wanted and stray signals. A capacitive joint will selectively attenuate the lower frequencies rather than the higher. All three mentioned defects will attenuate the (wanted) signal but in different ways and, quite often, in frequency dependent fashions.