The contrast of a lens is a measure of its ability to discriminate tonally between small adjacent areas within an image.
The ability of a lens to distinguish between different tones within a narrow range, such as highlight-to-highlight or shadow-to-shadow, is known as local contrast. It is this ability that is reduced by a soft-focus filter. Local contrast has nothing to do with the overall contrast within a scene. Without good local lens contrast, an image tends to appear muddy, flat and lifeless. Note that contrast can also be reduced by flare.
High-quality lenses have good resolution and good local contrast. Simpler lens designs, incorporating fewer elements, generally have higher contrast than those with numerous elements. To a large extent this explains why prime lenses achieve the highest levels of optical quality.