Lee, you and Helen are both right but with qualifications.
The toe is losing speed more rapidly than the shoulder, and so contrast goes up (in most cases), but imagine a film made of a blend of 2 emulsions. If the toe of both lose more speed, then you see higher contrast in the low density regions and lower contrast in the higher (mid scale) regions due to speed loss of the second component toe speed loss.
I know that this is a complex argument, but the result is higher contrast, normal contrast lower contrast normal contrast as you measure the curve, and the result is a bumpy curve. I have run into this building C41 films. They have 3 components in each layer.
The solution is to solve the reciprocity failure problem as much as you can, and that has been done in most modern EK films. In fact, it has been solved in most modern films.
Also, this assumes a 'straight' emulsion. I have seen emulsions which behave in the opposite direction due to other reasons such as spectral sensitivity and finish, and I have seen graded iodide emulsions which behave in a peculiar fashion as well. So, there is a general rule to use, but I would suggest individual tests on each film to be sure of the amount of change in speed, contrast and in the direction of change.
PE