Well, they're slower if you look at toe speed, but that isn't how they
are rated.

You can check Ilford's curves:

http://www.ilford.com/html/us_english/pdf/galerie_fb.pdf
http://www.ilford.com/html/us_english/pdf/ilfospeed.pdf

The first is for Galerie, and the grades have the same speed
and cross-over point, though toe speed varies. The second is
for Ilfospeed, where G1-4 have the same speed, however, G5
has a different composition and is much slower.

If you consider two Bell or Gaussian curves with the same average
grain size (same peak), but with different size distributions (width),
one can reasonably conclude that the one with the larger distribution
will have greater toe sensitivity, but lower contrast. If one takes
the curves, uses an X-axis where size decreases going to the right, and
transforms them to a cumulative frequency distribution form 0 to 100%,
you'd generate two "S" curves. They would cross-over at 50%, The one
with the broad distribution would have a long toe, faster toe speed,
and less contrast. If log exposure were plotted vs optical density
(a log-log plot), you'd have the H&D curves that one usually sees,
and are generated by experimental data. The resemblance of the two
isn't due to chance.

Now there is the G5 Ilfospeed outlier, which very likely is a very
fine grain slow emulsion with a narrow distribution. However it isn't
the norm. Graded papers with the same ANSI speeds are/were more common.

The early multigrade papers and very likely Ilford's current crop,
use the same size distribution (speed and contrast) for the high
and low contrast components. A portion is sensitized only to blue,
and another portion is sensitized to blue and green. With a low
contrast yellow filter, the component sensitive to only blue is not
exposed, and only the "1/2" sensitized to blue and green is exposed.
Fewer grains exposed = less density per unit exposure = lower contrast.
With a blue filter, both components are exposed, and the greatest
density per exposure unit results (highest contrast). Intermediate
filtration results in different ratios. (Preceding is Ilford's
explanation of how their MG papers work.)

I don't recall data on paper grain size or coverage vs pH.

You might want to check my "solarization" reply and link to Yurow's
page through the chemigram link page. He lists a number of properties
that are related to grain size, color, etc. Metol really is not a fine
grain developing agent per se (hydroquinone is, BTW), and the deep
cool blacks are due to large grains. The exposure increase or a
long development time doesn't hurt.

With films, "grain" is related the holes between clumps of developed
grains, and non-uniformity of grain distribution due to clumping.
At low pH's, the hardeners in the emulsion are more stable, and
with high ionic strengths, emulsion swelling is diminished. That leads
to less grain clumping, and more even distribution ("smaller" holes).
High sulfite and metol causes the grains to develop as filaments (increasing
size) which also "fills" the holes.

The "warm tone" paper developers (fine developed grain size) often use
hydroquinone as the only agent, and activation requires a high pH. The
situation is different from film. Actually the paper that may have had
the best coverage or highest reflection density reading was warm tone
Portriga. Unfortunately, one couldn't take advantage of that fact, since
the densities were so high that the eye couldn't distinguish one black
from a deeper black.