For "A is B" the explanation is simple.
Gensler's language has two types of "basic" formulae:
(i) formulae expressing relation between sets ("general categoris"): "All logicians are charming", translated as "all L is C"
(ii) formulae expressing the fact that an individual belongs to a set: "Gensler is a logician", translated as "g is L".
In this second case, g is the name of an individual; thus, we cannot quantify it with "all" or "some".
In the previous case, instead, L and C are names for sets and we have to quantify the first one in order to correctly express the relationship between them. If we say "Logicians are charming" (i.e. L is C) we have an ambiguous expression, because we do not know if we are asserting it of all or some Logicians.
I presume that he forbid the expression "all A is not B" as "ungrammatical" (non-wff) simply because it is already expressible as: "no A is B".