Remarks 33-39 address various issues regarding ostensive pointing. The following relates specifically to the first few sentences in the boxed text between Remarks 35 and 36 which address the ambiguity of the statement
That is blue.
when uttered while the speaker points at a blue vase. Is the speaker ascribing the color "blue" to the vase or indicating that the color of the vase is an example of the color called "blue"? Or more generally, when pointing to a location and asserting "That is f", is the speaker ascribing feature f to the location? ascribing f to an entity that currently occupies the location? indicating that a feature of either the location or its current occupant is an example of f? other?
One can view the use of names and words like "this" and "that" as verbal pointing, in which case such words are somewhat analogous to a type of computer programming variable called a "pointer variable" (alluded to in my discussion of Remarks 8 and 9). The value of a pointer variable is the location in computer memory (ie, the "address") of some other variable. To say that a pointer variable p "points to the variable x" is to say that the value of p is the location (in computer memory) occupied by the current value of x.
There is a issue with pointer variables that is similar to the ambiguity of verbal pointing, viz, how does one distinguish between the location of x (its address) and the occupant of that location (the current value of x). In the "C" programming language, this is done by prepending an asterisk to a pointer p to yield the composite symbol *p, which is interpreted as "the occupant of the location pointed to by p". Then the current value of the variable x can be retrieved from memory by writing either x or *p.
Given this analogy, what seems to be missing from verbal pointers such as "this" and "that" is a way of indicating the target of the speaker's verbal pointing. But such indicators aren't really missing, they're just not being used in the statement in question. To make clear what target is intended, a speaker need only augment the statement so as to make the target explicit. Eg:
That color is blue. or That vase is blue.
Assuming that whatever is in the location being pointed to is monochrome (colored object or ambient light), the first statement is unambiguous since it essentially says "the color of the occupant of the location is called "blue". However, the second sentence is ambiguous as to what feature of the occupant (in the example, a vase) is the referent (eg, it could be the vase's shape.) Thus, since the color of an object can be thought of as a feature that "occupies" the location constituted by the object, the second statement really should be:
That color on the vase is blue.
(Could it be that in some situations the problem isn't that language has "gone on holiday" but that the speaker has?)
To complete the analogy with programming pointers, one could express these augmented verbal statements symbolically by interpreting "that" as a pointer variable and indicating the intended target by prepending an identifier of the intended target. The resulting disambiguated statements can then be represented symbolically like this:
color-that = "blue" or color-vase-that = "blue"
These can be interpreted as "color-value in the location pointed to by 'that' " and "color-value in the location (on the object) in the location pointed to by 'that' ", respectively.
My programming days are ancient history, but I vaguely recall that if the variable x is itself an address, to call a value in the location pointed to by x you have to similarly expand the symbology to
*x = *(*p)
which in the "C" programming language is abbreviated to **p. In words, this is "the value in the location in the location pointed to by p". Comparing this with the symbolic description of the verbal pointers, the analogy appears to hold.
And just as verbal pointers distressed Wittgenstein, pointer manipulation distresses "C" programmers (at least it distressed this faux-programmer).