Wittgenstein's famous notion of a "family resemblance" is introduced obliquely in the opening discussion of language-games in the Philosophical Investigations. There, the question, "But how many kinds of sentence are there?" implies the eventual unstated answer, "Just look and see how they are related."
In order to bring this answer out implicitly, Wittgenstein 'tries' to reduce all utterance to the indicative mood by making explicit the framework of report which is supposed to insensibly accompany all utterances (i.e. "I want you to...," "I would like to know...," etc.). He rightly observes that this reduction does not help to "bring the different language-games together." The resemblance among them and their innate common referability to the indicative must be presupposed for them to serve as arguments of that frame made explicit.
The significance Wittgenstein sees in the possibility of reduction to the indicative (what solipsism means to say but can't) is doubtful. The analysis here of the common referability of different kinds of sentence completely ignores the findings of the previous sections. The test-cases of highly primitive languages make it more plausible that the common mood of utterance would be on a spectrum from invitatory to imperative. The teaching of solipsism itself could be expressed, "Invite this thought into your head: I am everything."
The above example shows that the correlative of invitation/command is not necessarily acceptance/obedience, at least insofar as the latter is taken to express a completely passive position. Acceptance/obedience must itself be reducible to invitation/command, not reciprocally doubling back on the speaker but experimentally or expectantly or hopefully transmitting the same mood or spectrum of moods forward into reality--urging it to respond to my will, soliciting its meaning, demanding its adherence to a physical model, expecting it to resonate with joyful surprise (as when I laugh at a joke)--and so on. The essence of language shows up clearly in the transmissions of its original mood in a way that renders not hopeless the project of scientifically comprehending the common aspect under which all appear (provided that science, too, can be reduced to the original mood).
Furthermore, the transitivity of acceptance/obedience can be transmitted back into the invitation/command, which itself must have listened to something more original.
Two clarifications: first, a 'spectrum' of mood does not repeat the problematic of an indeterminate diversity of functions. It is one mood with two poles--the orientation to one or the other of which is the whole field of ethics. Second: on this understanding, the equivalence of solipsism and realism (as treated in the TLP) has no place. The essence of language corresponds to a sol-"us"-ism, where the "us" is indeterminately defined but must include whatever is most original. The listener is as much the limit of the world as is the speaker. In the face of this contradiction of multiple limits, language ranges from a sociopathic campaign to regularize the conflict of limits according to my own, to a late, lyrical lament arising out of the pain of contradiction between world and self.