Philosophical problems are aberrations that obsess us and from which a grammatical therapy should free us. The grammatical therapy intends to show us how philosophical problems have been construed through undue wordplay. And understanding the construction of the problem, and understanding that this construction is aberrant with regard to our everyday language, should be enough to destroy the problem, or loosen the grip of the obsession:

My aim is: to teach you to pass from a piece of disguised nonsense to something that is patent nonsense. (Philosophical Investigation, I, §464)

As has been remarked, Wittgenstein's philosophy, considered as therapy, is akin to psychoanalysis, particularly Lacan's psychoanalysis. They both settle on the plane of language, as if our symptoms (even when these symptoms express themselves in the body) were formed, and should be resolved, on the plane of language. Lacan points to two processes, metaphor and metonymy, which represent in language what Freud called in dreamwork, condensation and displacement. In Wittgenstein's remarks, the processes at the basis of philosophical problems may be more varied or less precisely described, but all take place at the level of language. Then both therapists (the Lacanian and the Wittgensteinian) agree that making us see the origin of the symptom, its underlying construction, is enough to destroy it.

Most, if not all, the problems that Wittgenstein thus aims at destroying are related to fear. That is obviously why they are obsessive: that things could have a life of their own, that the external world could be an illusion, that 'I' do know what 'I' am. All these are related to certain fears, even though Wittgenstein does not seem to mention the fear factor, so to speak, below philosophical problems. That would be a difference with psychoanalysis.

However, as the fear factor is never mentioned in Wittgenstein's texts, we have no idea why we bother ourselves with these extravagant problems, and why acknowledging their extravagance would free us from these obsessions. The only reason for building these bizarre problems, and these bizarre solutions to them, would be our dislike for the irregularity of grammar: we can't bear to live in a city of language for which we have no map and which, anyway, is a maze. It is not exactly that we are lost. We are not lost as long as we remain in our usual neighborhood. But this small domain we cannot even overview, and we know it is surrounded by a maze where we would get lost. In a word, we are in the dark, even though we may find our beds.

So maybe the grammatical therapy would just be a preliminary for psychotherapy: just clean up the philosophical problem before going into the real stuff: why are you afraid of the dark, and what beasts do you fear would be coming out of the darkness? It is worth trying, I guess.

However, in the end, both therapies have the same drawback: they aim at destroying our phobias, and replacing the magical world of phobias with the plain world of common sense. Things have no life. The room around me is not an illusion. The unknown beyond would just look the same. Something has been lost (provided the therapy works), and lost before it could even be described.