I'd like to try to tidy up this discussion just a tad, especially for people who don't have time to read through all the comments and responses. We started out with the topic of verbal self-defense for kids, moved to the distinction between spoken hostile language and the two forms (online and offline) of written hostile language, and then took off on something we've been calling "hypertexting."
Proposed definitions of terms
Hypertexting, as I understand its use in this discussion, can be defined as follows: The process of producing a sequence of language which begins with a single topic and divides into one or more subparts, each of which may then divide into one or more sub-subparts, and so on ad infinitum. A common term for the language-object produced by hypertexting is "tree structure," or just "tree."
Proposed basic concept
If -- and only if -- the goal of the treemaker is to communicate a coherent meaning, the following constraint holds: No matter where the reader or listener is in the sequence of language, there must always be a clear path available for getting back to the top of the tree.
[Other possible goals include at least the following: to demonstrate skill in the use of language -- a sort of linguistic figure-skating; to have fun; to create an artistic display composed of language; or, as with the Boring Baroque Response, to deliberately prevent the communication of a coherent meaning. The constraint above does not hold, and wandering off into the never-nevers is perfectly acceptable on the part of both the treemaker and the reader or listener.]
There's no question about the ability of language users to do hypertexting in written language. Online, if you're writing an article about robots and you mention Detroit, or Louis Quatorze, you can add a link to information about Detroit or Louis Quatorze, and that information may itself contain links to information about Kansas City or rosebushes, which may itself contain ... and so on. If you're writing something more personal, you can add a link to information about the apron your Aunt Genevieve wore when baking Christmas cookies, or a recipe for cabbage soup, or a short essay on your synesthesia. The links provide a reliable path back up the tree. Offline, where there are no links, you can do a clunky and primitive and cumbersome form of hypertexting using numbered footnotes -- plus parentheses, brackets, and dashes -- to establish a path back up the tree. And then there are hybrids, like the book by Chomsky for which the footnotes are so extensive that they had to be put on the Internet, and the book has a "link" giving you the Web address for the footnotes. All of that is noncontroversial.
In spoken (and presumably in signed) language, however, things are far less clear. (At least they are far less clear to me.) A number of people have claimed in their comments to also be able to do hypertexting in spoken language. I couldn't do that, myself. Suppose I were preparing for some very important conversation; I could sit down and write out a tree structure for that conversation, laying out ways it might go and planning various ways that I might respond in each case. I often do that sort of planning before a committee meeting and just take the written material to the meeting as a reminder, for example. But that's not what the commenters are describing; their claim is that they can do that hypertexting in real time, as they're carrying on the conversation, without any need for advance planning or written language. If that's true -- and I have no evidence either that it is or that it isn't, but I am willing to apply Miller's Law and assume that it is true and try to imagine what it could be true of -- a number of questions arise.
Some of those questions are so intricate that they're difficult to ask, and I know of no technology that would make them answerable. For example: When people do spoken hypertexting, do they generate the entire tree structure in their minds all at once at the beginning of the sequence of language and then follow it (and elaborate it when necessary), or do they generate the tree as they go along? I don't plan to ask any questions of that kind.
There is a question I do want to ask, and that I think is answerable in the same way as the question "Can you do hypertexting in spoken language?" is answerable -- that is, answerable on the basis of the speaker's own subjective perceptions. And that question is: "When you are hypertexting as you carry on a conversation, are you at the same time able to listen with your full attention to the other person(s) participating in the conversation?"
It seems to me that hypertexting as you speak would require a great deal of self-monitoring and "link-accessing"; it seems to me that that would make it very hard to give your full attention to what other people in the conversation are saying. If your goal is not to communicate a coherent meaning, and/or not to have a coherent meaning communicated to you by the other speaker -- conversation, by definition, not being a one-way street -- whether you can listen with your full attention or not is irrelevant. But suppose you do want to come out of a conversation with an established coherent meaning as its result, even if that meaning is nothing more than what you and a friend want to eat for lunch. In that case, I would very much like to know whether those of you with this ability feel that you can "speak hypertext" and listen with your full attention at the same time.
Over to you...