As regards any possible case for encoding localizable sentences *as
characters*,
in my opinion, the train long ago left the station for that one.
Indeed, people have been devising systems for representing words and
sentences via ordinary numbers that worked just fine for at least 170
years.
<link to a book with telegraph codes>
You are reminding me of these books with pictures for travelers to point
at, such as "Picture Talk" by Langenscheidt and many products by
Kwikpoint. There seem to be quite a few books like that on the market
now, often small and laminated. I don't remember seeing them in the
1990s, so I'm wondering how old this invention is.
While all these approaches suffer from the problem of imprecision, their
relative usefulness (I'd /really/ want to see which of the codes in such
telegraph books were put to practical use, and how) comes about by them
not attempting to enumerate sentences but rather phrasal components to
stick together.
The closest to a modern, electronic thing I can think of would be an app
that walks you through a hierarchical menu with phrases and pictures. If
graphics like those from William's High-Logic threads are used, there
will be a steep learning curve, so one would really need pictures or
simple, iconic representations. So such an app would essentially become
a user interface for an MT system with a restricted range of expression.
Not obviously "not useful", but it might be a hassle, and
straightforward MT might be faster – but it'll depend on the intended
domain of expression. Note that this is not a model that enumerates
sentences, and it's far from Unicode. If the concepts and sentence
fragments are not used for showing to someone else or for input but
instead for transmission, one can assign them internal codes; still the
result would be outside of the scope of Unicode.
Stephan
