Loading ...
Sorry, an error occurred while loading the content.

170679Re: Plan B variations

Expand Messages
  • Alex Fink
    Mar 3, 2010
      On Wed, 3 Mar 2010 06:47:22 -0500, MorphemeAddict <lytlesw@...> wrote:

      >Has anyone actually used in a conlang the Huffman-like encodings that Plan B
      >illustrates?

      A while ago I had a sketchlang like this. I think it was around the same
      time that I joined this list; I remember folks' first reaction being "hey,
      that thing has a lot in common with Plan B" (which I didn't know of at the
      time).

      Mine was underlyingly just a bit string, with no particular chunking (by
      contrast Plan B really has four-bit elements). It was designed to be
      unambiguously parsable left-to-right without lookahead: this meant the
      lexicon was a prefix code and the syntax was Polish notation, with each word
      explicitly marked for its set of arguments. Beyond that I just organised
      the lexicon taxonomically, none of this indicating the word length by form
      which Plan B has.

      I also made a few spoken renderings. One gave each string of 7 bits a
      pronunciation, systematically but not in a way reflecting the bitstring
      well, and perversely ignored word boundary effects -- so this would have
      been completely unusable in practice, each word having seven unrelated
      allomorphs with a bizarre distribution and ridiculous sandhi effects where
      blocks spanned word boundaries. Another was based on 12-bit strings and did
      respect word boundaries, but I intentionally assigned dissimilar
      pronunciations to similar strings to avoid the "similar words sound
      confusable" issue that taxolangs usually have.

      Alex
    • Show all 16 messages in this topic