Print

Print


On Sun, 18 Mar 2018 19:39:19 -0400, Eyal Minsky-Fenick <[log in to unmask]> wrote:

>On Sun, Mar 18, 2018 at 7:10 PM, Herman Miller <[log in to unmask]> wrote:
>
>> Has anyone made any telegraphy codes for your conlangs? - anything from 
>> semaphore to Morse code, smoke signals, or whatever? I.e., not a "code" 
>> in the sense of cryptography, but an encoding of your language's writing 
>> system in some other medium?

Fun!  I like this sort of thing, transferring between media, but I haven't yet done it into a medium with this sort of engineers' minimalism.  Instead I've got my projects like changing cuneiform into a pencil-and-paper script with glyph compositions like modern Western scripts, mapping ASL into a spoken form as close to isomorphically as manageable, etc.

>> Recently I've been thinking about tap codes, where instead of long and
>> short beeps like in Morse code, only the timing of the taps is significant.
>> I'll use a single digit to repesent a series of taps at a constant rate,
>> e.g.
>>
>> 1 = .
>> 2 = ..
>> 3 = ...
>>
>> In this code, a word like "Tirëlat" could be represented as something like
>> "112 11 4 131 12 2 112", where numbers grouped together represent sequences
>> of taps with short pauses between them, and the spaces represent longer
>> pauses, i.e.
>>
>> . . ..  . .  ....  . ... .  . ..  ..  . . ..
>
>It's probably a better idea to use a prefix-free code instead of long/short
>pauses. Self-synchronizing codes are even better.

The thing is, this *is* a prefix-free code!  You just have to analyse it the longer pauses between phone-encodings as instances of 0, so /t/ = 1120, /i/ = 110, etc.  This makes it even better than your run-of-the-mill prefix code, in that you can tell without learning the code table where a letter stops: it stops after the first 0.

In a binary channel -- like "tap" or "no tap" is -- you can't get away with ignoring timing information completely.  The only possible messages are
  on off on off on off on off ...
going on to some number of repetitions.  (Starting with "off" instead of "on" is not different, because one of those two has to be the "not-in-use" state of the channel.)  So all you can control is the number of "on"s, which exponentially expands your message.  Cf David Peterson on why a phonology, whatever its medium, must have at least one binary contrast.

The other thing is, realistically, the first code to be invented in a given culture probably won't be one that has these sort of optimality properties like self-synchronisation, at least unless their technology of application lags behind their theory far more than ours did; they'll just create _something_ and it'll become a standard.  Morse code started out even more baroque than this: in its first introduction in the 1840s, there were contrastive space lengths *inside letters*!  E.g. O = dit, long pause, dit contrasted with I = dit, short pause, dit.  

>> I'm still working out the details about what this code might look like,
>> but here's what I've come up with so far. The "*" by itself is a general
>> modifier when followed by a letter; one of them at the end of a word means
>> that something more is still coming, and two of them with a pause between
>> them marks the end of a sentence. The letters don't represent any specific
>> language, but include many of the typical sounds of Azirian languages like
>> Tirëlat and Jarda. More common sounds like "i" and "l" are represented by
>> shorter codes.

Is Azirian literate culture really (a) so accepting of the primacy of speech over writing, as it were, that they would make the crosslinguistic equivalences on the basis of sound rather than letter?  (Or are all their scripts perfect alphabets without any deviations from one-to-one sound-spelling correspondences?)  (b) so "internationalist" that their frequency distribution would be consciously based on multiple languages, rather than just the language of the inventors?

Alex