20-09-2020, 02:57 PM
One more brief comment about my efforts on VCI, whatever one thinks of the results:
My hope is that it gives those of us who are not computer programmers or coders a way to visualize at least some of the things that Koen, Marco, Rene, Nick, Emma, and others are analyzing. It's one thing to say, "Here are the results of a program where [ok], [ot], [od], etc., are treated as single units." It's another thing to present a visual expression of that analysis with a transcription, alphabet, interpretation, or whatever you want to call it, where the reader can see [ok] as an actual single different letter, [ot] as an actual single different letter, [od] as an actual single different letter, and so on. That was one of my aims in developing the VCI thing, whatever label you want to give it.
I'm sure VCI can be amended and improved. As I admitted, the treatment of EVA [ar] was inconsistent. Perhaps [qo] and [qok] need to be distinguished again somehow. I discussed other more minor possible changes in my previous post. But the point is, at least this approach gives us a visual alphabetical way to see some of the results and effects of all of these very interesting statistical analyses that Koen, Marco, Rene, Nick, Emma, and others are performing (sorry if I left out anyone important). For some of you computer programmers and coders, just saying "This set of regular adjustments made the h2 conditional entropy 3.01 and kept the h1 entropy at 4.12!" may be all you need to grasp the significance of the analysis and its result. But for many of us, seeing what an alphabetical rendition of such a transformation will look like in a natural visual way, in reading some lines of the actual ms text where each of Koen's single units looks like an actual single letter, or in reading the 12 labels of a circle in the actual text in that way -- that is what shows the rest of us the actual impact and significance of "3.01 h2 / 4.12 h1" entropy statistics.
Perhaps we can find a better way to accomplish this aim, one that is more rigorously consistent with Koen's and Marco's analysis, but which still allows all of us to read the actual ms text in this new way, where the new single units look like actual single letters, and the whole thing looks like the letters of a natural language text just as much as we would expect a text with 3.01 h2 conditional entropy and 4.12 h1 entropy to actually appear.
Geoffrey
My hope is that it gives those of us who are not computer programmers or coders a way to visualize at least some of the things that Koen, Marco, Rene, Nick, Emma, and others are analyzing. It's one thing to say, "Here are the results of a program where [ok], [ot], [od], etc., are treated as single units." It's another thing to present a visual expression of that analysis with a transcription, alphabet, interpretation, or whatever you want to call it, where the reader can see [ok] as an actual single different letter, [ot] as an actual single different letter, [od] as an actual single different letter, and so on. That was one of my aims in developing the VCI thing, whatever label you want to give it.
I'm sure VCI can be amended and improved. As I admitted, the treatment of EVA [ar] was inconsistent. Perhaps [qo] and [qok] need to be distinguished again somehow. I discussed other more minor possible changes in my previous post. But the point is, at least this approach gives us a visual alphabetical way to see some of the results and effects of all of these very interesting statistical analyses that Koen, Marco, Rene, Nick, Emma, and others are performing (sorry if I left out anyone important). For some of you computer programmers and coders, just saying "This set of regular adjustments made the h2 conditional entropy 3.01 and kept the h1 entropy at 4.12!" may be all you need to grasp the significance of the analysis and its result. But for many of us, seeing what an alphabetical rendition of such a transformation will look like in a natural visual way, in reading some lines of the actual ms text where each of Koen's single units looks like an actual single letter, or in reading the 12 labels of a circle in the actual text in that way -- that is what shows the rest of us the actual impact and significance of "3.01 h2 / 4.12 h1" entropy statistics.
Perhaps we can find a better way to accomplish this aim, one that is more rigorously consistent with Koen's and Marco's analysis, but which still allows all of us to read the actual ms text in this new way, where the new single units look like actual single letters, and the whole thing looks like the letters of a natural language text just as much as we would expect a text with 3.01 h2 conditional entropy and 4.12 h1 entropy to actually appear.
Geoffrey