(12-08-2020, 01:36 PM)ReneZ wrote: You are not allowed to view links. Register or Login to view.
...
It is immaterial whether one transliterates e as 'e' or 'c'.
I agree, in terms of transliteration. It makes no difference. The computer doesn't care. However, in terms of a researcher's psychological perception of the text, it makes a difference.
Quote:Coming back to the title of this post, it is very tempting to map 'expectations' of the meaning of some glyphs to the way they are transliterated.
This is why I don't like mapping EVA-e to the "e" character when it is shaped like "c". It creates a psychological 'expectation' that EVA-e is a vowel. Some computational attacks have specifically treated it that way. Copying the shape is more 'expectation' neutral.
Quote:It is immaterial whether one transliterates Sh as 'sh' or 'Sh'.
If EVA-S S is used to transliterate a portion of one VMS glyph and EVA-s s is used to transliterate a completely different VMS glyph with different statistical properties, then it matters if they are treated the same because they are not the same. In the Basic EVA set, they are recognized as being different. In the Takahashi transliteration that uses capitalization consistent with the Basic EVA set, they are recognized as being different.
If they are remapped, they need to be remapped to two different slots, not collapsed into one (which is what happens if people are given the impression that they are the same).