[sword-devel] Musings about the Cherokee NT module

[David Haslam]

Has the reversibility of the Cherokee transliteration been thoroughly tested,
Chris?

While I was considering the back conversion it became apparent that the
Cherokee to Latin transliteration would include real instances whereby the
same Latin output would be generated by different Cherokee syllable
combinations.

My solution was to insert a (zero-width) U+2060 WORD JOINER in just those
identified combinations in order to prevent the Latin being incorrectly
parsed at the time of back conversion.

Before we look at the larger task of matching proper names, it would be an
interest exercise to test the two way conversion method involving icu-sword.

btw. Some transliteration capability is built into Xiphos 3.1.5 for Windows,
but Cherokee was not one of the languages covered when this version was
compiled for release.

David

PS. Of course, one could go beyond my minimalist approach, and insert a
visible divider between every syllable, but that seemed overkill to me. I
have done that also.  For the visible syllable divider, I made use of U+00B7
MIDDLE DOT. The presence of such syllable dividers reduced the relative
visual impact of ordinary spaces between words, so I converted ordinary
spaces to U+2003 EM SPACE for that alternative method, merely to improve
readability.

--
View this message in context: http://sword-dev.350566.n4.nabble.com/Musings-about-the-Cherokee-NT-module-tp4650474p4650478.html
Sent from the SWORD Dev mailing list archive at Nabble.com.