Ambiguous Spaces

[ReneZ]

This may not help, but I would argue that neither of the three approaches will lead to the correct answer.

Even if we assume (optimistically) that the majority of word spaces have been correctly identified in the various transliteration files, then many of the doubtful ones will have to go either way, and even some of the certain ones are wrong, and there will be word spaces completely overlooked.

I believe that any analysis approach that relies on word breaks is making dangerous assumptions.
That is: assuming that these breaks are indeed word boundaries.

For me, the only safe way is to take some arbitrary decision and always be aware that this will have errors.
The only hope is then that these errors are not prohibitive.

Example: the first paragraph above ends with "overlooked". If we had an incorrect space there, and it read "over looked" we would not have a real problem. Same if we had "inthe" and "ofthe" with missing spaces.

Only the ZL and GC transliterations have ambiguous spaces. If you look at Table 2 here:
You are not allowed to view links. Register or Login to view.
you will see the effect.

My recommendation is: use the RF transliteration, because it makes the problem invisible.