Solutions [discussion thread - moved]

[nablator]

A simple substitution based on the character frequency of the target language; then hammering whatever garbage you get into target-language words (I used a dictionary and edit distance).

There are some tricks, often used, that ease the "hammering" step:
- reduce the target alphabet: confusion of some vowels, similar sounding consonants, or allow the omission of some vowels (abjad),
- map similar-looking glyphs (with or without tiny differences) to different letters,
- creatively decide where to add and remove spaces,
- use not one but many dictionaries, several languages, dialects, also match roots of words (shorter = easier to match).