Aligning Parallel Sentences for Languages without a Written Form

Question

what are the old or current approaches for word alignment for parallel text? what is the best approach of word alignment for under resourced language or language without a written form?

so far i found about the Levenshtein distance algorithm, and a little bit about the IBM models and some others

if there is any suggestions or links that i can refer to that would be helpful

Answer 1

Without more specifics, it's hard to know what to suggest, but you may want to consider Facebook's Laser or Google's BERT, both have very good sentence embedding models that can be used for alignment tasks.