Every cat chases some dog.

There are, as well you know, two readings for that sentence. What does it look like in Thai? According to Google, like this:

สุนัขแมวทุก chases บาง

Maybe this gives us a clue about the training corpus. “Every cat loves some dog” goes through, and if you mouse over words in the resulting sentence, the original English word lights up through the magic of Javascript. This gives us a mapping like this:

Thukkhn_EVERY rạk_LOVES s̄unạk̄h_DOG mæw_CAT bāng_SOME

That looks non-projective to me. But a similar sentence might never have appeared in the training data. So let’s try something a bit simpler.”John loves Mary” and “Mary loves John” both come out as Cxh̄̒n mærī̀ rạk. “John loves to fish” is Cxh̄̒n chxb plā and “John loves fish” is Cxh̄̒n rạk plā. Don’t tell Mary.

This entry was posted in th. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *