This was the headline of an article in NY Times Feb. 11. When I looked at it, before reading the article, I realized it had two completely different possible interpretations.
(1) Americans who are like me.
(2) Americans who do like me.
Is this ambiguity unavoidable? It turns out the article was about (1).
Yes, semantic ambiguity in natural language is unavoidable.
There's a standard example in the theory of natural language processing. We'd like to be able to parse a sentence mechanically. Here's the subject, here's the verb form, here's the object. But that doesn't work. The example is,
Time flies like an arrow.
Fruit flies like a banana.
In the first sentence, the subject is "time," the verb is "flies," and then there's a comparison via "like." Easy to parse.
But in the second example, the subject is "Fruit flies." How would a program know that? It has to know that fruit flies are a type of insect attracted to fruit, and that therefore they would "like" a banana. In this case like is not the start of a comparison of two things; rather, it's the verb! And "fruit flies" must be read as one single thing, rather than a subject and a verb.
How could a syntax parser ever figure this out? The answer is that it can't. This is one reason Google translate is so bad. Natural language processing is a very difficult problem for computers. But an easy one for human toddlers. Let the computationalists explain that!