Natural language processing

Words can be ambiguous when context is omitted (or distorted).

Time flies like an arrow.
Fruit flies like a banana.

In this case, the former "flies" is a verb while the latter "flies" is a noun.

This is a linguistic example of syntactic ambiguity sometimes called a "garden path sentence".

Meaning: "I don't smell good" vs. "I don't smell well".

More: Word ambiguity

There was a farmer had a dog and bingo was his name, ...

Who is "bingo"? Aside: Was the farmer named bingo or was the dog named bingo?

Invitation: We would like to invite you over for dinner.

Who is the invitation from? Aside: Is the invitation from friends or cannibals?

How is a computer supposed to decide if a human cannot understand what is meant?

Consider the following sentence.

The man saw the woman with the telescope.

Who has the telescope? Does the woman have the telescope? Or, is the man using the telescope to look at the woman? So, truth can be elusive as even language can be ambiguous.

Natural language is inherently ambiguous. It depends on what the definition of "is" is. Bill Clinton's grand jury impeachment testimony, 1998.
When a human cannot understand the sentence, how can a computer be expected to understand the sentence?

More: Word ambiguity

Here is a paragraph of recommendation.

To whom it may concern: You wrote to ask me for my opinion of John, who has applied for a position in your department.

I cannot recommend him too highly nor say enough good things about him.
There is no other student of mine with whom I can adequately compare him.
His thesis is the sort of work you don't expect to see nowadays and in it he has clearly demonstrated his complete capabilities.
The amount of material he knows will surprise you.
You will indeed be fortunate if you can get him to work for you.

Source: Paulos, J. (1995). A mathematician reads the newspaper. New York: Basic Books., p. 43.

That is, while (many) computer languages are designed to be unambiguous, natural language is inherently ambiguous.

More: John Allen Paulos

Projects promoting programming in "natural language" are intrinsically doomed to fail. Edsger Dijkstra (computer scientist)

More: Edsger Dijkstra

Natural language processing is concerned with understanding natural, or spoken, language. Unlike programming languages, such as Pascal, C, Visual Basic, FORTRAN, and COBOL, which are very well defined, natural languages are ambiguous. For example, "Time flies like an arrow. Fruit flies like a banana." How would you program a computer to recognize that? English is especially ambiguous, providing opportunities for many plays on words jokes (also called puns). A part of a natural language processing system would be a system to convert voice to text. Again, the ambiguity of natural language creates problems. Commercials that claim to recognize speech in order to assist in making phone calls are based on very simplified natural language systems where as much ambiguity as possible has been removed. In some cases, the person must train the system with a limited vocabulary so that the system can recognize that particular voice.

Example: How would you interpret the following.

The man saw the woman with a telescope. Who has the telescope?

Give an example of the ambiguity of the natural language English.

One potential use of voice recognition and natural language processing is that of converting voice to text. Why convert voice to text? Well, think about all of the information that is available in audio form. How can one do a search to locate instances of certain phrases, expressed as text? This is a difficult problem that could be at least partially solved with the use of voice recognition and natural language processing.