I am not sure how hashing works exactly, all I know it is impossible to reverse it. All hash types have X amount of characters in them while the phrases and made up words can have an infinite quantity hence infinitely more times than the character counts in a hash. So do hashes repeat themselves?
Yes, it's possible because of the limited length, but it has a very little chance. Read a bit about hash collision, for example: http://preshing.com/20110504/hash-collision-probabilities/
The thing that you are asking about is called hash collision.
To make it extremely simple let's say your hash has 1 byte (=8 bits). This means that you can assign a distinct hash to at most 2^8 messages. If you had 2^8 + 1 (257) messages you would definitely experience a collision.
Now let's fast forward to todays situation. MD5 for example has 128 bits. This means it can encode 2^128 (3.4028237e+38) messages at most. So if you have a good algorithm (this is very important) your collision rate will be negligible and you will be able to assign a distinct hash to almost every of 2^128 distinct messages.