The primary purpose of the mapper is to emit data that the reducer will later aggregate into a final. In this trivial example, each occurrence of a word results in a (word, 1) pair since we're giving a single appearance of a word a score of 1. We very well could have emitted the word by itself (that is, 'amet') and put the score of 1 in the reducer; however, this would have made the code less general. If we wanted to give a heavier weighting to certain words, we'd merely change our mapper to output a different number based on the word and would leave our reducer code as-is.
The following code block shows how we would give the word amet a score of 10 while all other words count as 1. Of course, this is no longer counting ...
