Since a finite language is a finite set of words, one way to specify the language is to list all the words (although more efficient ways might exist). A recognition algorithm can compare the input word with each. While inefficient for large languages, it is still theoretically possible. Sometimes, there are more efficient ways.