It's hard to understand what you're asking for. You need to be careful about using $read without the -nt switches, as you can cause a line containing the text $version be replaced by the result of that identifier.
If you are wanting to find out how many lines of a text file is exactly a known word:
//filter -fw
versions.txt @test word | echo -a $filtered
Number of lines containing a string:
//filter -fw
versions.txt @test *word* | echo -a $filtered
(Note this finds password and sword)
Number of lines beginning with a word and a space and other words:
//filter -fw
versions.txt @test word * | echo -a $filtered
If you want to scan a file for whichever is the most common line, you could make a hash table or @window to contain each word found, along with a counter of how many times that word/line appears.
edit: instead of creating a junk @window for the output, you can use the bit-bucket $null alias:
//filter -fk
versions.txt null word | echo -a $filtered