• use strict and warnings and clean up the messages that result.
• Run your code through perltidy to clean up the indentation.
• Post your expected output. I don't understand what you are trying to accomplish.

A beginner needs to use strict;
An expert wants to use strict;

Hi, your shebang looks broken... Shouldn't it be #! |path|of|interpreter (1)?

your subject says "ignoring blank lines", your text says "leave out those lines which has one partner"?

So, what do you want to do?

What do you mean with "partner"? Are you talking about words (partner == word)?
Do you simple want to print those lines with two words?
Maybe only those lines with two words, where the words are the same?

The better you explain your problem, the better others can help you.

Just a guess:

#! /usr/bin/perl
use strict;
use warnings;

# read magically from given files or STDIN; see perldoc perlop (null f
+ilehandle)
while ( my $line = <> ) {
    # split each line
    my @elements = split /\s+/, $line;

    # two words per line (and words are equal)
    if ( 2 == @elements && $elements[0] eq $elements[1] ) {
        print $line;
    }
}

# just for testing
__DATA__
file
ass ass
array
rat rasa
arrow ant
sad
[download]

Not totally sure what you are after, but your code doesn't compare and isn't well laid out - see perlcritic and perlstyle. As a beginner (or anyone else) you should always use warnings and strict. You may also find diagnostics helpful - it will give you fuller explanations of warnings and their possible origin.

Making a few guesses as to what you wanted (two identical elements per line?) I came up with the following:

use warnings;
use strict;

use Data::Dumper qw/Dumper/; # very useful for debugging
++$|;
my @array = (<DATA>);

# check input
print "INPUT :\n" . Dumper \@array;

foreach my $line (@array){
    chomp $line; # remove trailing newline
        # ignore empty lines
    next unless (defined$line && $line ne '');   
        # split on space      
    my @t = split /\s+/, $line;
        # accept lines with two, identical elements
    if ( scalar @t == 2 && $t[0] eq $t[1] ){
        print "MATCH : $line\n";
    }
    else {
        print "NO : $line\n";
    }
}
# gives -
#INPUT :
#$VAR1 = [
#          'file
#',
#          '
#',
#          'ass ass
#',
#          '
#',
#          'array
#',
#          '
#',
#          'rat rasa
#',
#          '
#',
#          'arrow ant
#',
#          '
#',
#          'sad'
#        ];
#NO : file
#MATCH : ass ass
#NO : array
#NO : rat rasa
#NO : arrow ant
#NO : sad
__DATA__
file

ass ass

array

rat rasa

arrow ant

sad
[download]

There are shorter ways to do this using lookaheads and regular expressions but it is usually best to stick with something clear! HTH.

You could do this on the command line, a script hardly seems necessary.

$ cat spw841608.txt
ile

ass ass

array

rat rasa

arrow ant

sad

  jack sprat

cat on a mat

joe blow
$
$ perl -ne 'print if 2 == split' spw841608.txt
ass ass
rat rasa
arrow ant
  jack sprat
joe blow
$
[download]

I hope this is helpful.