Build a hash of keys keyed by cell number from the match file then run through the search file a line at a time looking for matches:

#!/bin/usr/perl
use strict;
use warnings;
use Text::csv;

my $csv_keys = <<CSV1;
blue,lagoon,moon
red,banana,sun
CSV1

my $csv_phrases = <<CSV2;
blue lagoon under a summer moon,still water,east of the sun
hundreds of lines,several entries separated by commas,thanks in advanc
+e
CSV2

my %keys;
my $csv = Text::CSV->new();

open my $keysIn, '<', \$csv_keys;

while (my $row = $csv->getline($keysIn) ) {
    for my $cellNum (1 .. @$row) {
        $keys{$cellNum}{$row->[$cellNum - 1]} = $.;
    }
}

close $keysIn;

open my $phrasesIn, '<', \$csv_phrases;

while (my $row = $csv->getline($phrasesIn) ) {
    for my $cellNum (1 .. @$row) {
        for my $key (keys %{$keys{$cellNum}}) {
            next if $row->[$cellNum - 1] !~ /\Q$key\E/;
            print <<MATCH;
Matched '$key' in cell $cellNum from keys line $keys{$cellNum}{$key} t
+o phrases line $.
MATCH
        }
    }
}

close $keysIn;
[download]

Prints:

Matched 'blue' in cell 1 from keys line 1 to phrases line 1
Matched 'sun' in cell 3 from keys line 2 to phrases line 1
Matched 'red' in cell 1 from keys line 2 to phrases line 2
[download]

It's assumed that the entire match string needs to match and that it should only be matched in the same "cell".