I'm trying to do a somewhat tricky search and replace in a tab delimited txt, and I'm getting odd results.

I want to check line by line if the contents of the first cell are repeated in the second, and if so, delete the contents of the first cell. Otherwise, the line should be reprinted with no changes.

The script I came up with does what I want it to, but in many lines, the last character of the first cell gets deleted. I don't see how this could be caused by an error in my regex, as it should either leave the whole line unchanged or leave the first cell empty; it's not supposed to mess around within the first cell. The strange thing is, I also tried sed and it produces similar errors. (BTW, I'm using Activeperl 5.10.0 on Win XP.)

Here's my script:

#!/usr/bin/perl
use strict;
use warnings;
open(IN, "<:encoding(UTF-8)", "in.txt") or die "Can't open input file:
+ $!";
open(OUT, ">:encoding(UTF-8)", "out.txt") or die "Can't open output fi
+le: $!";
while (<IN>) {
s/([^\t]*)\t([^\t]*)\1([^\t]*)\t/\t$2$1$3\t/;
print OUT $_;
}
close IN;
close OUT;
print "Replacements done";
<STDIN>;
[download]

And here's a snippet of my input file (note: my tabs seem to be converted to 4 spaces by the forum software)

abandon/to    abandon/to    
abandon/to    (a child)    
abandon/to    (a road)    
abandon/to    (claim)    
abandon/to    (motion)    
abandon/to    (one's appeal)    
abandon/to    (privilege on document)    
abandon/to    abandon rented premises/to    
abandon/to    steal or abandon/to  (personal property)    
abandon/to    vacate or abandon/to  (possession or occupation of a lan
+d)    
abandoned/to be    abandoned/to be    
abandoned/to be    appeal that is abandoned    
abandoned/to be    application that is abandoned    
abandoned/to be    motion that is abandoned    
abandoned motor vehicle site    abandoned motor vehicle site    
abandoned orchard    abandoned orchard    
abandonment    abandonment    
abandonment    (appeal)    
abandonment    (land)    
abandonment    (mines or minerals)
[download]

And the output I'm getting (see lines 3,5,6,7,13 etc. for the errors):

    abandon/to    
abandon/to    (a child)    
abandon/t    (a road)    
abandon/to    (claim)    
abandon/t    (motion)    
abandon/t    (one's appeal)    
abandon/t    (privilege on document)    
abandon    abandon rented premises/to    
    steal or abandon/to  (personal property)    
    vacate or abandon/to  (possession or occupation of a land)    
    abandoned/to be    
abandoned/to b    appeal that is abandoned    
abandoned/to b    application that is abandoned    
abandoned/to b    motion that is abandoned    
    abandoned motor vehicle site    
    abandoned orchard    
    abandonment    
abandonment    (appeal)    
abandonment    (land)    
abandonment    (mines or minerals)
[download]

Thanks for any help!
Update: solved.
As AnomalousMonk pointed out, I should have made sure the first matched substring starts at the beginning of the line, i.e. s/^([^\t]*)\t([^\t]*)\1([^\t]*)\t/\t$2$1$3\t/;