The concept is that the hash keys in this case are array references and hence they need to be treated accordingly, "@{$people{$person}}", this is visible by using the Data::Dumper module ...Here is my take at your post

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

my %people;
my @sentences = ("here i am","i am me");
$people{me}[0] = 0;
$people{me}[1] = "1985";
$people{me}[2] = "@sentences"; #interpolating...

@sentences = ("there she is");
$people{friend}[0] = 1;
$people{friend}[1] = "1984";
$people{friend}[2] = "@sentences";




foreach my $person (keys %people){     #each key references an anonymo
+us array.
        my $rowId = $people{$person}->[0];
        my $DOB =   $people{$person}->[1];
        my $sentence = $people{$person}->[2];
        addDOBToDatabase($rowId, $DOB);
        addSentenceToDatabase($rowId, $sentence);
        }
        
sub addDOBToDatabase{
    my @args = @_;
    print "ID:$args[0], DOB:$args[1]\n";
    }
    
sub addSentenceToDatabase{
    my @args = @_;
    print "ID::$args[0], Sentence: $args[1]\n";
    }

#Viewing data structures can help make decisions on the looping strate
+gy...

use Data::Dumper;
print "\n\UViewing the Data Structure through Data::Dumper:\E\n";
print Dumper(\%people);
[download]

Finally, here is a bunch of links that were very useful to me at clarifying the concept of references, perlref, References quick reference, Referencing in advanced data structures, perldsc and I don't understand hash references...

Updated:Added links

The code below is a guess at what you want to do - I have added parameters I feel would be necessary for your DB handling.

use strict;
use warnings;

my %people;
my @sentences = ("here i am","i am me");
  # Create $people{me} - it contains a HASH-REFERENCE
$people{me} = {ROW=>0,
       DOB=>1985,
       SPEAK=>[@sentences], # This element is an array-reference, crea
+ted from a copy of the array @sentences
      };

@sentences = ("there she is");
$people{friend} = {ROW=>  1,
           DOB=> "1984",
                   SPEAK=>[@sentences] # 
                  };

foreach my $person (sort keys  %people) # Use the KEYS function to get
+ hash keys, then sort them
{
  addDOBToDatabase( $people{$person}->{ROW}, $person, $people{$person}
+->{DOB});
  my $sentenceRow = 0;
  foreach my $sentence ( @{ $people{$person}->{SPEAK} }) # Walk throug
+h each element of the array-reference
  {
    $sentenceRow++;
       addSentenceToDatabase( $sentenceRow, $people{$person}->{ROW}, $
+sentence);
  }  
}
[download]

With help from Data::Dumper and perltidy, here is how you would normally write that

my %people = (
    'friend' => {
        'SPEAK' => [ 'there she is' ],
        'DOB'   => '1984',
        'ROW'   => 1
    },
    'me' => {
        'SPEAK' => [ 'here i am', 'i am me' ],
        'DOB'   => 1985,
        'ROW'   => 0
    }
);
[download]

Global symbol "@person" requires explicit package name at foo.pl line 
+18.
Global symbol "@person" requires explicit package name at foo.pl line 
+19.
Global symbol "%sentences" requires explicit package name at foo.pl li
+ne 22.
Execution of foo.pl aborted due to compilation errors.
[download]

use strict;
use warnings;

my %people;
my @sentences = ("here i am","i am me");
$people{me}[0] = 0; 
$people{me}[1] = "1985";
$people{me}[2] = \@sentences;

@sentences = ("there she is");
$people{friend}[0] = 1;
$people{friend}[1] = "1984";
$people{friend}[2] = \@sentences;

foreach my $person ( values %people) 
{
  my $rowId = $$person[0];
  my $dob = $$person[1];   
  my @says =  @{$person[2]}; 
  addDOBToDatabase($rowId, $dob);
  foreach my $sentence ( @says ) 
  {
    addSentenceToDatabase($rowId, $sentence);
  }  
}
[download]