Just to elaborate a bit on the hash approach, here's a very simple example of how you would populate the hash and then look up some value(s):

#!/usr/bin/perl

my %dict;  # the hash

while (my $line = <DATA>) {
    chomp $line;
    $dict{$line}++;  # instead of ++ you could also assign some value.
+..
}

my @inputs = qw(
foo
fooed
fooen
prefoo
postfoo
);

for my $input (@inputs) {
    print "found '$input' in lexicon\n" if exists $dict{$input};
}

# I'm using the special DATA filehandle here to be able to inline it..
+.
# That would be your DICTE handle supplying all the precomputed 385090
+ lines

__DATA__
foo
prefoo
bar
baz
...
[download]

BTW, am I understanding you correctly, that what you mean by 'analyse' is essentially to check whether some given $input is found in the lexicon?