why is the name `loop` treated differently in PostgreSQL

I issue the following DDL:

CREATE TABLE foo ( loopa INTEGER, loop INTEGER ); 

Then, examining the DDL in the server (using a client-side tool like DbVisualizer - sorry don't know the query to produce it directly) I see the following:

CREATE TABLE foo ( loopa INTEGER, LOOP INTEGER ); 

Observe the change in case for the column named loop but not for the one named loopa. Looking at the list of SQL keywords loop doesn't appear to be a keyword. Why then, this selective change?

This seems to be related to a problem I am having with a trigger trying to access a column loop in a table where I get:

Exception in thread "main" org.postgresql.util.PSQLException: ERROR: record "new" has no field "loop" 

(where I get the same error regardless of whether the code in the trigger function access the field as NEW.LOOP or NEW.loop)

UPDATE

The upper-casing seems to be an artifact of DbVisualizer. psql shows:

psql (9.1.13) Type "help" for help. foo-test=> \dS+ foo Table "test.foo" Column | Type | Modifiers | Storage | Description --------+---------+-----------+---------+------------- loopa | integer | | plain | loop | integer | | plain | Has OIDs: no 

Yet this still doesn't explain why the PostgreSQL engine apparently doesn't allow my trigger to access NEW.loop. When I rename the column (and the trigger code) to loopi everything appears to work fine. So, there is likely something deeper at work here than just a DbVisualizer glitch. I'll try to create a SSCCE later.

UPDATE 2

A SSCCE for the trigger problem has been created here.