Jess - the raw SQL you need (at its most basic) is:

select party_id, description from contacts where arch_short_name = 'contact.email';

The above will extract every email address. A more targetted query is:

select c.party_id as id, c.description as email, e.name, e.description as address from contacts as c join entities as e on c.party_id=e.entity_id
where c.arch_short_name = 'contact.email' and e.arch_short_name='party.customerperson' and e.active=1;

which grabs only active customers and provides their names and addresses also.

Attached is a jrxml report file that incorporates the above. You can create a template that uses this (remember to make its TYPE=Report so that it will show up in the Reporting|Reports list). You can then run the report and click the Export button to get a CSV file that can be imported into a spreadsheet.

NOTE: while testing this, I found what I think is a bug in the export code -- see http://www.openvpms.org/forum/exported-csv-data-has-extra-fields

As a result, I adjusted the report not to include the address - and this generates a CSV file which happily imports into Excel. [Note that if you are wondering why the column header for the ID column is "CID" rather than "ID", its because of a bug/funny in Excel which causes it to be confused if the first column is named "ID".