Oracle ODBC: Why are national characters changed to Latin equivalent in SELECT result

Question

I have Oracle 11 database to which I connect using both JDBC and ODBC. JDBC works well, but in ODBC all Polish letters in SELECT result are changed to Latin equivalent, for example ą -> a, Ó -> O etc. I tested it with my application and simple Python program that uses odbc module. The same value from database is returned as:

ZAMOŚĆ - by JDBC
ZAMOSC - by ODBC

My environment:

DB server: Oracle Database 11g Release 11.2.0.1.0 - 64bit Production

Client machine:

• Windows Server 2008 R2 64 bit
• Oracle clients in 32 bit and 64 bit versions in: c:\Oracle\Ora1120_32bit and c:\Oracle\Ora1120_64bit
• ODBC manager reports Oracle driver version as: 11.02.00.01

I think that all locales are set to Poland/Polish, but they are not visible by SET command line utility.

Anybody knows what ODBC or environment setting is responsible for translation of Polish letters into Latin equivalents?