There are three possible causes of this:
1. The data is like this in the source, before its gets to EDQ, for example the data has been garbled by ETL.
2. The Data Store in EDQ has the wrong character encoding configured. Only relevant for file-based data stores.
3. The repository database you are using with EDQ is not Unicode enabled.
Please check the above (I suggest in order).
Thank you Mike. The data store encoding is the culprit.