Why do the language codes in the metadata change?

When you submit Language property metadata to DataCite in XML and JSON format, you might have noticed that there is a normalization in place which is reflected in the REST API. This means that if there is a three-letter ISO 639-3 language code or a ISO 639-1 language code with a subvalue provided, the system will automatically update it to the matching two-letter ISO 639-1 code.

This is the expected behaviour at present, but we are aware that this can be confusing. We welcome feature requests, metadata schema feedback, and other suggestions through the DataCite Suggestions forum on GitHub.