Oxford Information Labs study for EURid highlights failure to support Greek script.
The research team behind the World Report on Internationalised Domain Names, recently conducted a report on behalf of EURid to investigate the performance of multilingual email addresses across leading email providers. The study found that despite internationalised domain names (IDNs) having existed for over 20 years, there continues to be no interoperable email client that can facilitate effective support in European language scripts. By this, we mean that there is no interoperable platform to allow European language speakers to use email in their native script.
EURid’s intention in commissioning this research was to identify a suitable email vendor to recommend to its network of registrars that supported all European language scripts. What we found was that whether the native script is Latin, Greek or Cyrillic – there is currently no product that offers fully interoperable support for internationalised email addresses across. Failures are particularly apparent in Greek script.
What is currently on the market?
The current market for the sending, receiving and replying to email addresses across European scripts is complex. There are specialist tools, such as XGen Plus, that we found worked reasonably well within a closed environment – ie when when both the sender and receiver operated on the same platform. When we began to investigate the more widely used providers, the household names of Outlook, Mac Mail, Thunderbird, Outlook Online and Gmail, we observed significant limitations to EAI at present for every script that we analysed.
What did our study involve?
Our main task was to understand and evaluate the process of sending, receiving and replying to emails across Latin, Cyrillic and Greek scripts, as these three scripts support all official EU languages. To do this, we established a control environment, and then measured the performance of two specialist EAI providers (XGen Plus and CMail Pro), when interacting with other popular commercial email providers. The initial intention of our study had been to simply evaluate XGen Plus and CMail Pro as specialist EAI tools. What became clear very swiftly, was that there are fundamental flaws at the foundational level of the email ecosystem where multilingualism is concerned – and urgent work is required across the email vendor environment to improve support for online linguistic diversity.
What did we find?
Firstly, for those using the Greek script domains, problems begun at inception. Notably, within Gmail, there is a transliteration error between Unicode and Punycode when domain names are communicated via DNS. An example of this is the handling of the Greek sigma, ‘ς ‘ (in the form used at the end of words), which when converted back from Punycode, is rendered as the regular sigma, ‘σ’.
As we continued to scrutinise the email process in more detail, it became clear that some internationalised email addresses were corrupted either on display, send or reply. Errors were encountered across all major commercial email providers. For example, Outlook was only able to successfully send three test messages, and one with partial success.
The study’s findings in relation to Greek script email addresses suggest that there may be incomplete or insufficient implementation of IDNA2008. IDNA2008 pertains to the application of certain Unicode characters to be used in domain names with the intention that internet users can use links in their native language. The IDNA2008 standard is essential for a proper handling of Greek script email addresses, and an inadequate implementation of the protocol is likely to be responsible for the particular errors encountered with Greek script.
What now?
Our findings are a reminder of what failures to support the use of multilingual domains really means for internet users worldwide. More than 4 billion people used email in 2020 - yet for those whose native script is not Latin, there is simply no effective way to participate in email using their own language.
Multilingualism is a core pillar to EURid and the .eu infrastructure. Our insights have already been shared with ICANN’s Universal Acceptance Steering Group, and we look forward to conversing between stakeholders to find an effective solution to work toward achieving multilingualism within global email systems.