The English language abilities of international students keep coming up as an issue. Despite reassurances from the regulator and universities, there is little publicly-available evidence to support a conclusion that the language issues are too infrequent or not serious enough to be a major concern.

Although we should always be careful with statistics on self-assessed abilities, they can provide a rough guide. If someone says that they do not speak English very well chances are that they will struggle in an English-speaking university.

One source of data on self-assessed English is the 2016 Census. The Census does not directly identify people on student visas, but I have used a number of filters – enrolled full-time in university, non-citizen, and arrived in Australia 2014-2016 – to identify likely international students. I also removed people who are monolingual English speakers. It’s likely that the data still includes some visitors to Australia and domestic students, such as those on humanitarian visas, but it should be dominated by international students.

As the chart below shows, the vast majority of likely international students say that they speak English well or very well. However, 9 per cent say that they do not speak English well.

Looking at the data by the most common languages spoken at home indicates that majorities of people who speak Nepali, Punjabi or Hindi at home say they speak English very well, and hardly any rate their proficiency as ‘not well’.  Speakers of Chinese languages are much less likely to say that they speak English very well, and much more likely to say that they do not speak English well.

As we might expect, students from former British colonies are more likely to have had the long-term immersion in English needed for full proficiency. Within the group who speaks a Chinese language at home, those born in Hong Kong, Malaysia or Singapore are much more likely than those born in mainland China to say they speak English very well. Sixteen per cent of students born in mainland China say that they do not speak English well.

The ‘not well’ group are a concern, but their overall numbers are at the lower end of what we might expect given the negative publicity around this issue. But the category of ‘well’ is likely to cover a range of abilities.

Using a more nuanced set of categories, a University of Melbourne international student survey found that a quarter of them described their English fluency as ‘moderate’ and 4 per cent as low or very low. The U of M has a lot of Chinese and few Southern Asian international students, so it is drawing on markets with less English ability. But it is also an elite institution that can be selective about which students it takes, so the moderate or low share at other universities could easily be higher.

Is moderate English really enough for successful academic study? Intuitively I am sceptical. We at least need a better evidence base.

Universities should report more extensively on the international students they admit. They submit some English language test information for immigration purposes, but it is not part of the separate higher education enrolment data collection.

With English on admission data linked to enrolment data, we could find out much more about the relationship between language ability and subsequent performance such as subjects passed or failed and course completion. A national data set would let us identify, and possibly investigate or explore, university-level results that differ substantially from the national norm.

Without more evidence and better regulation the English-language proficiency of international students is an issue that will not go away.