Just before Christmas, The Australian published an article about indigenous language use in Australia that drew on research conducted by the Centre for Aboriginal Economic Policy Research (CAEPR) at the ANU. The article began with this:

Aboriginal language ‘crisis’ exaggerated, says census

New census analysis reveals that the “crisis” in Aboriginal languages is overstated, debunking the notion that mother tongues are dying out.

In the Northern Territory the proportion of Aborigines speaking an indigenous language at home rose from 59.1 per cent in 2006 to 64.7 per cent in 2011.

For years now we’ve been hearing statistics like ‘a language dies every two weeks’ and ‘by the end of the century, half of the world’s 7000 languages will no longer be spoken’, and the top-end of Australia has been indicated as a hot-spot for language endangerment. Also, having worked on endangered languages over the last 7 years, I can say from my own experiences, and those of my colleagues, that the situation is critical and is not being overstated.

So how could the census possibly reveal that there isn’t a problem?

To attempt to answer this, I downloaded the original report and got reading. I quickly found that its findings don’t bear much resemblance to those of The Australian:

The main finding was that there has been a steady decline in the percentage of indigenous Australians who speak an Indigenous language between 2001 and 2011. Given the potential individual and community benefit of Indigenous language retention, this is clearly a negative finding.

How then, did The Australian come up with the overall impression that ‘mother tongues’ are no longer threatened?

The numbers that The Australian cites are correct, inasmuch as that is what the census data says. The number of Aboriginal people in the Northern Territory that self-reported as speaking an Aboriginal language rose from 28 974 in 2006 to 34 086 in 2011, which equates to 59.1% and 64.7% respectively (as reported above) of the overall indigenous population that reported speaking an Aboriginal language.

But if we look closer at the numbers and the breakdown of individual languages, it quickly becomes clear where the misunderstanding lies.

The census is simply not the best way to assess the health of Aboriginal and Torres Strait Islander languages

Alongside the report, CAEPR has published a spreadsheet of data (available here) showing indigenous languages and their speaker populations in both 2006 and 2011. One thing that’s strikingly obvious is the extraordinary increase in the number of speakers for some of the languages on the list. Some languages that had very few or no speakers in 2006 (reportedly) apparently experienced miraculous rejuvenation over the subsequent five years. Dalabon for example, exhibited a phenomenal rate of uptake, with only 7 people reporting as speaking the language in 2006 shooting up to a whopping 73 in 2011, a ten-fold increase. This, to be brutally honest, is impossible, and many of the languages in the list exhibit a similarly inexplicable increase in the number of speakers.

Respondents are clearly answering the language question based on their linguistic affiliation – the language they claim in virtue of being born into a particular linguistic group – rather than their actual linguistic ability. There’s nothing wrong with this, especially given that the census doesn’t otherwise ask for anything such as one’s ancestral language. Nor is this specific to Aboriginal people. This is something that many migrant communities and families have to grapple with when filling out the census. A 3rd generation Italian family might only speak Italian sparingly, with their grandparents for instance, but they’ll still put down ‘Italian’ as the language they speak at home, just as my father – a staunch atheist – always answered ‘Methodist’ under religion. It simply highlights the limitations of a simplistic, self-reporting survey such as the census for enumerating more complex questions such as language (and religion, for that matter).

Another methodological problem for the purposes described here, is the inclusion of non-traditional Aboriginal languages, such as Kriol (the most populous language in the list) and Yumplatok (or Torres Strait Islander creole; incidentally the second-most populous language on the list). Since the census is being interpreted to assess the linguistic health of Aboriginal languages overall, including these languages only serves to cloud the data. For instance, if a community switches from their traditional language to Kriol (as has happened many times in Australian history), but the size of that community grows over time, then the overall numbers would show an increase in the number of people speaking ‘an Aboriginal language’ rather than, more accurately, that the traditional language is no longer being spoken.

The census is simply not the best way to assess the health of Aboriginal and Torres Strait Islander languages for the issues mentioned above. Fortunately then, there is the National Indigenous Languages Survey. The first such survey, completed in 2005, was a huge national undertaking that attempted to get accurate figures for the state of indigenous languages. It found (emphasis mine):

The situation of Australia’s languages is very grave and requires urgent action. Of an original number of over 250 known Australian Indigenous languages, only about 145 Indigenous languages are still spoken and the vast majority of these, about 110, are in the severely and critically endangered categories.

The second National Indigenous Languages Survey is currently being conducted and when finished, will probably show that the situation hasn’t improved in the last 7 years, contrary to the claims of The Australian.

The census is an extremely useful enumeration of the population for a whole range of different reasons, but gathering accurate and useful information with respect to the health of Aboriginal languages is evidently not one of them.