Is Te Reo Māori Growing or Shrinking?
Measuring the Health of a Language
We are apparently living through a Māori language renaissance. Lorde has released an album in Te Reo. The weatherman peppers his broadcasts with Māori placenames. Enlightened bureaucrats now sign off their e-mails “ngā mihi”. The wānangas are booked out for years in advance. More than ever, people want to kōrero.
Yet for all this praise and celebration, there’s little discussion about whether the Māori language is actually growing; ebullience won’t save a language, but more speakers might. Are those people taking introductory Te Reo classes getting to the point where they can have conversations in the language? And are they passing this knowledge down to their children?
After a first course you might be able to introduce yourself and say how old you are. But you are far from being a fluent speaker. It takes a long time and a lot of effort to get to that point; it’s not always obvious when you have reached it.
So how do you determine whether someone is a speaker of a language or not? There are two ways. The first is to assess them: make them sit a test or have a teacher evaluate their performance. The second is to have them assess themself.
Some languages have well-established assessments administered by credible organisations based on standards like the Common European Framework of Languages (CEFL). This gives you a fairly objective way to determine your language ability. However, not everyone takes or even knows about these tests. In the case of Māori, standards and testing are not as developed and universal.
Our primary way of determining how many people speak Māori is therefore through surveys. But a survey response is subjective in a way that a formal test—whatever it measures—is not.
For example, in the census you are asked, “In which languages can you have a conversation about a lot of everyday things?” How much do you have to be able to talk about before you can say you speak a language? Different people will interpret this differently. Is it enough to be able to order a meal and navigate to your hostel? To be able to speak the language whenever it comes up in your life (which may not be very often)? To be good at using the language relative to your peers? To have passed several years of classes about the language in school or university?
The linguistics literature shows that there is a correlation between self-assessed language ability and externally-assessed language ability. In other words, people who rate themselves as being good at speaking a language tend to perform better on tests. But the correlation is weak and self-assessments are difficult to interpret.
This mismatch is partly due to a broader cognitive bias known as the Dunning-Kruger effect: when someone is bad at something, they tend to overrate their ability; when someone is good at something, they tend to underrate their ability. The effect, in short, is that while you may know what you know, you don’t know what you don’t know.
We’ve also seen evidence of the Dunning-Kruger effect in second-language self-assessment. A summary of the literature in 1990 by Patrick Blanche and Barbara J. Merino found that proficient users of a language tended to underestimate their ability, while overestimates typically involved less proficient users.Later statistical meta-analyses, one by Steven Ross in 2008 and a more recent one in 2021 by Minzi Li and Xian Zhang, agreed with that conclusion.
Self-assessment is less reliable when the students are less proficient in the language being assessed and when they have less experience in formal language learning and self-assessment. In fact, one 1989 study of inexperienced adult learners of Dutch found no correlation between self-assessment and external assessment.But in general, the meta-analyses confirm a correlation, albeit a weak one, i.e. between 0.25 and 0.5.
People are also better at accurately self-assessing the receptive skills of reading and listening than the productive skills of speaking and especially writing.They also give better answers when asked more specific questions—Can you deliver a pepeha? Can you describe the members of your family? Can you watch a movie without subtitles?—rather than broad questions—How well can you kōrero?
These studies tell us useful things about designing self-assessments in a student-teacher enviornment. But their lessons are not really applicable to a general survey of the population. Accordingly, what information we have about Te Reo is frustratingly vague.
Since the 1990s we’ve had three surveys for assessing the number of Māori speakers:
The census, run by Statistics New Zealand, typically every 5 years. Participants are asked to list languages in which they could “have a conversation about a lot of everyday things.”
Te Kupenga surveys of Māori health, also run by Statistics New Zealand in the same year as the census. Among other questions which are not related to language, participants are asked to evaluate how well they can read, write, listen, and speak Te Reo by placing themself into one of five categories for each skill: very well, well, fairly well, not very well, and no more than a few words or phrases.
The Health of the Māori Language surveys run by Te Puni Kōkiri in the 2000s, in particular, the 2001 report. Participants are asked to evaluate their reading, writing, listening, and speaking just like in Te Kupenga.
Because each survey is slightly different, straightforward comparisons are difficult. The census question is open-ended, so everyone is likely to interpret it differently. The TK and TPK categories, while also open to interpretation, are more specific, and so likely to be giving us more accurate results in the sense that relatively good speakers will rate themselves relatively better.
Since TK and TPK are conducted on samples of the Māori population, we can only determine what proportion of Māori fit into the categories. The census, by contrast, is taken by (almost) everybody in the country and can also give us (more-or-less) accurate figures on the total number of Māori speakers—but at the cost of being unclear about what a “Māori speaker” actually is.
Another difference is in the age of those being surveyed. TK and TPK only survey Māori that are 15 or older, while the census surveys everyone, including babies and young children. Because their census sheets will be filled out by a parent or guardian, the adult has to make an estimation about their child’s language abilities, an estimation which may not always be correct.
All that said, let’s look at the data. In the table below you can see the number of speakers across the last three censuses. Note that I am only considering Māori who can speak Māori; there is also a (much smaller) population of non-Māori who speak Māori.
Between 2013 and 2018 we saw a big increase in the number of Māori speakers. The Māori population also grew a lot in that same period. There could be several reasons:
More Māori were born
More Māori moved back to New Zealand from overseas (especially Australia)
More people with Māori ancestry identified as Māori
While the total number of Māori speakers grew by about 38,000, the proportion of Māori who can speak Māori did not grow very much. In fact, it has declined overall since 2006.
Since we don’t know how respondents are interpreting the question, it’s not possible to say anything about the level of Māori that is being spoken by these 160,000 speakers of Māori. For example, if they can only speak rehearsed Māori on “ceremonial” occasions (pepeha, whaikōrero, etc.) should we even consider this as relevant to the health of the language?
And while the proportion of Māori speakers may have declined, this could be because a larger, more statistically significant group of people of mixed ancestry who only speak English began to identify as Māori in the last census. Those people already existed though, and whether we count them as Māori or not, they are not necessarily relevant to the growth (or shrinkage) of those who could speak Māori in 2013.
Below are the percentage of Māori who can speak or understand Māori either “well” or “very well” from the TPK and TK surveys:
The Te Puni Kōkiri National Health Survey and Te Kupenga have methodological differences, but if you squint, the percentages for TK 2013 and TPK 2001 look fairly similar. TK 2018 is the outlier; it suggests that the proportion of Māori who could speak or understand Māori either “well” or “very well” has indeed declined.
7.5% of the Māori population in 2018 is approximately 58,000 fluent speakers of Māori. This is a much lower figure than the 160,000 people who self-reported as being able to have a conversation about a lot of everyday things in the census—something which, to my mind, implies being able to speak Māori well.
Why the discrepancy? Most people who speak Māori cannot do so with fluency. And as we’ve discussed, beginners are prone to overestimating their language ability. It’s possible that beginners are buoying the census figures. Being a larger proportion of the Māori population, they are statistically more significant; if they are growing, that might obscure the fact that the subpopulation of fluent speakers is declining. But that subpopulation may also be growing—we can’t say for sure.
Since Te Kupenga also breaks down the numbers by demographics, we can see which age groups were affected the most. No age group improved their ability to speak Māori “well” or “very well” between 2013 and 2018. Māori that were age 45-54 had a notable increase of 3.8% in understanding Māori either “well” or “very well” and a 12.7% increase in understanding Māori “fairly well.” In other demographics there were sometimes very slight increases, but these fell within the standard error.
An important part of a language’s health is whether or not it is being passed on to children, particularly through their parents. TPK 2001 doesn’t measure this, but the TK surveys again suggests that the proportion of Māori who speak Māori at home has declined:
If Māori is the main language in the home of those who can speak it either “well” or “very well”, we could expect children to grow up with some kind of fluency. Unfortunately, among these demographics, the figures either have a large amount of sampling error (>30%) or have been suppressed, which means there is almost no-one in this category, or that not enough people were sampled.
Since 2018 there are also new statistics on how people learned Māori—whether from home, immersion schools, kōhanga reo (language nests), and so on. It will be important to see how this changes in the next Te Kupenga survey, particularly among youth demographics, because it will give us a sense of which efforts are having the most effect in growing the language.
So is Māori growing or shrinking?
Te Kupenga gives the most detailed insight into the state of the language, but it only tells us how many Māori speak the language as a proportion of the Māori population. On that view, the proportion of fluent Māori speakers has declined between 2013 and 2018. Fluent speakers—those who can use the language for various tasks and might one day pass it onto their children—are invaluable to the language’s long-term health. It is not good if they are declining.
But according to the census the total number of Maori speakers (as well as the proportion) has grown, though we can’t say for certain how good they are at speaking Maori, how often they are speaking Māori, or whether they are passing down their language ability to their children. It could be that more people than ever can speak Māori, but only to a basic level. It could also be that the population of fluent Māori speakers is growing (or stable), but the proportion, as measured by Te Kupenga, has become smaller, due to a larger proportion of people now identifying as Māori in the census.
We can’t definitively conclude that Māori is shrinking. But there is little evidence that it is growing. Given that Māori has been in a long-term decline, this might give one an inclination to pessimism. But on the numbers we have thus far, a valid case could be made either way. In the end, I can only throw up my hands and say: e aua hoki!
“Likewise, many investigators concluded that their more proficient subjects tended to underrate their linguistic abilities (Ferguson, 1978; Evers, 1981; Achara, 1980, 1981; Heindler, 1980; Bournemouth Eurocentre, 1982). Conversely, overestimation cases involved weak students to a greater extent than high achievers (Ferguson, 1978; Heindler, 1980; Blanche, 1985)”. From Patrick Blanche and Barbara Merino. 1989. Self-assessment of foreign language skills: implications for teachers and researchers. Language Learning 39, 313-40.
Stephen Ross. 1998. Self-assessment in second language testing: A meta-analysis and analysis of experiential factors. Language Testing, 15(1), 1-20.
Minzi Li and Xian Zhang. 2021. A meta-analysis of self-assessment and language performance in language testing and assessment. Language Testing, 38(2), 189-218.
Anne-Mieke Janssen-van Dieten. 1989. The development of a test of Dutch as a second language: the validity of self-assessment by inexperienced subjects. Language Testing, 6(1), 30-46.
"The listening condition produced the strongest correlation (.486), followed by reading (.451), and speaking (.442). Writing had the weakest correlation with writing performance (.381).” From Li and Zhang 2021.
In New Zealand you can identify as more than one ethnicity on the census.