ONZE - Origins of New Zealand English Project

[ONZEMiner software has been renamed Labb-CAT Software]

The Origins of New Zealand English (ONZE) Project endeavours to chart the origins, features and changes of New Zealand English (NZE), and to apply the findings to theories of language and language change.

With a strong emphasis on acoustic analysis, sociolinguistic variation and speech perception, the project continues to make relevant and important contributions to the linguistic community, as well as documenting an important aspect of New Zealand’s society and identity.

The aim of ONZE is to not only document features, patterns and changes in NZE, but to use this information to make wider theoretical statements about language in general. We aim to combine the documentation of social patterns in NZE with refined acoustic analysis, and we consider the results in light of theoretical models to better understand how both the production and perception of language actually works. We are committed to scientifically robust studies, using the latest in language related technology and statistical software.

The main reason for the original genesis of the ONZE project was the discovery by Elizabeth Gordon of a remarkable set of recordings  - the Mobile Unit Recordings. These have allowed ONZE researchers to document the emergence of NZE amongst its very first speakers. The fact that the very first stages of the dialect are captured on tape make NZE a valuable testbed for theories of new-dialect formation, and a large number of papers have addressed this issue.

The ONZE corpora comprises of three collections:

The Mobile Unit (MU) archive is an historical archive collected by members of the NZ National Broadcasting Service between 1946-8. The intention was to broadcast pioneer reminiscences and stories from parts of New Zealand outside the main city centres.

The Mobile Unit was a large van which carried two heavy disc recorders. These were operated in the van by a producer and a technical assistant but the microphones were connected to long cables so that the interviews could be carried out in people’s homes, farms, schools and in local town halls. The recordings were captured on 16-inch discs made of an acetate coating on an aluminium base, with most of the discs recording 10-11 minutes per side.

About 300 MU speakers were recorded singly and in groups.The earliest recordings are of spokers born between 1851-1910. In general the quality of the MU interviews is good although in some there are extraneous noises such as ticking clocks, meowing cats or rattling tea cups. In some cases, people were interviewed alone, but in other cases group discussions were recorded, some with as many as eight speakers, making it difficult for researchers to identify all the participants. Some speakers spoke for extended periods of time and others only briefly. The interviews vary in their degree of formality, with some speakers reading from notes, others telling popular stories and still others engaged in casual conversations. Some of those interviewed sound quite nervous at first but then relax. Some seem completely oblivious of the fact they are being recorded while others do not relax at all.

This collection of recordings is particularly important because it contains speakers born around the time of the earliest organised European settlement into New Zealand. The earliest born speakers all sound quite varied, reflecting a range of dialects that form the input into New Zealand. However quite rapidly, a relatively homogenous dialect emerges.

An analysis of the phonetics of the MU speakers forms the basis of the book:

New Zealand English: Its Origins and Evolution

(Some of this prose is taken from: Gordon, Maclagan and Hay (forthcoming) The ONZE Corpus. In Beal, J.C.; Corrigan, K.P.; Moisl, H. (eds) (forthcoming) Models and Methods in the Handling of Unconventional Digital Corpora. Volume 2: Diachronic Corpora. Palgrave.)

There are approximately 140 speakers in the Intermediate Archive born between 1890 and 1930. They come from four different sources and many come from geographical areas not covered by the Mobile Unit.

  • Two researchers, Nicola Woods and Sandra Quick, recorded some of the descendants of the original speakers in the Mobile Unit archive. These recordings provide continuity with the MU archive and were collected in the 1990s. The interviewees were often able to add additional information about their ancestors recorded by the MU. These interviews were designed to parallel the interview style of the original MU recordings. Unlike some of those interviewed in the MU archive, none of these speakers reads from prepared notes. 

  • Rosemary Goodyear, a history student at the University of Otago, collected oral history accounts of early memories of old New Zealanders in Dunedin and Christchurch for her M.A., and later for her PhD thesis. She was especially interested in accounts of childhood, both schooling and childhood games.

  • Lesley Evans, a volunteer radio programme maker, collected interviews of ‘interesting people’ aged 60-100 living in Christchurch. Over 100 interviews were recorded in the speakers’ own homes, and lasted from half and hour to over two hours. These were later edited down to 20 minute broadcasts called ‘Life Stories’ for Plains FM, a community radio station. These interviews were recorded between 1990 - 1993 and acquired by the University of Canterbury in 2000.

  • The NZ Broadcasting Service collected recordings of 13 speakers from the West Coast of the South Island (born 1880-1910) between 1960 and 1985. These recordings were broadcast on National Radio programmes and were obtained from the Radio New Zealand Sound Archives in 2004. They vary in length from 10 minutes to an hour with the speakers telling stories of the small towns where they lived and interesting incidents in their lives.

This archive has been collected since 1994 by students in the Ling 303 New Zealand English course at the University of Canterbury. The archive is structured according to a speaker quota and constitutes a judgement sample. Approximately equal numbers of men and women are included, equal numbers of younger (age 20–30) and older (age 45–60) speakers and equal numbers of speakers from higher social classes (categorised as professional) and from lower social classes (categorised as non-professional).

This structure sets up eight cells in the database. In order to create an archive of New Zealand English (NZE) speech, all the speakers were born in New Zealand, and none spent significant periods of time outside the country. No control was kept over the birthplace of the speakers. Most of them come from the Canterbury region, but some come from other parts of the country. The speakers were born bewteen 1930-1984. Because these recordings were collected by students, the interviewing skills and quality of the recordings vary.

The Canterbury Corpus recordings were deliberately collected to be part of an archive of NZE. Each speaker reads the NZE word list designed to give examples of all the phonemes of New Zealand English and to provide information about sound changes in progress. Student interviewers also engage their subjects in conversation for thirty minutes. The aim was to obtain material that is as close to casual speech as possible.

The primary research focus at ONZE is variation and change in New Zealand English, this topic has numerous projects.

The availability of ONZE corpora make it possible to study the entire history of individual variables in NZE. A large number of sociolinguistics variables have been studied by researchers in the ONZE lab.

  • The short front vowel shift, including FLEECE
  • Vowel mergers
  • Linking and intrusive /r/
  • Phrase-final /t/
  • Morphosyntactic variation
  • Regional variation - Darfield
  • Maori English

As part of the work on variation and change in NZE we are interested in how the speech perception mechanism functions in the face of such variation, This has led to a series of perception experiments.

  • NEAR/SQUARE merger
  • KIT

The short front vowel shift, including FLEECE

New Zealand English is well-known amongst sociolinguists for its front short vowel shift – resulting in high realizations of the TRAP and DRESS vowel, and a central realization of KIT. The high TRAP and DRESS often lead to miscommunication with speakers of Northern Hemisphere varieties of English, and the central KIT vowel is one of the main features that sets NZE aside from Australian English.

Work from the ONZE project shows that the initial stages of this push-shift began in very early NZE, with speakers born in the 19th C, already raising their TRAP and DRESS vowels. The main period for KIT centralisation was the early 20th Century. DRESS continues to rise today, and now overlaps in F1/F2 space with FLEECE. FLEECE has begun to diphthongize in reaction.

Vowel Mergers

We are very interested in a number of vowel mergers occuring in NZE, and these have been studied in the ONZE corpora, in real-time production studies, and using perception experiments

Linking and intrusive /r/

Linking /r/ is an /r/ which is present in spelling, and which can be pronounced across word or affix boundaries when the following sound is a vowel. Examples of linking /r/ include “more apples”, “car engine”, “fearing” and “honourable”. Intrusive /r/ occurs in similar positions, but is not present in spelling. Examples include “law/r/and order”, “ma/r/and pa” and “draw/r/ing”.

Intrusive /r/ evolved later than linking /r/, but it is currently believed that they are now the same thing. In New Zealand English, a new vowel is also now attracting intrusive /r/, as in plough/r/ing, and now/r/and then.

Phrase-final /t/

The visit of Gerry Docherty to ONZE in 2006 sparked an interest in the behaviour of /t/ in NZE, and we began a collaborative project which, in the first instance, looks at variation in phrase final /t/.

Morphosyntactic variation

While the biggest focus of ONZE is on phonetic variation, we have also looked at some morphosyntactic variation, and we expect this focus to increase in the next few years, with a new collaboration between Joan Bresnan and Jen Hay. Morphosyntactic variables which have been studied include agreement in existentials, the prefix ‘un’, the dative alternation, preterites and possesive ‘have (got)’.

Regional variation - Darfield

Despite the fact that ONZE holds over 1000 hours of spoken NZE, this is not sufficiently varied to study the question of whether there are emerging regional variants in NZE. A project was conducted which aims to investigate the development of regional variation in NZE.

A large-scale, in-depth survey of the Canterbury region was surveyed by collecting and documenting rural and urban vernacular data firstly in the Darfield area. Data collection has taken place where a local farmer used her personal network to fill a sample of 20 locally born and raised speakers. This sample will be equally stratified by gender and age, straddling two age groups: 19-30 and 45-60. To give a better view of the current linguistic situation, the sample is to be extended by a further 10 speakers (5 men, 5 women) in the 70-85 age bracket.

Hour-long interviews are conducted, in which participants are encouraged to tell stories about themselves and their experiences in the area. This is a well-established method in sociolinguistics, allowing the researcher to tap into more vernacular styles of speech . At the end of the interview the participants read the standard word-list from the Origins of New Zealand English project.

Maori English

Jeanette King and Margaret Maclagan are leading this project which also involves the related MAONZE project.


Despite the fact that young NZers claim they can’t hear the difference between the vowels in NEAR and SQUARE, our experiments show that they are actually quite good at identifying which variant is being produced. By associating NEAR/SQUARE tokens with photos of different speakers, we have shown that listeners are much more accurate at the task when they think they are listening to someone who is expected to make the difference (i.e. an older speaker, or an upper-class speaker). Further experiments have revealed that listeners’ ability to distinguish between these vowels is affected by their recent exposure, either to speakers of different dialects, or to lexical items associated with particular dialects.


Kate Drager's MA thesis synthesized a continuum from TRAP to DRESS, and showed that where a listener places a boundary along this continuum depends, among other things, on the perceived social characteristics of the speaker.


Following work by Nancy Niedzielski, we have been asking participants to listen to natural speech, and then match particular vowel tokens to points on a synthesized continuum. Our results for KIT show that participants hear a more “Australian” KIT vowel if “Australian” is written at the top of the answer-sheet, and a more “NZ” KIT vowel, if “New Zealander” is written at the top of the answer-sheet. Surprisingly, this result was robust despite the fact that our participants reported they all knew they were listening to a New Zealander.

To test the idea that mere exposure to the concept of Australianism could affect speech perception, we re-ran the experiment with a stuffed kangaroo in the room in one condition, and a stuffed kiwi in the other. This produced similar results. We have also tried exposing participants to ‘good ‘ and ‘bad’ facts about Australia, and find that these conditions shift perception to different degrees.

The ONZE Project is based in the Department of Linguistics at  the University of Canterbury. 

The Principal Investigator of the ONZE Project is Professor Jen Hay.


ONZE Project
Linguistics Programme
University of Canterbury
Private Bag 4800
Christchurch 8140
New Zealand

You can also email us.