Māori Pronunciation Aid Tool
The Curious Minds development team:
Prof Catherine Watson, Isabella Shields, Department of Electrical, Computer and Software Engineering
Assoc Prof Peter Keegan, Te Puna Wananga
Dr Justin Hui, Mechanical and Mechatronics Engineering
Brooke Ross, University of Auckland
Background
The contact between Māori and English over the last 180 years has affected the Māori language. The MAONZE project (https://maonze.blogs.auckland.ac.nz) has been investigating sound change in Māori from a corpus of 60 bilingual speakers (men and women), whose birthdates span over 100 years. The corpus comprises three groups. The first group of speakers, historical elders, were born in the late nineteenth century and recorded mostly in 1946-48. The second group are present day elders born between 1920-40, and the third group are two sets of young speakers (L1 and L2) born between 1970-90.
The MAONZE study focused mainly on the vowels, investigating sound change in the long and short monophthongs and the five most common diphthongs /ai, ae, au, ou, ao/. The acoustic analysis was mainly based on the first and second formants, and also on vowel duration.
The results showed that there were changes in the vowel quality and duration over time. We have not only disseminated the findings to research audiences, but also to the larger Māori speaking community including language teachers. In general the academic audiences were predominantly interested in the mechanisms and motivations of sound change. However the Māori community were excited about the systematic presentations of the vowel spaces of the elders compared to the young speakers, and the fact there were measurable differences. This coincided with what the community was hearing, but had been unable to quantify. The MAONZE project received requests to create a platform to enable Māori speakers to compare their speech with the current day elders in the MAONZE corpus. This led to the development of MPAi.
What initial development work was done for the MPAi tool? Was there a co-design process involving whānau?
The MPAi project has been under development for 17 years. It has been led by Catherine Watson and Peter Keegan (Waikato-maniapoto,Ngāti Porou) (https://newsroom.co.nz/2017/09/14/academics-and-elders-merge-skills-for-unique-mori-language-tool/ ) We have developed various versions of the platform. The platform has been developed on a variety of languages including TCL/TK and Python. We have also experimented with version that sat on computers, and others that were web-based. In a major study published in 2017 we received feedback from 22 participants, mainly Māori medium teacher trainees. Whilst we got a lot of very positive feedback, it was clear that people struggled to interpret the technical displays. Further studies in 2019 and 2020 confirmed this.
In 2023 we were awarded a Curious Mind grant developing a platform to learning the science of speech through the Māori Pronunciation Aid (MPAi). We decided that if people understood some of the background around speech production and speech acoustics, they would find the MPAi displays easier to interpret. We are currently entering into a testing phase of the Book/MPAi.
What are the main objectives and research questions your team aims to address?
Our research aim for this project is to see whether there are changes in people production of the vowels in te reo Māori, when people use MPAi. We need to stress here – we aren’t showing whether people are speaking “right” or “wrong” but rather we want to show them what they can adjust to change the production of vowels.
Another aim of the Curious Minds project is to make the link between the science of speech production and speech acoustics, and learning a language.
What challenges were encountered during development and investigation? How were they overcome?
It is tempting, in a pronunciation aid, to include lots of speech processing techniques. However, these highly technical displays are of little use to the average language learner. They require too much specialist knowledge to interpret. A participant in the 2017 noted trial “A lot of trust will be given to the application, so it needs to work well before [it’s] released.” Whilst this statement is true for any system that provides automated feedback, it is particularly crucial for a language which is seriously under resourced and is undergoing revitalization. Therefore, we are very careful about what we release, and what we say MPAi will do.
What key public engagement activities (e.g., events, workshops, media) has the team conducted to raise awareness and revitalize the indigenous language?
We have organised a number of Māori speech technology huis bringing together academics and industry leaders in 2021,2023 (https://speechresearch.auckland.ac.nz/2021/04/30/maori-technology-workshop/, https://speechresearch.auckland.ac.nz/2023/06/28/maori-speech-hui/), and 2024.
In addition we have been presenting at conferences There is list of our publications at https://maonze.blogs.auckland.ac.nz/mpai-the-maori-pronunciation-tool/ – but we haven’t updated it since 2019.
What is the CeR’s role in delivering the MPAi tool to the public?
This booklet contains a number of activites, and includes various version of MPAi, to help people with their pronunciation. Working with the eResearch Centre has been very useful for us, because the developers at the Centre were able to take what we have developed and make it robust. Whilst many in the Curious Minds team have a technical background, there is a huge difference between having a system working on a specific machine, with all its idiosyncrasies, and have a solid app in which the user experience is consistent across many different platforms, and is also secure. We have also appreciated the HCI expertise that the Centres developers have. The final outcome for the Curious Minds project can be found https://www.canva.com/design/DAGGkZnYEBo/4lT1haX4lenGwclLl2VdRA/view
