Members of the KIN Group partnered with other PhDs from different MAI (Māori and Indigenous PhD groups at different tertiary institutes) sites from as far away as Otautahi to do something towards helping revitalise our reo rangatira. We entered a team of 8 in the kōreromaori.com Hōpu kōrero Māori competition and it actually was a lot of fun. It was fantastic that many educational organisations and whānau groups got behind the competition and some even utilised the fundraising opportunity.
The purpose of koreromaori.com is to teach computers and other technological machines the sounds (phonemes) of the Māori language with as many different voices as possible. The process required reading a variety of pre-determined te reo Māori language sentences and earning points based on an accumulation of the minutes and hours of sentences read. Points were allocated and judged according to their accuracy of pronunciation and the actual words in the sentence. From experience my experience and feedback from our team, accuracy when reading was a challenge. The successful team capturing the largest amount of accurate sentences and accruing to the highest amount of time reading, would win a cash prize. The competition lasted for 10 days and ended today. Whilst winning wasn’t our objective, we were pleased to make a koha toward the kaupapa.
I found the innovative nature of the competition really inspiring, knowing how much people could get paid to do this as actual work. I could imagine that when Apple were gathering a variety of voices and pronunciations to get Siri up and running, they wouldn’t have had to call out to volunteers from the community to read sentences. After looking on the web for information about Apple’s processes for capturing a new language, I learned they netted different people with various dialects and accents reading sentences, paragraphs, and word lists much in the same way we were doing in the competition. It wasn’t clear whether those people were paid for their contributions however I think it is unlikely that they were not paid.
After the first phase of capturing a new language, Apple releases it as a feature of iOS and macOS dictation. It is then usable for voice to text recognition, and people can record themselves with the microphone on the iPhone and it comes out as typed text to send as a message. As such Apple is able to gain anonymous speech samples gathered from a wide demographic of people. Apple then pays people to transcribe the voice samples, and uses them as a verified combination of voice and text data to be used on the language platform.
Learning about the difficulties of trying to get any language platform together it’s with gratitude that our small contribution has made a difference to teach computers te reo rangatira. A big mihi to the sponsors and drivers of this kaupapa, who have tried such novel methods to revitalise our te reo Māori. Congratulations to the winners of the competition and to everyone that was able to lay down even one sentence towards this cause.
Toku reo toku ohooho, toku reo toku mapihi mauria
My language is my awakening, my language is the window to my soul