3
RESEARCH, INNOVATION AND ENTREPRENEURSHIP AT THE UNIVERSITY OF WAIKATO
Vanuatu – so many languages to study
THE ground work has been done and
now the hard work is set to start for
Julie Barbour.
Dr Barbour, a senior lecturer in linguistics at
the University of Waikato, received a $345,000
Marsden Fast Start Award in 2011 to complete
the world’s first large-scale comparative study
of “mood systems” in Vanuatu languages.
She spent 2012 working with Vanuatu
authorities, villages and residents to organise
and authorise the research and is now
launching what could prove to be some fairly
arduous fieldwork. One masters student is
already into a three and a half month stint
on a remote island and in June Dr Barbour
will move with her family – including two
young children – to Port Vila to begin the
massive data collection process needed for
such an ambitious project.
For her PhD Dr Barbour examined
the Neverver language of Vanuatu, recording
and documenting its grammatical system. Her
new research is looking at a specific element
of that grammatical system – mood marking
– and comparing Neverver with other
Vanuatu languages.
More than 100 different languages are
spoken in Vanuatu, many of which have never
been written or described. “In English we use
tense, we describe events as happening in the
past, present or future. In these languages
they seem to describe events on the basis of
whether those events are real or unreal,” says
Dr Barbour.
“When I was studying Neverver, one of
the things I noticed was the mood system
was completely different to the system we use
in English.”
The study will be a world first inter-
island study of the functions of “grammatical
mood” in Vanuatu’s languages. Dr Barbour’s
research is a continuation of the work of her
PhD supervisor Professor Terry Crowley, who
dedicated years to recording and documenting
the many Vanuatu languages and was carrying
out more research when he died in 2005.
Dr Barbour says the project is centred on
the island of Malakula, the second largest in
Vanuatu with a population of about 28,000
and with more than 24 distinct languages.
“It is quite a small geographic space with a
large number of languages. The hypothesis is
they were originally from a single language,”
she says. “But what I find interesting is
the extent of the diversity in the way the
languages have developed. We are finding
all these differences in the organisation of
grammatical structure.”
The initial fieldwork for her project
involves recording native speakers talking
on a range of subjects, such as ‘what I did
last weekend’.
“I give them scenarios but they choose
what to talk about so they have control over
the information they provide. Once we get the
recordings, the transcription process begins.”
Using a native speaker, the recordings are
translated first into Bislama, a form of pidgin,
and then to English. “It’s sound by sound
translation,” she says. “It’s a very slow process
but every bit of data feeds into the project.”
It can take up to an hour to translate one
minute of recorded conversation, although
it does get quicker over time, she says. “Then
we go back and ask questions, such as if we
wanted to change this sentence from describing
something that happened to something that
didn’t happen, how would this change? To
understand how mood systems work we have
to look at the use of mood in context, not just
sentences but the whole context.”
The changes to questions allow researchers
to hone in on different elements of the
grammatical system, such as negation, and
only when the fieldwork has been completed
on the different languages can the comparative
part of the research take place.
Dr Barbour says the sheer scale of the
project is daunting but she’s excited by the
challenges ahead.
“The data is so rich and new to the
linguistic world, we know nothing about some
of these languages.”
Ultimately she will publish results about
how mood systems work in Vanuatu compared
to the rest of the world.
“That’s what we’ll get to but you have to
do the ground work first.”
jbarbour@waikato.ac.nz
LANGUAGES GALORE: Dr Julie Barbour is studying “mood systems” in different Vanuatu languages.
A University of Waikato engineering student spent
his summer helping to scale-up a project that turns
chicken feathers into fibre.
Geoffrey Wood picked up a Summer Research
Scholarship to work with chemical engineer Dr
JohanVerbeek to come up with novel ways to refine
the process of turning chicken feathers into a fibre.
Every year New Zealand produces about 50,000
tons of chicken feathers as a by-product of the meat
industry. The feathers are washed, cut and sorted
producing resilient feather-fibre. Potential uses could
be as a substitute for fibre-glass or ceiling panels or
packaging – anything requiring rigidity or strength.
“My role was to build a fibre classification device
to separate the different grades of fibres,” says
Geoffrey. “It’s basically a large cylindrical tube with
a fan at the bottom. The idea being that when you
put in a range of feathers the really light, fluffy ones
– the best ones for fibre – are collected at the top.
“I’ve also worked on refining the cleaning
process, looking at different ways to remove the
grease and blood from the feathers we receive.”
Dr Verbeek says the project will play a big
role in investigating whether the feather-fibre is
commercially viable.
“The ultimate goal for us is making something
like a glass fibre, but first we need to upscale the
process we’re using and make it cost-effective.”
jverbeek@waikato.ac.nz
Feathers to fibre
A software package that correctly adds macrons to
Māori texts is more accurate than a human.
The Māori Macron Restoration Service,
commonly known as the Macroniser, allows users to
simply paste or upload files in te reo Māori and have
the macrons inserted in the correct places.
It was developed at the University of Waikato
by masters student John Cocks as part of research
relating to the digitisation of the university’s Pei
Jones collection, which consists of books, manuscripts
and taonga amassed during the lifetime of the late
Dr Pei te Hurinui Jones (Ngāti Maniapoto).
He was one of Māoridom’s leading scholars, and
an advocate for the correct use of the macron to
signify a lengthened vowel.
Dr Te Taka Keegan from the Faculty of
Computing & Mathematical Sciences supervised
the research and says the macroniser is “more
accurate than most humans”, with about one error
per 300-400 words. “For most stuff people do
that’s great. People typing don’t get anywhere near
that rate.”
Dr Keegan says the macroniser looks at words in
context. “It looks at the words before and after and
decides when it should make the call. Essentially it’s
statistics and probability.”
The software is based around a large corpus of high
quality Māori language texts and it compares words
before deciding whether a macron is needed or not.
Changing the world, one macron at a time
It has proved so successful it is now being used
by the Ministry of Justice’s National Transcription
Service (NTS) for transcribing theWaitangi Tribunal’s
Te Rohe Pōtae district inquiry hearings in Te Kuiti,
which are approximately 40% te reo Māori and
60% English.
NTS Integration Manager Lile Ramsay says
the court reporters responsible for transcribing the
hearings have found the tool invaluable.
“They found they were spending an inordinate
amount of time macronising te reo, and then
we discovered the University of Waikato’s highly
effective macroniser and we have utilised this with
excellent results.
“The court reporters are able to transcribe te reo
and then another person takes responsibility for the
cutting and pasting into the macroniser, and what
would normally take a day or two for one day of the
hearing, is down to roughly 30 minutes for one day.”
Dr Keegan says the next step for the macroniser,
which is available through the University of
Waikato’s Greenstone Digital Library, could be to
make it bi-lingual, allowing text using both English
and te reo Māori to be correctly macronised.
To make it bi-lingual, developers would work
through the process of language detection and
incorporate that into the macroniser, which would
then analyse the correct language and insert macrons
as required.
tetaka@waikato.ac.nz
DR TE TAKA
KEEGAN:
“The macroniser
is more accurate
than most
humans.”