ELDP Project Highlight: Matukar Panau
Today on the ELAR blog, we are featuring ELDP grantee Danielle Barth’s project ‘Matukar Panau‘. Matukar Panau is a highly endangered Oceanic language spoken near Madang, Papua New Guinea (PNG). Although most children are no longer learning Matukar Panau, current speakers (approx. 300) form a vibrant community of multilingual speakers in dense social networks. As an Oceanic language on the PNG coast, Matukar Panau has many interesting Papuan features. Danielle’s ELAR collection ‘Matukar Panau’ can be accessed here.
Impact on community/speakers highlight
A highlight for me this work period was to interview people about what they felt to be historical events, including World War II and interactions with James Yali, an important figure in PNG history. Community members understand that they have unique perspectives to offer on world history and want their stories to be heard and be part of the archive. We had many discussions of what an archive is. I was called to interview Komik, the (by far) eldest Matukar community member.Komik Tabag is an older deaf woman who was a young girl in world war 2, so she must be in her late 80’s or early 90’s. Her perspective on the war was different from what I’ve heard from another community member who lived through the war. In 2011, Peter Ratan Barui recounted his life story, which included a mention of the war. He has since passed away.
I recorded a story from Komik in May 2019. Komik Tabag talks about her memory of the Japanese landing on Matukar beach. She describes herself at that time as before her breasts developed, so she was probably around 12-13. She says the people of the village were afraid of the Japanese, and so the village leader gave all the unmarried girls to the men of the village. It is implied through her account that this was to protect the girls from harm from the Japanese soldiers. The Japanese destroyed the village’s canoes and the villagers ran away into the jungle until the end of the war. Later Yali let them know that the war was over and held a ceremony. Peter Ratan Barui (apx. 1938-2014) in 2011 explained his age in terms of world war 2, that he is old because he was born before the Japanese came. He explained that he saw the Japanese, he got papaya for them, he climbed coconut trees and give them coconuts to drink from and he carried cargo for them. He interacted directly with Japanese soldiers and received small payments from them for his help.I have also heard from other community members, also men, that as boys they helped both Japanese and American (this perhaps could also mean Australian) soldiers, brought them food in return for small trade. This is a different take than that of Komik, a woman for whom war meant fear of assault and an early marriage.
Advancement of current technologies has made a substantial impact on the general field of linguistics. For example in the field of phonetics, the automatic segmentation of phonemic segments and alignment with orthographic transcription has allowed researchers to extract massively more information than was possible by doing the same manually. This allows an increase in the scope of studies conducted. However forced alignment, as this procedure is called, has mostly been on majority languages for which there are existing acoustic models developed from large corpora, including English (Fromont and Watson 2016, Labov et al. 2013), Spanish (Goldman and Schwab 2011), and French (Brognaux et al. 2012, Milne 2014). While there has been some application of forced alignment to underdocumented and minority languages (e.g., DiCanio et al. 2012, Coto-Solano and Nicholas 2018), it is not common. Often this is because a lot of data is required before reliable acoustic models can be created and that is lacking for many under-documented languages. However, more recently it has been shown that success in forced alignment can come even with less data through enhancement of available methods (Fromont and Watson 2016, Gonzalez et al. 2018).
The implementation of forced alignment to a Matukar Panau, utilizing the Montreal Forced Aligner (McAullife et al. 2017), was improved using a post-hoc algorithm (Gonzalez et al. 2018). The force-aligned data created a large sample of vowels (over 60,000) to conduct the first acoustic analyses of this language, considering the social conditioning of Matukar Panau vowels and, in particular, the impact of clan as a sociolinguistic variable (Barth et al. in press). Matukar has four clans in two pairs of aligned clans. Our analysis of the cardinal vowels showed that the pairs of clans differed from each other in vowel production. Further, within one pair, measurements were quite similar which reflects a relatively harmonious relationship between the two clans. Within the other pair, and an offshoot group that is nominally still part of one of those clans, there is much more variable production. We think that this may reflect the discord within the clan and may be caused by distancing behavior that is manifesting itself in the linguistic production. Until now, clan has not been a significant part of understanding the linguistic production of syntactic structures, the lexicon or the morphology of Matukar Panau. Therefore we are excited to see clan affecting the phonetics which will lead us to learn more about clans and how social networks affect phonetic production.
As a new scientific highlight, I have a currently on-going collaboration with Australian tech company Appen (appen.com). They are working with my data to create essentially a spell-checker that will run on my ELAN data to standardise spellings and identify spelling errors and word break errors. This is especially helpful as I have multiple people from the Matukar community transcribing data, as since the project has been going on since 2010, I myself have made different decisions regarding word breaks and spelling in the last decade. This should make the data much more harmonious and better for corpus work and produce nicer community materials. Appen is doing this work as ‘giving back’ to the academic community, and it is a major advantage to have advanced technological resources put towards editing Matukar Panau data.