The Linguist

26 The Linguist Vol/63 No/1 thelinguist.uberflip.com FEATURES Jill Wigglesworth, Jane Simpson and Nick Evans report on the findings of a fascinating project studying South Pacific languages, and how they ensured the research was ethical and useful for all Australia and its surrounding regions are rich in languages. Many are highly endangered, few have been documented, even fewer have rich documentation. In 2012, researchers from several disciplines across Australia formed a group to carry forward research in the language sciences, and particularly on the languages and their speakers in our region. We established an eight-year research programme (2014-2022), the Australian Research Council Centre of Excellence for the Dynamics of Language (CoEDL), designed around four major themes: Shape, Processing, Learning and Evolution. Central to our research ethos was the assumption that for every newly discovered linguistic phenomenon (the job of the Shape programme), we need to understand how speakers use it in real time (Processing), how children can learn it as a first language or in a multilingual environment (Learning), and how it arose (Evolution). Each theme required meticulous and wide- ranging collection of language data. Much fieldwork was undertaken, including the Canberra Longitudinal Child Language Project, which provided rich material for understanding which childhood factors affect later language development, 1 and 'Sydney Speaks', which offered longitudinal data on Australian English in Sydney. 2 Such was the extent of CoEDL data collection that the number of languages in the Pacific and Regional Archive for Digital Sources in Endangered Cultures (PARADISEC) rose from 836 in 2014 to 1,342 in 2022. Work on the 'transcription bottleneck' was just as important, as we improved the technology for transcribing and annotating field recordings. Shape The emphasis on linguistic diversity meant that researchers were on the lookout for previously undiscovered ways that languages can be built. 3 A key paper showed, on the basis of a vast new database (Grambank), that it is the most endangered languages which have the most unexpected linguistic structures. 4 The grammars and vocabularies of these languages need to be set against what we know about other languages in order to determine their significance. This is where new methods of analysis and comparison come in to ensure that claims of uniqueness are well- founded. The rarer the phenomenon, the greater the need to make the data available for scrutiny through trackable data citation. One intriguing aspect of Yelmek, a Papuan language, was uncovered by Tina Gregor. 'Suppletion' refers to the phenomenon by which words are not formed in a predictable way (e.g. adding '-ed' for the past tense) but do something totally off the wall (using 'went' instead of 'goed'). Since our ability to form an infinite number of new sentences depends on using regular rules, languages can only get away with a few such 'suppletive' sets. Languages have long been known to supplete for number (is/are) and person (am/is), but Yelmek does something no other language is known to do: it suppletes for the gender of objects. 'He held her' is poyopoa but 'he held him' is peŋepea – the stems -oyopo and -eŋepea are different according to whether 'he' is holding a man or a woman. Yelmek has dozens of pairs like this. A further finding was the phenomenon of 'engagement': the use of inflections on verbs to indicate whether the speaker thinks their conversation partner is paying attention to an action they are talking about. This was first reported in a Colombian language, Andoke, by Jon Landaburu. A major survey by Nick Evans, Henrik Bergqvist and Lila San Roque brought together evidence from other South American languages (e.g. Kogi) and Papuan languages (e.g. Marind) to set up a logical framework for exploring this phenomenon. 5 Since then, scientific interest in engagement has caught on, as witnessed by a special issue of Open Linguistics, including an important article by CoEDL researcher Bruno Olsson on techniques for understanding engagement based on video-recorded interaction. 6 Processing Australian languages – famous for their free word order – offer a special challenge in the realm of processing. How do speakers plan out their sentences as compared to speakers of languages that use different word orders? Early CoEDL discussions led to work using mobile eye-trackers to discover how speakers of Murrinhpatha, which has free word order, plan their sentences. The findings indicated that speakers begin planning their sentences much earlier than English speakers, probably because they need a holistic overview of the situation before uttering the first word. 7 Learning In the realm of learning, unfamiliar structures again raise questions. How do children learn them? Do complexities not found in English and other well-studied languages pose special problems? How do children learn linguistic Research revelations

• Cover