Revitalization at a distance: Engaging digital archives for language reclamation∗

Claire Bowern, Sue Hanson, Denise Smith-Ali, and George Hayden

February 10, 2021

1 Introduction

Many languages of the world are endangered (Campbell & Belew, 2018) and many communities are interested in language work with linguists. What types of projects can be done with remote collaboration? How can remote collaborations benefit both communities and linguists? and, since remote collaborations cannot ever replace in person language reclamation and revitalization, how can remote collaborations proceed without taking time away from the vital work that is done in person? In this paper, we describe collaborative work in progress on reclamation projects between Australian Aboriginal and Torres Strait Islander communities in Australia and the Chirila (digital lexical materials for Australian languages; Bowern 2016) project. These projects were created in response to a need for language work which was respon- sive to community needs in different regions of Australia, but where the work could

∗This work was funded by NSF Grant BCS-1423711 “Typology as a Window on Prehistory”. Kado Muir, Luxie Hogarth-Redmond, and Geraldine Hogarth, provided language consultations and are coauthors of grammars through the bootcamp project for and Kuwarra (respectively). Matt Tyler, Andy Zhang, Anaí Navarro, Ryan Budnick, Sasha Wilmoth, Akshay Aitha, Sarah Mihuc, Sarah Babinski, Kate Mooney, Omar Agha, Tom McCoy, Joshua Martin, and Lydia Ding were all members of the bootcamp projects over the years. Thanks to Juhyae Kim for comments on a draft of this paper and to the audience of the NASI conference in September 2020 for their questions and feedback.

1 also be done remotely: that is, aiding reclamation work without taking attention away from local collaborations within families or schools. Such work does not in any way re- place language work on the ground, and community-internal language work, but it can augment existing projects and continue connections while we cannot meet in person, as well as provide undergraduate linguistics students with in depth training in language documentation.1 The main language reclamation goal for the Chirila project has been to improve the accessibility of Aboriginal archival materials for Indigenous communities. We have proceeded in several ways. One is through informal data sharing and community-led collaborations. For example, we act as a clearinghouse for “gray literature” – language materials which may be hard for communities to access because they were only circu- lated in arenas where academic linguists have access (cf. Nordhoff & Hammarström, 2012a,b). We work in conjunction with Language Centres and groups such as Living Languages (formerly known as RNLD) to provide information stored in academic repos- itories about Australian Indigenous languages, so that members of Australia’s (Commonwealth of Australia, 1997) can recover cultural and linguistic heritage. More formally, we have conducted grammar “bootcamps” to work with linguistic fieldnotes. The aim of the bootcamps is to make archival resources for grammatical data more easily usable by Aboriginal communities in their language programs. They are nicknamed “bootcamps” because we work intensively on the materials for a month as a summer research workshop. Such fieldnote materials are often copious but are often impenetrable. Such materials may also contain extensive examples of the language but not include information about linguistic generalizations, which makes it more difficult

1Note that also this paper describes work done in the years 2014–2018, this model is likely to be especially relevant now that fieldwork travel is halted due to the Covid-19 pandemic. It ismore important than ever to protect elders and that includes academic researchers not conducting activities which may (even inadvertently) bring or reintroduce Covid to Indigenous communities.

2 for them to be used in pedagogical programs, such as school lessons. In this paper, how the bootcamps work and the types of materials that are par- ticularly suited to this style of collaboration. We describe the types of materials that are produced and how these can feed into community language programs. We describe how the bootcamps use collaboration “at a distance” to train university students in ethics and community partnerships, and comment on how the projects have worked from both the community and university perspective. We stress that this is only one possible model of remote collaboration. It was designed for the needs of the Goldfields Language Centre and the Budjar Language Centre. It also fit well with Yale’s Linguistics Department and the goals of our linguistics program. It was appropriate, given the language materials available and the expertise of the project leaders (who are the authors of this paper). In this case, learner’s guides (or sketches) and dictionary editing were the current language priorities for the Language Centres, and that was work which the remote participants (at Yale) were qualified to work on. Other partnerships could well involve different projects or modes of collaboration. It is fundamental to the success of partnerships like these that they grow from actual community needs.

2 The bootcamp concept

The concept of the grammar bootcamp is simple. Several undergraduate students work together (and with a supervisor, this case, Claire Bowern) to write a sketch grammar of a language from existing materials. They do this in collaboration with language community members and a local linguist (where possible), to make linguistic general- izations and write them up as far as possible. Sue Hanson and Denise Smith-Ali are Senior Linguists for the Goldfields Language Centre and the Noongar Boodjar Lan- guage Cultural Aboriginal Corporation espectively). It is called a ‘bootcamp’ because

3 this is done intensively: the aim is to write a book-length description in a single month.

2.1 How the project arose

The bootcamps originated as a collaboration between Sue Hanson and Claire Bowern (two of the authors of this paper). Sue is a linguist, now based at the Goldfield Aboriginal Language Centre in Kalgoorlie, . Claire is based at Yale University in the USA. Regional language centres in Australia are typically the main (and sole) coordination point for all linguistic work for the Indigenous languages of the region. This might involve everything from coordinating language help at the local schools, to language documentation, to coordinating with local and regional authorities for signage, to fielding interpreter requests, and helping external researchers. They are typically funded with bare-bones federal funding (which requires extensive record keeping and audits) and are usually staffed with some permanent staff and oneor more volunteers; they are usually overseen by a council or board of elders; indigenous representatives of the languages of the region. Sue had projects which needed linguistic expertise but because of the complexity of running a language centre that represents multiple languages, and the many demands on her time that coordinating that centre entails, there was more work to do than time to do it. Claire has undergraduate students in the US who are looking for research projects, but cannot necessarily travel to Australia (which would also place more of the supervision burden back on Sue). We wanted to find a way to work that would a) result in materials that were wanted and needed by the Indigenous communities we are working with; and b) provide a meaningful training experience for undergraduates, to encourage them to continue work with ethical language documentation.2

2One might reasonably point out that this project centers linguists rather than community mem- bers, and focuses on the linguists and linguistic work rather than the reclamation activities. This is true. We see it as important for academic linguists to be trained in the ethics of what they do, and that has to start as early as possible in a person’s linguistics career. This paper also describes a small piece in a much larger set of work by regional language centres in Australia, run by language

4 Previous stages of the Chirila project (Bowern, 2016) focused on lexicographic ma- terials, and in particular, the creation of a database of lexical items in Australian languages. Materials were identified and permission obtained to add them tothe database.3 In the course of this lexicographic work, it became clear that another type of collection is also found in archives: the materials of field workers who conducted extensive work on languages but were, for a variety of reasons, unable to write up and publish the results of that fieldwork. In the course of identifying archival collections for processing, Claire has also been approached by linguists offering to provide mate- rials for student projects. That is, there is a growing awareness within the field that unpublished fieldnote collections represent an opportunity for student researchers to gain valuable experience in language documentation. Once the general set of principles was worked out, Sue suggested projects to com- munities and we worked with those who were keen to participate. So, while the project was not directly “community-initiated”, it was initiated through local needs, and the partnerships were with communities who wished to work in this way.4

2.2 The languages involved

Four bootcamps were conducted over four years (2014–2017) with languages from the Southwest of Australia. The language and community locations for the bootcamps are given in Figure 1. Ngalia, Tjupan, Kuwarra, and Cundeelee Wangka are all Western Desert/Wati languages. They are closely related to one another, though they show differences in communities and based on their priorities. The bootcamps assist with one aspect of that work. 3The process of consultation depended on the materials. For some languages with 19th Century materials and no current descendants, there were no representatives from whom permission could be sought. Other permissions involved a combination of contacting linguists and asking them to check with the groups they had worked with, along with talking with regional language centres and giving presentations at conferences with substantial presence of Indigenous language leaders. 4For further details on community-oriented work in Australia see, for example, Wilkins (1992), Rice (2006), and Yamada (2007).

5 Figure 1: Language locations for grammar bootcamps both lexicon and grammar, enough that having different sketch materials for each va- riety is needed. Noongar is in a different subgroup of the Pama-Nyungan family. The collaborations were facilitated directly through the Goldfields Aboriginal Language Centre. The Noongar work was also done in collaboration with the Noongar Budjar Language Centre, part of the Noongar Budjar Language Cul- tural Aboriginal Corporation, overseen by Denise Smith-Ali (Noongar), Sue Hanson (Goldfields) and Claire Bowern (Yale).

2.3 Types of materials

Archival sources, of course, take many forms. Across material in Australian archives, there are collections which have very different composition, clarity of metadata, number of languages, legibility of handwriting, inclusion of audio and video, and accompanying transcriptions. Some corpora have been digitized from print or analog audio sources; others were created from digital media. (See Babinski & Bowern 2021 for some discus- sion of the ways in which digital corpora vary.) The linguistic documentation we are working with is, for the most part, digitized and word processed. That is, though we are working with corpus materials and field notes, they are notes which are relatively easy to begin working with quickly, having

6 been typed in a consistent orthography, and translated. We have particularly worked with print materials, draft dictionaries, and elicited sentences. We have also worked with published sources that contain example sentences, as well as theses and prelim- inary drafts of sketch grammars.5 That is, we are building on preliminary analyses rather than beginning a documentation project from scratch. We have also had ac- cess to audio recordings for some of the languages. Some material that was used in the bootcamps was been handwritten but much was already available as digital tran- scribed text. Where textual material had not been transcribed, our first priority was to type it and add it to the collection. We focused on collections that were already close to ready to use (because they had already been partially processed) so that we could spend most of the time working on analytical questions and writing. The materials that comprised the bootcamp corpora were from a combination of published sources, institutional archives, and personal collections. For example, these are the materials worked on for the Ngalia grammar in 2015:

• a dictionary of about 2000 words, with 1000+ sentences illustrating various head- words.

• Approximately 20 pages of interlinearized text, from one speaker.

• A learner’s guide, written by Sue Hanson and Kado Muir, of about 30 pages, and distributed for the Kalkatjurra Cultural Centre, the Ngalia Foundation, and the National Trust of Australia (WA).

Languages for bootcamps were suggested by Sue Hanson, based on her knowledge of the region, the priorities of the language centre, and the materials that would be most suitable for long-term work.

5These include Blyth (1988); Glass (2006); Goddard (1982); Hansen & Hansen (1978, 1992); Hef- fernan (2000); Miller (1972), amongst others.

7 2.4 How students were recruited

Advertisements were sent to many universities with linguistics programs to invite stu- dents to apply. The program was also advertised through social media. Students were required to have taken classes in phonology, syntax, and morphology, and either field methods or typology. They also needed to have experience in writing research pa- pers, since they were required to analyze data and write it up quickly. Applications were particularly encouraged from rising seniors or seniors who had just graduated and were starting graduate school the following year, since they are the groups most likely to have the prerequisites and also the most likely to gain the most benefit from the research experience.6 Students were selected by application. Two letters of recommendation commenting directly on the students’ skills at analysis and writing were requested. Students also wrote a letter of application detailing their prior language experience, why they want to work on this project, and their future plans. Participants were selected from the applicants based on their classroom and research experience, writing skills, and their future plans (giving preference to strong writers with research experience who intended to go to graduate school7). Students were paid a stipend, travel costs, and living expenses while in New Haven. The budget for the bootcamp came from the National Science Foundation’s Research Experience for Undergraduates program, and as such, applications were limited to US citizens. Language workers were also compensated for their time.

6Setting the prerequisites in this way inevitably excludes applicants. Since the program is funded by the NSF’s REU program, it is restricted to US citizens (per NSF rules). By including substantial linguistic coursework and some research experience as prerequisites, that excludes applicants whose linguistics programs don’t offer a large range of courses, or for whom access to research opportunities at the undergraduate level is difficult. However, since the premise of the bootcamp is that the students have the skills and training to be able to analyze data quickly and to write up the results clearly and effectively, it’s not a first research experience. 7To my knowledge, all but one of the participants in the program are currently in graduate school or have graduated from doctoral programs.

8 2.5 Why write a grammar in a month?

The bootcamp program brings benefits to communities, linguistics, and the student researchers. For the language communities involved, it makes materials on their lan- guages more accessible. It brings them out of archives and in a format where the documents can be incorporated into other language practices. It does not replace other language activities in the community but it augments and makes other things possible. It also provides communities with access to additional linguistic partnerships. We have not emphasized the potential for long term partnerships between students and communities here, because most of the students we have worked with are not in the position to commit to a multi-year project right at the start of graduate school. Both Claire and Sue have long-term partnerships with communities throughout the region (particularly Sue). There is a benefit to linguistics as well, since we learn more about the languages of the world. For the undergraduate students, they get intensive training, research, and writing experience with a real project, not a problem set with cleaned-up data. They have a CV line which may help in their applications to graduate school. They also benefit with engaging directly with community members about language concerns and the broader context of ethical linguistic research from the beginning of their studies.

3 Procedures

3.1 Timetable

The following description generalizes over the 4 years of bootcamps. Some of the indi- vidual details differed from year to year but this setup describes the general procedures. As the name “bootcamp” implies, the month is intensive. The students and Claire meet everyday, twice a day, as a group.8 From approximately 9am to 10am, we discuss

8After the first week, this was changed to a less frequent schedule, depending on the year.One

9 the day’s plan and any questions from the previous day. Students will work on their materials from 10–3, analyzing the data and drafting sections of a sketch grammar. From 3pm to 4:30pm, we read each other’s work and review the questions still to be answered. For Ngalia, all three students worked on the same grammar, taking the lead on different topics. For the 2016 bootcamp, there were five students, working onthree languages between them. In 2017, all students worked on Noongar but on different topics (phonology, morphology, and dialect variation in the lexicon). The differences reflected the extent of materials (more for Ngalia, less for Cundeelee, Yulparija and Kuwarra). We worked on a 2–3 day cycle after the first day of discussion of introductions, general information about Australian languages, and procedures. On the morning of the first day, we assigned topics. These “topics” ended up as sections or subsections within the grammar. At the morning meeting we would discuss the scope of the section, the types of material to be covered, and the like. Students were encouraged to consult other grammars and sketches to familiarize themselves with the types of structure they might see in the data. Then, from roughly 10am to 3pm, they would work on their sections, producing a handout with the relevant data (example sentences), morphemes, a provisional analysis, and points to be noted (and questions to cover). At the afternoon meeting, we reviewed what had been found, clarifying any ques- tions. Over the next day, they write up a first draft of their findings, including many examples and making a note of any questions that remain. Claire gives them feedback on this that afternoon, and they revise as needed. They then start a new topic on the next morning. Some topics (such as verb morphology) may take several sessions and year we met once per day with check-in over email in the afternoon after the first week. Another year we met twice a day several times per week, but had every second day for longer sustained writing. For days when we had elicitation meetings, which were held in the evenings due to time zone differences, we started late the next day.

10 are broken down into smaller topics. Others, such as numerals, may be finished more quickly, and so a new topic is started or earlier topics revisited and augmented. The aim is to appropriately “chunk” the material so that the topics are always feasible to do in the short amount of time available before moving on, but not so narrow that the sketch ends up incomplete or unbalanced. A further consideration is that the materials we work with are not so copious that they necessarily contain large numbers of examples about any individual phenomenon. That is, for some parts of the language, we might only have one example that illustrates the point. That limits the conclusions we can draw about those topics. Maintaining connections and keeping the language communities in the loop about how materials were going was important. We held regular (weekly or twice-weekly) sessions in the evening with Sue, Denise (for Noongar), Kado Muir (Ngalia), Geral- dine Hogarth and Luxie Hogarth-Redmond (Kuwarra). Some of these were clarifying questions about grammar; others were reports of findings, and discussions about how to proceed, as well as general chats around life and getting to know each other. The clarifying questions were aimed at filling in gaps in paradigms or syntactic questions. For example, for some verbs, we had the past tense and present tense form, but needed to know the future tense form. For other verbs, we had the present tense but needed to know the past tense form so that we could record which verb conjugation class it belongs to. In other cases, points in the dictionary needed further clarification. For example, some words were glossed in English but the English word is ambiguous as to whether it is a noun or a verb. Of course, over the course of the month, the students learned more about the languages they were working on and needed to revise their earlier writing. They were encouraged to make notes about additional points of analysis at the relevant point in the grammar, but to concentrate on one section at a time. In that way, we struck a

11 balance between keeping track of revisions and completing sections in a timely manner. This part of the bootcamp also provided explicit instruction in writing and how to plan and draft effectively.

3.2 Topics for the sketches

All bootcamps began with a list of topics that formed the skeleton of the sketch gram- mar. This list was compiled based on what types of information are typically feasible to find in basic language materials. It includes topics that one would expect tosee, based on other sketch grammars. In all cases, students were encouraged to expand the list of topics as they worked on the materials. The broad topics are given below.

1. Language background, speakers, inven- 11. Agreement tory of existing materials, language location 2. Phoneme inventory, minimal pairs, 12. Pronouns phonotactics 3. Possession 13. Interrogative pronouns/questions 4. Case marking: core cases 14. Numerals 5. Case marking: locational/oblique cases 15. Negation 6. Other nominal morphology (e.g. deriva- 16. Adverbs tional marking, number marking) 7. Verbal inflection 17. Copular clauses 8. Adjectives 18. Subordination 9. Word classes 19. Focus/topic marking 10. Word order 20. Other morphology not covered here

For some of the sections, students were given a guide that they filled in. For example, for phonology, they were asked to describe the phoneme distinctions, stress realization, to extract syllable types, and the like. A big weakness of this model for language reclamation materials is that it does not cover some of the topics that are most useful to community-oriented language use, such as greetings, conversational turns, or interjections. This is in part because our material does not contain examples of this type. We focused the grammar sketch on describing the structures which are present in the original materials we are working from.

12 3.3 Software

We collaborated in a shared Google Docs folder (docs.google.com). Everyone in the project had access to each other’s work and students were encouraged to work collaboratively wherever possible, to comment on each other’s sections, to ask for help, and to support each other.9 Google docs was also useful for sharing work between participants in Australia and those in the USA, as the Language Centre linguists could see what the Yale-based linguists were writing. The data was mostly in Toolbox format; a backslash-coded text file format read by the program Toolbox. Working collaboratively on Toolbox files is difficult, since one cannot edit the files simultaneously. Because we were all in the same room most of the day, it was possible to have one person edit the file at once (others made “read only” copies from the main file). We also edited the file as a text document where the additional features of Toolbox parsing were not required. However, this did slow down work at times, because only one person could make edits at a time. At present, there are no effective collaboration tools where multiple people can work on the same files simultaneously. Where we also had audio, this was shared through Dropbox.

4 Project Outcomes

The outcomes for the bootcamps were 6 sketch grammars for community use, on the Ngalia, Tjupan, Cundeelee Wangka, Yulparija, Kuwarra, and Noongar languages. Ngalia has been accepted for publication with Pacific Linguistics and comments from reviewers are being addressed. The Tjupan, Cundeelee, and Kuwarra grammars were produced for community use rather than academic publications and so are not generally available. The Noongar material is currently being edited for a chapter for a forth-

9One of the points that is emphasized in the bootcamp is effective and respectful collaboration, which includes being comfortable sharing work in progress.

13 coming Handbook of Australian Languages10 and the bootcamp produced considerable additional material which the Noongar Budjar Language Centre can use. In addition to the grammar work, an additional outcome was extensive work on a few of the dictionaries. Those dictionary files were part of the input to the grammar work. In the course of working on the grammar and analyzing examples sentences, we were able to make progress on various aspects of dictionary work, including spelling standardization, making new headword entries from words in the example sentences, adding additional relevant examples; editing, expanding, and refining glosses based on the example sentences; and adding grammatical information such as paradigmatic information about verb forms. We were also able to compile lists of suggestions for priorities for future work.

5 Further discussion

In this section, we provide further discussion about the bootcamp approach and its efficacy. We also address participant reflections on the bootcamp program andpoints to keep in mind for future bootcamps.

5.1 Questioning the “bootcamp” approach

One might legitimately question whether a bootcamp approach is appropriate for lan- guage work. After all, it prioritizes speed over other considerations, which could lead to sloppy analyses, errors, and with more time a better grammar could be written. Naturally we agree that “faster isn’t better”. Working to a strict deadline like this brings up the substantial possibility of errors. It also invites cursory analyses and overlooking generalizations which can only be found with a deeper knowledge of the language and extensive work with speakers and language owners, often over many years. None of that is possible in a month-long bootcamp. There is also limited time to

10The Yulparija sketch was not completed.

14 revise and the students begin working on the language with very little previous back- ground in Australian languages. Aspects of the language which would be obvious to Australianists with experience in documentation of these languages (such as suppletive ergative allomorphs that vary between -lu and -ngku, for example) can be puzzling to those encountering the variation for the first time. We have a number of methods in place to avoid problems of this type. First, by working collaboratively, everyone is able to catch each other’s errors.11 By working regularly with speakers of the languages (where possible) we are able to check analyses that we might be unsure about. Sue, Denise, and Claire also have extensive experience with the languages of the region and are able to point out possibilities that can be checked. Time is built in to allow students to go back and revise. Secondly, because this is the only project students are working on during the month, they are very focused and able to work without distractions. The very first grammar project, in 2014, was run during a semester, with students meeting every two weeks as a group and working independently in pairs at other times. It was less successful simply because the students involved were doing many other activities: other classes, extra-curricular activities, and so on. They were simply unable to devote the sustained attention that lets them make rapid progress and it showed in first draft of the resulting material. More examples were missed, fewer connections between sections were made, and the overall grammar was more cursory than the bootcamp grammars which were completed in a month of sustained work. Some sections were unfinished and others were inconsistent with material presented in other sections. Working together and focused on a single grammar made for a much better grammar. Thirdly, we argue that “accessible but some gaps” is much better than “currently

11Errors are inevitably made but by working collaboratively we emphasize that errors are not penalized, but simply fixed. We stress this because this approach is rather different from the “grading” systems that feature in much academic work, where students must work alone and compete for grades.

15 inaccessible but might be in the future”. The materials that the bootcamp is working on would otherwise not be available if it weren’t for these projects. There is always more work to do on language documentation, and always more possibilities for community- oriented language work. We expect that these grammars can be improved on in time, and hope that this happens. Fourthly, we point that having more time does not always result in a “better” book. Some grammars take years, even decades, to write. That might be because the writer has so much information about the language that it takes years for them to describe. Or that the morphology is so irregular and so complex that a parsimonious analysis requires extensive research. Or that new data is collected and incorporated. Or that the grammar is written piecemeal, a little at a time. None of these is the case here. The written sources on which the grammars are based are rich, but they are not so extensive that thousand-page grammars could be written from them. If moving substantially beyond the topics of a typical sketch grammar, we would very soon run into missing (or absent) data. These languages, like all others, are complex and nuanced systems, but the basic phonological, morphological, and syntactic systems (that we have information about in the written materials) do not require hundreds of pages of detailed sub-paradigms. The languages of the Western Desert region are sufficiently similar to one another that some clues to each variety can be gleaned by looking at closely related languages, but different enough that they are distinct varieties in their own rights and thatone should not assume that each variety has the same phonological, morphological, and syntactic structure.12 In summary, the bootcamp is not just about getting things done fast, no matter

12This is not all that language is, of course. But it’s what we’ve focused on for sketch grammars that summarize information for later use in school programs where most of the students come from these cultures.

16 what. It’s about doing the best job we can, and bringing materials to a point where they are as useful as we can make them for language programs and communities, given the real constraints of time and material availability.

5.2 Impact on speakers

While the primary purpose of the bootcamp was to provide linguistic materials in com- munities, in reality there were several ways in which the bootcamp program provided an impact on communities and language speakers (and language workers). This section describes this impact on the language speakers who participated in the program. The fact that linguistics students from another country wanted to work on a lan- guage added a great deal of esteem to the language work, from the speakers’ perspec- tive. The speakers were amazed that people from another country were aware of their language and valued them so much that they would like to work on the languages with them. This comes in a context with the State Government does not officially recognize the languages of Western Australia. Secondly, the bootcamps enabled speakers to see that locally based linguists and international linguists all work the same way, using similar tools and methods, and asking similar questions. It reinforced the work of the local field linguists as speakers saw that international linguists work to the same protocols and standards. Thirdly, the project provided reinforcement to the speakers that Indigenous Aus- tralian languages are valuable, important, and valued. That is, the project created a great deal of social capital for the speakers in their local communities. Finally, linguistic fieldwork is slow. It can frustrate speakers as they are anxiousto see resources produced quickly. The bootcamps enabled a fast output of specific and relevant materials that fed back into local language programs. It made speakers feel that progress was being made.

17 5.3 Student reflections

Following each bootcamp, students provided feedback on the program. This section summarizes their responses and further contextualizes the bootcamp program. Students liked the format where they were directed towards particular topics. They felt it was satisfying to be in charge of a particular subtopic or subproject, but they also felt well supported. Depending on the year, some wished they had had more op- portunities for direct collaboration, rather than working in parallel on different topics. Other students made an effort to be collaborative and sought out feedback on their work from other students in the bootcamp. Claire’s impression was that collaboration worked very well but the students had to be on board; it could be encouraged and facilitated but not mandated. Students reported something of a tension between perfectionism and the need to keep writing on new topics. Some would like liked some more time to polish what they were doing, while others were satisfied that they had been able to produce reasonably complete drafts within the time frame. This tension is, we think, unavoidable, since the pace of work would not have been sustainable for more than a month. Students reported feeling noticeably more tired by the end of the month. Several students commented on how important they felt it was to do manual in- terlinear glossing, which forced them to make sure they understood the morphology of the language and let them compare and test their analyses as they worked. That is, the quasi-automatic parsing that Toolbox provides is not necessarily the nest way to understand a language, though it does facilitate rapid interlinearized text. There were comments about the role of previous analytical materials. Some found the materials useful, but others commented that the grammars of related varieties did not help them in their analyses, because there were enough discrepancies that made

18 things more confusing than helpful. That is, they were unable to separate partial de- scriptions from genuine differences between varieties. That is, some of the descriptions appeared to describe different phenomena, whereas in reality they were describing the same phenomena, but drawing slightly different conclusions due to differing amounts of data. The students also commented that they learned a great deal about good documentation practices through having to work with materials collected under many different circumstances, which outweighed the confusion around analyses. Software was something of an issue. There was a steep learning curve with using Toolbox, and learning how to use Toolbox while also using it to analyze data was a problem. In future bootcamps, it would be advisable either to make sure that students have experience with the software tools that are being used, or provide a “pre-training” before the camp starts, so that they are not trying to learnt he software at the same time as conducting analyses. While the bootcamp idea was not to provide a great deal of “pre-work” beyond some articles on the languages of the region, more preparation is likely to make the first part of the bootcamp go more smoothly. Other problems are inherent to current language documentation. At one stage, we had three different sources of data which were collated in three different programs. Working between them was difficult as the programs have limited interoperability. This is, we think, a general problem in language documentation and one that fieldworkers constantly have to negotiated. Finally, students also commented on the reasons they applied to the program. Having a funded summer “internship” was important to some. They were looking for work with endangered languages but did not want to cold-call communities, reinvent work that was already underway or otherwise waste valuable time of elders when they could be useful doing something else. They liked the interaction with students from other institutions, and the methodological and theoretical pluralism that came from

19 recruiting students from different departments. Students also had different comments on the role of linguistic theory in their work. Some enjoyed the opportunity to work on language data without being pressured to make a particular theoretical point, or to work within a particular linguistic theoretical framework. Others enjoyed the opportunity to learn about languages they previously were not familiar with, and commented on how it made them think about those the- oretical questions. They appreciated being able to work on many different areas of a language without having to just specialize in semantics, phonology, or another subfield.

6 Conclusions

In conclusion, the bootcamps worked well for the particular combination of community members and linguists who were collaborating on these materials. We do not see this as the only way to write a grammar, nor the only way to do long-distance collaboration. This is a particular model that capitalizes on a need around these language centers and on the skills of those involved. It was designed as a program that benefited those involved and other communities may also benefit from something like it. We stress, however, that there are communities and languages where this program would not be appropriate. For example, it relies on working on language off-country by outsiders. Some communities do not want that. It is also based on linguist-centered models of language description, albeit in a way that makes those materials accessible to teachers for lessons. There are languages where that material is already available. Or there are languages where the existing documentary material is not yet in a form where this scale of project would be feasible. This model is good for ‘long-distance’ collaboration. That is, it capitalizes on the strengths of both the sites and the people involved. It does not outsource language revitalization activities which would be better done in the local community (running

20 lessons over the internet, for example, would be inappropriate). However, it should be pointed out that this collaboration is limited and relied to some extent on prior connections (Claire and Sue’s connections, and Sue’s in person connections in southwest Western Australia). It worked in part because we already knew each other. I imagine it would be more difficult to work like this from scratch.

References

Babinski, Sarah & Claire Bowern. 2021. Contemporary digital linguistics and the archive: An urgent review. ICLDC 2021.

Blyth, Noel. 1988. Wangka Western Desert base dictionary. Wangka Western Desert Base Dictionary. ISBN 073161738X, 1988. Part (1) Wangka, 42 pages. Part (2) Grammar. Part (3) English, 30 pages.

Bowern, Claire. 2016. Chirila: Contemporary and Historical Resources for the In- digenous Languages of Australia. Language Documentation and Conservation 10(1). 1–45.

Campbell, Lyle & Anna Belew. 2018. Cataloguing the world’s endangered languages. Routledge.

Commonwealth of Australia. 1997. Bringing them home: Report of the national inquiry into the separation of aboriginal and torres strait islander children from their families. Tabled in Parliament, May 1997.

Glass, Amee. 2006. Ngaanytjarra learner’s guide. Alice Springs: IAD Press.

Goddard, Cliff. 1982. Case systems and case marking in Australia: a new interpreta- tion. Australian Journal of Linguistics 2. 167–196.

21 Hansen, K.C. & L.E Hansen. 1978. The core of grammar. Alice Springs: IAD Press.

Hansen, K.C. & L.E Hansen. 1992. Pintupi/Luritja dictionary. Alice Springs: IAD Press.

Heffernan, K, J. & Heffernan. 2000. A learner’s guide to Pintupi-Luritja. Alice Springs: IAD Press.

Miller, Wik. 1972. Dialect differentiation in the Western Desert Language. Anthropo- logical Forum 3(1). 61–78.

Nordhoff, Sebastian & Harald Hammarström. 2012a. Cataloguing linguistic diversity: Glottolog/langdoc. In Dh 2012, 296–298.

Nordhoff, Sebastian & Harald Hammarström. 2012b. Glottolog/langdoc: Increasing the visibility of grey literature for low-density languages. In the 8th international conference on language resources and evaluation [lrec 2012], 3289–3294. ELRA.

Rice, Keren. 2006. Ethical issues in linguistic fieldwork: an overview. Ethics in Academia 4(1–4). 123–155.

Wilkins, David P. 1992. Linguistic research under Aboriginal control: a personal account of fieldwork in Central Australia. Australian Journal of Linguistics 12. 171– 200.

Yamada, Racquel-María. 2007. Collaborative linguistic fieldwork: Practical application of the empowerment model. Language Documentation & Conservation 1(2).

22