1.1 Arrival in Sula
Travel from Hawaiʻi to remote areas in Sula is not for the listless, but it is worth each of the five flights, the overnight sea voyage, and the longboat jaunts. The approaching horizon’s white sand and verdurous peaks are enough to mute stabbing rib cramps as the dugout longboat slams over every swell. And the growing aroma of nutmeg, cacao, and cloves drying on straw mats under the sun makes sitting between a cow and a bouquet of live chickens hardly noticeable. Crowds gather on the beach to greet arriving longboats. Villagers hope for news of loved ones who labor in distant parts, while laughing children leap into the sea from thick branches that twist out beyond the surf. Arrival is met with a cup of a freshly brewed drink; it is milky and sweet but also sharp. Coffee grounds, ginger, and chopped kenari nut dance in a medley of the very flavors that gave the Spice Islands their name. Each sip is a sensual history lesson that illustrates why these spices were once valued above gold. This traditional beverage often fuels hours of jungle trekking to come for passengers en route to periphery communities—communities where Sula is spoken as it has been for generations, places that seem to exist out of time and whose inhabitants seem unaware of the global community at its doorstep.
1.2 Language in Context
Sula is a remarkable language with a hidden record of humanity’s distant past woven throughout its vocabulary, and it is a language whose grammar at times challenges current understandings of how language works. Also, while Sula unmistakably belongs to the Austronesian language family, many of the words have unknown origins. Comparing these words to other languages might reveal connections with civilizations that were previously unknown, and it could result in redrawing part of the map of human migration—or at least help answer some of the persistent questions about how and when humans settled in parts of Oceania. Sula’s phonology is also remarkable in that the sounds show a type of historical change that is not known in any other language; these are sound changes that some of the most widely accepted linguistic theories of the past hundred and fifty years have deemed impossible. This is discussed in Chapter 2: dialects, part 1 and in the phonology section of Chapter 3, phonology).
Sula has neither a formalized orthography nor a print literature, and Indonesian Malay has displaced it from several of its traditional spoken domains and mostly supersedes the language in newer, non-traditional domains such as political meetings and digital communication. I was the only linguist researching Sula during the course of my work, and I am also the first to conduct in-depth research specific to the language. Previously, Sula language data was limited to two short wordlists that are over a century old (Holle c. 1900 via Stokhof 1980, Wallace 1869), a 35-year-old article on genetic grouping (Collins 1981), and a short, unpublished grammatical sketch written as an undergraduate thesis (Umaternate 2013). Even the most basic descriptive materials were lacking, such as a dictionary, grammar, or available texts. Although not ‘undocumented’ in the sense that no information whatsoever existed, when I began my research, Sula was and it still is unarguably an under-documented language (e.g., Blust 1981, Collins 1981, 1983, Esser 1938, Grimes 1992). From a practical standpoint Sula was certainly undocumented in the sense that the academic literature did not contain enough basic information to help me begin to learn to speak prior to setting off on my first fieldwork trip, and information was unavailable to answer even the most fundamental questions about the language’s typology, structure, or lexicon. The academic and non-academic literature alike was insufficient to even populate a basic Wikipedia entry for the language, as is shown in Figure 1.
Figure 1. Screenshot of the English Wikipedia entry
The French entry for Sula also includes the following list of eight vocabulary items pulled from Collins 1981 (albeit with errors): ‘house’, ‘egg’, ‘pig’, ‘hear’, ‘chicken’, ‘seven’, ‘two’, and ‘burn’.
My research has demonstrated that Sula is primarily transmitted orally and that is not represented in the formal education system. As an orally transmitted language, working with Sula presented me with challenges for studying that I had not encountered with any of the languages I had previously studied. I relied heavily on guidance from experienced fieldworkers and from a number of books for planning my fieldwork and anticipating what to expect. Beyond that, I was in a situation with the language where there was very little I could study ahead of time to prepare myself to communicate in the Sula Language, and even if I did speak the language, Sula would—not surprisingly—be an ineffective language for the rest of my tasks in Indonesia, and getting to and from Sula is not always straightforward. While there is an airport (read ‘landing strip’) on Sanana island, it has been closed due to a land dispute for many years, and even when it was open, it served irregular and unpredictable—terrifying—flights on tiny aircraft to a few other remote islands in Eastern Indonesia. During my first fieldwork trip, I spent a full day in Bali trying to find a travel agent who could both speak English and who was aware that a place called ‘Sula’ existed in Indonesia. Getting to Sula requires first going to Ambon or Ternate. From there, even when the airport on Sanana was open, it was still more reliable to travel by crowded passenger boat, and that ride takes almost a full day.
On arrival in Sanana, many of the people in the port area are not ethnically Sula and do not speak or understand the language, and since English becomes far less useful after leaving Bali, I quickly came to realize that it would not be a practical intermediary language for me; to get anything at all done I would need to learn Bahasa, and fast. Because of this, my first fieldwork experience was a crash course on both languages—luckily I am not shy and I had a lot of motivation to learn since even things that might have distracted me required me to communicate. By the end of my third month when it was time for me to return to Hawaiʻi, I was not having any deep philosophical discussions, but I had become functionally proficient in both tongues at a rudimentary level, and I had collected a mountain of data so massive that I have honestly still not finished going through. After returning, I enrolled in my first Indonesian course and began combing through data, decoding Sula, actively spending time participating in the East West Center’s large Indonesian community, and frequently embarrassing myself as I learned to differentiate what was Sula, Bahasa, and local Moluccan Malay. While I will never be a great scholar of Bahasa, I did manage to learn enough to find the answers I needed in news articles, government publications, and academic papers. And my subsequent field trips were not without communication challenges, but they were highly productive and they went much smoother.
Figure 2. The Sula Archipelago
The Sula Archipelago is in the Maluku Utara province of Eastern Indonesia. It lies at the boundary of the Molucca Sea (north of Taliabu and Mangon) and the Banda and Ceram seas (to the west and east of Sanana Island).
Sula is an indisputably threatened language; however, the degree of threat is not yet determined. My research has identified the region with the most severe language attrition to be the town of Sanana, where over a quarter of the native Sula population resides. While it is rare in Sanana to hear young people conversing in Sula, in some of the less populous, remote villages throughout the islands, the language is still commonly used. And even in Sanana it is not uncommon to hear the language spoken among older adults. All Sula speakers are also native speakers of Malay, but whereas Malay is a dominant world language with a rich, cross-meda literary tradition, written Sula is found almost exclusively in graffiti, on handmade signs, and (occasionally) in political slogans (Appendix A). The Sula language is increasingly pushed out of traditional communicative domains in favor of Malay, and it is not chosen as the primary medium of communication in newer domains like text messaging and social media.
1.3 The Past
The Sula Archipelago has been listed among the least studied regions in Indonesia (Collins 1981, 1982). It consists of three main islands: Mangon, Taliabu, and Sanana, located at the geographic center of Maluku, Indonesia. The combined land area is roughly 60% of the Hawaiian Islands, but Sula has less than 10% of Hawaiʻi’s population, the largest portion of whom being recent, non-Sula speaking immigrants from Sulawesi. Since prehistory (at least since the arrival of Austronesians), the Sula are reported to have relied primarily on protein from the sea as well as chicken, pigs, cuscus, and deer, and their staple starches have been sa (sago palm starch), nui (coconut), sisa (sugarcane), suk (breadfruit), and fia (various banana varieties). I have seen kat (taro) growing wild around populated areas, but I’ve never seen it consumed, or heard it referred to as part of the local cuisine. When I pointed it out and asked an elderly consultant, Ismael Duila, I was told that ancient people might have eaten it, but that it wasn’t eaten anymore. Descendants of a primitive dog breed, as, are also on the islands, but the introduction of Islam complicated people’s relationship to the animals. It is unclear whether they were a food source, companions, or both. Today they are useful to farmers for keeping wild animals away from fields, and some might still use dogs for hunting, as the term dol as ‘to hunt’ exists in the language, and it literally translates to ‘bring/invit + dog’.
The early spice trade opened Indonesian islands to more traffic and trade, and that brought many other animals and plants to the island including goats, cows, buffalo, jackfruit, tea, and durian. The later spice trade and European expansion introduced many New World and African staples such as tomatoes, cassava (yuca), potatoes, chili peppers, peanuts, coffee beans, and maize. Traditional foods remain the staples of today, although imported rice, wheat, and refined sugar have largely taken the place of sago in larger villages.
Since the spice trade began, the economy of the Sula people has been based on tropical forest agriculture: cenke (cloves), kemiri (candlenut), ipa (kenari nut), mina kau (cinnamon), paa fat (nutmeg), and paa ful (mace). Even today, that appears to be true for the ethnically Sula population; however, the Bajo (Sama-Bajau) population supplies fish to fishmongers from other areas, and the northern Sula islands has remote mineral mines and logging operations. These are reportedly not locally owned ventures, and they do not primarily employ an ethnically Sula worker base.
Although there is a scarcity of archaeological work on the Maluku region and none at all to my knowledge on Sula specifically, there is a good deal of geographically peripheral data (Spriggs 1998). To that, whoever the inhabitants of Sula may be, it is certain their ancestors had to pass through peripheral regions along the way.
1.3.1 The Past. Pre-Austronesians
Sula is a Malayo-Polynesian language within the greater Austronesian language family, and the Sula people bear the hallmark features of Austronesian culture. Much has been written on the Austronesian people and their routes of migration, and this dissertation will help to fill in some of the Austronesian picture in and around Sula, but it must be noted that half of the story of Maluku’s habitation predates the arrival of Austronesians by tens of thousands of years. And half the story may mean literally half; nearby coastal New Guinean and island populations of Papua New Guinnea (PNG) along with other Melanesians have in recent studies been shown to have a 50:50 mixture of identifiable Austronesian and First Sundaland People genes (Chambers and Edinur 2015). Additionally, Pääbo (2014) finds genetic evidence of nearly 5% Denisova hominin DNA among populations in PNG. Pääbo hypothesizes that this admixture occurred in Indonesia prior to arrival on New Guinea, and this is corroborated by other studies that have found that various modern Melanesian populations (and Aboriginal Australian genomes) also contain a significant percentage of Denisovan DNA (Harmon 2012).
Cooper and Stringer (2013) make a strong argument that Denisovan admixture happened east of the Wallace line. The conclusion is based on inability to find Denisovan DNA among the Tianyuan bones recovered in China or other hunter-gatherer (Negrito) groups on peninsular Malaysia or the Andaman Islands. In fact a migration route proposed by Birdsell (1977) has been widely accepted as one of the two likely paths to Sahul that has First Sundaland People crossing the Wallace Line into Sulawesi and then passing directly across Sula (Figure 3) (e.g. Cooper and Stringer 2013, Lourandos 1997). Along this line, Cooper and Stringer state Denisovan admixture first occurred in Eastern Indonesia with genetic signals weakening as they fan out from the region.
The climate of Eastern Indonesia is not optimal to fossil preservation, and even the well-studied, abundant Homo floresiensis remains elusive to DNA extraction. Because of this, we still lack strong evidence as to when Denisovans arrived in Southeast Asia, but it is thought that they and Neanderthals both descend from a group of H. heidelbergensis who left Africa 300,000–400,000 years ago (“Why Am I Denisovan?” 2019). If this is the case, Sula’s earliest ancestors might have arrived a very long time ago indeed.
Figure 3. Migration routes
The two main hypotheses for the route First Sundaland People took into New Guinea and Australia migration (Lourandos 1997)
At a minimum, humans first inhabited Australia 40,000 years ago (Hiscock 2008), and it is likely that this migration happened earlier: approximately 50,000 to 70,000 years ago during periods when sea levels were especially low, joining islands into larger landmasses and shortening the distances between them. Cane (2013) suggests that the first wave of modern humans to reach Australia could have been 70,000 years ago, and that the migration might have been triggered by the Toba supervolcano eruption.
Although the Sula are unmistakably Austronesian, many of their ancestors were not, and much of the language’s vocabulary does not appear to be Austronesian either. Though it is mostly speculative, it is fascinating to wonder which words might have entered the language via non-Austronesian ancestors when considering Sula vocabulary, and if Denisovans had spoken language, it boggles the mind to consider whether some of those words could trace back geological ages before modern humans even evolved.
1.3.2 The Past. Austronesian expansion
Much can be speculated but little is known about the nature of habitation in Sula’s deepest past. However, a more recent wave of immigration is responsible for most of Sula’s languages and culture; and this was of course the Austronesian expansion.
European chronicling of Austronesian languages began as far back as the 1519–1522 Magellan expedition, during which Antonio Pigafetta collected vocabulary from many languages along their circumnavigational route (Fox 2004). Much later, Hadrian Reland (1706–8) put forth a hypothesis for a common ‘Malayan’ language that ranges from Madagascar through Indonesia and farther east. Fox also notes that a link between Polynesian language numerals and Indonesian and Malagasy numerals was later identified by a chronicler on Captain Cook’s second voyage, but it was Lorenzo Hervas y Panduro in 1784 who first synthesized these observations into a cogent hypothesis for a definable group encompassing all of the Malay and Polynesian languages from Madagascar to Rapanui (Hervas y Panduro 1880). Franz Bopp was first to use the term “Malayo-Polynesian” to specifically refer to a linguistic grouping, although Humboldt (1836) is generally given credit (Fox 2004).
Fast forward another century for an attempt at a reconstruction of Proto-Austronesian by the Comparative Method. . Dempwolff (1934–1938) first accomplished this feat analyzing just eleven Austronesian languages. However, as none were from Formosan branches of Austronesian, it is more appropriate to consider Dempwolff’s work to be a reconstruction of Proto-Malayo-Polynesian rather than of Proto-Austronesian.
Robert Blust (1978) took Dempwolff’s foundation and built an edifice on top of it, ushering in a modern era of Austronesian comparative linguistics. Blust recognized the Formosan languages as primary branches of Austronesian, and he provided the following structure for the Austronesian family tree:
Figure 4. Austronesian family tree as proposed by Blust (1978)
Austronesian Family
_____________________|_______________
| |
Formosan Language branches Malayo-Polynesian Languages
______________________________|___
| |
Central Eastern Malayo-Polynesian Western Malayo-Polynesian
______________| ____________
| |
Central Malayo-Polynesian Eastern Malayo-Polynesian
________________________| _________
| |
South Halmahera–West New Guinea Oceanic
Blust has identified ten separate primary branches of the Austronesian family; nine of these are Formosan languages found only on Taiwan, while the tenth, the Malayo-Polynesian branch, exists primarily outside of Taiwan (e.g. Blust 1999, Blust 2013). One representative of Malayo-Polynesian is also found in Taiwan, Tao (Yami) (e.g. Blust 2013). It is spoken on Orchid Island and is thought to have been a back migration from the Philippines.
Blust used comparative linguistics to demonstrate what has come to be the prevailing theory of Austronesian expansion: the ‘Out of Taiwan’ model. Around 5,000 years ago, and probably coinciding with the invention of the outrigger canoe, which provided seaworthy stability, a group of Proto-Malayo-Polynesians exited Taiwan and settled the northern Philippines (e.g. Blust 2013:749, 1999). From there, they settled the rest of the Philippines and continued on to Borneo and the rest of Indonesia and Malaysia as well as in to Micronesia and Melanesia.
The linguistic and genetic evidence points to groups from Melanesia then branching off into Polynesia a few thousand years ago, and linguistic and genetic evidence shows that a group of Austronesians from Borneo sailed all the way to Madagascar, settling the island around 1,500 years ago (e.g. Ricaut et al. 2009). Along the way, Austronesians mixed with pre-Austronesian populations and by the time of European contact in the early 1500’s, Austronesians had spread across more than half the planet: from Madagascar to Easter Island. Sandwiched within those macro-level Austronesian migrations, a group of Central-Eastern Malayo-Polynesian moved into Central Maluku and some of them were the direct ancestors of the people who now inhabit the islands of Buru, Taliabo, and Sula—where three close sister languages have evolved.
1.3.3 The Past. Sultanate of Ternate to present
The Sultanate of Ternate (Kerajaan Gapi) traces back to 1257 and its first leader, King Baab Masyhur Mulamo. At the time, the region was the world supplier of cloves, and as such, the kingdom of Ternate and the neighboring kingdom of Tidore grew wealthy and powerful. Still today walking through rural areas, you will frequently encounter footpaths lined with woven mats where seeds, nuts and spices lay curing in the sun. Although Ternate and Tidore enjoyed great success, much of their wealth was used to fund ongoing wars between each other. During the mid fifteenth century, King Marhum converted to Islam and changed the royal title to Sultan. The religion spread throughout most of the region. In remote Sula villages, even where there are no established businesses or official buildings, there is almost always a mosque with an electrical generator that broadcasts the call to prayer via cassette recording.
In 1512, Sultan Bayanullah welcomed a convoy of shipwrecked Portuguese sailors under Francisco Serrão, in an attempt to forge strong international ties, yet the alliance never grew strong, and by 1575, then Sultan Baabullah Datu Syah expelled the Portuguese. This period was the height of the sultanate’s power at a time when the influence stretched from the southern Philippines to Ambon and from parts of Sulawesi to Papua.
Sula was within the sultanate’s jurisdiction, but it is unclear precisely when it was incorporated or what the political situation was at the time. It has been reported to me that Sula’s village heads were chosen by the sultanate during this period, and that vestiges of this legacy are still present in the family naming system practiced widely throughout Eastern Indonesia known as fam (referred to as marga outside of Maluku). Many fam in Sula are bimorphemic, beginning with Uma-, and as such, I have been told by two consultants that the fam, Umaternate ‘house-Ternate’ does not necessarily indicate blood relations between people with that name but rather it represents descendants of any of the sultanate’s hand-selected (village leaders).
In 1606, the Spanish captured the Portuguese fort on Ternate. The Spanish were allied with Ternate’s rival Tidore Sultanate, and this gave an inroad for the Dutch to ally with Ternate. With Dutch help, the Spanish were pushed out by the 1660’s, but Dutch allegiance came at great cost to the Ternate Sultanate, who had to ceed control of much of its territory to the Dutch East India Company. By the late seventeenth century, Ternate was in a subordinate position to the Dutch, and in 1914, the territory was officially annexed by the Netherlands–all the while Sula remained under Ternate’s administration. This all would change when the Dutch lost Indonesia to Japan during WWII. Following a brief stint under the Imperial Japanese Navy, Ternate was incorporated into Maluku province under a newly independent Indonesia, and in 1999, Maluku was split and the province North Maluku was formed which included Sula and the Moluccan islands to the North. At that time, Ternate was made capital, and Sula once again fell under its administration until 2010 when the capital was relocated to Sofifi on Halmahera.
1.4 Determining vitality
1.4.1 Determining vitality. Speaker numbers and vitality introduction
The annual reference publication, Ethnologue, lists a 1983 estimate of 20,000 native Sula speakers (Eberhard et al. 2020). This estimate is many years out of date, and it is not well explained. In an effort to capture more up-to-date data on language use, in 2014 I conducted a survey to establish a more reliable estimate of the current speaker base, their age ranges, and their geographic distributions. The study considered data from the Badan Pusat Statistik Kabupaten Kepulauan Sula (Central Statistics Agency of Sula Regency) and the results of a questionnaire that was implemented by myself and assistants in sample communities (Appendix B).
Unfortunately resources were limited throughout my research, and this became an especially challenging obstacle during the vitality survey, as it meant I could not only not afford to travel to each community, but I could not personally travel even to each region—particularly on Taliabu Island. The Sanana Badan Pemberdayaan Masyarakat dan Pemerintahan Desa (BPMPD, ‘Community Empowerment and Village Government Agency’) was instrumental for helping me identify regional demographics and helping to set targets for surveying based on where Sula populations do and do not primarily live throughout the archipelago. When possible, I visited communities myself and, in keeping with local customs, I would met with the kepala desa (KD, ‘village head’) who, along with his or her entourage, was almost always eager to work with me directly on the survey. In a few cases the KD was either busy or disinterested, and another group of individuals was suggested for me to work with.
For areas I was unable to visit personally, I relied on the creativity of connections I made during fieldwork. The BPMPD and other civil servants make regular check-ins at communities around the archipelago, and I befriended medical professionals stationed by the central government in Jakarta in remote hub communities. These resources facilitated collecting at least the most basic quantitative survey answers for regions when all else failed, but luckily extended family networks allowed me to go beyond that for most regions (especially for Taliabu and Northern Mangon regions), because residents from all communities make periodic trips to larger towns on Mangon and Sanana to stock up on supplies and sell spices. I was therefore able to leverage family relationships of my consultants from Sanana and Waitina to introduce me to relatives from remote locations when they visited the larger towns. Additionally, telephone interviews were possible for certain parts of southern Mangon and Taliabu where cellular signal was available. While the quality of on-site interviews has the bonus of visual observation and followup questioning about what I observed, I believe the remote interviews still sufficed for attaining basic information within the margin of error. Appendix C lists the survey results by region. A spreadsheet containing full results and formulas is available in the archive.
Results demonstrate a forty-year long period of dramatic communal language attrition, and yet over twice the number of speakers identified by Simons and Fennig. For the Ethnologue figure to have been correct in 1983, there would have to have been a population explosion resulting in hundreds of thousands of additional, ethnically-Sula people on the islands today who cannot speak the language. It is my academic opinion that there were many more Sula speakers in 1983—likely over 40,000.
If the population growth rate of Sula mirrors that of Indonesia, the number of ethnically Sula in 1983 would have been 48067—sixty percent what it is today. The rate of Sula speakers over 50 years of age today stands at 75% even in Sanana city (the area with the most attrition). In 1983, this demographic would have represented only about 25% of the total Sula population (mostly children, because Sula have a low average lifespan). Of the group representing those ≥50 years-old today, only 25% do not speak the language. They would have comprised about 6.25% of the population in 1983, and even if we assume that none of these people alive today are siblings and both of their parents also did not speak the language, that would put the maximum non-speaker base of the Sula population in 1983 at 18.75% and the lowest estimate of Sula speakers at 39,054. The Ethnologue data is un-cited, so I cannot comment specifically on what it represents or where it came from, but if it is an accurate figure, I believe it is being listed in the absence of important contextual factors.
Sula language vitality is determined in part using an augmented and modified version of the UNESCO scale of language endangerment (Brenzinger et al. 2003). This augmented scale deviates from the standard UNESCO scale in that it looks further into topics of mass-media communication and access to digital communication. It considers twenty factors correlated with language health to establish a baseline relationship between access to digital communication and loss of native language. This topic is becoming increasingly relevant as the Internet expands into the developing world, but it has not yet been researched in Sula.
While far from perfect, the findings establish a quantifiable baseline of current Sula language use and the reach of digital communication in the region. Results of this survey cannot be taken as a definitive statement of Sula language vitality though, because the region has begun to undergo significant investments in public infrastructure (both ad-hoc and formally sponsored).
Perhaps the two most significant infrastructure expansions in Sula’s history came about within the last few years. They are, first, the creation of a coastal road that now places remote villages that were historically reachable only by longboat within only a couple hours of the town center by public car. The other significant infrastructural development is the proliferation of battery-powered, smart devices, and burgeoning access to cellular broadband Internet. It is too soon to know what effect these changes will have on language vitality, but history as a model gives us cause for concern when factors isolating a bilingual community from an economically privileged, dominant community are suddenly removed. Each development project places the Sula language’s rural strongholds in increasing daily contact with non-Sula speakers—both virtually and face-to-face. And remote villages are being opened to migrations from non-Sula populations—groups who already outnumber the Sula population in parts of the archipelago. These changes will have unknowable but likely profound impacts on the language and its future spoken domains.
1.4.2 Determining vitality. Newcomer communities
Malay has been a regional lingua franca for generations, and newcomers to the region by and large have no cause to learn Sula; they speak their native languages with others from their region, and Malay is spoken with people from other populations—including the native Sula population. A historical example of this is seen in the Bajo (Sama–Bajau) community on Northern Sanana island. This group is recognized as ‘newcomer’ population; however, they have resided in Sula for so many generations that I was unable to even receive a general estimation of their arrival date. The Bajo are surrounded by Sula communities, and they interact with them on a daily basis. Many Bajo children even attend school with Sula children, yet Bajo language is spoken in Bajo village, and most of the Bajo population report to be unable to speak Sula. Excepting ethnically mixed families, interactions between Sula and Bajo are reportedly conducted in Malay. More recent waves of migrants from Java and Sulawesi are reported (and have been observed) to demonstrate an even lesser degree of Sula language adoption.
Because of the tendency for island newcomers not to adopt the language, it is necessary in estimating the number of Sula speakers to work from population numbers representing only the ethnic Sula population. Most of the census population figures, however, represent the regency’s total population (all communities, irrespective of ethnic background). Further complicating the problem, in 2013, Taliabu island was removed from the Sula Island Regency and a new regency was created for it, the Taliabu Island Regency. This administrative change presents a complication, because Sula groups have settled several communities along the southern coast of Taliabu, and data representing them was not necessarily available in parallel form from Sula’s statistics office.
To find the limits of the ethnic Sula population’s range, I brought a map of the archipelago with me to the regional development office in Sanana and inquired about the ethnic makeup of each village on the islands. Many of the villages were known unambiguously, and some others were determined based on the fam distributions for each village and folk histories. I was unable to personally visit every village to verify the office’s determinations; however, their determinations involved a half dozen representatives discussing the topic until they reached consensus. In a few cases where the representatives did not all agree, they would telephone an external person that they identified as more qualified to answer about a disputed area. Once consensus was reached, I referenced census data that I acquired from the regional statistics office and isolated the figures based on Sula population distribution. The resulting determination of the Sula population’s range can be seen in Figure 5.
Figure 5. Sula language range
Striped areas represent regions on the archipelago where Sula settlements can be found. There are also smaller numbers of Sula living and working in communities that are majority diaspora from Sulawesi, particularly in the northwestern part of Mangon and on Taliabu.
There are limitations with my approach to estimating population. For instance, it does not account for non-Sula residents who have married into the community or Sula residents who have married out of the community. Admittedly, I am making an assumption in concluding that these two figures should roughly cancel each other out, but shy of surveying every village door-to-door, I could not find a way to more accurately estimate this figure. Another limitation with my approach is that I had no way to estimate the numbers in diaspora communities on other islands, most notably the town of Namela on the island of Buru, which is said to be a Sula settlement. It would take a dedicated trip to Buru to determine what percentage of Namela’s population does in fact descend from Sula, and of them, what percentage can indeed still speak Sula. Hence, I am limiting this study to the primary range of Sula speakers, which fall within the sixteen political regions listed in Figure 6.
Figure 6. Political districts
Listed are the political districts within the Sula Island Archipelago where significant Sula populations reside. Mangon is written as “Mangoli” in the following graphic, as the official political division names are written that way. The district names also include directional terms from Bahasa Indonesia (timur ‘east’, barat ‘west’, utara ‘north’, selatan ‘south’, tengah ‘middle’).
Speaker number estimations are limited by census data granularity, which is available for districts but not available village by village (a matter of great frustration to me over the years). The census was conducted village-to-village, door-to-door, so a sub-specified dataset does likely still exist somewhere, and could likely be dug up from storage if I pursued informal means such as providing monetary incentive to key gatekeepers, but as a researcher, this option was neither ethically nor financially viable in exchange for what should technically be publicly available data.
1.4.3 Determining vitality. Taliabu island population
An unspecified but sizable population of Sula people live on Taliabu. They have primarily settled along the southern coast, with very few Sula people living on other parts of the island. I am working under the assumption that this latter group will fall within the margin of error for the island. My contacts in the local government informed me that the Badan Pusat Statistik Kabupaten Kepulauan Sula (Sula Islands Regency Central Bureau of Statistics) does not tally people by ethnic background. Instead, I was informed, “about half” of the people in the southern part of Taliabu are Sula. Needless to say, this was an unsatisfying response, so I set about to find a more accurate way to come up with the figures. While I could not personally visit each village, I was able to ask a network of civil servants and extended family members off-island to begin qualitatively inquiring with villagers in hard-to- reach regions, and after more than a year, I finally received enough responses from each region to extrapolate figures that I am confident in.
The results of this study, while far from exact, fall within the broad, non-quantifiable margin of error that is inherently present in all language vitality surveys—that is, being that language-shift scenarios include a gradation of language gain, language attrition, and varying proficiency levels across geographical and age ranges, numerical representations of multi-factorial, fluid situations are unreliable by nature. That said, this study creates a more reliable, not to mention updated and novel, determination of Sula’s language vitality, by interpreting quantitative and qualitative survey data.
1.4.4 Determining vitality. Gathering cellular reach data
This research includes (1) a rough mapping of the islands’ centralized and ad-hoc power grids and (2) a point-sampling of the reach and strength of cellular coverage on the islands (see Figure 7). These findings are primarily intended to help understand the correlation between access to digital communication and language attrition, and secondary applications of this data include aiding the implementation of language revitalization and maintenance efforts, humanitarian aid, disaster relief, and community development efforts.
Figure 7. Map of cellular and data availability on Sanana
Below is a section from a GIS map of cellular and data availability on Sanana island. Cellular readings were made for both of the region’s service providers’ strength of voice and data signal. These are compared to assessments of linguistic vitality in each region.
1.4.4.1 Determining vitality. Gathering cellular reach data. Method for determining cellular reach
Cellular signal strength was sampled and recorded at all session sites and at various other spots around the islands that I crossed during fieldwork. To gather the data, I used a cellular iPad and an Android phone. Readings were made for both Telkomsel, and Indosat providers (the two carrier networks servicing the area). Readings measured voice and data strength separately, and they were accompanied by textual metadata about reading locations (e.g. reading taken on bridge just north of Pastina)—this was to enable location correction when device’s built in GPS readings were in error. To ensure robust and reliable data (i.e. ensuring readings were not anomalous), consultants at each site were questioned about each area’s access to reliable cellular and digital communication signals and whether the day and time of collection was typical.
These cellular signal readings were used to answer parts of an augmented UNESCO framework for determining language endangerment (Brenzinger et al. 2003). These measurements correlate language use and transmission with access to digital communication in order to help reveal the effects of rapid digital infrastructure development in endangered language communities.
In addition to using this data to help determine language endangerment, it has been digitally mapped using GIS technology to establish a coverage baseline that can be referenced in the future in measuring the correlation between language vitality and the reach of digital infrastructure. Findings at this point demonstrate (1) a gradation of language attrition that is positively correlated with access to reliable digital communication, (2) that the correlation is stronger on the island of Sanana than on Mangon, and (3) that it is likely affected by increased urbanization and an increased proportion of non-Sula populations in Sanana town.
1.4.5 Determining vitality. Evaluative Factors of Language Vitality
Sula language vitality is determined using a heavily modified and augmented version of the UNESCO guidelines for assessing language endangerment (Brenzinger et al. 2003). The augmented version deviates from the UNESCO original in that it looks deeper into the role that access to mass-media and digital communication plays in community language attrition. The framework considers twenty factors correlated with language health among minority language communities to help establish a baseline relationship between language vitality and factors relating to access to digital communication. The factors for the modified framework are explained step-by-step below. Much of the first half is borrowed verbatim from the UNESCO document, and the second half are newly added factors chosen to gauge the reach of mass-media and digital communication. The complete form is found in Appendix D.
Table 1. Augmented Evaluative Factors of Language Vitality
1. Intergenerational Transmission (5 – 0)
2. Absolute number of Speakers
3. Proportion of Speakers within the Total Population (5 – 0)
4. Shifts in Domains of Language Use (5 – 0)
5. Response to New Domains and Media (5 – 0)
6. Availability of Materials for Language Education and Literacy (5 – 0)
7. Governmental and Institutional Language Attitudes and Policies, Including Official Status and Use: (5 – 0)
8. Community Members’ Attitudes towards Their Own Language (5 – 0)
9. Type and Quality of Documentation (5 – 0)
10. Electric. Public utility service area (YES / NO)
11. Electric. Access to electrical power in home (% of community)
12. Electric. Community cooperative generators (or public utility access points) that people can use to charge or power personal electronic devices (e.g. mobile phones, reading lights) (YES / NO)
13. Electric. number of hours of electrical power daily
14. Communication. reach of television broadcast signal (including satellite dish when available) (YES / NO)
15. Communication. Is there a community television (YES / NO)
16. Communication. percentage of community with access to the television (i.e. live close enough to a community television for it to be a practical regular activity)
17. Communication. Scale for cellular voice/SMS signal (0 – 5)
18. Communication. Scale for cellular data signal (0 – 5)
19. Communication. percentage of adults with a smartphone
20. Communication. percentage of adults with a ‘dumbphone’ (SMS)
1.4.5.1 Determining vitality. Determining endangerment level
On-site determinations were established at several fieldwork sites for each factor, and figures were extrapolated based on other known information about communities that were either not visited or that were visited prior to implementation of this survey (these figures are marked with an asterisk). For sites that were not visited, when possible, values were determined either by questioning members of a target community at another location, by questioning civil servants whose work requires them to visit remote communities, or by questioning government medical workers who were stationed in remote communities (figures determined this way are marked with a double asterisk). A lot of general helpful census data was also found in official government census publications (Indonesia 2016, 2018). It should be noted that much of the survey terminology was borrowed directly from the Brenzinger UNESCO scale, and as such could be misinterpreted if results are quoted out of the context of this study. within this study, each determination considers a particular factor of the Sula language confined to a particular site sample location and as such, these determinations are not a statement about the Sula language as a whole. This is especially the case with regard to the word ‘safe’. In Factor 1 (Intergenerational Transmission) for example, a location could be marked as #1 safe: The language is used by all ages, from children up, but that determination is not a statement on the Sula language but rather only on that particular factor within a particular subset of the Sula community. It does not necessarily mean a heck of a lot that intergenerational transmission is strong within an impoverished community of 50 that is surrounded by a more affluent community of five million who have abandoned the tongue altogether.
The purpose of this study is to provide more granular information about factors correlated with language vitality, as vitality itself is not the sort of topic that lends itself to meaningful quantification. At its core, determining language vitality is predicting the future, and as on the futures exchange, different analysts can often come to wildly divergent conclusions based on the same dataset depending on how each interprets and weights the data. Language vitality is a calculation of dissimilar units, and there is thus no straightforward way to perform the calculation that could be generalized to other languages. Take again Factor 1 (Intergenerational Transmission), and imagine a scenario where that factor scores badly but every single other factor in the language scores well, the language would still most likely be quite endangered since young people, for whatever reason, had not learned it. Likewise, even if all other factors were strong—including intergenerational transmission—if the birthrate had plummeted or some external force led to few children in the community, the language would again be quite endangered. At their extremes, many of the factors correlated with vitality can trump all of the other factors and cause language loss. When not at an extreme, each factor exerts itself on a language’s vitality, but this happens in a nonlinear and non-hierarchical relationship between the other factors, tugging back-and-forth at one another and leading each language toward an often difficult to predict outcome.
Considering the relationship among the many correlated factors that were measured as well as the many more that I did not identify or attempt to measure (e.g. the rise of a nationalistic strong-man dictator, a catastrophic crop failure, etc.), it is not likely that I can make a definitive prediction regarding Sula’s future viability, but with that said, I do still attempt a determination about Sula language vitality. However it comes with the caveat that my determination is of slightly less academic utility than my explanation of how I reached the determination, and it is of far less academic utility than the data that the study discovered and compiled about each of the correlated factors that the determination is based upon; this data is information that can be meaningfully quantified, and it is thus more likely to benefit other research.
Brenzinger et al. are strong in pointing out that their original scale not be used for a straight quantitative, numerical assessment of vitality, and they rightly note, “Languages cannot be assessed simply by adding the numbers; we therefore suggest such simple addition not be done.” This is absolutely correct, however thoroughly unsatisfying, as it leaves us in the same place that we were prior to the exercise (albeit with an organized set of relevant data). It is necessary then that researchers can look at the relevant data, make a vitality assessment, and explain the grounds for their assessment in a straightforward manner that enables other researchers to improve upon it. Here now, that is what I will attempt.
In reaching this estimate, it is important to tally village results and factor them for the speech community as a whole. Determining the values for each community or region individually and then averaging them together will generate incorrectly-weighted data and a false result. This would happen, because Sula is divided into numerous, sparsely populated communities that the regional government has grouped into 20 regions in the Sula Archipelago with wildly varying populations (16 of which contain significant populations of ethnically Sula). Of the 16, a few have much larger populations than the others, but only one of them, Sanana, could potentially pass the threshold of critically endangered according to the speaker base factor. Even if this were the case and Sanana generated a weight of (3) vital (which it does not), it would be up against 15 communities weighted as (1) critically endangered. Averaged together, this would generate a score of 1.125 (critically endangered) even though the sum of Sula speakers could theoretically be in the neighborhood of 75,000 individuals—well beyond the threshold of vitality (which again is not the actual case).
Likewise, determining the Proportion of Speakers within the Total Population (factor 3) on a village-by-village basis could result in most of the sub communities generating a weight of (3) vital (even for a language that is clearly critically endangered), if they do not have many non-speakers in each community, yet on the granularity level of the entire community there could be large concentrations of non-speakers that are spread throughout in important locations that serve as hotbeds of daily interaction. With that in mind, I consider the following categories for vitality among the 20 factors that were surveyed:
1.4.5.1.1 Determining vitality. Determining endangerment level. Category 1: speaker base
(assign 1, 2, or 3 by averaging A, B, C below):
(1) critically endangered: if sum of A, B, C below is 1–3
(2) endangered: if sum of A, B, C below is 2–6
(3) vital: if sum of A, B, C below is 7–9
A. absolute Speaker base (assign 1, 2, or 3)
Communities need to have a large enough population base to remain viable against random external threats like plagues and warfare, so I consulted an epidemiologist colleague at the Arizona Bureau of Epidemiology and Disease Control who considered the topic from an epidemiological point of view and we settled on the following guidelines:
Factor 2. Absolute number of Speakers:
(1) critically endangered: a population size less than 3,400 (2,500 adults) is possibly unable to bounce back from a single pandemic or population culling episode
(2) endangered: a population size between 13,400–27,000 (10,000–20,000 adults) is possibly unable to bounce back from two to three pandemics or population culling episodes over three generations (e.g. cholera, warfare, and an earthquake within three generations could reduce the population and social structures so dramatically that the survivors must disband and assimilate into other communities)
(3) vital: a population size over around 27,000 (20,000 adults) is likely to bounce back from three to four pandemics or population culling episodes over three generations
The absolute number of speakers was determined by starting with the estimated size of the ethnic Sula population in each region:
Table 2. Estimated population that is ethnically Sula in each region
Community Population
Mangoli Tengah 6,112
Mangoli Timur 4,870
Mangoli Utara Timur 4,311
Mangoli Barat 401
Mangoli Utara 11,23
Mangoli Selatan 5,216
Taliabu-Barat 3,000
Taliabu-Selatan 3,000
Tabona 1,500
Taliabu-Timur-Selatan 2,000
Sulabesi Barat 5,364
Sulabesi Selatan 4,782
Sanana 23,060
Sulabesi Tengah 6,317
Sulabesi Timur 3,522
Sanana Utara 5,533
Census data from the Badan Pusat Statistik Kabupaten Kepulauan Sula provided a population breakdown of Sula Regency by age for Mangon & Sanana islands (Indonesia 2016, 2018). It is assumed that the breakdown of the ethnically Sula regions of Taliabu would more closely match this data than the data for Taliabu at large, because much of Taliabu’s population have very different lifestyles from the Sula populations. From this data it was determined what percentage of the total population corresponded to each age range.
For Sula, the intergenerational transmission categories: (0) extinct: None speak the language, and (1) critically endangered: Very few speak the language could be excluded offhand. Similarly, it was not difficult to assign an estimated average speaker percentage for the categories of (5) safe: All speak the language (excepting extraordinary exceptions) and (4) unsafe: Nearly all speak the language (albeit with a margin of error). Field surveys and discussion with the regional development office led to the determinations of 95% and 85% respectively. Categories (2) severely endangered: A minority speak the language and (3) definitively endangered: A majority speak the language were a bit trickier to quantify, but percentage estimates were possible based on the survey results from sample sites around the archipelago and observation of speech trends.
For areas that earned an intergenerational transmission score of two, the region’s population is reduced to 23%. This estimation is based on Sula’s age demographics and the following age to speaker ratios:
75% of people aged 50 and older are speakers
25% of people aged 30–49 are speakers
15% of people aged 15–29 are speakers
10% of people under 15 years old are speakers
The following calculation resulted in the 23% figure:
((11,969 • .75) + (25,317 • .25) + (25,557 • .15) + (36,353 • .1)) ÷ 99,196 = 23%
11,969 is the number of people over fifty
25,317 is the number of people thirty to forty-nine
25,557 is the number of people fifteen to twenty-nine
36,353 is the number of people under fifteen
99,196 is the total population
For areas that earned an intergenerational transmission score of three, the region’s population is reduced to 44.6%. This estimation is again based on age demographics, however it is only broken down into three age ranges. This difference is due to the fact that areas marked as three are less urban, and a higher percentage of middle aged people are speakers and the difference in ability to speak among different age ranges of young adults is not as pronounced. Also, many more children are able to speak in these areas, so a granular divide between children and young adults was not necessary. The following age to speaker ratios represent communities marked with an intergenerational transmission rate of three:
Age Range% who are speakers
>4595
20–4440
<2030
The following calculation resulted in the 44.6% figure:
((16,736 • .95) + (36,304 • .4) + (46,156 • .3)) ÷ 99,196 = 44.6%
16,736 is the number of people over forty-five
36,304 is the number of people twenty to forty-four
46,156 is the number of people under 20
99,196 is the total population
Table 3. Speaker numbers by region
Sula Transmission estimate of Sula speaker
Region Score community population number
Mangoli Tengah 4 (85%) 6,112 5,195
Mangoli Timur 4 (85%) 4,870 4,140
Mangoli Utara Timur 5 (95%) 4,311 4,095
Mangoli Barat 2 (23%) 401 92
Mangoli Utara 3 (44.6%) 1,123 501
Mangoli Selatan 4 (85%) 5,216 4,434
Taliabu-Barat 3 (44.6%) 3,000 1,338
Taliabu-Selatan 3 (44.6%) 3,000 1,338
Tabona 4 (85%) 1,500 1,275
Taliabu-Timur-Selatan 4 (85%) 2,000 1,700
Sulabesi Barat 3 (44.6%) 5,364 2,392
Sulabesi Selatan 5 (95%) 4,782 4,543
Sanana 2 (23%) 23,060 5,304
Sulabesi Tengah 3 (44.6%) 6,317 2,817
Sulabesi Timur 4 (85%) 3,522 2,994
Sanana Utara 4 (85%) 5,533 4,703
Absolute number of Sula speakers in archipelago:46,861
B. relative speaker base (assign 1, 2, or 3)
It is safest for a language to be the majority language or the only language in its own territory.
Factor 3. Proportion of Speakers within the Total Population
(1) critically endangered: (<2)
(2) endangered: (2-4)
(3) vital: (>4)
The relative speaker base factor considers the number of Sula speakers against the total population in the region (including the non-Sula ethnic groups). To discern this value, locations were surveyed throughout the region, and values from (0) extinct to (5) safe were assigned to each and then generalized to the rest of the region in cases where it was not possible to survey each village within a region. The numbers assigned to each region were then weighed against the region’s percentage of the ethnically Sula population and added together. This resulted in a value of 2.86 on the 0–5 scale, placing the value for the language at large at (2) endangered. This calculation also provided a way to verify the accuracy of the estimates, because the absolute number of residents in the regions under consideration is known from census data (106,385). That means that comparing the estimated absolute number of Sula speakers, 46,861, to the total population, 106,385, should also land on (2) endangered which, luckily, it does. Sula speakers comprise 44% of the total population, making the speech community a minority of the total population, and the UNESCO guidelines indicate that earns a score of (2) severely endangered for the factor: A minority speak the language.
C. transmission:
Vitality requires children learning the language and surviving to adulthood.
Factor 1. Intergenerational Transmission
(1) critically endangered: (<2)
(2) endangered: (2-4)
(3) vital: (>4)
The method of determining total intergenerational transmission has more or less already been explained: values were determined for each region, weighed against each region’s percentage of the total Sula population, and then added together, resulting in an estimated value for the total Sula ethnic population. The calculated value for Factor one is 2 endangered.
Overall category 1 (speaker base) score: 2.3
1.4.5.1.2 Determining vitality. Determining endangerment level. Category 2: language pervasiveness trends
(assign 1, 2, or 3 by averaging A, B below):
(1) critically endangered:
(2) endangered:
(3) vital:
A. Domains (assign 1, 2, or 3 by averaging factors 4, 5, 6, 7, 9 below):
It is best that a language is expected in enough domains that it will be used frequently and individual speakers will remain primed to that code and likely to select it during their next communicative interaction. Furthermore, it is vital that speakers use the language enough that the next generation is sufficiently exposed to learn it completely. Average the following five factors to assign 1, 2, or 3 for the Domains subcategory.
Factor 4. Shifts in Domains of Language Use (based on a 0–5 scale)
(1) critically endangered: (<2)
(2) endangered: (2-4)
(3) vital: (>4)
Domain shifts were determined for each region, weighted, and combined as with previous factors. The resulting value is 2.104, earning a score of (2) endangered for Factor 4.
Values for factors five, six, seven and nine did not vary among regions, so no calculations were necessary.
Factor 5. Response to New Domains and Media
(1) critically endangered: (<2)
(2) endangered: (2-4)
(3) vital: (>4)
Factor 5 score: 1 critically endangered
Factor 6. Availability of Materials for Language Education and Literacy
(1) critically endangered: (<2)
(2) endangered: (2-4)
(3) vital: (>4)
Factor 6 score: 1 critically endangered
Factor 7. Governmental and Institutional Language Attitudes and Policies, Including Official Status and Use
(1) critically endangered: (<2)
(2) endangered: (2-4)
(3) vital: (>4)
Factor 7 score: 2 endangered
Factor 9. Type and Quality of Documentation; and inasmuch as they limit spoken domains
(1) critically endangered: (<2)
(2) endangered: (2-4)
(3) vital: (>4)
Factor 9 score: 1 critically endangered
Subcategory A. Domains score: 1.4
B. Attitudes (1, 2, 3):
It is vital that the next generation actually uses the language, and poor language attitudes drive trends away from that.
Factor 8. Community Members’ Attitudes towards Their Own Language
(1) critically endangered: (<2)
(2) endangered: (2-4)
(3) vital: (>4)
Values were minimally different among regions for Factor 8 and the score was determined in the same manner as previous values.
Subcategory B score: 1 critically endangered
Overall category 2 (language pervasiveness trends) score: 1.2
1.4.5.1.3 Determining vitality. Determining endangerment level. Category 3: globalization
(assign 1, 2, or 3 by averaging A, B below):
A. Compatibility:
IT brings a flood of content capable of drowning out all but a few well established tongues. It is vital that a language be compatible with new media communication—e.g. if a language needs a special orthography that is not easy to use and universally supported, it will have no chance of establishing a foothold within global IT media. New media compatibility is judged as follows:
(1) critically endangered: The language is generally unwritten and not easily compatible with an established, dominant language orthography
(2) endangered: The language is generally unwritten but easily represented by an orthography that the community is already familiar with
(3) vital: The language has its own, well-attested orthography that is already vigorously used across new media
Overall category A score: 2
B. Access to interactive and non-interactive global media:
The flood of dominant global language content will not affect communities until it reaches them. The degree to which that content has made its way into communities is determined from factors: 13. Number of hours of electrical power daily; 14. Reach of TV broadcast signal; 15. Community accessible TV; 16. Percentage of community with access to the TV; and 17. Availability of a cellular voice/SMS signal. There are six additional correlated factors that I encountered during fieldwork whose connections to vitality are perhaps somewhat less readily apparent; they are broken down below:
Public utility service area (10):
Public power is cheaper and more reliable than community generated power and this makes people more functionally able to consume and participate in new media
Access to electrical power in home (11):
Whether or not global media extends into the home affects how deeply it can pervade private communicative domains.
Community public power points (12):
Power points that community members can use to charge or power personal electronic devices (e.g. mobile phones, reading lights). These may be powered either by cooperative generators or the public utility. Where electricity does not enter the home, having power points in public spaces enables people to charge mobile devices.
Scale for cellular data signal (18):
Whereas a basic cellular signal enables community members to make phone calls and send simple text messages, it does not open the door to global media content. For that, a data signal is required.
Percentage of adults with a smartphone vs ‘dumbphone’ (19 & 20):
As with the difference between access to cellular vs data signal, smartphones help open the gateways to global media in ways that dumbphones do not.
The sub category of factors associated with access to interactive and non-interactive global media is coequal to the category of compatibility, but it is pluri-factorial, containing sub-determinative factors. Moreover, in the case of minority versus dominant languages, the weights of these factors are reversed (i.e. easy access to a flood of new media bolsters English—because that media itself largely represents English. On the other hand, a flood of global media is more likely to suppress minority languages by establishing an important new communicative domain in which the minority languages play no part). The access to interactive and non-interactive global media category is calculated as follows (for minority language communities):
B1. Access to electricity for media devices
(scale 0–11 points possible):
(1) critically endangered: 9–11
(2) endangered: 5–8
(3) vital: 1–4
11a. Percentage of community with access to electrical power in home
(If not in a public utility service area (10). For communities in a public utility service area, skip to 11b.)
+3 >70%
+2 30–70%
+1 10–29%
+0 <10%
11b. Percentage of community with access to electrical power in home
(If in a public utility service area (10))
+5 >70%
+4 30–70%
+3 10–29%
+2 <10%
Factor 11 weighting has a six point spread (0–5). We multiply each regional population estimate and the weight assigned to it and add the results. This will generate an estimate that looks like the number of people who have electricity in home, but it is not. This figure is a normalization of the data measuring the degree of influence in-home electricity has on the community; it must then be converted to the 0-5 scale. In our case, the corrected population estimate is 80,111 (i.e. governmental district population minus population of non-Sula regions), so the values for the scale are this:
+0 (0–13,351)
+1 (13,352–26,703)
+2 (26,704–40,055)
+3 (40,056–53,406)
+4 (53,407–66,758)
+5 (66,759–80,111)
This means that if the normalized data result were 3,024, the scale value assigned would be zero. If it were 18,000, the value would be one. If it were 35,600, the value would be two, and so on. In reality, the normalized data result is 36,167.5, so the factor 11 scale value assigned is two. Because regions surveyed vary in population sizes, it is necessary to weight the data before combining it.
12. Community electricity that people can use to charge or power personal devices
+3 YES
+0 NO
Because YES is assigned a value of three, and NO is given zero, the spread is four (0–3), and the tally of the values of each region will fall on the following scale based on the corrected population estimate of 80,111.
+0 (0–20,027)
+1 (20,028–40,055)
+2 (40,056–60,082)
+3 (60,083–80,111)
The tally of Sula-populated regions marked as YES is 71,018, and that lands the Sula community as a whole at three on the scale.
13. average number of hours of electrical power daily
+0 N/A
+1 evening only
+2 evening–night
+3 24 hr
The values for each region’s average daily access to electricity also fall on a four point scale (0–3), and the sum of the weighted values will follow the same scale as above. The tally of the weighted values is 53,689. This results in the Sula language community receiving a normalized value of two for factor 13.
B1 score
The total score for B1 is determined by adding the values determined for each factor and comparing that number to the corresponding value on the following scale:
(1) critically endangered: 9–11
(2) endangered: 5–8
(3) vital: 1–4
The sum of values determined for each factor is seven (2 + 3 + 2) and that corresponds to a B1 total score of two.
B1 score: 2
B2. Access to passive media (television) (scale 0–5 points possible):
(1) critically endangered: >4
(2) endangered: 2–4
(3) vital: <2
14. TV broadcast reception (including satellite dish if available)
+1 YES
+0 NO
Because YES is assigned a value of one, and NO is given zero, the spread is two (0–1), and the tally of the values of each region will fall on the following scale based on the corrected population estimate of 80,111.
+0 (0–40055)
+1 (40056–80,111)
The tally of Sula-populated regions marked as YES is 64,302, and that lands the Sula community as a whole at one on the scale.
15. Is there a community TV?
+1 YES
+0 NO
Because YES is assigned a value of one, and NO is given zero, the spread is two (0–1), and the tally of the values of each region will fall on the following scale based on the corrected population estimate of 80,111.
+0 (0–40055)
+1 (40056–80,111)
The tally of Sula-populated regions marked as YES is 69,518, and that lands the Sula community as a whole at one on the scale.
16. percentage of community with access to the TV
+3 >70%
+2 30–70%
+1 10–29%
+0 <10%
The scale ranges 0–3, so the weighted tally will fall on the following scale:
+0 (0–20,027)
+1 (20,028–40,055)
+2 (40,056–60,082)
+3(60,083–80,111)
The weighted tally representing the percentage of the Sula community with access to a community TV is 63,627, and that lands the Sula community as a whole at three on the scale.
B2 score
The total score for B2 is determined by adding the values determined for each factor and comparing that number to the corresponding value on the following scale:
(1) critically endangered: >4
(2) endangered: 2–4
(3) vital: <2
The sum of values determined for each factor is five (1 + 1 + 3) and that corresponds to a B2 total score of one.
B2 score: 1
B3. Access to active media (Internet) (scale 1–20 points possible):
This subcategory is simple to calculate, because all of the units are alike. Values can be tallied for each region and then that total for each region can be weighted against its population before adding those values together to reveal the final B3 value for the Sula community at large. This value is compared to the following 1–20 scale to find the total score for B3:
(1) critically endangered: (13.34–20)
(2) endangered: (6.68–13.33)
(3) vital: (0–6.67)
17. Scale for cellular voice/SMS signal
(0) no access
(1) signal is available in certain locations of the village sometimes
(2) signal is always available in certain locations of the village
(3) signal is available in most locations of the village intermittently (asynchronous comm: send now, delivered later)
(4) signal is available in most locations of the village most of the time (semi-synchronous comm)
(5) signal is available in all locations of the village nearly all of the time (synchronous comm)
Table 4. Scale for cellular voice/SMS signal results
Region: Totals
Mangoli Tengah 1
Mangoli Timur 3
Mangoli Utara Timur 3
Mangoli Barat 1
Mangoli Utara 1
Mangoli Selatan 1
Taliabu-Barat 3
Taliabu-Selatan 3
Tabona 1
Taliabu-Timur-Selatan 2
Sulabesi Barat 5
Sulabesi Selatan 0
Sanana 5
Sulabesi Tengah 2
Sulabesi Timur 0
Sanana Utara 5
18. Cellular data signal
(0) no access
(1) signal is available in certain locations of the village sometimes
(2) signal is always available in certain locations of the village
(3) signal is available in most locations of the village intermittently (asynchronous comm: send now, delivered later)
(4) signal is available in most locations of the village most of the time (semi-synchronous comm)
(5) signal is available in all locations of the village nearly all of the time (synchronous comm)
Table 5. Cellular data signal results
Region: Totals
Mangoli Tengah 1
Mangoli Timur 1
Mangoli Utara Timur 0
Mangoli Barat 0
Mangoli Utara 0
Mangoli Selatan 0
Taliabu-Barat 3
Taliabu-Selatan 0
Tabona 0
Taliabu-Timur-Selatan 0
Sulabesi Barat 4
Sulabesi Selatan 0
Sanana 4
Sulabesi Tengah 1
Sulabesi Timur 0
Sanana Utara 3
19. Percentage of adults with a smartphone
+7 >70%
+5 30–70%
+3 10–29%
+1 <10%
Table 6. Percentage of adults with a smartphone results
Region: Totals
Mangoli Tengah 5
Mangoli Timur 3
Mangoli Utara Timur 1
Mangoli Barat 3
Mangoli Utara 1
Mangoli Selatan 3
Taliabu-Barat 5
Taliabu-Selatan 1
Tabona 1
Taliabu-Timur-Selatan 3
Sulabesi Barat 5
Sulabesi Selatan 1
Sanana 5
Sulabesi Tengah 3
Sulabesi Timur 1
Sanana Utara 3
20. Percentage of adults with a ‘dumbphone’ (SMS)
+3 >70%
+2 30–70%
+1 10–29%
+0 <10%
Table 7. Percentage of adults with a ‘dumbphone’ results
Region: Totals
Mangoli Tengah 3
Mangoli Timur 3
Mangoli Utara Timur 1
Mangoli Barat 3
Mangoli Utara 2
Mangoli Selatan 3
Taliabu-Barat 3
Taliabu-Selatan 2
Tabona 2
Taliabu-Timur-Selatan 3
Sulabesi Barat 3
Sulabesi Selatan 1
Sanana 3
Sulabesi Tengah 3
Sulabesi Timur 1
Sanana Utara 3
Adding B3 factors (17–20) for each community results in the following regional tallies:
Table 8. Factors 17–20 tally
Region: Totals
Mangoli Tengah 10
Mangoli Timur 10
Mangoli Utara Timur 5
Mangoli Barat 7
Mangoli Utara 4
Mangoli Selatan 7
Taliabu-Barat 14
Taliabu-Selatan 6
Tabona 4
Taliabu-Timur-Selatan 8
Sulabesi Barat 17
Sulabesi Selatan 2
Sanana 17
Sulabesi Tengah 9
Sulabesi Timur 2
Sanana Utara 14
These values were reduced according to each community’s relative percentage of the entire Sula population as follows:
Table 9. Values relative to percentage of entire Sula population
Region: Totals population % weighted totals
Mangoli Tengah 10 x 0.0763 = 0.7629
Mangoli Timur 10 x 0.0608 = 0.6079
Mangoli Utara Timur 5 x 0.0538 = 0.2691
Mangoli Barat 7 x 0.0050 = 0.0350
Mangoli Utara 4 x 0.0140 = 0.0561
Mangoli Selatan 7 x 0.0651 = 0.4558
Taliabu-Barat 14 x 0.0374 = 0.5243
Taliabu-Selatan 6 x 0.0374 = 0.2247
Tabona 4 x 0.0187 = 0.0749
Taliabu-Timur-Selatan 8 x 0.0250 = 0.1997
Sulabesi Barat 17 x 0.0670 = 1.1383
Sulabesi Selatan 2 x 0.0597 = 0.1194
Sanana 17 x 0.2879 = 4.8935
Sulabesi Tengah 9 x 0.0789 = 0.7097
Sulabesi Timur 2 x 0.0440 = 0.0879
Sanana Utara 14 x 0.0691 = 0.9669
The sum of the weighted totals is 11.126 for Sula as a whole. This results in a B3 value of two when compared to the 1-20 scale:
(1) critically endangered: (13.34–20)
(2) endangered: (6.68–13.33)
(3) vital: (0–6.67)
As with previous determinations, totals for each region were multiplied by the region’s percentage of total Sula population and combined to reach a B3 (Access to active media) score of two for the greater Sula community.
This completes the calculations for section B, resulting in an overall subcategory score for B (global media) of: 1.33 (average of B1, B2, B3). It also completes calculations for the survey of factor correlated with language endangerment, yielding:
Overall Category 3 (Globalization) Score: 1.667
(average of A, B) and a determination of Sula’s language endangerment:
1.4.6 Determining vitality. Conclusion
Total Sula Language Endangerment score: 1.73 (average of categories 1, 2, 3)
This value places Sula at the safer end of the critically endangered category, however, that is still not a comforting category to land in. As stated earlier, the modified UNESCO framework has considered twenty factors correlated with language health among minority language communities to help establish a baseline relationship between language vitality and factors relating to access to digital communication. It is my hope that returning to this question at regular intervals over the next ten to fifteen years will reveal determinative relationships between these factors and language health that can be quantified and generalized to provide better forecasting abilities for language vitality in general.
I agree with Brenzinger et al. that the framework should not be used to plug in values and generate a one-size-fits-all determination of viability, and in fact I had a strong suspicion ahead of time what the result would be for Sula. In addition to reasons discussed earlier, the decision to use this modified framework to calculate Sula’s level of endangerment was to more clearly explain my reasoning and retroactively quantify what was at heart a qualitative determination. By modeling my logic this way, I am establishing a level of transparency of thought such that others need not ‘take my word for it’. My reasoning can be evaluated and its flaws and the areas for improvement can easily be identified. Moreover, the transparency enables other researchers to restructure the points according to their evaluative criteria.
1.5 Online presence and the effects of globalization
Sula currently has a limited online presence, but I have been involved in an informal project for the past several years to boost its presence, particularly on social media. The goal of our project is to mitigate the language shift that is likely to accompany the coming digital transition, so that Sula’s communicative domains are not supplanted by Malay and English. Digital content included in the digital footprint includes instructional videos, education videos, song and story performances, and videos of the location and environment. Going forward, minority-language maintenance models will have to incorporate creation of digital footprints within modern communication domains.
There are two Internet users from the developing world for every one in the developed world, yet four billion people in the developing world still do not have Internet access (ICT Data and Statistics Division 2015). Digital infrastructure development is rapidly underway in Eastern Indonesia though. This is a mixed blessing for Sula, whose communities now find themselves on the precipice of a digital revolution, and this revolution will be a significant one: it could be the death knell for a language on the verge of becoming moribund or a great democratizer that bolsters linguistic diversity.
Considering the challenges toward building an online Sula-language footprint, two Sula stakeholders have been working with me to deliberately generate Sula discourse online. The first of the stakeholders is Ida Ryberg Umage, who is ethnically Sula but was raised in Surabaya. She now lives in Sweden working with the children of Middle Eastern and African refuges. Ida’s two children were raised in Sweden; they do not speak Sula and have never been there. But, Ida is passionate about connecting her children to their Sula heritage from abroad, and she has been instrumental in generating online discussions and helping me translate difficult vocabulary and generate online discussion about Sula related media.
Facebook was chosen by default as the communication platform. Sula is an impoverished region, and Facebook partnerships with Indosat and Telkomsel mobile providers have enabled most anyone in Indonesia to access Facebook content even without purchasing a cellular data plan. The service seems to be available pending unallocated network bandwidth, so it is slow and intermittent, but importantly, it is free and it works. Marlia Banapon, an English instructor and proponent of Sula maintenance and revitalization, is a Facei tribe member living in Sanana. Throttled behind Sanana’s low-bandwidth data, Marlia can only generate and upload still image and textual content, however she is able to view and comment on others’ content and generate Sula language discussions (Figure 8).
Figure 8. Sample of Sula language conversation in a digital domain
This conversation centers around the question: who are the most famous people from Sula?
The low number of Sula speakers who have access to broadband Internet connections makes it challenging to generate online Sula-language content. Nine years ago, there were three known Sula speakers with broadband Internet access and a public online presence. Two were university students who have since returned to Maluku. By 2015 there were only Ida in Sweden and her sister, Sanna Tabona in Surabaya—neither of whom are first-language speakers. Other Sula speakers with a known public online presence occasionally popped up online with cellular broadband access when they travelled to major Indonesian cities, and some—particularly diaspora living on Ternate and Ambon—had regular but slow/Intermittent non-broadband Internet access. But within the last few years, the number of Sula with broadband access has increased to where I could not count them. These are individuals living or traveling outside of the archipelago to regions whose wireless networks have been upgraded, and that network upgrade is getting closer to Sula by the day. I will not be surprised if the archipelago achieves reasonable broadband coverage by the time this dissertation goes to print.
So far, the community’s offline status has provided an opportunity to deliberately generate Sula media and discussions in advance of widespread broadband proliferation and the likely de facto establishment of a Malay language norm for digital communication. As of 2015, virtually all Sula residents with Facebook accounts were on each other’s friend lists, and Ida, Marlia, and myself had positioned ourselves centrally within that group to give our content near universal exposure. Today this is no longer the case, and we can only hope that the online Sula language footprint we established prior to broadband Internet arrival will have helped to position the language as modern, relevant, and compatible with new domains of communication so that it can find a footing in the sea of global media. Our project has sets in place a scenario that will, over the course of the coming decades, test the question: can language shift be mitigated by establishing an endangered language footprint that sets a communicative-code inertia prior to the arrival of new information technology?
1.6 Chapter conclusion
The previous pages have established the context of the Sula language, place, and people in terms of history, geography, population numbers and distribution, and the external global pressures facing the community. These are pieces of Sula’s puzzle that are necessary for future research into the language and for establishing strategies to bolster the language against the external pressures that threaten its ability to survive.