24. Speak of the Devils:
Discourse and Belief in Spirits on Rotuma
[Published in Spirits in Culture, History, and Mind, edited by Jeannette Mageo and Alan Howard. New York: Routledge, 1996]
Rotuma is a small volcanic island located 465 km north of Fiji, on the western fringe of Polynesia. Physically, linguistically, and culturally, Rotumans reflect influences from Melanesia, Micronesia, and particularly the Polynesian islands to the east: Tonga, Samoa, Futuna, and Uvea.
Following first European contact in 1791, Rotuma was intensively intruded upon by whalers, renegade sailors, traders and missionaries. The first missionaries arrived in 1839, and reflecting a prior political cleavage, approximately two-thirds of the people converted to Wesleyanism, the remainder to Catholicism. Exacerbated by national rivalries between English ministers and French priests, a series of skirmishes between the sides led the chiefs to petition Great Britain for cession in 1879.
From 1881 to 1970 Rotuma was governed as part of the Colony of Fiji with a resident commissioner (later a district officer) acting as governor, magistrate, and usually, medical officer. The chiefs of Rotuma's seven semi-autonomous districts met in council but were relegated to an advisory role. When Fiji gained independence in 1970, the Rotuman chiefs opted to remain part of the new nation and, not without controversy, affirmed their decision after the military coups of 1987. While the position of district officer has been retained, the powers of governance have been reallocated to the Rotuma Council, composed of district chiefs and elected district representatives.
The population of Rotumans reached a low of around 2,000 following a devastating measles epidemic in 1911. From then on it steadily increased and presently numbers around 11,000. Only about 2,600 individuals now live on the island, however, down from a peak of 3,235 in 1966. The bulk of Rotumans have settled in Fiji's urban centers, although substantial enclaves have formed in Australia and New Zealand as well.
Rotumans place a high value on formal education and have done extremely well in occupational pursuits, being vastly overrepresented in the professions, managerial positions and high government office. English is spoken by all but a few elderly individuals and is the language of instruction from the third grade on. Although copra remains Rotuma's only significant export, remittances and a free flow of goods onto the island have resulted in an uncommonly high standard of living for such an isolated Pacific Island.
Over the past thirty years I have noticed that talk about spirits on Rotuma has diminished significantly. During my initial period of fieldwork on the island, in 1960, people spontaneously brought up the topic on innumerable occasions, whereas on my recent field trips (1987-1991, 1994), mention of spiritual agency was conspicuously absent, even on occasions that would seem to call for it. This raises some interesting questions: What conditions have led to this change in discourse? What does the change imply about Rotuman beliefs? Are Rotuman spirits headed for cultural oblivion? Before confronting these issues directly, I summarize accounts of spirits, mostly provided by European visitors who obtained their information by talking with Rotumans.
Categories and Concepts
The most general Rotuman word for spirits is 'atua, which Churchward defines as "dead person, corpse, ghost." He adds that "The last is its commonest meaning, ghosts being very material beings...to the Rotuman mind" (Churchward 1940:352). But such a simple, concise definition fails to do justice to the complexity of usage in Rotuman discourse. European visitors to Rotuma have, from the beginning, had difficulty coming to grips with such concepts, in part, it seems, because they have been more concerned with logical consistency and systemization of beliefs into religious theologies than with the contexts in which the concepts were used. That early European visitors to the island were disconcerted by the lack of systemization in Rotuman discourse about spirits is evident in their accounts (see, for example, Lesson 1838-9:437; Bennett 1831:478).
The conclusions of early commentators were likely a consequence of responses to a discourse format Rotumans found unfamiliar. Instead of discussing spirits in the abstract, Rotumans talked about spirits in rather specific contexts--when telling stories, expressing apprehension or a sense of foreboding, attempting to explain anomalous occurrences, coping with uncanny feelings and unnatural sensations, etc.
Not surprisingly, early European accounts of Rotuman "religious beliefs," or, as they were frequently designated, "superstitions" (which suggests irrationality as well as inconsistency and incoherence), fall back on anecdotal information. But Rotumans learned, rather early on, I suspect, to provide coherent accounts that were more satisfactory to European interrogators. Thus Churchward obtained a verbatim account (in Rotuman) obtained from Mesulama Titifanua in response to queries concerning the meaning of various terms. With regard to the word 'atua, Mesulama replied:
As soon as a human being dies he becomes an 'atua. It was held by [our] forefathers that it was the spirit ('ata) of the person that was the 'atua, and that he was able to go about. In their time, moreover, they were in the habit of summoning their dead to come to them that they might converse. This they did, at times, [just] because they loved their dead friends so much. They also had great confidence in them when they wanted to know various things, asking their 'atuas to tell them. Especially did they trust in [the 'atuas of] their prematurely born children. They said that the 'atua that had more power to deliver than any other was [that of] a child prematurely born (Titfanua 1995:123-124); translation by Churchward).
Some aspects of the meaning of 'atua proved confusing to Europeans. For example, as Churchward's definition indicates, 'atua refers to a corpse as well as a ghost. This usage suggests a being devoid of "spirit." Hocart wrote that "any Rotuman will tell you that atua is a dead man (famör ala). It is actually used of the dead body, and once children playing with human bones told me that they were "the bones of atua" (sui ne atua)" (Hocart 1915:129). To make matters even more complicated, 'atua, modified by possessive markers used with edible items, is used in reference to a person killed in war or defeated in a wrestling match. 
Rotuman attitudes toward 'atua, and their efforts to control spirits' powers, were an endless source of fascination to European visitors. Here, too, outsiders encountered a range of propositions that failed to meet their criteria for a reasoned (and reasonable) religion. Gardiner reflects typical European perceptions in his discussion of relations between 'atua and human beings:
Long before the advent of the missionaries to Rotuma, the religion of its people seems to have degenerated into the grossest superstition and a mere belief in atua, a generic name for all devils, spirits, and ghosts. It is also used for the soul, as we understand it. These atua were ever ready to punish and prey on any one who did not propitiate them with plentiful gifts of food and kava. Each hoag ["section of a district or village under the authority of a sub-chief"] had its own atua, but several hoag might acknowledge a big atua over all, while they each had their own atua. At the same time, so long as they propitiated their own atua, no great harm could happen to them, unless a greater atua laid a curse on them, causing sickness, etc.; the atua, though, could only affect them personally, and had little or no power over their crops. This atua might be termed "the god of the hoag," but there was also an inferior class of atua, who might be called "devil spirits," whose sole delight it was to go about causing sickness and death ... Their dwelling-places were in trees, stones, and rocks ... but some were said to enter into men ... The still inferior class of atua, but a class with little or no power of itself alone, would best be termed "the ghosts of men." They could be to some extent called up at will by the relations to assist them against their enemies and to cure them of sicknesses of a certain class, supposed to be due to the influence of soul on soul (Gardiner 1896:466).
The compound form sur'atua is used in reference to possession [sur = "to enter"]. According to Churchward, the term signifies a person into whom the spirit of a deceased person has entered. It can also be used adjectively to designate a person who has such visitations periodically, i.e., a spiritual medium, and in noun form to refer to a "seance." When a person was possessed by an 'atua they were said to take on the appearance, mannerism and voice of the deceased person who entered them (Churchward 1940:317). In contrast, the phrase to'ak'atua, meaning "to utter messages alleged to come from the spirit of a deceased person," does not imply possession, only the use of a living individual as a medium by the spirit (Churchward 1940:334).
The phrase re'atua [re = "to do"] is used in reference to actions designated to harness the powers of 'atua through ritual transactions. Various forms of cursing, attempts to gain advantage in conflicts or disputes by appealing to dead ancestors, and invoking the healing powers of 'atua were so labeled.
'Aitu is a second term used in reference to spirits. Churchward defines 'aitu in its noun form as "god, object of worship; shark, stingray, or other creature regarded as the habitat of a god;" as a verb "to regard as divine, to worship" (Churchward 1940:348). The distinction between 'aitu and 'atua was unclear to many European commentators, some of whom treated them as synonymous (Russell 1942:249). This may have been in response to the fact that certain kinds of 'atua are considered to be 'aitu, as in the case of a group of wandering spirits known collectively as sa'aitu (alternatively as la' ti'ta "the big travelling company"). Titifanua described sa'aitu to Churchward as the souls of uncircumcised men who, in times of war, helped one side or the other to victory (Titifanua 1995:123-124). Other informants described companies of sa'aitu as composed of the spirits ('atua) of dead chiefs, or persons who died suddenly (MacGregor n.d., box 1).
A key to understanding the distinction between the two concepts for spirits lies in the fact that 'ait[u] is often used as a verb in such compounds as 'ait'aki ['aki = instrumental suffix] "to deify, to treat as divine," and ro'aitu "to pray" (Churchward 1940:349, 301), while 'atua is never used as a verb. This suggests that while 'atua is used as a generic term for spirits, including free-roaming malicious ones beyond human control, 'aitu is reserved for spirits who have been brought into the human moral order through various forms of binding, a point made by Vilsoni Hereniko (1991, 1995). Whereas Churchward glosses the sa' in sa'aitu as cognate with the Tongan ha'a and Samoan sa, meaning tribe, family or class ([Titifanua 1995]1939:469), Hereniko argues that it derives from sa'a "to weave." He thus glosses the term sa'aitu as "woven gods."
Hereniko derives his inspiration from a Rotuman myth entitled "'Aeatos," in which malicious 'atua are neutralized by being caught in woven nets. He points out that sa'aitu served human interests, albeit at their own discretion. Pushing his case further, he convincingly argues that apei "fine white mats," which are central items of ritual exchange at weddings, funerals and other life-crisis events, are themselves given a god-like status in Rotuman culture. They can be thought of as containing (and constraining) spirits in their own right, a view supported by the fact that apei are consecrated through the ritual sacrifice of a pig.
Hereniko's analysis helps explain why the god of the sau and mua, Tagroa sir'ia, was an 'aitu.  The sau was described as the "king" of Rotuma by European visitors, while the mua was considered to be a "high priest"; their main roles were to perform cyclical rituals designed to insure the prosperity of the island (see Howard 1985). Tagroa sir'ia was prayed to for food, for rain, and for success in island-wide enterprises. He could be called on to avert hurricanes or other calamities. Significantly, Gardiner was told that Tagroa "does not concern himself with the doings of the 'atua" (Gardiner 1898:467).
Whereas the phrase to'ak'atua refers to circumstances in which the spirit of a deceased person speaks through the mouth of a medium, to'ak 'aitu refers to a condition in which an 'aitu speaks, unheard by others, to an entranced recipient who then relates the messages to a waiting audience (Elisapeti Inia, personal communication).
In an ironic twist of fate, the English Methodist missionaries and French Catholic priests who translated the Bible selected alternate terms for God and devil(s). Methodists chose 'aitu for God and relegated the concept of 'atua to "devils," while Catholics chose the reverse. Whether this exacerbated already existing tensions between the two groups is problematic, but it did provide a basis for adherents of the conflicting faiths to heap verbal scorn on one another. It also may have contributed to the confusion of the the two concepts.
Tu'ura, Ape'aitu and Tupu'a
The term tu'ura was used to designate a being that hosted an 'atua. According to Titifanua, tu'ura
really means what they [the ancestors] referred to when they said, "Such and such a woman has become a tu'ura living in an owl [lit. The woman, she has tu'urad to an owl; note that tu'ura, though fundamentally a noun, is used also as a verb]; while such and such a man has become a tu'ura living in a cat." [Thus] we sometimes use the expression "an 'atua cat," the reason being that, when we say this, we think of an 'atua as having entered into the cat....[Our] forefathers said that animals into which 'atuas had entered as tu'uras had a different shape from other animals, and they were able to distinguish an animal into which an 'atua had entered as a tu'ura (Titifanua 1995:125; translation by Churchward).
Several of MacGregor's informants considered human beings who hosted an 'atua (or an 'aitu) to be tu'ura as well. In fact some were unable to distinguish between tu'ura and ape'aitu, a term Churchward defines simply as "priests" (1940:174). MacGregor speculates that ape'aitu may have been prophets, while tu'ura were mediums engaged by families to communicate with their deity. He comments that the terms were being used synonymously at the time of his visit to Rotuma in 1932. 
Regarding the role of tu'ura, MacGregor cites an informant by the name of Varamua:
When the tu'ura has been asked to prophesy or tell of [the reason/outcome of someone's] sickness, he goes to his house...and beats his drum to call the god. Then when he feels "very strong" or possessed with the god he eats uncooked taro and pig, even the head which is tabu to the chief, and takes kava, all of which is presented by the person who wished to consult the god. When he has eaten and had kava which are for the god (not the tu'ura) he becomes the mouthpiece of the god and answers questions as to sickness, prospects of a coming war (MacGregor n.d., box 1).
Like Churchward, Gardiner equated ape'aitu with priests (and priestesses) insofar as they officiated at invocations and acted as mediums for local gods.
Another term associated with spirits is tupu'a, which Churchward translates as "immortal man; rock or stone reputed to be such a person petrified" (Churchward 1940:337). Certain rocks were thought to be tupu'a, and their spirits could be called upon by persons acting as mediums.
Rotuman spirits were thought to occupy a wide variety of niches, according to type. Tagroa sir'ia was located in the sky, ancestral ghosts took up their abode in various off-shore locations under the sea, while other 'atua were said to dwell in trees, rocks, cemeteries, and isolated localities on the island. Some spirits were free-roaming and could appear anywhere in the form of animals or apparitions. 
The most general term for the abode of spirits is 'oroi, which means "to be hidden from view" or "hidden from knowledge, mysterious, unknown, unknowable" (Churchward 1940:360). In contrast, the material world is referred to as ran te'isi "this world."
Every district but one had a named location offshore to which the souls of the dead migrated (see map). The best known of these, or more accurately, the most talked about, was Li'marä'e [li'u = "deep sea" + marä'e = "open space within a village where gatherings are held"], off the west end of the island. The route to the 'oroi regions went westward, through the village of Losa (Russell 1942:249). Gardiner states that Li'marä'e was "full of cocoanuts, pigs, and all that man could wish for...[and that] Any things buried with the body would be taken by its ghost" (Gardiner 1898:469).
Bush areas were considered likely habitats for spirits, as were wells and certain kinds of trees (see Eagleston 1832:401-402 for an interesting example). One particular spirit, a legendary figure by the name of hanit e ma'us "wild woman of the bush" sometimes appeared in the form of a succubus, a beautiful temptress who lured men into sexual liaisons in order to capture their souls. In general, it seems, the further removed one was from human habitats--from places under cultural control--the more likely one was to encounter spirits.
During my 1960 field trip a number of people reported having seen 'atua. They told me that the road to Losa, which passes through the bush, is especially plagued by spirits, and that the night of a quarter moon is worst. A large hefau tree on the way was supposedly inhabited by a whole family of 'atua. Individuals also reported having seen the figure of a woman standing facing a mango tree along the same road. She was described as nude, with long black hair down to her waist, and white skin. Those I interviewed claimed to be very much afraid of her. Another story concerned a large bird that flies through the air which, upon landing, assumes a human form. Some people claimed that the figure resembled someone they knew who had died (1960 field notes).
The Actions of the Spirits
Numinals were conceived to perform a wide range of activities, from malicious mischief to overseeing the prosperity of the island in Tagroa Sir'ia's case. Left to their own devices, most 'atua were considered to be wanton destroyers of human beings. They engaged in a constant effort to steal the souls of humans and to feast on their bodies. These were the beings to which the English term "devils" was readily attached.
'Atua were said to lure human beings by presenting themselves as attractive paramours, particularly in dreams. Women were especially vulnerable to malicious spirits who sought to enter their vaginas when they were urinating, sometimes causing miscarriages. When outside, women were instructed never to urinate in an open space; instead they should relieve themselves near a rock or tree. If a woman is impregnated by an 'atua, she will give birth to something resembling fish instead of normal children and will likely die soon afterwards. Women who suspect they have been impregnated by an 'atua can go to a native healer in order to drive away malignant spirits. Pregnant women must be especially wary of female 'atua who try to capture the souls of unborn children.
Not all free-roaming spirits were evil, according to a recent account by Ieli Irava, a respected Rotuman educator. He contrasts 'atua in the form of dogs, cats and owls, all of which were associated with malignant spirits, with stingrays, sharks and turtles, which were benevolent toward humans. He states that all good spirits were sea creatures, and speculates that being seafarers, Rotumans "put great emphasis on the good sea spirits in the hope of receiving good weather during their sea journeying or fishing expeditions" (Irava 1977:10).
The souls of recently deceased individuals were said to make their presence known through the cries of birds, an owl's flying by, or other unusual events. In other instances they appeared in dreams. This indicated the spirit was restless, and it was common for relatives to go to the cemetery to implore it to rest (Howard 1960; see also Russell 1942:251). The spirits of the newly dead were considered to linger in the vicinity for five days, after which a ceremony was held ending the death taboos.
Ancestral spirits presumably remain sensitive to the actions of their descendants, and use their powers to punish bad behavior, especially disobedience to chiefs (Russell 1942:251). They also rectify wrongs and invoke justice. Just about every Rotuman can tell a story about someone who had committed an egregious act and received his just desserts soon thereafter. Land disputes between relatives are prototypical, the underlying assumption being that spirits who are common ancestors of the disputants will punish the party in the wrong, or perhaps both parties if they share the blame. Justice is distributed in the form of luck, with those in the right prospering, those in the wrong suffering ill-fortune. The consequences of wrongdoing may follow directly from the transgressions, or they may be called for by an aggrieved party in such forms as "the land has eyes," or "we shall see who is right."
The most feared curses are from the lips of chiefs, since they have channels to more powerful spirits. Chiefs usually call for immanent justice (pu'aki) when serious offenses have been committed within their domains and no one accepts responsibility. A number of cases have become classics and are told with relish, though often without explicit reference to intervention by spirits (see Howard 1990:270 for an example).
Interactions with Spirits
The most common way of dealing with unbound spirits is to avoid them. In 1960 Rotumans were reluctant to go out at night. If out after dark, they walked quickly past cemeteries, and stayed away from localities spirits were said to inhabit. Despite the tropical heat, windows were often closed at night to keep out marauding 'atua.
If avoiding malevolent spirits was impossible, making loud noises was a way of frightening them off. At other times spirits have to be mollified by ritual. For example, any incident in which blood is shed should be followed by a ceremony called hapagsu. The ceremony involves consumption of ritual foods, including a sacrificial pig prepared in in an earthen oven. The goal is to placate the spirit or spirits who caused the event, so as to avoid a recurrence. Hapagsu are performed following surgical operations, as well as after accidents; in addition, they are held for prisoners returning from jail (Howard 1960).
Early commentators noted that each locality in Rotuma had spirits who were propitiated and were supposed to look after the interests of the local group. Such spirits generally took the form of animals and were treated with totemic respect. According to Gardiner:
The "hoag gods" were usually incarnated in the form of some animal, as the tanifa (the hammer-headed shark), juli (sandpiper), olusi (lizard), mafrop (gecko), etc. Should a man by any chance have happened to kill one of the particular animal which was his atua, he would have had to make a big feast, cut all his hair off and bury it, just in the same way as a man would be buried. (Gardiner 1898:467-468)
Gardiner states that in warfare each ho'aga would propitiate its own 'atua, rather than invoking Tagroa sir'ia, since "such small matters did not concern him and, as he was the god of both sides, it was quite unnecessary" (Gardiner 1898:471).
The most critical propitiation of spirits were rites performed in association with the sau and mua, for it was upon them that the welfare of the entire island depended. Such rites included a good deal of ceremonial feasting, kava drinking and dancing. Dancing was an especially important means of communicating with the spirits, and of exercising a degree of control over their activity (see, for example, the myth of Kirkirsasa in Titifanua 1995:86-91, and Hereniko's [1991,1995] interpretation of the myth ).
Changing Contexts of Discourse about Spirits
Missionization, followed by the establishment of a British colonial administration in 1881, resulted in the elimination of nearly all public contexts in which traditional spirits were propitiated. By 1874 the institution of the sau and mua had disappeared, and along with it ritual observances to Tagroa sir'ia and the spirits of dead chiefs. The missionaries, depending on their denomination, labeled 'atua or 'aitu "devils," and used the other term as synonymous with the Christian God. This did not, however, result in the suppression of talk about spirits. Even devout Christians continued to speak about 'atua as if their presence on the island were unquestioned, and Rotumans continued to perform rituals, such as kava ceremonies, pig sacrifices, and healing routines that presupposed the potency of spirits, in some instances invoking them directly through chants and prayers. Rotuman legends continued to be told, and were a source of knowledge about the antics of spirits. Furthermore, adults used children's fear of the spirits as a control mechanism (Irava 1977:10).
In 1960, talk of spirits was still relatively common. The topic came up regularly as people spoke about their experiences in remote parts of the island, about their dreams, and about the cultural past. I was told numerous personal accounts involving possession, and legendary stories liberally sprinkled with reference to the antics of 'atua. I was warned of the danger of tempting 'atua by certain actions, such as wearing red at sunset, particularly in the village of Losa, and was instructed to stay away from cemeteries at night. Spirits had, in other words, a strong social presence, and were talked about openly and frequency.
I returned to Rotuma in 1987 for a brief visit, then for longer stays over the next three years,  and have been struck by the fact that talk about spirits is much rarer now. Only occasionally were the antics of 'atua used to explain events, and I learned of only one recent instance of possession. When I suggested the possibility that some anomalous creature might be an 'atua people laughed, not in the nervous manner that signifies a defense against possible ridicule, but in the hearty manner of a good joke. People no longer close their windows at night, youngsters are regularly seen congregating in cemeteries, leisurely sitting on gravestones, and my wife and I were given absolutely no warnings or proscriptions for avoiding harmful 'atua. They had, in effect, lost much of their social presence.
Socioeconomic Changes 1960-1990
In many respects Rotuma has changed considerably in the past thirty years. What struck me most on revisiting the island were the physical changes, like the airport, which opened in 1981, and a wharf built in the 1970s. Another significant change was in housing. Hurricane Bebe destroyed almost all native style, thatched, houses in 1972. For the most part they have been replaced by concrete houses with corrugated iron roofs (Rensel 1991).
Far more of the island is electrified now. On my first visit only the government station and the Catholic church compounds at Sumi and Upu had generators. Now most villages provide electricity for a few hours per day, and many individual households have their own generators. Perhaps the biggest change in daily life, however, has been brought about by the presence of motor vehicles. In 1990 the district officer estimated around 300 motorbikes on the island, and the number has increased since. In addition, some thirty to forty privately owned cars and trucks are in use, along with vehicles regularly operated by the government and local business organizations. This means that everyone has ready access to some form of motor transport.
Another major improvement has been the construction of bush roads out of what were previously footpaths. This permits people easier access to their remote gardens by motor vehicles. In my recent visits it was rare to see men carrying baskets on a shoulder pole, or on horseback.
Educational levels have increased. In 1960 only a small number of adults on Rotuma were educated beyond standard eight (U.S. eighth grade). Today, most younger adults have at least completed Form III (U.S. ninth grade). This educational advancement, along with increased opportunities for travel, lends a greater air of worldly sophistication to Rotuman social life. People read more, are better informed, and less prone to accept authority in an unquestioning fashion.
Why the Silence, and What Does it Mean?
Taken at face value, my experience on Rotuma suggests a dramatic change in spirit discourse. Talk, or at least public talk, about numenals has substantially diminished, raising several significant questions. Has the Rotuman world view demystified in the interim? Has some social threshold been passed, making talk about spirits unacceptable? To what extent do changes in talk reflect changes in belief?
Let us begin by considering the possibility that Christianity has contributed to a demystification of the Rotuman world view in recent years. Christianity has been a persistent foe of Rotuman spirits since the early nineteenth century. The missionaries actively undermined, and eventually brought about the demise of the religion of the sau and mua, which incorporated a wide range of spiritual beings. They sought to focus discourse on God and Biblical heroes, leaving ghosts and local spirits on the social periphery. But the missionaries did less to deny the existence of Rotuman spirits than to relabel them as "devils." These changes were well-entrenched by the early twentieth century, and certainly by 1960. They cannot, therefore, account for alterations that occurred between 1960 and 1990. In both periods Rotumans spent a great deal of time engaged in church-related activities: saying prayers, singing hymns, and discussing the Bible. In 1960, however, Rotuman spirits retained a place in the islanders' discourse they have since lost.
What then of the physical, social and economic changes described above? Have they served to demystify the Rotuman world view? I suspect they have but can only guess at the reasons. Increased levels of education, and exposure to "scientific" discourse, may well have resulted in a progressive substitution of naturalistic for supernatural explanation. Furthermore, as Rotumans have come into more intense contact with others, they may have found naturalistic forms of discourse to result in more acceptance, and so resort to them more often. Education and exposure to heterogeneous cultural contexts may have nurtured a more cosmopolitan form of talk, if not personal viewpoint.
One could also make a case for a change in the nature of experience brought about by different material and socioeconomic conditions. It was striking to me, for example, that many of the places on Rotuma that used to be remote, and associated with uncanny experiences, are now within easy reach by motor vehicle. As a result, men spend less time going through the bush to their gardens, and can go in daylight instead of pre-dawn hours to reach their remote plantations in time to work during cool mornings. Vehicles also allow them to go in groups more easily. They therefore spend less time alone in contexts associated with spiritual encounters. Malicious 'atua, the kind one had to worry about and guard against, formerly were regarded as much more dangerous when one was away from one's home village--away from one's territory. This was perhaps a corollary of few opportunities for travel; most people visited other villages or districts only rarely. Now transportation is accessible to everyone, and many people travel to other localities on a daily basis. Perhaps it is easier to entertain a mystified view of other places when one rarely sees them than when they become familiar.
Physical changes may have also affected world view by shifting the focus of experience from the natural to the manufactured world. In the past, building a home or making a canoe required individuals to spend time outdoors collecting materials. Now the cement, lumber, nails and paint are delivered by truck. Appliances of all types command attention, including radios, tape players and videos, providing new contexts for entertainment. Machines and spirits are uneasy companions at best. Motorbikes, trucks, chainsaws, and radios make loud noises, and noise, it will be recalled, is one way of warding off malevolent spirits. The main point, however, is that Rotumans appear to pay much less attention than before to natural phenomena, to the anomalies and sounds of nature that provided the raw material for talk about spiritual encounters. Paying close attention to nature is much less compelling when one (thinks one) has the means to control it, or at least to mitigate its more serious threats.
A drop in death rates may have had a similar effect. Illness and death are two contexts in which spiritual explanations are likely to be invoked, especially when people feel a lack of control over healing processes. As I have written elsewhere (Howard 1979), Rotumans always seem to have entertained alternate explanatory models for illness and death, one naturalistic, the other based on the actions of spirits. The balance between these two shifted noticeably toward the former with the introduction of wonder drugs in the 1950s, although supernatural explanations maintained a definite presence up to 1960. A radical decrease in infant and child mortality, and unanticipated deaths in general, may have also lessened the need to resort to supernatural explanations. In addition, I suspect that an expanded medical facility, with several Rotuman nurses trained in western medicine, has shifted the balance even further in the direction of naturalistic discourse. As access to the clinic has increased because of improved transportation, and people come there more frequently to have illnesses treated, they talk about such matters in a context that discourages supernatural explanation. In the clinic setting, talk about spirits is distinctly devalued.
Finally, the changing role of chiefs may well have had a demystifying effect. Chiefs were previously seen as the embodiment of community authority, and in earlier times, as important conduits to the world of spirits. Their authority, in large measure, was perceived as based on divine activity. To the extent that chiefs were so viewed, spiritual sanctions were at their disposal, and a chief's subjects had reason to fear his wrath. But in today's society chiefs are seen more as politicians than as divinely inspired incarnations of the community. They are no longer awe-inspiring, and are talked about in quite different terms, even ridiculed on occasion. In my recent visits, people expressed little fear of chiefly anger or curses.
Along with the politicalization of chieftainship has gone a diminished sense of community, of which some spirits, at least, were representations. Increased household and individual autonomy, facilitated by greater access to money through remittances and wages, has allowed individuals to refuse participation in community activities with much less penalty. Indeed, one of the topics Rotumans often talk about today, when contrasting the present with the past, is the absence of community spirit. To take Durkheim's point, a diminished sense of community may be related to a diminished set of collective representations in a spiritual mode.
Additionally, politicalization, not only of chieftainship but of the entire Rotuman community (Howard 1989), has diminished people's attention to their ancestral culture (in contrast to many other Pacific Island societies where 'kastom' is of focal concern). They are now more engaged in thinking and talking about current economic and political issues than about past glories; correspondingly they pay less attention to their ancestors, who were known in the form of 'atua. From what I can ascertain, grandparents tell their grandchildren far fewer stories about the past, far fewer myths and legends than before. These were primary contexts in which the exploits of spirits were recounted. Contemporary children may be growing up hearing less about the spirits, and having less to tell about them.
In sum, there is good reason to believe that the Rotuman world view has been demystified, and that a threshold has indeed been passed. Put differently, social contexts in which spirits are relevant may have changed so radically that spirits only rarely enter the realm of human conversation. Perhaps it is not so much that they have disappeared from the scene as they now find far fewer contexts of existential relevance in which to emanate. As the contexts for talking about spirits have narrowed, and people have less information to share, evaluations of talk about spirits may have also changed. What was once clearly acceptable talk within a community-based discourse may now be relegated to the idiosyncratic, seen as indicative of an individual's views rather than a culturally preferred form of expression.
I am not arguing that spirit beliefs are incompatible with modernization. Spirit beliefs thrive in many cultures far more developed than Rotuma. Even in urban Bangkok, for example, Thais maintain spirit houses. But Thailand has a well-established great tradition in which spirits play an integral part. They are woven into literature, dance and art in a way that gives them a visibility absent on Rotuma. As a result, spirits play an important role in nearly every Thai's socialization, and reminders of their immanence are recurrent in daily experience. I suggest that spirits on Rotuma may have been particularly vulnerable to modernizing influences precisely because they lack such objectification in artistic productions and performances. Instead, they were relegated to the margins of public experience following conversion to Christianity, and relied primarily on talk to maintain their presence in the cultural milieu. Thus any factors reducing talk about spirits significantly diminishes their relevance.
Talk and Belief
What does this apparent change in talk suggest about Rotuman beliefs? Have they changed accordingly? I find these questions much more difficult to assess, since I view the relationship between talk and belief as highly problematic. There was no word for belief in the Rotuman language prior to European intrusion, so the missionaries introduced a Rotumanization of the term, pilifi. The closest Rotuman equivalent is aire "true, correct." The antonym of aire is siko "false, untrue or incorrect." However, aire and siko are used primarily as terms of affirmation or denial of a speaker's claims (whether about events, rights and obligations, or other phenomena). They are not ordinarily used in reference to an individual's personal convictions about what is metaphysically true or real. Thus aire is used to signify agreement, and siko to signify disagreement, with a speaker's statements. This usage is similar in form to that described by David Gegeo and Karen Watson-Gegeo for the Kwara'ae in the Solomon Islands, and as they point out, it reflects a distinctive theory of truth (Gegeo and Watson-Gegeo In Press). As among the Kwara'ae, Rotuman discourse suggests an implicit link between potency and truth, such that to say something will occur puts one's mana to test. If events unfold in the way stated, the speaker's potency as a social being is affirmed and his credibility enhanced; if they do not his potency is rendered dubious and his credibility diminished. Whether the speaker has control over the outcome or not is not at issue; the statement itself puts his mana at risk. As the Gegeos point out, since mana derives from spirits, such notions of truth reflect an assumed linkage between the world of humans and the spirit world (Gegeo and Watson-Gegeo In Press).
My skepticism concerning the relationship between talk and belief has been fueled by two types of experience on Rotuma, one of them personal. When I was on the island in 1960 I distinctly remember walking down a path on a dark night with some Rotuman friends. As we passed certain places they expressed apprehension and stepped up their pace. They may or may not have spoken of 'atua, but it was clearly 'atua they feared.  And I, too, experienced fear, perhaps by emotional contagion. This presented me with a dilemma. Did I believe in ghosts? If by "belief" one understands a commitment to a proposition about the existence of numinal beings, I would have to say that I was very skeptical, and hence a nonbeliever. But if an observer were to use as evidence my actions and expressions of fear then they might indeed conclude that I believed.
Nor am I the first European to have such experiences on Rotuma. Hugh Romilly, who was the British Crown's Deputy Commissioner to Rotuma in 1880, just prior to Cession, claims to have seen the ghost of a Rotuman man who had been murdered. He wrote a book about his experience, titled A True Story of the Western Pacific in 1879-80. After describing in vivid detail his encounter, Romilly concluded, "I am not a believer in ghosts. I believe a natural explanation of the story to exist, but the reader...must find it for himself, as I am unable to supply one" (Romilly 1882:82).
The question such reports raises is: can we assume that when Rotumans express apprehension in the face of uncanny experiences, or when they report seeing apparitions, that this constitutes prima facie evidence of a commitment to the proposition that spirits exist?
The second type of experience that raised questions about the relationship between talk and belief is an inversion of the above. Although talk about spirits is now greatly diminished, people still tell stories about the "power" of certain objects, about immanent justice and about occult happenings that imply supernatural activity, but they leave out explicit reference to any presumed agency. They commonly attribute ill-fortune to failures to perform customary ceremonies, but reference to the mechanism involved is conspicuously absent. For example, several people reported that when the maternity hospital was opened some years ago, a proper ceremony was not performed, and that subsequently four women, each young and attractive, died in childbirth.  When the new radio-telephone installation failed on initial attempt, my wife and I were told something similar--the ceremony, which we attended, had not been conducted properly. In other words, the logic of causality remains consistent with supernatural explanation, but spiritual agency is omitted from accounts.
Reconsidering the Nature of Spirit Belief
With a few notable exceptions (e.g., Hahn 1973; Leach 1954, 1967; Needham 1972; Saler 1967, 1977) anthropologists have treated the relationship between belief, ritual practices, emotional expression, and talk about spirits as unproblematic, each regarded as a reliable indicator of the other. Perhaps this is consequence of focusing on intracultural regularities and a relative neglect of intragroup variation. We have tended to treat belief as an undifferentiated category, as if any two statements about the actions of spirits can be given equivalent weight. But as a number of social scientists have pointed out with considerable vigor, there are multiple dimensions to belief that need to be taken into consideration if we are to comprehend relationships between thought, knowledge and social action (for a summary of viewpoints see Bar-Tal 1989:16-20).
I use the concept of belief in reference to meaningful propositions (statements) that can be considered either true or false. To believe is to accept the truth of a proposition; to disbelieve is to reject it (Goodenough 1963:155, Saler 1967:30). Belief is ultimately a mental phenomenon, located in individual minds, and is subject to verification only through inference from talk or action. Nevertheless, beliefs may be shared, and groups may commit to certain propositions collectively, justifying the notion of group beliefs. Beliefs may be considered the building blocks of individual and cultural knowledge insofar as knowledge encompasses all the beliefs accumulated through personal experience, thinking, and social interaction (Bar-Tal 1989:5, Black 1973, Goodenough 1963). Underlying beliefs are presuppositions, or axiomatic assumptions about the nature of reality.
Perhaps anthropologists have so often skirted issues associated with belief because of the thorny methodological problems posed by the concept. How can one identify a belief with assuredness? In western culture we take for granted that people articulate their beliefs, in both internal and external dialogues. Belief is a distinct category in western discourse. Yet we often question statements professing beliefs, regarding them as insincere, self-serving, or inconsistent with an individual's behavior. We note that people frequently express contradictory propositions, and that affective expressions often are in opposition to what an individual avows. Furthermore, we often take actions and talk as indicative of beliefs even when people do not, and perhaps cannot, articulate the propositions directing them. When our concerns are with changing beliefs, these difficulties are compounded.
The problems of identifying beliefs are even more complex when dealing with non-western cultures (Needham 1972).  What do we do in cases, such as in pre-missionary Rotuma, where no emic category approximates our notion of belief? One solution is to rely exclusively on answers to explicit questions. The presumption is that beliefs are consciously reasoned phenomena, and can be articulated with a little thought. When employed with caution this approach has served some ethnographers quite well, but its pitfalls are many. Not the least is the possibility of forcing informants to engage in an unnatural form of discourse, resulting in their clumsily articulating propositions they may never have acted upon, nor talked or thought about--propositions that are artifacts of a unique, atypical situation.
Another possibility is to ignore talk entirely and to rely exclusively on behavioral manifestations of belief. The notion behind such an approach is that acceptance of a proposition is equivalent to a bet on its effectiveness in the world of action (Price 1969:254-7). If an individual takes vitamin pills with meals we infer a belief in the effectiveness of vitamins. This approach has the advantage of allowing a researcher to identify unconscious, or unarticulated, adherence to propositions. It also eliminates ambiguities posed by individuals' saying one thing and doing another. While an "acting-as-if" strategy sidesteps the issue of inconsistencies between talk and action, it has the disadvantage of placing in limbo large areas of knowledge and understandings that are central to worldview, though only indirectly reflected in behavior, if at all. For the the study of spirit beliefs such an approach is too limiting.
A third approach is to analyze discourse in its various guises, both verbal and non-verbal. At the core of this perspective is the notion that beliefs impose form on, and shape the content of, communication structures. Talk in normal (socially recurring) contexts is privileged, although responses to an ethnographer's queries are not excluded. Rituals, artistic performances, visual representations, literary texts and other forms of communication are also examined for what they reveal about propositions, both explicitly and implicitly. Contexts for communications are taken into account, and are made integral to resulting descriptions.
In approaching the question of spirit beliefs, the first approach, based on explicit interrogation, generates descriptions such as those obtained by Hocart, MacGregor and Churchward, reliant on lexical discriminations made by informants. Words for spirit types are defined in terms of appearances, motives, and activities (usually with the implicit assumption that the views of a "knowledgeable" informant are representative). Additional questions may be asked about how concerned humans deal with each spirit type, generating descriptions of rituals, prayers, avoidances and so on. Monitoring changes in beliefs so constructed relies on recognizing changes in vocabulary (expanded, reduced, or altered semantic content), as well as transformations of activities explicitly related to spirits. The resulting emphasis is on alterations in content and form rather than on strength of beliefs (or changes in other dimensions of belief; see below).
The "acting-as-if" strategy leads naturally to an assessment of the degree to which spirit beliefs play a role in shaping social action. Time spent engaged in spirit-related activities provides a ready measure of commitment to spirit beliefs, and of changes in belief. Choices of where to build one's house or plant one's garden, of when to travel the road or go on a fishing expedition-- if based on the prospective actions of spirit beings--are indicative of belief intensity. From this perspective spirit beliefs compete with other considerations for primacy in directing action. As they get stronger they play a more significant role in accounting for behavior. Talk is only important only insofar as it helps to identify the extent to which spirits enter into deliberations, or otherwise influence actions.
From the discursive perspective talk is far more central, although still only one of many ways in which spirit beliefs can be comprehended and monitored. Whereas the "acting-as-if" view focuses on the effects of belief in action, discourse analysis places its emphasis on the reinforcement of beliefs through modes of communication. As a general proposition (which I believe), the more explicitly spirits are represented in communication structures (including speech, art forms, ritual performances, etc.) the stronger the case for inferring belief. My assumption is that strongly redundant cultural messages make certain propositions so central to discourse that meaningful communication generally depends on their acceptance. Individuals who are exposed to an intense socialization in which spirits are multiply represented and dramatized are particularly likely to take for granted the truth of propositions about spirit phenomena.
I find the discursive approach especially appealing for understanding the Rotuman case because the island did not develop representations of spirits in visual forms. This means that aside from certain rituals and dances, communication concerning spirits relied heavily on talk. Although a case can be made that spirits inhabit fine white mats (giving them a special place in Rotuman ceremonies; Hereniko 1991, 1995), the association is not explicit. The paucity of alternative forms for representing spirits thus renders talk especially important to the perpetuation of belief. For this reason I regard a reduction in talk about spirits on Rotuma as of profound significance. Without talk about them, the spirits have no other means to insure a place in Rotuman culture.
Cultural Change and Dimensions of Belief
My purpose in this section is to explore dimensions of belief that might be relevant to understanding the changes I observed on Rotuma between periods of fieldwork. I do not intend to be exhaustive, nor to present a specific theory of belief change. Rather, I wish to illustrate the multiple possible ways to account for such change once we unpack the complexity that lies behind the commonsense notion of belief.
We can begin by distinguishing types of belief, based on what may be considered their underlying presuppositions. From this standpoint existential beliefs can be differentiated from those whose presuppositions are evaluative or pre/pro-scriptive. Existential beliefs address the question of existence, as in "Do you believe in (the existence of) ghosts?" Existential assumptions have been fundamental to Western discourse about spirits. That such is not everywhere the case is nicely illustrated by an incident related by Torben Monberg concerning a discussion he had with people on the Polynesian outlier of Bellona. He had great difficulty getting the Bellonese to understand his question: "Do you believe that the gods propitiated by the residents on the nearby island of Rennell really exist?" The reply was straightforward enough. Of course they exist; who would be stupid enough to propitiate gods that do not exist (Monberg, personal communication)!
Evaluative beliefs concern problems of good and evil and their variants (e.g., moral/immoral, ethical/unethical, holy/unholy). Beliefs concerning the disposition of categories of spirits are of this type. Some types of spirits are seen as basically benevolent and trustworthy, others as malevolent and untrustworthy. Such judgments influence which spirits ought to be propitiated and which guarded against. The distinction between existential and evaluative beliefs was central to comprehending statements made by Rotumans, in response to questions during my 1960 visit, as to whether or not they believed in 'atua. They invariably answered negatively, although it was very clear that they behaved (and in other contexts spoke) as if 'atua were present on the island. It became clear after a while that what they were talking about was propitiation. The most common statement after denial was, "I believe in [the Christian] God." In other words, propitiation, rather than existential reality, was at issue. This use of "belief" (the question was asked in English) corresponds to statements such as, "I believe in my doctor," or "I believe in capitalism"; it signifies trust versus mistrust rather than existence versus nonexistence (see Price 1969:76-77).
Prescriptive and proscriptive beliefs relate to a presumed causal connection between certain kinds of actions and likely consequences. The belief that performing a specific ritual will bring blessings from the gods is prescriptive; the belief that stealing will bring misfortune is proscriptive.
Sources of belief can also be distinguished. Here we are concerned with what kinds of information are considered to be evidence in support of or against particular propositions. Information obtained from direct sensory experience can be contrasted with information obtained from interpersonal interactions and from impersonal media (books, motion pictures, radio, television). That we do not always believe our own sensory impressions is affirmed by the fact that many people believe the world to be round, despite the fact that our senses tell us otherwise. Sources of knowing may have quite different weightings in different cultural settings. Whereas in modern cosmopolitan contexts beliefs are largely structured by formal education and mass media (it is on the basis of the authority of scientists that most of us do believe the world is round), in many other cultural contexts interpersonal transactions are key to the formulation of beliefs. One can make a subdistinction here between relationships based on hierarchy versus those that are not. Where relationships are structured hierarchically the pronouncements of those in authority are given great weight; in more egalitarian contexts evaluations are based on a variety of considerations having to do with the credibility of a speaker. As pointed out above, Rotumans, like the Kwara'ae, relate truthfulness to the mana of a speaker (which is based on perceptions of his/her efficacy in making things happen). A Rotuman who conveys information that proves to be false, even though it clearly originated elsewhere, quickly loses credibility and is labeled unreliable. In other words, the overall credibility of a speaker is far more important in evaluating evidence than the intrinsic merits of the information conveyed. So whereas one person's report of an encounter with a spirit may be regarded as prima facie evidence, someone else's account may be dismissed out of hand.
Another set of dimensions can be applied to the characteristics of an individual's beliefs. Beliefs differ, for example, in degree of precision, or the clarity of their formulation. Some can be stated precisely (e.g., if X then Y; the meek shall inherit the earth), while others are only vaguely held, and cannot be easily articulated. In large measure precision results from discourse that places a premium on such considerations as logical consistency. Early European visitors to Rotuma were dismayed by an apparent lack of precision in answers to their questions about "religious" beliefs (but this may well have been the result of both language difficulties and inappropriate framing of questions).
Centrality is another dimension that a number of social scientists have identified (Rokeach 1960, 1968; Bem 1970; Bar-Tal 1989), although there is no agreed-upon definition. I use the term in reference to the degree that a given belief is basic to an individual's worldview, as indicated by the extent to which it shapes perceptions and other beliefs. A change in a core belief is likely to necessitate a change in one's whole perspective, whereas a change in a peripheral belief may have little effect on one's outlook. Assessing the centrality of spiritual beliefs would seem to be critical for determining how readily they might be altered or dropped when conditions change.
Conviction refers to the degree of confidence with which one holds a belief. Some beliefs are held quite tenuously, as indicated by expressions of doubt or a willingness to entertain alternative propositions. Others are held with great assurance, as if firmly anchored in unassailable evidence. Closely related to conviction, but distinguishable from it, is commitment, which indicates tenacity in the face of no or contrary evidence (a word frequently used in English to express commitment as I have defined it is "faith"). Commitment is especially relevant when a belief is regarded as indicative of something of great importance to an individual, like membership in a group. Indeed, the less evidence there is to support a belief, the more powerful it becomes as an indicator of commitment to a group that professes it. Potency, which I use to denote the degree to which a belief motivates action, is an additional dimension. People may hold certain beliefs with great conviction, even though they may have little motivating potency (e.g., Venus rotates around the sun). Other, less vigorously held beliefs, may be strongly motivating, at least in certain contexts (e.g., God helps those who help themselves).
Integration of beliefs into coherent structures has also been widely discussed (Rips 1990). An array of propositions held by an individual may logically cohere, forming a structure or schemata, or they may have minimal interdependence, with specific propositions supported more by external considerations than their relationship with one another. At a lesser level of integration, certain beliefs may be coupled such that one is perceived as implicating the other.  For example, on Rotuma, belief in immanent justice and belief in the power of 'atua were coupled in the past insofar as spirits were said to be agents of justice.  The question this poses is: under what conditions can beliefs become uncoupled?
When dealing with beliefs at a group level, some additional dimensions need to be considered. What forms of expression are beliefs given, for instance? Are they articulated in speech, in ritual, in artistic productions? How explicit are the renderings of specific propositions in various media? Degree of redundancy for specific beliefs must also be considered. How frequently are various beliefs given expression, and how many channels are employed in representing them? A belief that is voiced through limited channels (e.g., a single ritual) may be jeopardized if those channels are blocked, whereas one given expression in multiple channels is likely to be less vulnerable. Multiplicity of beliefs refers to the extent to which the total pool of propositions available to a population includes, in any given context, alternate or contradictory possibilities. In general, the more extensive the pool of propositions people can draw from, the more problematic group beliefs become. In small, culturally isolated groups the distinction between individual and group beliefs may be minimal, whereas in large, culturally complex groups the distinction can be so great that identifying any "group beliefs" at all may be difficult. In the process of transition from isolated small-scale society to heterogeneous cosmopolitanism, the locus of belief shifts from the group to the individual.  Thus, whereas in former circumstances Rotuman statements about spirits could reasonably be considered reflections of a cultural world view, today they are more appropriately seen as indicators of an individual's personal beliefs.
It is one thing for people to share beliefs as a result of common experience, but quite another for a group to make shared belief an index of membership. Even in culturally complex settings, certain beliefs may come to be regarded as indicative of a particular group's character (e.g., ethnicity, sophistication, political commitments, solidarity). This, however, necessitates an objectification of belief (Bar-Tal 1989). The fact that Rotumans had no lexeme for belief is strong evidence they did not objectify it prior to European intrusion. The contrary teachings of Methodists and Catholics, however, firmly introduced awareness of shared beliefs as a powerful basis for defining group membership and allegiance.
The necessity of recognizing the complexities behind our common sense notion of belief is especially important if we are to account for the type of change I have described for Rotuma. At issue are the ways in which worldviews are constructed, how knowledge systems (which are composed of beliefs) function in various cultural contexts, and the interrelationships between beliefs, emotions, thinking, and social action. I do not intend to provide a singular theoretical explanation for what I have observed on Rotuma, but rather wish to mirror the complexity of the phenomena by reflecting on a variety of possibilities, all or none of which may have contributed to change.
Let me begin by dismissing one possibility--that Rotumans have shifted en masse from a firm belief in spirits to disbelief (i.e., from acceptance to rejection of the proposition that spirits exist).  Merely refraining from talk about spirits does not in itself signify disbelief, and no other evidence at my disposal suggests outright rejection. More to the point is that Rotuman beliefs in spirits were primarily prescriptive/proscriptive rather than existential or evaluative. Rituals were the main channels for expressing belief; talk was secondary. One could therefore make a strong case for the continuance of spirit belief on the grounds that rituals persist, and that failure to conduct rituals properly is still a culturally appropriate explanation for misfortune. The decline in talk about 'atua may be little more than a minor concession camouflaging core beliefs.
It is also possible that Rotuman spirit beliefs were never central in the sense discussed above, and involved low levels of conviction and commitment. This would be consistent with a view of Rotuman beliefs as having a primarily social, as opposed to a psychological, function. Beliefs can perform social functions adequately without much conviction or commitment, especially in ritual contexts. If this were the case, the uncoupling of belief in 'atua from belief in immanent justice would not be surprising. Whereas 'atua provided a mechanism for justice in the past they are not necessary for it to occur (the Christian God or fate can be substituted, but even they are not necessary, since the mechanism need not be explicit).
Other processes may have been involved. Rotuman beliefs seem to be strongly context dependent. Perhaps the number of contexts for which spirits are relevant has been drastically reduced, thus removing beliefs concerning them to the peripheries of experience. This may have led to a diminished consciousness, and hence a reduction in talk about spirits. Sociolinguistic processes also might have contributed. Even if no significant changes in personal beliefs had occurred, it is possible that conversational rules changed, inhibiting the mention of spirits. Maybe, for example, the social costs of mentioning spirits has increased, so that people are increasingly reluctant to express their personal viewpoints (in much the same way that American politicians are more reluctant to express bigotry, regardless of their beliefs, because of political costs). Whether this signals a temporary shift in talk patterns or a deeper change in Rotuman explanatory discourse remains to be seen.
Less talk about spirits also could have resulted from social changes implicating others' credibility. A weakening of authority structures, with which beliefs in spirits were associated, may have been involved. Chiefs and elders had a strong stake in perpetuating spirit beliefs since spirits were said to punish disrespect and disobedience; a decline in the credibility of such authorities might have contributed to less attention being paid to reports of their experiences in favor of greater reliance on one's own. Or it may be that those with a Western education are now granted greater credibility, and that they are reluctant to talk about spirits because it suggests backwardness.
None of these possibilities implies finality. It is quite conceivable that under other conditions spirit beliefs will regain centrality for at least a portion of the Rotuman population. Although contemporary Rotumans have not resorted to objectified beliefs to define their identity vis-a-vis others, they may find it expedient to do so in the future. If so, talk about spirits may again rise to prominence, and play a significant role in defining what it means to be Rotuman.