To home page

To culture

To Culture

To Legend of Raho

To legend of Raho

Aspects of Rotuman Myth

[This segment is adapted from A. Howard, 1985, History, Myth and Polynesian Chieftainship: The Case of Rotuman Kings. In A. Hooper and J. Huntsman (eds.), Transformations of Polynesian Culture. Auckland: Polynesian Society .]

The first record of Rotuman oral narratives was collected by Fr. Trouillet ca. 1873. The account is in the form of a history of Rotuman chieftainship, beginning with the founding of the island by a chief from Samoa named Rao (Raho), and ending in 1868 when the last sau 'king' held office. The document is remarkable for its chronological ordering of fabled events, and sets a useful framework for the study of Rotuman mythology.

In the years following Trouillet's initial account several other visitors to the island collected texts of Rotuman narratives, many of which provide alternative versions or elaborations of Trouillet's. These include accounts by Romilly (1893), Gardiner (1898), Hocart (1913), MacGregor (1932), Titifanua and Churchward (1995) and Russell (1942). Of the published accounts, only that of Titifanua and Churchward includes Rotuman texts; the rest appear only in English.

We must recognise these texts for what they are--residues of living performances, recorded by individuals who had particular notions about what was worth recording. They provide virtually no information about the teller and the context of performance, let alone the way the stories were learned and transmitted. One suspects that they are responses to specific questions such as: "Where did the Rotumans come from?" and "Do you know any other interesting stories about the old days?" In such a context they would likely be stripped of elaborations that would mark performance before indigenous audiences. Nor do we know how the story-tellers classified their tales. Rotumans distinguish between three types of narrative. The most mundane is covered by the word rogo, which as a verb translates as 'to be reported, heard of, generally known' and as a noun as 'report, news' (Churchward 1940:300). The word, sometimes used in its reduplicated form, refers essentially to events witnessed by the teller or reported by presumed reliable sources. A second category is hanuju 'tale, story, especially a fictitious one'. Churchward speculates that this is probably a corruption of hagnuju, from haga 'to feed' and nuju 'mouth' (Churchward 1940:216). The implication seems to be of frivolity, of stories meant merely to entertain. This contrasts with the third type, fäeag tupu'a 'legend, myth' (Churchward 1940:189). The first word, fäeag, refers to speech, the second, tupu'a, is translated by Churchward as follows:

tupu'a, n., (in mythology) immortal man; rock or stone reputed to be such a person petrified: image, statue, idol, doll; star or constellation, esp. as a point for steering by; point of the compass, direction, bearings. (Churchward 1940:337)

One of Hocart's Rotuman consultants in 1913 specifically stated that, in contrast with hanuju, fäeag tupu'a are "true" stories. Whether this referred to a belief in the literal occurrence of events as described is problematic. I suspect that the reference was rather to a notion of structural truth--that these stories truly represent the nature of Rotuman collective experience, in much the same way as an icon represents true religious experience.

Yet another problem is that we know very little about the degree of variation and its correlates. There is some evidence from Hocart's notes that Rotumans of his day discussed specific myths in order to arrive at a consensual account, but where we do have multiple versions there are significant discrepancies. What makes the problem of interpretation somewhat more acute is that the collection of texts covers a period of some seven decades, and we cannot distinguish genuine cultural changes through time from individual and geographical variation within a given time period.

We are thus in a position somewhat akin to that of an archaeologist who is attempting to reconstruct a material order from a fragmented artefactual record, divorced from its performative context. The problem is one of making reasonable inferences, based upon certain assumptions about human behaviour, the utility of various artefacts, the consistency of particular patterns, and so forth. Just as archaeologists have found ethnographic observations (i.e., ethno-archaeology) and comparative analysis useful for grounding such assumptions, analysts of myth can benefit from like efforts. Based upon my own field work on Rotuma, supplemented by the observations of others, and a comparison of Rotuman cultural patterns with those of other Polynesian societies, I have arrived at the following set of working assumptions concerning Rotuman oral narratives:

1. Stories are constructed out of an extensive array of semiotic codes, which are transmitted in a variety of media. In addition to codes embedded in the string of words from which written texts are constructed, there are expressive codes embedded in speech and gesture, spatial and temporal codes, and a number of other performative codes available that lend meaning to oral narratives when they are told in vivo.

2. Both digital and analogic codes are used. Digital codes are constructed out of basic oppositions between such concepts as land/sea, male/female, person/spirit, east/west, raw/cooked and so on. As Lévi-Strauss has noted, digital codes generate mediating categories, such as beaches, mountains, birds, male-like females, demigods, etc. Digital codes are used in Rotuman myth (and perhaps all myth) to convey messages about basic categories of social construction, e.g., the components of chieftainship, the distinctions between men and women, between humans and gods, and so on. Analogic codes involve changes in degree of states, such as emotion, potency, acceptability and the like. They are used to convey messages about limits and boundaries, and about the implications of variation within categorical parameters. They are most conveniently embedded in expressive media such as gestures and voice intonation, and hence are more prone to being lost when stories are transformed into written texts. Perhaps this is one reason why digital codes have received so much more attention from armchair analysts.

3. In Rotuma, the codes used to construct oral narratives are generative. That is to say that they are subject to a set of meta-rules, or story syntax, that allows for the production of a range of acceptable variation for any particular story. Generative possibilities are influenced by variations in social settings in which stories are told, time allotments, relations of teller to audience as well as the personal and social characteristics of the story-teller. This differs from some societies in which at least a significant segment of oral literature is ritualised, embedded in chant and otherwise restricted to singular versions. It means that attempts to find "correct", "official," or "consensus" versions of Rotuman myths are unwarranted, and raises some questions about the relationship between cultural codes and individual usage. This makes it imperative to examine the full range of available texts before attempting inferences about the meaning of any particular narrative.

4. The written texts recorded by visitors to Rotuma that provide the basis for our analysis are restricted to certain codes and therefore only represent partial semantic structures. Their full meaning has been lost, and it is possible that performative codes significantly altered, perhaps even inverted, some of the meanings in the written texts (as, for example, an ironic tone of voice inverts meaning in English). Corollary to this, the full meaning of key symbols, metaphors and metonyms cannot be recovered from such residual texts. At best they can be inferred from the contextualisation of their usage. For example, in Rotuman myths the opposition between chiefs and commoners is so consistently associated with geographical directions that I feel quite confident in making inferences based upon them. More problematic are inferences to be drawn from proper names of persons or places. In some instances the overt meaning is blatantly obvious, in some it is somewhat suggestive but thoroughly ambiguous, while in other instances there are no grounds whatever for making an inference. This underscores the importance of examining the full range of available texts for consistency of usage so that at least core features of semantic units can be inferred with some degree of conviction. Fortunately, a considerable degree of redundancy occurs in the texts, between as well as within codes. It is reasonable to assume that the messages of greatest concern were the most redundant, and that they were the least likely to be nullified or drastically altered by unrecorded performative codes.

5. Although the texts of narratives are often written in the idiom of history, they do not appear intended to do the work of history in our usual sense, i.e., to accurately record significant events of the past in correct chronological sequence. Whether or not certain incidents related in the narratives are based on actual events, they are processed through such a powerful codification system that their validity as history must be regarded as extremely problematic. A more defensible view is that chronological sequencing provides a mechanism by which structural oscillations are explored in their various permutations (see Howard 1986).

6. Rotuman myths appear to reflect a preoccupation with cultural dilemmas associated with relations between humans and supernatural beings, on the one hand, and between chiefs and commoners, on the other. Since gods and chiefs merge conceptually at certain levels of contrast, these two themes can be considered as derivative from a single overarching cultural problem, namely the problem of the genesis and control of mana 'potency'. At a more basic (often implicit) level the concern is with the continual regenesis of life--with the fertility of the land and the people. The fundamental issue is one of harnessing the mana of the gods in the service of this goal.

7. Fäeag tupu'a 'myths' seem to owe their sense of drama to the fact that they involve explorations of basic structural properties of the cultural system. In contrast with rogo or hanuju, which deal with variations within the received structure, myths explore the consequences of altering the parameters of structure. They thus probe structural properties, and examine the possibilities for structural transformations. Within myths the possible effects of adding, deleting or altering the value of categories can be played with, a process which has the effect of providing a visibility to key aspects of the cultural order that might not be elsewhere apparent (except, perhaps, in ritual). It must be added that the overall effect of myth in Rotuma appears to have been conservative in so far as it focuses on the negative consequences of breaching fundamental principles of structure, e.g., of violating the rules of the use of power (see Howard 1986).


Churchward, C. M.
1940 Rotuman Grammar and Dictionary. Sydney: Australasian Medical Publishing Co.

Gardiner, J.S.
1898 Natives of Rotuma. Journal of the Royal Anthropological Institute 27:396-435, 457-524.

Hocart, A.M.
1913 Field Notes from Rotuma. Manuscript in Turnbull Library, Wellington, New Zealand.

Howard, A.
1986 Cannibal Chiefs and the Charter for Rebellion in Rotuman Myth. Pacific Studies 10:1-27.

MacGregor, G.
1932 Field Notes on Rotuma. Manuscript in the Bishop Museum, Honolulu.

Romilly, Hugh
1893 Letters from the Western Pacific and Mashonaland, 1878-1891. London: Nutt.

Russell, William E.
1942 Rotuma. Journal of the Polynesian Society 51:229-255.

Titifanua, M. and C.M. Churchward
1995 Tales of a Lonely Island. Institute of Pacific Studies, University of the South Pacific. (First published in Oceania vols. VIII and IX (1938-9), then as Oceania Monograph No. 4, Tales of a Lonely Island (1940).

To home page

To culture

To Culture

To Legend of Raho

To legend of Raho