- 154

University of California, Santa Cruz

Scattered through the dictionaries and grammars of Oceanic languages are fragmentary references to words tabooed because of their association with things sacred or proscribed. Although word tabooing seems to have been widespread, and some cases (e.g. Tahiti) are fairly well known to specialists, the phenomenon has received little comparative study. The distributions of word tabooing in the Pacific have not been worked out. For those cases described in the literature, evidence is usually meager. It is seldom possible to estimate the frequency of tabooing or the types of forms selected; the mechanisms whereby particular forms were selected for tabooing are not known in detail; and there is seldom much evidence - 155 on how replacement forms were chosen. In almost all cases, detailed evidence is lacking on word tabooing as a sociological phenomenon.

Yet at this point it seems possible that word tabooing was characteristic of some or all early Austronesian speakers in the Pacific. That possibility, in turn, has profound implications for comparative study of Oceanic languages and the use of linguistic evidence in reconstructing Pacific prehistory.

Why is the evidence so meager? Largely because the very forces that led to the occasional recording of word tabooing were leading at the same time to its disappearance. Evidence is needed of word tabooing as an ongoing process, not an ancient custom; yet in most of the island Pacific that process came to a halt decades ago.

The potential importance of word tabooing in Oceanic linguistics derives from the possibility that this process has significantly accelerated vocabulary differentiation between genetically related languages. To the extent that “basic vocabulary” forms were tabooed, lexicostatistical analyses of Austronesian languages could be affected. Such tabooing could create a spurious impression of long divergence or skew datings, or in some cases could even hide genetic connections. 3 If tabooing affected less common, or “cultural”, forms, leaving a basic vocabulary largely untouched, this could create an impression of a layer of common Austronesian forms superimposed on a substratum of genetically unrelated languages.

To illustrate such possibilities, let me simply note three problems in current Oceanic linguistics: first, the problems of reconciling retention rate constants of glottochronology with archeological time scales; 4 second, the greater lexical differentiation in Melanesian languages than grammatical structure and other evidence would lead us to expect; 5 and third, substratum and pidginisation theories of Melanesian languages. 6

As a social anthropologist, not a linguist, I must leave to others the comparative study of word tabooing in Austronesian languages and assessment of the implications of word tabooing for subgrouping of Oceanic languages. However, my recent field work among the pagan Kwaio of Malaita, British Solomon Islands, among whom word tabooing is still very much an ongoing process, gives me an opportunity to supplement the present meager evidence on word tabooing, and particularly to examine it as a social phenomenon.

About 6,000 Kwaio speakers inhabit the mountainous central area of Malaita. Kwaio is one of some ten Melanesian languages spoken in belts across the 100-mile-long island. Kwara ? ae and ? Are? are, spoken in the adjoining zones, are almost completely unintelligible to Kwaio speakers. No adequate comparative analyses of Malaita languages, or attempts at sub-grouping, have yet been carried out.

In pre-European times, the Kwaio were a “bush” people, growing taro - 156 in upland swiddens. They lived in tiny hamlets scattered across named descent group territories. Introduced diseases have thinned the ranks considerably and in recent decades about half of the remaining Kwaio have become Christian and have moved to the narrow coastal strip. 7 However, a sizeable pagan population still lives in the high country along the east coast and carries on largely unchanged by direct outside influence. It is this pagan population, particularly inland from Sinalagu, that I studied for two years, and among whom I observed the word tabooing patterns I will describe.

Word tabooing is a reflection of Kwaio beliefs about the sacred, and hence to comprehend it one must begin with an outline of Kwaio religion. Further, since the restrictions on linguistic usage apply to particular actors on the social scene, they can be interpreted only within the framework of Kwaio social structure. To separate “religion” from “social structure” distorts Kwaio conceptualisation of their social universe which is populated with both human actors and their ‘ancestors’ (adalo). 8 These ancestral spirits, as unseen actors on the social stage, enter almost every phase of Kwaio life. Social relations of the living are often defined and ordered in terms of ancestors. Thus in sketching the “social” and the “religious”, an attempt will be made to show as well their interdependence and articulation.

The Kwaio landscape is carved into small territories composed of a patchwork of land tracts conceptually grouped under a focal shrine. Each territory is, or once was, the spatial locus of a descent group or lineage defined in terms of descent from the ancestors founding the territory. Relations with descent group ancestors are mediated by a priest at the focal shrine and various subsidiary shrines and at the sacred men's house he maintains.

Recruitment to these lineages is normally patrilateral, reflecting a strong ideology of agnation. However, a strong secondary interest in the maternal group can be converted to a primary affiliation by long residence, especially in childhood. Since in each generation double, if asymmetrical, relationships and rights are traced through each parent, the cumulative result is a wide cognatic extension of secondary land rights and relations to ancestors. Looked at another way, each individual acquires, in addition - 157 to a usually unambiguous descent group membership, two bundles of relationships to lands, kin, and ancestors traced through grandparents, great-grandparents, and beyond, and passes a single bundle to his children.

This combination of discrete descent groups and wide-spread cognatic ties to lands and ancestors has a paradoxical result. On the one hand, each descent group comprises a separate land-owning corporation and above all a ritual community linked to common ancestors. But on the other, neither land rights nor ritual relationships are exclusively exercised by group members: a cluster of outsiders cognatically descended from the founding ancestors, but primarily affiliated elsewhere, have secondary interests in the territory and its ancestors.

From ancestors comes the “power” (nanamaŋaa, metathesised and nominalised “mana”) whereby the living attain success, “good living”, successful gardening, acquisition of wealth, successful fighting, etc. Ancestors convey this power when they are pleased with the living; they are pleased when their descendants follow proper ritual procedures, observe taboos (especially those pertaining to the ritual uncleanness of women), and raise propitiatory pigs consecrated to particular ancestors. When ancestors are displeased, especially as a result of violation of a taboo, they cause misfortune, sickness, and death. Which ancestor is causing misfortune, why, and what can be done about it (usually sacrifice of expiatory pigs) is determined by divination.

Ancestors vary in the degree to which they are ‘sacred’ (abu, Oceanic “tapu”). Just as living Kwaio exemplify the “big man” system of Melanesia, so ancestors are ordered in terms of their power and influence within a group of related ancestors. Each descent group might be thought of as joined with what I have elsewhere called an “ancestral kin group”. Within this “group”, one (or sometimes two) ancestors will be most powerful. Sacrifices to the entire “group” are usually made through this “leader”.

The process whereby an adalo rises to prominence can only be inferred by extrapolation. Individuals who were prominent in life as major priests, successful fighters, important feast givers, etc. 9 are “important dead” when they depart the ranks of the living. As such, their death leads to elaborate ritual observances. Such a deceased person has a promising start toward being a powerful adalo: but as in life, only a start. How powerful he becomes depends not only on the attribution of special power to him, which reflects the power he had in life, but also the outcome of divination and the success of his descendants. The rise to power of an adalo is in practice also contingent on the proliferation of his descendants.

The most powerful adalo appear on genealogies some twelve to fifteen generations back. Through the proliferation and geographical spread of their descendants (particularly through out-marriage of women), they have come to be venerated at a number of shrines by many different descent groups, through different priests. The few most powerful adalo are claimed as ancestors by more than half of the people within a wide - 158 region. One adalo was common to 78% of my census respondents. It should at this point be noted that this adalo, and a number of other very prominent ones, were women in life, to whom extraordinary powers are attributed.

To provide sufficient background for a discussion on word tabooing, it is necessary to go briefly into some detail on the fo?ota system. The propitiatory pigs consecrated to an ancestor, are called fo?ota (‘offering’). An adalo is given a special name, a kind of pseudonym, that is used for the consecrated pigs, and because the original name is sacred this fo?ota name comes to be used for the adalo in lieu of the name used in life. In the case of the most powerful ancestors, the fo?ota name may come to refer not to a single ancestor but to a family cluster of ancestors who distinguished themselves in life by collective action; for example, father and his sons, a set of brothers, or a mother and her daughter. A pig bearing the fo?ota name is consecrated to them as a unit to elicit the power they convey collectively. The rise of an adalo to sacredness is reflected in the generation span between death and the time fo?ota pigs raised for the adalo become so sacred that only adult male descendants can eat them. Before this time, the adalo is minor and both ancestor and consecrated pigs are only minimally sacred. A sacred priest becomes sacred to his grandchildren; a ritually adult man becomes sacred to his great-grandchildren; others become sacred, if at all, to their great-great-grandchildren. A minor adalo, usually the parent or grandparent of the person sacrificing to it, is important for sentimental reasons and as an intermediary to more distant and powerful ancestors.

An individual's ritual position is defined largely by the fo?ota pigs he raises. An adult man usually “keeps” from eighteen to fifteen fo?ota to which he may periodically consecrate pigs, though at any one time he is likely to be raising pigs consecrated to only some four or five. When the consecrated pig is fully grown, it will be sacrificed to the ancestor by the priest and will be ritually eaten by the adult men descended from that ancestor. In sociological terms, descent group members will share a common cluster of descent group ancestors and partake of these sacrifices together, augmented by some cognatic descendants of these ancestors affiliated elsewhere. Descent group members are also related to different ancestors through their mothers, mothers' mothers, fathers' mothers, and so on, that differentiate them ritually. These non-agnatic ancestors, and sacrifice to them, provide a ritual expression of genealogical segmentation within a descent group. Thus, an ancestor traced through a man's FaFaFaMo will be common to a whole descent group segment; an ancestor traced through his FaMo will be common to a much smaller segment. Finally, in the case of those adalo that have come to be common to large segments of Kwaio society and are propitiated at different shrines by different priests, the sacrifice of large fo?ota pigs at mortuary feasts provides an occasion for men of these larger descent categories to express their unity by sacred commensality. 10

A final principle regarding adalo is also necessary to an understanding - 159 of word tabooing. This is that adalo are concerned only with action of their descendants. Thus, if a shrine is desecrated, or a ritual violation committed by an outsider, it will be the descendants of the ancestor who are punished. From this derives the principle that the descendants claim ritual compensation for such offences from those who commit them, usually in the form of expiatory pigs. An offence against an ancestor by a non-relative is an offence against living descendants as well.

A. Principles of Tabooing. The name of a person, in Kwaio culture, is associated with the ‘essence’ (to ?ofuŋana) of that person. The name of an adalo thus acquires sacredness roughly corresponding to the sacredness of the adalo.

• (1) The basic principle whereby linguistic forms become taboo is that usage of a form associated with that name can impinge on the sacredness of the ancestor. Such infringement is, as earlier noted, the concern of the ancestor and his descendants; it is the descendants who have the right to fa?aabua (render sacred, and thus forbidden) use of a linguistic form in their presence. From the Kwaio stand-point, it is usually the adalo himself who prohibits use of his name, and the living who implement this prohibition.

There are several permutations on this basic principle. The first two reflect the attribution of extreme sacredness to an adalo.

• (2) The fo?ota name of an adalo may become sacred so that a linguistic form associated with it is tabooed. This can be considered to be a secondary level of tabooing, since the fo?ota name acts as a sort of pseudonym. Still another set of pseudonyms is used for pigs raised for purification (siuŋa), as contrasted with propitiatory fo?ota pigs. These names are rarely tabooed.
• (3) On a tertiary level, certain euphemisms are used in lieu of fo?ota names in direct address (e.g. in divination) to very sacred adalo. Some restrictions may be placed on linguistic forms contained in these euphemistic expressions.
• (4) Certain “shrines” which are the traditional burial places of very sacred adalo are used for “high sacrifice”. The name of the shrine might rarely become tabooed in secular usage, leading to a word taboo.

A fifth principle reflects the association between a name and the person who bears it, but not extreme sacredness of the adaio with that name.

• (5) The name of a person who died within the lifetime of those now living may become prohibited in use among his surviving relatives in respect to his memory, even though no special powers are attributed to him, and association between the name and a linguistic form may lead to restricted use of that form as well.

Two subsidiary principles—only the first involving association of a name with a linguistic form—reflect the dangerous sacredness of two feared diseases, leprosy and tuberculosis.

- 160

• (6) A person who dies of xuu, “leprosy” or fonumela (a disease category partly overlapping with tuberculosis), and pigs consecrated to him, become sacred immediately. The name of such an adalo acquires sacredness early, and may lead to tabooing of a form.
• (7) The very terms for “leprosy” and “tuberculosis” are restricted in use, because of the danger of “contagion”. Especially in dwelling houses, euphemisms or alternative terms must be used; a sequence of these have been employed in recent decades, then discarded as they acquired dangerous power.

A final mechanism is recent:

• (8) Association of a Kwaio linguistic form with a word that has sexual or “dirty” overtones that restrict its use in English or “polite” Pidgin English may lead to the avoidance (though not tabooing) of the Kwaio form. 11

B. Kwaio Naming. Before examining the associations between names and linguistic forms that lead to tabooing something should be said about Kwaio names. Most Kwaio names comprise two word bases, sometimes connected by particles. The most common form is a verbal base plus a noun base (usually corresponding to an English transitive verb plus direct object): “hit corpse”, “spear fish”, “drink dew”, “string money”. Less common are two noun bases connected by particles that in common usage of the name are omitted (“money-shrine”, which might be “money for the shrine”); an intransitive verbal base plus noun base, connected by particles in the full form (“look-pig”, for “look for the pig”); a single noun base (“panpipes”); a combination of noun bases; or some other grammatically possible combination. Originally these names are based on some event in the recipient's early life, but they are often passed down through several generations. An individual receives at least two names in life, one for childhood and a second for adulthood. However, many people go through several names in a lifetime and have two or even three in use at a given time.

A frequent pattern is abbreviation of a name. Usually, for a man or a woman, the abbreviation is {-xa}, added to the first word base: e.g. Go →ubisu, ‘drink dew’, →Go →uxa; ?Ai ?aniaba ?e, ‘insist about shrine’,→Aixa. Other particles are rarely used, particularly {-ta}, e.g. Galuoŋa →Galuta.

Some contradictions might seem to exist between the sacredness of an ancestral name (or the sentiment connected with the name of a deceased kinsman) and the passing of a name from generation to generation. Four points bear mention in this connection: (1) Acquiring the name of a deceased relative, in a sense, reactivates his status or continues that essence associated with the name, and hence is complementary to tabooing of the name 12 in expressing esteem or respect for the dead. (2) Such sequences of names are usually given in that span of generations—parental to great-grandparental—where deceased persons are not yet - 161 highly sacred. When the original bearer of the name reaches ancestor-hood, 13 the name will no longer be passed down. (3) Such a name is given only as an adult name. As a childhood name, it could be used in a playful curse by an age-mate, with unfortunate results. (4) Use of an associated linguistic form might be tabooed even though a living individual bears the original name. This, however, would be rare.

fo?ota names are generally similar, except that they usually involve some incident or characteristic of a deceased person's later life. They consist somewhat less often of a single noun base.

Finally it should be noted that there are some word bases, mainly noun bases, that are extremely common in names. This, as will be seen, affects the working out of word tabooing mechanisms. There follows a calculation of the percentage of incidence of some common forms used in men's and women's names, from a sample of 3681 names.

The interesting aspects to note from the table are two: 1) that Kwaio name segments seem to be asexual in their incidence of occurrence; and 2) that these ten common name segments order themselves into four fairly tight groups.

C. Linguistic Associations Leading to Tabooing. Here will be analysed the purely linguistic associations between names and linguistic forms that cause the latter to be tabooed. In the next section, the factors will be examined that actually precipitate or inhibit tabooing of a form.

(1) A name is a word base.

The simplest and most direct relationship between name and language form occurs when the name consists of a single word base (sometimes with bound particle). This can occur with either an adalo name or a fo?ota name. Examples:

- 162

(2) A name is a modified form of a word base.

Here, the base is manifest in the name in some morphologically modified shape (often a reduplicated form). In this case, the base itself is usually tabooed, though not necessarily in all the constructions it enters into. The derivative form in the adalo's name is also tabooed.

(3) A name is one of two component bases in a compound form, and another base is substituted for it. Here the adalo name could be homonymous with the segment of the compound form. Example: Gola, an adalo name, led to tabooing of /madagola/, ‘hot coals’, and its replacement by /madalafa/.

(4) The name contains two bases, one of which is then tabooed. Here. one must account for the tabooing of one of the bases, but not the other. Examples:

(5) A form is tabooed because of phonological similarity to a name. Most often, here, the name is not morphologically related to the tabooed form. Examples:

- 163

(6) A name consisting of two bases may be tabooed so that either base can be used separately, but they cannot be conjoined. In this construction, one form could not be used, and some substitute would be required; or some separating “buffer” would be introduced. Examples:

(7) A name is a morphological modification of a word base, and that morphologically complex form is tabooed while the base and other derivative forms are not.

Since most names, or fo?ota names, contain one or more word bases—or could otherwise lead to tabooing of a linguistic form according to the principles outlined above—the problem here is why some names lead to word tabooing but most do not.

This depends on a number of interacting factors.

• (1) Attribution of sacredness to the bearer of the name, as an adalo, is a basic prerequisite. Already noted are the factors—including importance in life, mode of death, generation removal, success and proliferation of descendants, and the outcome of divination—that affect the degree of sacredness. An adalo without “power”, in its positive and negative aspects, requires no taboo.
• (2) Since tabooing is a decision made by descendants of an adalo, some precipitating incident usually takes place. Given an association between a name and a linguistic form that could be tabooed (see below), attribution of increased sacredness to the adalo with that name creates the conditions for tabooing. What finally triggers off such action is usually the discovery, through divination, that use of the name or associated word has caused illness or misfortune. This, in turn, is possible only when such has been proferred to a diviner as a possible cause, and usually only after other expiatory action fails. Use of the potentially tabooed word in a dwelling house, in

- 164

• play, in some sacred context, or especially in a “curse” or “swear”, may lead to this possibility being put to the test of divination. Sometimes the suspicion that a “swear” might have occurred among children, though no adults have witnessed it, will lead to this possible cause of the misfortune being put to the test. However, it should be noted that often dozens of possible reasons why an ancestor might be causing an illness are put up before one is selected through divination as correct. Usually this is done more or less randomly, since the common method involves breaking knotted leaves. From the Kwaio point of view, one should note, it is the ancestor that imposes the taboo due to some infringement on his name, and communicates this through divination.
• (3) The name itself—from the word bases of which it is composed or the constructions to which it is related—plays a major part in determining whether a taboo is imposed. The data are such that no decision model can be constructed. 14 Rather, one can simply list the principles that impinge on such a decision. These are:
• (a) The frequency with which the word base(s) in the name occur in other people's names. While names are sometimes changed in accordance with a taboo (see the example of Laubasi, below), changes of name by many people could not be expected. The very common bases in men's and women's names listed above are not tabooed, and some rather less common ones would also cause difficulty.
• (b) The frequency with which the associated linguistic form is used. The most common or culturally central noun bases would usually not be tabooed. However, as will be noted, this principle is modified by some others.
• (c) The availability of an already roughly synonymous form for replacement. If there are two common words in use, the tabooing of one will cause relatively little disruption.
• (d) The cultural contexts in which the potentially tabooed forms are used. Words used in contexts that are ‘secular’ (mola) are more likely to be tabooed than words that are used in “sacred” contexts (e.g. in ritual procedure).
• (e) Forms that are largely grammatical in function do not appear to be subject to tabooing even though some are technically bases, not particles (e.g. { ?ani-}, ‘with-’).
• (f) There appears to be a core of vocabulary that is “basic” in terms of resistance to tabooing, and this resistance is only indirectly based on frequency of use. These forms include numerals, body parts and substances, pronominals, terms of spatial orientation and some others (see below). They do not, generally, include natural objects and phenomena (and thus overlap but are not isomorphic with lexicostatistic item lists).
• (4) If an adalo becomes extremely sacred, this sacredness may override all other considerations.

- 165

• (5) Some extremely serious precipitating incident may have a similar overriding effect. This would be especially true if it were a drastic “swear” or if the linguistic usage were divined to be the cause of some very grave misfortune. It should be noted at this stage, however, that the linguistic factors already outlined, restrict the likelihood that use of the linguistic form will be put to the test of divination.
• (6) Finally, as will be seen, there are several degrees of severity with which the taboo can be imposed. Given a situation where one's adalo bears a name not well suited to tabooing, and seems to be causing trouble because of its usage in speech, one may choose to impose a non-stringent taboo that causes little difficulty to others.

To return to the centrally important question, in terms of Austronesian linguistics, of a possible taboo-resistant “core vocabulary”, the effects can now be examined of Kwaio word tabooing among the 100 items on Swadesh's revised lexicostatistic item list, grouped according to his semantic categories. Table 2 shows the frequency, in each category, with which (A) the Kwaio evidence from the Sinalagu area shows a form to have been tabooed for at least some people; (B) the form has been tabooed only in some particular construction; and (C) there is no evidence that the form has been tabooed.

Of the twelve tabooed words, two—/manu/, ‘bird’, and/mela/, ‘red’—are tabooed throughout the entire East Coast of Kwaio territory. It is - 166 worth noting that in Ray's list of cognate forms for “birds” in the South-east Solomons, the Kwaio form /laŋasi/ and the unrelated form from neighbouring Kwara ?ae are the only non-cognate forms. 15

In summary, if the name of an adalo whose sacredness is increasing is associated directly with a linguistic form not common in names or highly frequent in usage, for which some substitute form is available, and the form has no ritual associations and is not part of a “core” vocabulary, then chances of tabooing are very strong. As these factors are reversed, the probability of tabooing drops markedly.

When a linguistic form is tabooed, how is a replacement form derived?

The institutionalisation of word tabooing in Kwaio culture appears to produce and require the availability in the language of far more alternative monolexemic labels for the same cultural segregates, 16 i.e. “synonyms”, than any other language with which I am familiar. For “bamboo fire-tongs” there are five alternative lexemes. For “pudding bowl” there are four, for “leaf stopper for cooking bamboo” there are three. These almost random examples of alternative lexemes for cultural objects correctly suggest that when the decision to taboo a word base or construction is made, potential replacement forms are very often known and available. This, however, only pushes the problem a step further back. How are new forms produced in the first place? How is it that new forms subunia ‘sprout up’? In this section, I will outline those processes I have been able to isolate.

(1) Productive morphological devices generate from a semantically related word base, a semantically equivalent form. Examples:

- 167

(2) Semantic generalisation from a narrower form. Frequently this elevates to a higher level of a taxonomy, a form previously used only in a lower included contrast set. 17 Example:

(3) Semantic shift of a fairly radical sort (though based on some analogy, or a common derivative or metaphorical meaning). Examples:

(4) Use, in a compound form, of a substitute word base with some semantic overlap or resemblance. Examples:

(5) A loan word is borrowed from another Malaita language or dialect. Examples:

(6) A word is borrowed from Pidgin English.

(7) A new term is coined that modifies an existing form phonologically.

- 168

(8) A descriptive phrase or compound form is created. These may be semantically endocentric or exocentric. 18

(9) An alternative productive morphological device is used to create a semantically equivalent construction.

(10 A radically new term is coined.

A full compilation of tabooed Kwaio terms, the reasons for taboo, and replacement forms will shortly be published. This compilation will be keyed to the present analysis, so that the materials presented here for illustration can be viewed against the full range of Kwaio data.

The reader is asked to return, now, to thinking about the Kwaio social scene. Recall the descent groups, clustered around territories; the skeins of cognatic kinship linking outsiders to these descent groups; and the chains of cognatic descent linking groups to common ancient ancestors. Place in this landscape the tabooing of speech forms so they cannot be used by, or in the presence of, descendants of ancestors with the associated names. A host of new questions arises. How do people know what forms they cannot use, and with whom? How are the taboos first imposed? What restrictions, exactly, are placed on usage? What happens when a rule is violated? How, in short, are the abstract principles of word tabooing worked out in concrete social situations in such a way that communication is maintained?

1. Knowledge about Taboos. My first introduction to word tabooing came when, after five months in Kwaio, I set out to visit a distant settlement: “Don't use folia, xu ?i, or ele at Maaxona's village” I was warned. “They - 169 are taboo for his wife, Fele ?i, and if you use them you will have to give a pig right away as compensation.” (At this stage I was still far from fluent in the language, and I was advised that such an error on my part probably would be overlooked.) Such advice—about what words cannot be used and with whom—is the major mechanism whereby word tabooing functions in Kwaio society. Part of the socialisation process consists in learning about these restrictions. By the time a child is able to visit around the neighbourhood and interact fairly freely with adults, he or she has a cognitive map of the restricted usages in his neighbourhood, reinforced with parental reminders. But beyond one's own neighbourhood, one encounters taboos that are unfamiliar. Parents and other kin will know some of the taboos likely to be encountered, say, on a five-mile trip to a feast in the interior, and will provide warnings. But often these warnings are conveyed on the spot. As a party of visitors enters an unfamiliar settlement, they may be warned not to use the particular form.

2. Types of Taboos. It is the right of the descendants of the ancestor from whom the taboo derives to place restrictions on usage. As already noted, these are usually thought to be placed by the ancestor himself, through divination. There are, however, several general forms of restrictions that can be placed, with some local and individual variation within each category.

• (1) To prohibit usage of a form such that violation calls for compensation is to faa ?aia. The restriction may take somewhat variant forms, according to:
• (a) The physical setting where the restriction applies. Use of a form may be prohibited only in the settlement or dwelling house of a descendant, or may apply in any setting.
• (b) The type of interaction in which the restriction applies. The restriction may prohibit usage only in direct interaction with a descendant, or within hearing of a descendant, or in the same house, etc.
• (c) The consequences to the descendants if the taboo is violated. This generally places the descendants, and the ancestor, in a condition called siuŋa ?a, a state of ritual impurity in which the descendants are subject to punishment (illness, death, or other misfortune). Some expiation is thus required. But in addition, there are special restrictions in some cases—so that, e.g. the descendants must destroy their dwelling house in which the violation occurred, and move to a new clearing.
• (d) The compensation required from the offender. This may be shell money but more often is a pig (usually a piglet) or shell money to buy a pig. Sometimes compensation is required immediately, before the offender leaves; sometimes it can be paid later.
• (2) A lesser restriction can be applied such that, though people are not supposed to use the restricted form, no compensation is demanded if they do. Here the descendants themselves abstain from usage, and would have to expiate a violation. Outsiders would still

- 170

• be advised (though not warned so severely) not to use the form. Two terms used for this lesser restriction are loŋo?esia, ‘hear but let drop’, and malafofoŋo, ‘ignore’ (lit., “like-deaf”).
• (3) A partial restriction or warning can be given, especially among the descendants themselves, not to use the form in some social context or construction, or not to use it in a “swear”. To impose such a partial restriction is to fa ?aabu gulagula, ‘partially taboo’.
• (4) Finally, given a form that has been tabooed for descendants of an ancestor, some or all of these can partially withdraw the taboo. They may specify that the name is sacred when used in the “shrine”, and that the word cannot be used as a name (i.e. preceded by the masculine person-marker la, ‘mister’); but that the associated linguistic form can be used freely. Such a form is wala, ‘desacralised’, and the action is fa ?awalaa, ‘desacralise’. Again, the degree and form of desacralisation can vary. Sometimes this is merged with (2), in that descendants abstain from usage themselves but desacralise the term for others.

3. Initial Imposition of a Taboo. It should be recalled here that the ancestor himself usually imposes the taboo by punishing the usage (causing his descendant to be sick), then communicating the taboo in divination. A usual sequence leading to tabooing would approximate the following:

• (a) A descendant of an adalo whose name is not restricted gets sick. Some recent use or abuse of the name, either by a member of the group or a visitor, is divined to be the cause.
• (b) The “priest” of the kin group, or a close senior kinsman, attempts to pacify the adalo and rectify the situation without tabooing of the associated word.
• (c) If a pig must be given in expiation, the descendants whose close relative is sick provide it. A pig would not be demanded from an outsider unless a taboo had already been actively imposed.
• (d) Once it has been determined that the ancestor is insisting on the taboo—and that might take two or three illnesses or misfortunes—the descendants themselves begin to avoid the associated speech usages. Word quickly gets around the neighbourhood that these usages are to be avoided.
• (e) Only after sufficient time has elapsed for knowledge of the tabooed form to have spread (weeks or months) would the descendants claim compensation if an outsider used the form in the settlement. If the descendants continue to malafofono (‘ignore’) the accidental or careless use of the form by an outsider, they may simply be extending into a permanent arrangement, this transitional stage.

One should emphasise here that while many people and descent groups have forms they themselves do not use, the number of people known to make a major issue out of usage by an outsider—especially when the descendant imposing the taboo is himself visiting another place—is much smaller. When such a person appears, others will often be warned - 171 about this. “?ifi no ?ona genilabi ŋai ai”, ‘In that house there is a geni labi’, is a frequent warning. Geni labi is a denomination of shell money, the amount (about $3.00) one must usually pay for a piglet as compensation for violating a word taboo.

4. Word Tabooing in Space. Because a descent group shares an array of common ancestors, its members come to share a set of tabooed forms. Thus a tabooed form is associated with a place and a kin group, and this serves as a means for remembering a taboo and to whom it applies.

Two processes lead to the dispersal of a taboo beyond this group. The first process is descent group segmentation, through which genealogical segments hive off and eventually establish new territories. This is a slow, long-range process. More interesting to us, because it occurs so often, is the out-marriage of a girl from one descent group into another. When this takes place, she may introduce the taboos from her group among her affines. A girl who marries from a place with many strong taboos, or a place quite far away, can cause a considerable stir amoung her new neighbours. In the longer run, her children and grandchildren can start a new local cluster within which the taboos are observed. In the case of those ancestors common to large proportions of the population in a region, this process of out-marriage, proliferation, and out-marriage again has gone through several cycles.

Within a region of some ten to twenty square miles, most Kwaio adults have a great deal of information about descent groups and territories and the cognatic relationships of particular individuals. Beyond this range, they think in terms of larger geographical categories: in the case of the Sinalagu area I studied, the main ones were Lafea, the very broken and mountainous central part of the island, Ulu, the harbour and adjacent mountains to the north-west, and ?Olobuli, the harbour and adjacent mountains to the south-east. 19 In these peripheral areas, there are significant variations in vocabulary, many of which resulted from word tabooing, and some from “normal” processes of lexical diversification. Kwaio necessarily treat the word taboos, or possible taboos inferred from differential usage, from such a peripheral area, with less efficiency. That is, they can no longer differentiate what is taboo for whom. To the extent they modify their speech at all, they do so for everyone.

Until events of recent years, they seldom encountered such situations. Mortuary feasts, marriages, and markets would sometimes bring them into contact with people from afar, but interaction with them was always strained and rendered as safe as possible by magical protection against the malevolent spirits or evil magic they might bring. Adalo were too busy facing off against one another, in a game much like that of electronic countermeasures in modern strategic bombing, to worry much about word taboos.

However, the Maasina Rule (“Marching Rule”) movement of the late 1940s and subsequent developments have led to regular meetings of large groups from distant areas. In the weekly “Sub-District Committee” - 172 meetings of recent years, 20 many of which I observed, Kwaio living many miles apart frequently interacted. Often a delegation from Ulu would encounter a delegation from ?Olobuli. Kwaio themselves were not sure what to do about word taboos in such circumstances. If only two or three people were involved, others who knew what forms were taboo might warn, in a whisper, “you'd better not say—to so-and-so.” But when the delegations were large, they generally each used their own versions of Kwaio vocabulary quite freely. People from ?Olobuli would use the word folia, ‘buy’, in conversation; people from Ulu, to most of whom this is very taboo, would use the alternative term sugaa, which is rarely used at ?Olobuli and is taboo for some people there. People from Sinalagu, who use the two forms interchangeably, either did that or, in some cases, used the two differentially with people from each place.

5. Some Empirical Cases. From the many instances where word tabooing shows up in our field notes, two cases will add some empirical substance to the generalisations herein advanced. The first is interesting in that it shows a series of fairly extreme events that attracted much attention. The second is revealing as a normal and unspectacular sample of word tabooing as a regular ongoing process.

The first begins with a woman named Dauli ?i, whose agnatic descent group is Talanilau, inland from Sinalagu, but who grew up with her maternal kin at Nu ?u inland from ?Olobuli. When she married a man from another Sinalagu descent group, some years ago, three forms associated with her ancestors from Nu ?u were tabooed for use in her settlement: asuaa, ‘mid-morning’, (funu) funu, ‘fight’, and fee ?ai, ‘(single) tree’. She claimed compensation when they were used in her home, but not elsewhere.

Her oldest son Liŋeafaxa is now a teenager, and something of the Kwaio version of a juvenile delinquent. In early 1966, Liŋeafaxa was visiting the house of a kinsman near the coast when an elderly gentleman named Fa?atalo came into the house for a visit. Someone else then came in and asked Fa?atalo when he had left home. In his reply, he used the term asuaa (‘mid-morning’). Liŋeafaxa then announced that he had ‘sworn’ (taa) against anyone using asuaa or funu in his presence anywhere.

Fa?atalo was angry that such a swear had been made unannounced—since the form of Liŋeafaxa's swear had made the usage an offence against Fa?atalo's ancestors as well as Liŋeafaxa's. The litigation and “Subdistrict Committee” investigation dragged on for some weeks, and ranged from a general discussion of the legitimacy of word tabooing to investigation of the swear. The swear was interesting. Liŋeafaxa, it turned out, had been hanging around with another wild young man (and frequent accomplice in pig theft), Laitala. Laitala, though not yet 20, had succeeded his father as “priest” at an important shrine, and sacrificed several important fo ?ota, including one called Te ?ealo. The young men had been cooking taro. Liŋeafaxa jokingly referred to the taro corm he was cooking as “te ?e alo ŋai agu”, (‘my one taro’—alo is the Kwaio reflex of proto-Eastern Oceanic *ntalo, but is usually used not for taro in general or a single plant or corm but for a unit of 100 taro corms). Laitala angrily - 173 rebuked Liŋeafaxa for using this sacred fo?ota name in jest, and said “I don't use asuaa and funufunu with you.” Liŋeafaxa retorted with the very serious swear “A person uses asuaa and funufunu with me, and his ancestors crawl into his anus.” Nobody else was told about this swear.

Fa?atalo was rightfully incensed at having unwittingly offended his own ancestors, though he was prepared to give a pig as compensation to Dauli ?i's ancestors. The matter was more or less resolved, in the end, when Liŋeafaxa's father gave two pigs as compensation to Fa?atalo's ancestors and Fa?atalo gave one as compensation to Dauli ?i's ancestors.

An interesting aspect of the investigation was consultation with two important priests from ?Olobuli who were much more directly descended from Asuaa and Funufunu than Dauli ?i. They said that though these names were sacred when used in the “shrine”, and they avoided using them themselves, they did not claim compensation if someone else used them: “they are words other people use all the time, and we shouldn't try to stop them.” Confronted with this, Dauli ?i and her husband asserted that “these adalo were stronger with affines than with direct descendants”—an argument that was greeted with scepticism.

The second case was more routine. A girl named Dedei ?a, the daughter of an old priest and sorcerer at the edge of Lafea, married a young man named Seda from a descent group near the coast. Several forms were taboo for her father and other kin. However, it was not clear that these ancestors would enforce them in such a far-away place. But one by one, the major forms taboo for Dedei ?a's people were found in divination to be the cause of illnesses, or simply changed; thus an entire local cluster of kin began to avoid usage of several common forms not associated with their own ancestors. Here two examples will suffice. First, one of Dedei ?a's ancestors was named Laubasi (“seize bow”). Seda's FaFaBrSo was also named Laubasi. As a precaution, shortly after she came to live with them, his name was changed to Maiamae (a name he had used earlier in life). Second, the wife of one of Seda's kin visited them shortly after birth of their child (only a few months after the marriage). She used the term gwaea, ‘hold in the arms’, a form taboo for Dedei ?a's people. When the baby got sick several days later, this was divined to be the cause of illness; the “priest” sacrificed a pig in expiation, and members of the local group began to avoid use of the term.

Available information strongly suggests that the Malaitan languages comprise a single subgroup within a cluster of South-east Solomons Austronesian languages. This South-east Solomons cluster itself appears to comprise a subgroup of fairly closely related languages within the tentatively demarcated subgroup that has recently been referred to as “Eastern Oceanic”. These languages provide an ideal laboratory for intensive comparative study of “Melanesian” languages. In such a task, which I feel deserves a high priority in Oceanic linguistics, 21 full use will - 174 have to be made of the comparative method, analysis of shared innovations, and other refined techniques of subgrouping and reconstruction. Furthermore, they will have to be applied to entire lexicons and grammars, not simply short word lists. An important first step in this task will be reconstruction of a *Proto-Malaitan sound system and vocabulary. It is worth considering briefly the possible effects of word tabooing in this task.

The occurrence of word tabooing was noted, and its effects on linguistic diversity observed, in a missionary grammar of Kwara ?ae, 22 a language spoken in the zone adjacent to Kwaio:

In the task of subgrouping and reconstruction, such diversity need not in itself be a stumbling block. Shared morphological or grammatical innovations should still be effective markers of relationship, and in languages with as many cognate forms as these, lexical diversification greater than “normal” would not in itself upset reconstruction of sound correspondences and a lexicon. A good start in this direction has been made by Richard Levy and Nathan Smith.

However, one element of word tabooing, at least in Kwaio, could well lead to complications in use of the comparative method. This is the coining of replacement forms based on minor phonological modification of the original, tabooed, forms. Since such changes are apparently relatively unsystematic, they could cumulatively present to the analyst numbers of forms that appear to be cognate but do not fit regular patterns of sound correspondence.

Even where no active taboos are known, Kwaio often contains two or more interchangeable terms for the same common object, act, or phenomenon, that differ only in minor phonological variation (usually consonantal shift). Many of these probably arose initially through tabooing. But whatever their origin, they help to provide a reservoir of replacement forms when these are required. Thus, in Kwaio, there exist sets like the following:

- 175

Some patterns begin to emerge here, as in the alternation of /m/ and /n/. In other cases, such as the last two, the alternative form corresponds to the cognate form in a neighbouring language (Kwara ?ae, Kwarekwareo, Laŋalaŋa, ?Are ?are), and probably derives from borrowing. Such alternative forms are apparently common in other Malaitan languages for which lexical evidence is available (To ?abaita, Lau, Kwara ?ae and Sa ?a). If such sets were characteristic of *Proto-Malaitan, the original empirical existence of doublets and triplets would make the task of reconstruction more formidable, but would render more comprehensible some of the seeming phonological irregularities that show up in Malaitan lexicons.

The Kwaio data bring home forcefully how little is known about word tabooing in the Pacific. Without evidence that is lost forever we will never be able to make really effective comparative analyses. What is known suggests that Kwaio was by no means extreme in the development of tabooing. Sociologically, Kwaio stands toward the pole of fragmentation, lack of formal political organisations and fluidity. Kwaio word - 176 tabooing, with its localisation among small descent groups, clearly reflects its sociological pattern. With larger territorial-political groups and centralised authority elsewhere in the Pacific, word tabooing seems to have shifted correspondingly.

The writers have reflected already on some implications for the historical linguist of the Kwaio evidence. Here some cultural-historical speculation may be useful as well. Perhaps early “Eastern Oceanic” speakers in the Pacific practised some form of word tabooing; alternatively some element in these ancestral cultures may later have given rise to word tabooing in various parts of the Pacific. It would be useful to account not only for tabooing due to association with the names of ancestors, as in Kwaio, but also for quite different principles, such as tabooing due to association with the name of a relative in a tabooed kinship category. 23 Perhaps the early Oceanic form of tabooing was quite specific, but derivative forms have simply shifted as the foci of sacredness and proscription have changed and diverged. On the other hand, the original basis of Oceanic tabooing might lie in as general a principle as a strong association between a name and the person bearing the name. This, combined with the concept of tapu, could then lead to the separate evolution of word taboos in many places where extreme tapu became attributed, for varying reasons, to named individuals. But to speculate further along these lines would be premature until such evidence as can be gathered, in the field and library, is systematically put together.

This evidence will doubtless be fragmentary. Particularly in linguistics, word tabooing will probably be encounted with increasing frequency, but by indirection, as a hypothetical construct that accounts for statistics and reconstructions that do not accord with other evidence 24 or as an explanation for lexical differentiation. But to end with an optimistic note, it is possible that the recent acceleration of intensive social anthropology field work in the Pacific will turn up new evidence of word tabooing as an ongoing social and linguistic process, even in areas where it was thought such evidence could no longer be found.

Finally, evidence on Austronesian word tabooing can usefully be arrayed with evidence on similar processes in other language families and other parts of the world. Regularities in linguistic and sociological process will probably emerge from such study, and they, in turn, may well prove important to other questions of cultural theory. The recent suggestions by Lévi-Strauss 25 regarding naming as a central manifestation of La Pensée Sauvage open intriguing new vistas for exploration.

• CAPELL, Arthur, 1943. The Linguistic Position of South-Eastern Papua. Sydney Australasian Med. Publ. Co.
• — 1962. “Oceanic Linguistics Today”. Current Anthropology, 3(4):371-428.

- 177

• CODRINGTON, R. H., 1885. The Melanesian Languages. London, Oxford University Press.
• CONKLIN, H. C., 1962. “Lexicographic Treatment of Folk Taxonomies”, in F. Householder and S. Saporta (eds.), Problems in Lexicography, International Journal of American Linguistics Publication 28(2):119-41.

• DECK, N., 1933-34. “A Grammar of the Language Spoken by the Kwara ?ae People of Mala, British Solomon Islands”. Journal of the Polynesian Society, 42:33-48, 133-44, 241-56; 43:1-16, 85-100, 163-70, 246-57.
• DYEN, I., 1963. “Lexicostatistically Determined Borrowing and Taboo”. Language, 39:60-66.

• ELBERT, S. H., 1953. “Internal Relationships of Polynesian Languages and Dialects”. Southwestern Journal of Anthropology, 9:147-73. (esp. 161, 169.)
• EMORY, Kenneth, 1963. “East Polynesian Relationships”. Journal of the Polynesian Society, 72(2):78-100.

• FOX, C. E., 1924. Threshold of the Pacific. London, Kegan and Paul.

• GRACE, G. W., 1964. “The Linguistic Evidence”, in “Movement of the Malayo-Polynesians: 1500 B.C. to A.D. 500”. Current Anthropology, 5:359-406.
• GREEN, R., 1966. “Linguistic Subgrouping within Polynesia: The Implications for Prehistoric Settlement”. Journal of the Polynesian Society, 75:6-38.

• HAAS, M., 1951. “Interlingual Word Taboos”. American Anthropologist, 53:338-344.

• KEESING, R. M., 1967a. “Statistical Models and Decision Models of Social Structure: A Kwaio Case”. Ethnology, 6:1-16.
• — 1967b. “Christians and Pagans in Kwaio, Malaita”. Journal of the Polynesian Society, 76:82-100.
• — 1968. “Chiefs in a Chiefless Society: The Ideology of Modern Kwaio Politics”. Oceania, 38:276-280.
• — 1969a. “Kwaio Fosterage”. Forthcoming in American Anthropologist.
• — 1969b. “Shrines, Ancestors, and Cognatic Descent on Malaita”. Forthcoming in American Anthropologist.

• LÉVI-STRAUSS, C., 1962. The Savage Mind. (Translation of La Pensée Sauvage.) Chicago, University of Chicago Press.

• RAY, S., 1926. A Comparative Study of the Melanesian Island Languages. Cambridge University Press.

• SWADESH, M., 1955. “Towards Greater Accuracy in Lexicostatistic Dating”. International Journal of American Linguistics”, 21:121-37.