fish. An animal that inhabits the water. (Samuel Johnson, English dictionary, 1755)
The whale, the limpet, the tortoise, and the oyster … as men have been willing to give them all the name of fishes, it is wisest to conform. (Oliver Goldsmith, ca 1760)
This chapter investigates the semantic histories of five Proto Oceanic terms, *ikan, *pinaŋoda, *sisiq, *manuk and *manu-manuk, each of which is a candidate for the status of a high-level generic or ‘life form’ in the POc taxonomy of animals.1 In certain contemporary Oceanic languages reflexes of *ikan have a range of reference that encompasses fish and other large water-dwelling animals (cetaceans, dugongs, turtles and crocodiles), reflexes of *pinaŋoda encompass reef invertebrates, or sea animals other than fish and large water-dwelling animals, reflexes of *sisiq refer to gastropod (snail-like) shellfish, reflexes of *manuk refer to birds and bats, and reflexes of *manu-manuk to insects and other creepy-crawlies (spiders, centipedes, worms, etc.). However, in other Oceanic languages these etyma have reflexes with quite different ranges of reference. For example, in some cases the *manuk reflex includes not just birds and bats but all flying creatures, in others it includes all land-dwelling creatures or all creatures other than humans, in still others it includes humans, and there are some where it includes all objects. In many languages the sisiq reflex refers just to a small class of small gastropods (nerites and similar) while in others it includes all gastropods or all shellfish.
The aim here is to determine, as far as possible, the range of reference of these five terms in POc, their place in the taxonomy, and the nature of the changes that have occurred in daughter languages. The task of reconstructing their semantic histories raises some questions of more general interest: Are the variations in the meanings of major generics across daughter languages random or do they follow an implicational scale, a natural sequence of semantic extension or contraction? Why have such marked differences arisen in the semantic scope of these terms? Can some of the changes be attributed, for example, to differences in the range of animal life encountered by different communities?
Before we examine the Oceanic terms in question it is necessary to discuss some issues of method and conceptual framework. The semantics of folk taxonomies is a notoriously tricky domain, not one that one should enter without a map.
A first consideration is the nature of the analytic apparatus needed to make sense of folk taxonomies and the task of choosing between competing frameworks. This leads into questions about how different kinds of classificatory terms differ in their semantic structure and their role in the taxonomy. Another concern, of a more practical kind, relates to which kinds of semantic information, which components of lexical meaning it is feasible to deal with when trying to reconstruct the semantic history of classificatory terms. Finally, there is the problem that most definitions of classificatory terms in Oceanic dictionaries, our main sources of data, suffer from serious shortcomings.
A well-formed taxonomy is a system of semantic relations between lexical units that has the following properties:2 (i) Certain taxa stand in a class-inclusion or superordinate-hyponym relation. Thus, the statement X is a fox terrier unilaterally entails that X is a dog, which in turn entails X is an animal. That is, fox terrier is a hyponym of dog, which in turn is a hyponym of animal, (ii) Certain lexical units stand in a relation of incompatibility or contrast. Thus the statement X is a fox terrier entails X is not a labrador, alsatian or poodle, and the statement X is a dog entails X is not a cat, seal or bear, (iii) the hyponyms of a term X are natural kinds of X. Thus, it can be said that a fox terrier is a kind of dog and a dog is a kind of animal. A natural kind typically has many defining features (see §1.2.2) but all other features are subordinate to the one that defines its place in the taxonomy, the fact that it is a kind of something (a sparrow is a kind of bird, a dog is a kind of animal, a trout is a kind of fish).
The relation implied by ‘natural kind’ is not identical to ‘class-inclusion’ or ‘hyponymy’. Not all hyponyms are natural kinds. Some are ‘nominal kinds’, e.g. spinster, bachelor, mare, ewe, ram. Unlike natural kinds, contrasts between nominal kinds can be readily defined, approximately, in an analytic way in terms of just a few features, e.g. bachelor = man who has never married, spinster = woman who has never married, ewe = adult female sheep, ram = adult male sheep. It follows that the semantic difference between nominal hyponyms of the same set can be expressed analytically, as e.g. the difference between bachelor and spinster or ram and ewe. It is not a dominant feature of nominal terms that they are kinds of something. Although one can say a spinster is a woman, a mare is a horse, a ram is a sheep it is questionable whether it makes sense to say a spinster is a kind of woman, a mare is a kind of horse, or a ram is a kind of sheep.
Natural kinds cannot readily be defined analytically. For example, the difference between dog and bird cannot be reduced to one or two distinguishing features: the two differ in indefinitely many ways. It is not clear how one should answer a question like: What is the meaning of dog? Apart from mentioning that dogs are a kind of animal, should a definition say that they have four legs and a tail, are furry, are carnivorous, bark and howl, have a very keen sense of smell, are pack animals, that people have selectively bred them to create many breeds, and keep them as pets, and for hunting game, rounding up sheep and cattle, and guarding property, that they are not eaten (at least in English-speaking societies), and so on? What about the meaning of ‘bird’? Should a definition mention all and only the universally valid morphological attributes, that birds have feathers and two legs, two wings (some only vestigial) and a beak, and that they lay eggs? Which if any behavioural characteristics should it mention: that birds typically fly (not all do), that they make nests (not all do), that they have diverse calls, that many male and female pairs mate for life, that they are typically wild creatures but some species are domesticated, and so on?
Dictionary-makers typically baulk at the task of defining all the things that may distinguish natural kinds and are content to pick out just a few salient characteristics. For example, The Macquarie Dictionary defines dog, in its primary sense, as ‘a domesticated carnivore, Canis familiaris, bred in a great many varieties’. The definition of cat is almost identical, save for the Latin name: ‘a domesticated carnivore, Felis domestica (or F. catus), widely distributed in a number of breeds’. And dictionary definitions of terms of folk generic and life-form rank largely reflect scientific knowledge, not folk knowledge. Macquarie defines bird as ‘any of the Aves, a class of warm-blooded vertebrates, having a body more or less completely covered with feathers, and the forelimbs modified so as to form wings by which means most species fly’. As Anna Wierzbicka has repeatedly pointed out, such definitions are quite unsatisfactory as an account of folk knowledge about what these kinds of animals are.
Folk taxonomies must have at least two levels but can have many levels, as indicated by the sequence: berkshire > pig > animal > creature, or pinnacle floribunda > floribunda rose > rose > flower > plant. One way of distinguishing the levels is to refer to them as primary, secondary, tertiary, etc. or as first-order, second-order, third-order, etc.
Instead of speaking of levels in a folk taxonomy in this way, Brent Berlin and his associates propose that taxa at each level be given rank status in a way that broadly parallels the taxonomic ranks of Western biology (Berlin 1992, Berlin et al. 1973). Building on earlier work, Berlin (1992) also specifies some general principles of nomenclature. While the system of naming practices is distinct from the taxonomy, the two are systematically related so that we can often infer the taxonomic relations between particular taxa from the way they are named. A key nomenclatural distinction is between uninomials (or primary lexemes) and binomials (or secondary lexemes). Binomials are multiword names in which one element modifies the head noun, and where the head noun can stand alone as a generic name, e.g. rock crab, mud crab, hammerhead shark, white-pointer shark). A uninomial consists of a single morpheme, such as herring or perch, or a compound or phrase that is idiomatic, such as the type of hammerhead or leatherjacket, where no element can stand alone as a generic term (there is no category of fish called head or jacket), or the type of sea hare, sea cow and seahorse (the creatures so named are not a kind of hare, cow or horse).
The following is a summary of the system of rank distinctions in folk taxonomies, and generalisations about nomenclature given in Berlin (1992).
Most taxa belong to one of three ranks: life form, folk generic or folk specific.
Life-form. A life-form is a taxon that (i) distinguishes a distinctive morphotype which (ii) includes many (sometimes hundreds) of lower order taxa which share the characteristic morphology and ecological adaptation of the type, (iii) is not included in any other taxon other than kingdom, and (iv) is named by a uninomial. Examples of English life-form taxa are fish, bird, snake, tree and flower. The nature of categories that rank as life-forms may vary across speech communities, usually in ways connected with regional differences in flora and fauna and in the ways of life of communities. For example, for people in the central Pacific, shellfish are a much more numerous and more important category than snakes, or even birds, but the reverse is the case in many regions of the world.
Folk generic. A folk generic (or folk genus) is a ‘natural’ category in several senses, one perceptual, the others linguistic. First, the members of this category are usually marked off from non-members by multiple characters of morphology and behaviour or ecological adaptation that will be evident to any close observer. Second, unlike folk species, the category is named by a uninomial rather than a binomial. Examples of English folk generics are rose, oak, beech, blackbird, pig, dog, frog, toad, mullet, trout, whale, crocodile. Third, the folk generic is the usual way of referring to a particular plant or animal if its identity is known. Depending on various factors, a folk genus may correspond to a single species in biological taxonomies, to a number of species or a genus, or to a number of genera or families. Many folk generics are subtaxa of life forms but Berlin recognises a category of unaffiliated generics, first-order taxa that do not belong to any life-form but are not themselves life-forms.
Folk species. A folk genus sometimes further divides into folk species (or folk specifics), usually just a few taxa which contrast in a limited number of features with other members of the generic. Such specifics are usually the lowest-level taxa. Berlin (1992) holds that folk species names usually consist of the generic name plus a modifier, e.g. red mullet vs grey mullet, or mako shark vs hammerhead shark. Bulmer (1970, 1974), however, finds that a fair number of animal folk species, among the Kalam people of the New Guinea highlands, are named by primary lexemes (uninomials).
Three other ranks are sometimes distinguished in folk taxonomies.
Kingdom. A single primary (first-order) taxon - one that subsumes all life-form taxa and other taxa in the relevant semantic domain or class. Names for taxa of kingdom rank will be uninomials. For some English speakers, whose perceptions have been influenced by scientific taxonomies, plant and animal qualify as kingdom rank taxa, subsuming various life forms. In vernacular speech, traditionally, these two terms have narrower meanings. However, in most societies, folk classifications of living things lack overt taxa of kingdom rank.
Intermediate. A taxon which (i) is a member of a life-form (and so is never a first-order taxon) and (ii) comprises a small number of folk generics that show marked similarities to each other and often correspond to botanical or zoological families. If an intermediate taxon has a name, it will be a uninomial. However, Berlin says that intermediate taxa are generally not named, i.e. they are usually ‘covert taxa’, whose unity is indicated by other means than sharing a class name. For some English speakers, horses, zebras and donkeys form such a covert category, as do hawks, falcons and eagles.
Varietal. Under certain conditions, folk species can divide into lower order taxa, contrasting in subtle ways with other varieties. Distinctions between varietal taxa mainly apply to domesticated plants and animals - and so may be motivated mainly by cultural considerations. The names are usually compounds.
For larger animals and some kinds of plants, at the level of the genus and often at the level of the species, the distinctions made by communities living close to nature tend to be similar to those made by Western biologists. The similarities at the genus/species levels are usually explained as follows. These categories are marked by many distinctive features and are as obvious to any close observer of nature as they are to highly trained biologists.
At higher levels the methods and motives underpinning folk and scientific classifications tend to diverge and therefore so do the categories. The higher you go, either in a folk taxonomy or a scientific taxonomy, the less the taxa are likely to conform to ‘natural kinds’. Western biologists aim to capture the evolutionary relationships of organisms, a concern peculiar to science. Ordinary people, on the other hand, impose higher order categories on cultural grounds, be they pragmatic or cosmological. Sometimes the grounds for grouping different kinds of organisms are broad likenesses in form and behaviour. For example, it is fairly common to find a taxon that subsumes both birds and bats, or both fish and whales. In other cases, social factors (e.g. ritual restrictions and taboos) and technological or economic factors (e.g. techniques used to obtain or process foods) peculiar to a society influence the grouping.
Wierzbicka (1984, 1996) has advanced methodological objections to certain claims made by anthropologists, linguists and psychologists about the nature and membership of categories in folk taxonomies. She points out that people’s intuitions about category membership can’t reliably be established by asking them simple questions like ‘Is X a kind of Y?’ or by asking them to sort cards with pictures. Instead you need to find out how people talk spontaneously about such matters and to apply a variety of linguistic tests designed to bring out people’s tacit knowledge.3
Wierzbicka has argued persuasively for a distinction between true generic taxonomic terms and collectives, a distinction that will prove to be relevant to our discussion of Oceanic generics. For her, collectives are not taxonomic terms, at least not in the same sense as true generics. She discusses the properties of English terms like animal, bird, fish and flower, in popular speech, that clearly mark them as being ‘super taxonomic categories’, or ‘life forms’. Each of these terms stands for kinds of things that are alike, a category based on perceptual similarities, something you can draw a picture of schematically, something you can count individually. She contrasts taxonomic generics with collectives like livestock, game (animals), fruit, grass, vegetables, cutlery and furniture. Unlike true generics, these don’t stand for kinds of things that are alike. A collective is a heterogeneous collection of kinds that share a common function or some other feature(s). Wierzbicka says that you can draw a generalised picture of a fish or bird but not a generalised picture of fruit or cutlery. One can have an exchange such as Look at that colourful fish there? - Which one? - The one with yellow stripes, but Look at that big livestock there? - Which one? - The brown and white one, is bizarre. You can’t count individual objects using a collective: *I saw one fruit/livestock/furniture/cutlery. I saw one fruit is only acceptable in a more abstract sense: I saw one kind of fruit, not in the sense I saw one individual fruit (e.g. one apple)
This distinction between natural kinds and collectives will prove to be relevant in our discussion of Oceanic generics.
For any generic term, the assumption that there is a single correct definition of its semantic range that can be uncovered by careful research is itself questionable. The fact is that members of a speech community often do not agree completely as to the scope of generics. The kind of variation is of the kind predicted by prototype theory (Rosch 1973, 1975a,b). Generally people agree on the membership of typical members of a class but disagree when it comes to marginal members. The definitions given in dictionaries tend to understate the amount of variation within the speech community.
Consider, for example, English speakers’ use of the term fish. The Shorter Oxford English Dictionary (SOED) makes a sharp distinction between popular and scientific definitions of this term.
Fish 1. In [popular] language, any animal living exclusively in water, including cetaceans, crustaceans, molluscs, etc. In scientific language any vertebrate animal provided with gills throughout life, and cold-blooded: the limbs, if present being modified into fins.
Johnson’s definition and Goldsmith’s observations, quoted at the head of this chapter, show the prevalence of this popular usage in 18th century England. However, a mere two-way contrast between popular and scientific usage is surely an over-simplification. A survey of several hundred Australian informants shows that while everyone agrees on including prototypical fish there is much variation in regard to what other creatures count as fish. Some English speakers exclude eels, sharks, rays, seahorses and certain other ‘atypical’ fish from this category while others include whales and dolphins and even crayfish and jellyfish.
A similar problem arises with ‘shellfish’. Most English dictionaries give definitions similar to the following, taken from the SOED.
shellfish any animal living in water whose outer covering is a shell, whether testaceous, as an oyster, or crustaceous, as a crab.
The problem here is that for many native speakers of English, especially those who are not from the USA, shellfish refers only to edible molluscs with shells and does not include crabs, lobsters and prawns.
A further difficulty is that directly questioning informants about classification - ‘Are butterflies a kind of insect?’, ‘Are eels a kind of fish?’ - is likely to yield unreliable answers. While these are legitimate questions there are other, more reliable ways of getting at class membership. One such consists of spontaneous utterances that contain information about class membership: ‘What a beautiful insect! Is it a beetle?’, ‘Kiwis are strange-looking birds’. Lexical evidence is another valuable source, e.g. is it possible, or usual for a referent to be named by a true binomial expression, consisting of a generic with an identifying modifier: sulphur-crested cockatoo, rock cod, oak tree, white pine, grey wolf? Unacceptability or unacceptability of particular logical relations is yet another useful test. Thus, I saw cats and dogs is an acceptable conjunction in English but I saw animals and dogs is not. All English speakers would probably accept They netted ten fish and a turtle, while rejecting They netted ten fish, including two turtles. But They netted ten fish and two eels or They netted ten fish and two sharks might yield diverse judgments.
It is important to recognize that speakers often distinguish two or more senses of generic terms, some with narrower reference, others with wider reference. English animal is an example. There are some contexts, in popular speech, where animal denotes a category that includes, roughly, mammals, four-legged reptiles and frogs but not birds, fish or invertebrates. There are other contexts where it includes all creatures other than humans. And there are still other contexts where it includes humans (‘Humans are different from the rest of the animal kingdom’).
Although I see the merit of Wierzbicka’s argument that definitions of a folk category should try to capture as fully as possible folk perceptions and knowledge about that category, the data to hand for Oceanic languages do not allow us to pursue such an objective (see §1.2.7). My concern with definitions in this chapter is more modest: to find the referential range of a category, the kinds of animals that speakers of a language include in it, where ‘kinds’ can be defined in biological terms (whether this be simply the species, genus, family, etc. recognised by biologists, or features of morphology or behaviour that allow us to identify biological categories, such as a folk taxon that consists of birds and bats).
This objective might seem straightforward enough. But it is not, for reasons outlined in the following section.
For anyone wishing to compare taxonomies of flora and fauna in Oceanic languages a serious problem is the poor quality of definitions provided in dictionaries and other sources. Ideally, the taxonomies of a language should be inferable from information given in the dictionary.4
The definitions we find often betray the fact that the definer has not carefully investigated the range of reference of a term or variations in the way different speakers use that term.
For example, many Oceanic dictionaries give a bare gloss, ‘general term for fish’, or simply ‘fish’, without mentioning which kinds of creatures are included. Such a simple gloss is unhelpful for two reasons. First, the meaning of the English term ‘fish’ is itself imprecise: as noted above, it does not have the same range for all native speakers of English. Second, if a dictionary simply defines term X as ‘fish’ we have no assurance that the dictionary-maker has systematically investigated the semantic range of X.
A slightly different sort of gap is exemplified by an entry in Dixon’s (1970) dictionary of Dobuan, a Papuan Tip language, which glosses iyana as ‘fish, including alligators’. Leaving aside the detail that alligators but not crocodiles are found in New Guinea waters, we are left wondering what else besides fish and crocodiles counts as iyana. Lithgow’s dictionary of Muyuw, another Papuan Tip language, is a bit more informative, offering yin ‘fish, turtles, crocodiles, but not shellfish’. But again, we are not told whether, say, eels, cetaceans, dugongs, and octopus count as fish.
In the case of major generics, different sources for the same language often offer conflicting definitions. The standard dictionary of Motu (Lister-Turner and Clark 1954a) glosses gwarume simply as ‘fish’. However, Nigel Oram, who looked into the matter in some detail, obtained a different and narrower definition of this term from his informants, namely ‘fish, excluding sharks, rays, eels, and also excluding cephalopods’ (Oram, field notes). Capell’s (1941) dictionary of Standard Fijian (based on the Bauan dialect of Eastern Fijian) glosses ika as ‘fish’, overlooking the fact that the Fijian term has a broader sense that includes turtles, dolphins and whales (P. Geraghty, pers. comm.).
Fox’s (1955) dictionary of Gela (Gela) defines iga as a class name for any kind of sea creature: ‘sea creature: fish, mollusc, crayfish, whale, squid, sea anemone, etc.’. However, in his study of Gela fishing the zoologist Simon Foale (1998) distinguishes two uses of iga: ‘bony and cartiligenousfish’ vs ‘all fish, whales, dolphins, dugongs, turtles, crocodiles’. Neither definition exactly matches Fox’s. (Foale’s definition of the second, extended sense does not include squid, (shelled) molluscs, crayfish or anemones.) Do such differences reflect variation among informants or less than exhaustive enquiry on the part of the authors, or a bit of both?
A further problem is that dictionaries, quite understandably, seldom provide a corpus of examples showing how a term is used in everyday discourse. Such a corpus would, among other things, help us decide whether a particular term is a true taxonomic generic rather than a collective.
With all these sobering considerations in mind, let us turn to the several POc terms and their reflexes.
POc *ikan, roughly glossable as ‘fish’, is an extremely well-supported reconstruction, which continues PAn *Sikan ‘fish’ and PMP *ikan ‘fish’ (Blust 1995). What is less clear is what kinds of creatures counted as *ikan for POc speakers, and/or whether *ikan had two distinct senses, one with narrower and the other with broader ranges of reference.
| (1) | typical fish |
| (2) | 1 + sharks and rays |
| (3) | 2 + cetaceans (whales, dolphins) and dugongs |
| (4) | 3 + eels |
| (5) | 4 + turtles |
| (6) | 5 + crocodiles |
| (7) | 6 + cephalopods (octopus, squid, etc.) |
| (8) | 7 + decapod crustaceans (crabs, crayfish, prawns and their relatives) |
| (9) | 8 + other aquatic invertebrates (molluscs with shells, sea hares, nudibranchs, echinoderms, sea urchins, sea cucumbers, jellyfish, etc.) |
All Oceanic languages have a generic term that can be translated roughly by English ‘fish’ and typically this term is a reflex of POc *ikan. However, the scope of this generic varies considerably from language to language. A first comparison of definitions across the group indicates that if ‘typical fish’ are taken as the core members of the class, extensions to include other kinds of creatures can be ordered along an implicational scale roughly as in Table 8.1, bearing in mind that while most creatures of types (1-9) occur throughout much of the Pacific, certain animals, such as dugongs and crocodiles, have more restricted ranges.
Thus, the hypothesis is that if a language includes only one other group of animals in the generic for ‘fish’ it will generally be sharks and rays and the next group to be included will usually be cetaceans together with dugongs (if present) and after that will come eels, then turtles, and so on. Let us refer to those generics whose semantic scope is restricted to (1) as ‘grade 1’, those restricted to (2) as ‘grade 2’, and so on.
If this hypothesis were correct it would mean that the definition of Dobuan iyana as ‘fish, including alligators’ entails that sharks, rays, eels, cetaceans, turtles and cephalopods are also subsumed under iyana. In the same vein, the definition of Muyuw yin as ‘fish, turtles, crocodiles, but not shellfish’ implies that eels, cetaceans, dugongs, and octopus count as yin.
The following discussion compares dictionary definitions of reflexes of POc *ikan for various groups of Oceanic languages. These comparisons are generally followed by tables that make the dictionary definitions directly comparable by reformulating them in terms of the presence or absence of particular classes of referents.
Relatively detailed definitions of reflexes of *ikan are available for many Central Pacific languages.
| Fijian and Rotuman | ||
| Rotuman | iʔa | ‘fish, turtles, whales, alligators, etc but not crabs or shell fish, etc’ |
| Wayan (W. Fijian) | ika | (1) ‘typical fish, true fish. This category includes sharks, rays and eels. syn. ika dū “true fish’”, (2) ‘fish, including cetaceans, turtles and, for many speakers, octopus and squid’ |
| ika-bula | ‘turtle (generic)’ | |
| ika dū | ‘the green turtle, Chelonia mydas’ (lit. ‘true fish’) | |
| Bauan | ika | ‘fish, including eels, plus turtles, cetaceans, cephalopods’ (P. Geraghty pers. comm.) |
| Polynesian | ||
| Tongan | ika | ‘fish, turtles, whales but not eels, cuttlefish or jellyfish’ |
| ika mui | ‘male turtle’ | |
| ika matu | ‘tender part of turtle near front paddles’ | |
| Niuean | ika | ‘fish, whales and dolphins’ |
| Niutoputapu | ika | ‘fish (including eels), turtles, cetaceans, and cephalopods’ |
| Tahitian | iʔa | ‘fish, porpoise, whale, turtle, octopus’ |
| Hawaiian | iʔa | ‘fish or any marine animal, e.g. eel, oyster, crab, whale’ |
| Ifira-Mele | ika | ‘fish, whale, porpoise but not turtles, eels or octopus’ |
| Tikopia | ika | ‘general category with primary reference to fish but including allied creatures, e.g. turtle, cetaceans’, [turtles are classified both as ika and as paka ‘crabs and other shelled creatures’] |
| Rennellese | ika | ‘fish, turtle’ |
| Samoan | iʔa | ‘fish plus turtles and whales (and perhaps others)’ (Milner), ‘the general name for fishes (except the bonito), molluscs and crustaceans’ (Pratt) |
| iʔa-mānu | ‘whale’ (lit. ‘floating fish’) (Milner) | |
| -iʔa-sā_ | ‘turtle’ (lit. ‘sacred fish’) (Milner, Pratt) | |
| iʔa-vai | (1) ‘eel’; (2) ‘fresh-water fish’ (Pratt) | |
| Tokelauan | ika | ‘fish, turtle’ |
| ika-fafine, ika-fua | ‘female turtle’ | |
| ika-tagā | ‘male turtle’ |
The only Central Pacific dictionary that distinguishes between a narrow sense of the *ikan reflex, including only ‘true fish’ and a broader sense, including other large marine creatures, is Wayan Fijian. The Wayan dictionary contains these example sentences: Na siihiana ika ni qwara, ei tarn tia ika dū. ‘The octopus is a fish living in holes; it is not a true fish. Na ika dū ei vakalaulau. ‘True fish have tails.’
The semantic range exhibited by these definitions is summarised in Table 8.2. Here and in later tables the following conventions are used to indicate whether or not the class of referents listed at the head of each column is included in the meaning: ‘+’ = included, ‘-’ = excluded, ‘(+)’ = there is disagreement among speakers, and/or certain members of the class are included, others excluded, and ‘?’ indicates that sources do not say whether or not the referent is included.
| typical fish | shark, ray | whale, dolphin | eel | turtle | octopus, squid | crustacean | shelled mollusc | Grade | |
|---|---|---|---|---|---|---|---|---|---|
| Rotuman | + | + | + | + | + | ? | - | - | 5 or 6 |
| Fijian | |||||||||
| Bauan | + | + | + | + | + | + | - | - | 6 |
| Wayan 1. | + | + | + | + | + | (+) | - | - | 5 or 6 |
| Wayan 2. | + | + | - | + | - | - | - | - | 2,4 |
| Polynesian | |||||||||
| Tongan | + | + | + | - | + | - | - | - | 4 |
| Niuean | + | + | + | ? | - | - | - | - | 3 or 4 |
| Niuatopu. | + | + | + | + | + | + | - | - | 6 |
| Tahitian | + | + | + | + | + | + | - | - | 6 |
| Hawaiian | + | + | + | + | ? | + | + | + | 8 |
| Samoan | + | + | + | ? | ? | - | - | - | 3-5 |
| Tokelauan | + | + | ? | ? | + | - | - | - | 5 |
| Tikopia | + | + | + | - | + | ? | - | - | 3 |
| Rennellese | + | + | ? | ? | + | - | - | - | 5 |
| Ifira-Mele | + | + | + | - | - | - | - | - | 3 |
Note: Crocodiles or dugongs are not present in Triangle Polynesian or Fijian waters, although they are known to some Polynesian Outlier communities in Melanesia.
| typical fish | shark, ray | whale, dolphin | eel | turtle | octopus, squid | crustacean | shelled mollusc | Grade | |
|---|---|---|---|---|---|---|---|---|---|
| PPn | + | + | + | ? | + | (+) | - | - | 5 or 6 |
| PCP | + | + | + | + | + | (+) | - | - | 5 or 6 |
The Polynesian languages cited here agree fairly well in their range of referents. Leaving aside cases where the definition is uninformative on this point, reflexes of *ikan always include not only typical fish but also sharks, rays and cetaceans. Most (six out of eight) also include turtles. Most (all but one) exclude crusteaceans and molluscs with shells. The main area of disagreement concerns eels (three languages include them, three exclude them) and cephalopods (three include them, six exclude them). The most reasonable historical inference is that Proto Polynesian (PPn) *ika embraced sharks and rays, cetaceans and turtles. The status of eels and cephalopods in the PPn taxonomy is less certain.
Rotuman and the Fijian languages agree in including eels and cephalopods, as well as sharks and rays, cetaceans and turtles, and in excluding crustaceans and molluscs with shells. Table 8.3 proposes likely semantic profdes for PPn and Proto Central Pacific.
Details follow of reflexes of *ikan in five Micronesian languages for which quite informative definitions are available. The semantic range exhibited by these definitions is summarised in Table 8.4.
| Kiribati | ika | ‘fish: all swimming marine creatures including turtles, dugong, whales and porpoises, but not squid or octopus’ |
| Marshallese | ek | ‘fish’ (Abo et al. 1976) However, in the English-Marshallese finder list, under ‘fish’, the authors list terms for ‘dolphin’, ‘giant octopus’, and ‘jellyfish’ as well as ‘shark’, ‘eel’, but not ‘whale’ or ‘turtle’ |
| Carolinian | īx | ‘generic term for fish’ (Jackson and Marck 1991). However, in the English-Carolinian finder list, under ‘fish’, the authors list terms for ‘porpoise’, ‘whale’, ‘octopus’, ‘sea urchin’, ‘starfish’, ‘sea cucumber’, as well as ‘eel’, ‘shark’ and ‘ray’, though not ‘turtle’ |
| Woleaian | ixar | ‘fish’ (Sohn and Tawerilmang 1976). However, in the English-Woleaian finder list, under ‘fish’, the authors list terms for ‘dolphin’, whale’ as well as ‘eel’, ‘shark’ and ‘ray’, though not ‘turtle’, ‘octopus’ or ‘squid’. |
| Puluwatese | yīk, yikan | ‘fish, porpoises and whales but not squid’ |
| typical fish | shark, ray | whale, dolphin | eel | turtle | octopus, squid | crustacean | shelled mollusc | Grade | |
|---|---|---|---|---|---|---|---|---|---|
| Kiribati | + | + | + | + | + | - | - | - | 5 |
| Marshallese | + | + | + | + | ? | + | - | - | 4, 6 |
| Carolinian | + | + | + | + | - | + | - | - | 4+ |
| Woleaian | + | + | + | + | ? | ? | - | - | 4+ |
| Puluwatese | + | + | + | ? | ? | - | - | - | 3+ |
All the languages sampled typically exclude turtles, molluscs and crustaceans. Some definitions include eels and cetaceans but others are indeterminate.
Informative definitions are available for a few Oceanic languages of the Solomon Islands. Among the following, all belong to the SE Solomonic group except Marovo, which is NW Solomonic. For each of Lau and Gela, two definitions are given, from different sources.
| Arosi | iʔa | ‘fish, including sharks, rays; also porpoises, and possibly other marine animals’ |
| Lau | iʔa | ‘fish, any sea creature’ (Fox 1955) |
| Lau | iʔa | (1) ‘typical fish, sharks and rays’; (2) ‘extended to include large marine animals such as porpoises, whales, dugongs’ (Akimichi) |
| To’aba’ita | iʔa | ‘fish, sharks, rays, dolphins, whales, dugongs, turtles’ (F. Lichtenberk pers. comm.) |
| Gela | iɣa | ‘sea creature: fish, mollusc, crayfish, whale, squid, sea anemone, etc.’ (Fox) |
| Gela | iɣa | (1) ‘bony and cartiligenous fish’; (2) ‘all fish, whales, dolphins, dugongs, turtles, crocodiles’ (Foale) |
| Marovo | iɣana | ‘generic for all free-swimming creatures, e.g. fish, eels, stingrays, dolphins, whales, squid, jellyfish’ |
It can be seen that the witnesses largely agree in including sharks, rays, eels, turtles, cephalopods and crocodiles, virtually all free swimming aquatic creatures. Fox’s definitions for Lau and Gela are extreme in including all marine organisms.
| typical fish | shark, ray | whale, dolphin | eel | turtle | crocodile | octopus, squid | crustacean | shelled mollusc | Grade | |
|---|---|---|---|---|---|---|---|---|---|---|
| Arosi | + | + | + | ? | ? | ? | ? | ? | ? | 3 |
| Lau (Fox) | + | + | + | + | + | + | + | + | + | 9 |
| Lau (Akm) | + | + | + | + | ? | ? | ? | ? | ? | 4+ |
| To’aba’ita | + | + | + | + | + | - | - | - | - | 5 |
| Gela (Fox) | + | + | + | + | + | + | + | + | + | 9 |
| Gela (Foa.) | + | + | + | + | + | + | ? | ? | ? | 6+ |
| Marovo | + | + | + | + | + | + | + | - | + | 7 |
There are few informative definitions for reflexes of *ikan in Vanuatu languages.
| Mota | iya | ‘fish’ |
| NE Ambae | iya | ‘typical fishes, rays and dolphins, but not sharks and whales (which are grouped together), eels or cephalopods’ (C. Hyslop, pers. comm.) |
| Raga | ixe | ‘fish, including whales and porpoises but not eels, seasnakes, octopus, squid, cuttlefish, jellyfish or balolo worms’ |
| cf. also: | ||
| Anejom | namu | ‘fish, including sharks, eels, cetaceans, turtle, octopus and squid, and crustaceans but not molluscs with shells’ (J. Lynch pers. comm.) |
| Paamese | mesau | (1) ‘fish (syn. ai)’ (2) ‘any sea dweller incl. turtles, dolphins, shellfish’ |
| Tamambo | manji tarusa | ‘fish and shellfish’ (manji ‘animal, including fish and birds’, tarusa ‘sea’) |
Leaving aside the under-informative Mota and Tamambo definitions, we get Table 8.6.
| typical fish | shark, ray | whale, dolphin | eel | turtle | octopus, squid | crustacean | shelled mollusc | Grade | |
|---|---|---|---|---|---|---|---|---|---|
| NE Ambae | + | (+) | + | - | - | - | - | - | 3 |
| Raga | + | + | + | - | ? | - | - | - | 3 |
| Anejom | + | + | + | + | + | + | + | - | 7 |
| Paamese | + | + | + | + | + | + | + | + | 8 |
Definitions are given for *ikan reflexes in four Admiralties languages. The initial n- they show is a reduced form of the POc article *na which has fused with the following noun.
| Titan | nii | (1) ‘fish’, (2) ‘fish, porpoise, dugong, turtle, crocodile’ |
| Lenkau | nik | ‘fish, excluding porpoise, whale, turtle, crocodile’ |
| Penchal | nii | (1) ‘fish’, (2) ‘fish, turtle, porpoise, whale, octopus’ |
| Loniu | ni | ‘general term for fish’ |
Discarding the Loniu definition as under-informative, we get the comparative breakdown in Table 8.7. It can be seen that the definitions for Titan and Penchal agree in including sharks, rays, cetaceans and turtles, but disagree over cephalopods.
| typical fish | shark, ray | whale, dolphin | eel | turtle | crocodile | octopus, squid | crustacean | shelled mollusc | Grade | |
|---|---|---|---|---|---|---|---|---|---|---|
| Titan | + | + | + | (+) | + | + | _ | — | _ | 6 |
| Lenkau | + | + | - | (+) | - | - | _ | _ | _ | 2,4 |
| Penchal | + | + | + | (+) | + | - | + | - | - | 5,7 |
Of the Western Oceanic reflexes that follow, only the definitions for Kilivila, Muyuw and Motu are reasonably informative. The latter are all Papuan Tip languages. It can be seen from Table 8.8 that two of the three Papuan Tip witnesses agree in including sharks, rays, eels and crocodiles but disagree over cetaceans and turtles.
| North New Guinea | ||
| Manam | ika | ‘fish’ |
| Yabem | i | ‘fish’ |
| Papuan Tip | ||
| Dobuan | iyana | ‘fish, including alligators’ |
| Kilivila | yena | ‘fish, including sharks, rays and eels, but excluding aquatic mammals, turtles or cephalopods’ |
| Molima | iyana | ‘fish’ |
| Muyuw | yin | ‘fish, turtles, crocodiles, but not shellfish’ |
| cf. also: | ||
| Motu | gʷarume | ‘fish, excluding sharks, rays, eels and also cephalopods’ (N. Oram field notes) |
| Meso-Melanesian | ||
| Tolai | en | ‘fish’ |
| Nakanai | ia | ‘fish’ |
| Roviana | iɣana | ’generic name for |
| typical fish | shark, ray | whale, dolphin | eel | turtle | crocodile | octopus, squid | crustacean | shelled mollusc | Grade | |
|---|---|---|---|---|---|---|---|---|---|---|
| Kilivila | + | + | - | + | - | ? | - | - | - | 2,4 |
| Muyuw | + | + | ? | + | + | + | ? | ? | - | 6 |
| Motu | + | - | - | - | - | - | - | - | - | 1 |
There is reason to think that POc *ikan was polysemous, having a narrower sense, roughly ‘typical fish’ and a broader sense, which included various other aquatic animals. Among our sources, such a distinction is made in just a few contemporary languages (chiefly Wayan Fijian, Gela and Lau of SE Solomonic, and Titan and Penchal of the Admiralties) but it is also tacitly present in the descriptions of certain Micronesian languages.
As to a broader sense, if only the more detailed dictionary definitions are considered, we find that in a high proportion of languages across different major subgroups the reflex of *ikan refers to sharks and rays, eels, cetaceans and turtles, and also to dugongs and crocodiles in places where these animals are present. I conclude that this was also the case in POc; that is, *ikan had at least grade 5 scope. Turtles were ‘fish’ in POc. The status of octopus and squid is less clear. Other kinds of water-dwelling invertebrates, such as molluscs in shells and crustaceans, have much weaker claims for inclusion. The balance of the evidence suggests that while free swimming creatures with bones were *ikan, invertebrates that crawl, burrow or stick to rocks probably weren’t. However, in a few languages (Lau, Gela, Marovo) the scope of *ikan has been widened to include all free swimming sea creatures. In Lau and Gela it may been extended to all marine (and presumably freshwater) creatures (grade 7), although different sources are in conflict on this point. We must allow that in some Oceanic languages the semantic range of *ikan has reduced to grade 4, 3 or 2.
The data loosely support the implicational scale given in Table 1. Some simplifications may be in order. For example, among languages spoken where both crocodiles and turtles are common, no language is known to include turtles but to exclude crocodiles or vice versa. Thus, categories 6 and 7 can perhaps be merged. Only one language in the sample, Anejom, includes crustaceans but excludes shelled molluscs, so an argument can be made for merging categories 8 and 9. However, we have little information about the status of other marine invertebrates included in category 9, such as jellyfish and sea hares.
Why do speakers of various Oceanic languages (like people in many other places) place certain other kinds of animals in the same class as typical fish? Most likely the reasons were diverse. Some non-fish creatures look and behave rather like fish, e.g. whales, dolphins and dugongs to some extent resemble fish in the way they swim and in being completely aquatic. But what about turtles? Apart from being free-swimming bony animals that spend a lot of time at sea, turtles do not look or behave like fish. What they have in common with certain fish, perhaps, is that everywhere in the Pacific turtles are prized game, caught at sea. That leaves crocodiles, which resemble sharks in being fearsome creatures but otherwise share little in common with fish or turtles other than being free-swimming bony creatures.
At any rate, the two senses of *ikan reflexes pose a problem for Berlin’s system of taxonomic ranks. If the ‘typical fish’ category counts as a life form, what is the status of extended categories such as ‘all water creatures’ or ‘all creatures that are free-swimming’ or ‘typical fish, eels, cetaceans, turtles and crocodiles’? Clearly, categories that contain creatures of such diverse appearance do not fit well with Berlin’s requirement (§1.2.3) that members of the same life form taxon share a distinctive morphotype or with Wierzbicka’s analogous requirement (§1.2.4) that one should be able to draw a generalized picture of the type. In order to convince Wierzbicka that in a language L, the extended sense of *ikan reflexes, encompassing turtles, crocodiles, dolphins, etc., is a true major generic in the same way that the narrower, sense applying to ‘typical fish’ is, we would have to show that speakers of L use the extended generic not just as a collective but also to refer to individual turtles, crocodiles and dolphins, e.g. ‘See that ikan there. It’s a turtle.’ and ‘We caught ten ikan and two of them were turtles.’ Unfortunately, this kind of data is almost completely lacking for the languages in our sample.
More light might be shed on the scope of POc *ikan if it could be shown to contrast with another generic having a complementary meaning. There is a candidate. Clark (1991) provides an insightful reconstruction of the history of a POc term, *pinaŋoda, that referred to a class of sea creatures that evidently stood in contrast to *ikan. The following discussion is based largely on his account.
Dictionaries of some Polynesian languages explicitly mention a two-way contrast between a reflex of PPn *ika, denoting fish and other large aquatic vertebrates such as dolphins and turtles and a term, reflecting PPn *fiŋota, that denotes forms of sea life other than *ika, but especially molluscs and other invertebrates found on the reef. Compare, for example, the Tongan and Samoan entries in the following comparison:
| Polynesian | ||
|---|---|---|
| Tongan | fiŋota | ‘sea creature other than ika (fish, turtles, etc.) especially shellfish, but also crustaceans, cephalopods, jellyfish, eels, sea-snakes, sea cucumbers, starfish, seaweed’ |
| Samoan | fiiŋota | ‘general term for sea animals except fish, cetaceans and turtles, especially molluscs, crustaceans, echinoderms and sea cucumbers’ |
| Niuatoputapu | finjota | ‘shelled marine animals that creep or crawl: Crustacea, molluscs and echinoderms’ |
| East Futuna | finjota | ‘coquillage’ [molluscs with external shells] |
| Tokelauan | fiiŋota | ‘general term for shellfish’ |
| Ifira-Mele | finjota | ‘shellfish’ [molluscs with external shells] |
| Takuu | fiinota | ‘various types of shellfish and seafoods gathered by women’ |
Cognates outside Polynesian are scarce, except in Papuan Tip languages. In the following, g for POc *ŋ in Dobuan and Molima is unexpected (but regular in Kilivila).
| Papuan Tip | ||
|---|---|---|
| Dobuan | igoda | ‘shellfish’ |
| igo-igoda | ‘any inedible shellfish’ | |
| lo-igoda | ‘to gather shellfish’ | |
| Molima | igoda | ‘to collect shellfish’ |
| Kilivila | vigoda | ‘shellfish (live, or shell only)’ |
A problem with many of these definitions is that for some English speakers the term ‘shellfish’ refers only to molluscs with shells while for others it also includes crustaceans. There is reason to think that the authors of the dictionary definitions cited here generally had in the mind the former, narrower definition.5
Clark argues convincingly that PPn *fiŋota and Papuan Tip vigoda, igoda are reductions of POc *p-in-aŋoda, the nominalization of a verb, *paŋoda, by means of a well-attested nominalising infix *-in-6 The primary sense of *paŋoda was ‘to gather seafood (other than fish) on the reef’ with the additional sense ‘to fish, catch fish’. This verb is widely attested in Oceanic. Selected examples are given below. In some witnesses, especially in Western Oceanic, POc *ŋ is unexpectedly reflected as a stop, g.
| Meso-Melanesian | ||
| Teop | yagana | ‘to fish, (N) fishing’ |
| SE Solomonic | ||
| Bugotu | vaŋoda | ‘hunt for shellfish on the reef’ |
| Gela | vaŋoda | (1) ‘collect food on the reef’; (2) ‘anything of the reef: starfish, crab, octopus, sea egg’, (3) ‘shell of mollusk or crab’, (4) ‘gatherer of reef food’, (5) ‘fisher’ |
| Micronesian | ||
| Marshallese | yaged | ‘go fishing’ |
| North and Central Vanuatu | ||
| Mota | vaŋona | (v) ‘catch fish with a line, get shellfish’, (v) ‘things got to eat on the reef or by fishing, things for a relish to go with vegetable food’ |
| Namakir | ba-vaŋot | ‘to fish’ |
| Nguna | pa-vaŋoda | ‘to fish, to gather shells’ |
| Paamese | vaŋor | ‘gather shellfish on reef’ |
| Anejom | (a)haŋej | ‘forage on the reef’ |
| Central Pacific | ||
| Rotuman | haŋota | ‘to fish, fishing excursion, shoal of certain fish’ |
| Polynesian | ||
| Tongan | faaŋota | ‘to fish or search for any kind oifiŋotd’ |
| Niuean | faŋota | ‘gather shellfish on the reef’ |
| Samoan | faaŋota | ‘to fish; fishing’ |
| Rennellese | haŋota | ‘to fish or gather shells, especially by women on the reef’ |
A few languages retain a reflex of *pinaŋoda in which the infix is transparently preserved.
| Meso-Melanesian | ||
| Tolai | winagonoi | ‘generic name for sea-shells’ |
| North and Central Vanuatu | ||
| Nguna | vinagoda | ‘shell, shellfish’ |
Can we conclude that POc *pinaŋoda was a major generic, a life-form of the same rank as *ikan? Almost certainly not. Clark concludes that *pinaŋoda was probably not a biological taxon in the strict sense but a collective, a term for a diverse category of things, namely ‘seafood gathered on the reef’ in which shellfish, i.e. molluscs, were the focal members. Unfortunately, sources for contemporary Oceanic languages seldom provide discourse examples to show how speakers talk about reflexes of *pinaŋoda. Whenever examples are given they show the noun being used to refer to a non-specific entity, an undifferentiated collectivity. Given the heterogeneity of the types of edibles subsumed under this term it would be surprising if it behaves, grammatically, as anything other than a collective.
Clark remarks that in some contemporary Oceanic languages *pinaŋoda reflexes appear to have become a biotaxon, specifically a generic for shellfish. This has happened in at least two Polynesian languages (East Futunan and Tokelauan), one Vanuatu language (Nguna), and two Papuan Tip languages (Dobuan and Kilivila). However, as sources do not give examples showing how speakers of these languages use the shellfish generic we cannot be sure whether is a genuine life-form taxon or a collective.
Oceanic speech communities that have access to reefs and beaches typically distinguish many kinds of molluscs. Speakers of Wayan Fijian, for example, have approximately 140 names for kinds of bivalve and gastropod shellfish, of which about 100 are gastropods. This is a domain, then, where one might expect to find one or more high-order taxonomic categories in Oceanic languages, a point made by Clark (1981).
We have already seen that a reflex of *pinaŋoda ‘seafood gathered on the reef’ has developed into a generic for shellfish in Dobuan, Kilivila, Tolai and Nguna, as well as in certain Polynesian languages. Clark (1981:268, 1991:79) also cites, as Polynesian examples of a generic that encompassed both bivalves and gastropods, Marquesan (Ua Pou dialect) puukava, Tahitian puu and puupuu, Rennellese hatu ʔatua, and East Futuna and Tokelauan figota.
However, no widespread cognate set has been found that consistently serves as a general name for shellfish. In fact, most Oceanic languages do not have such a taxon. It is more common to find a generic that refers to edible gastropods, or at least to a large class of edible gastropods, and a generic that refers to a large class of bivalves. But, once again, hardly any such terms have widespread cognates.
Perhaps the strongest candidate for a POc generic in this domain is *sisiq, which may have had variants *siseq and/or *sisuq7 Reflexes of *sisiq vary greatly in range of reference across Oceanic languages.
| North New Guinea | ||
| Takia | sisei | ‘generic term for all varieties of sea shells’ |
| Papuan Tip | ||
| Dobuan | sisi-yara | ‘a bivalve, Asaphis diehloma’ |
| sisi-we | ‘common whelk’ | |
| Molima | sisi-ʔalo | ’a brown bivalve, small and edible, found in tidal flats |
| Motu | dudu | ‘name of a shellfish’ (*z > u unexpected) |
| Sinaugoro | dudu | ‘cone shell’ (*z > u unexpected) |
| Meso-Melanesian | ||
| Nakanai | e-sisi | ‘a shellfish’ |
| Roviana | sise | ‘general name for a number of small shells’ |
| Southeast Solomonic | ||
| ’Are’are | sisi | ‘a shellfish’ |
| sisi- | ‘first element in compound names for certain molluscs, e.g. sisi-nanara, sisi-nikau’ | |
| Arosi | sisi- | ‘first element in compound names for certain molluscs, e.g. sisi-apir ‘limpet’, sisi-nahu ‘small gastropod sp. in streams’ |
| To’aba’ita | sisi- | ’first element in compound names for a few mollusk species, e.g. sisi-afufu, sisi-ʔeiria |
| Lau | sisi-afufu | ‘mollusc sp.’ |
| Gela | hihi-vuhi | ‘freshwater snail sp., Nerita brevispina’ |
| North and Central Vanuatu | ||
| Tolomako | sise, sisi | ‘shellfish’ (scope uncertain) |
| Mota | sese | ’a smooth shell, turris; in which the gatou ’hermit crab lives |
| NE Ambae | hihe | ‘Nerita sp.’ |
| Raga | hiha | ‘sea snail, winkle’ |
| Uripiv | (na)ses | ‘periwinkle (?)’ |
| SE Ambrym | ses | ‘shellfish of family Neritidae’ |
| Nakanamanga | siisa | ‘snail, incl. nerites (sisa karo)’ |
| sususa | ‘maculated moon snail, periwinkles, mitres’ | |
| S Efate | ses | ‘nerite’ |
| New Caledonia | ||
| Cemuhi | ti | ‘generic term for gastropods’ |
| Fwai | thik | ‘trochus, but used generically for gastropods’ |
| Nemi | thik | ‘trochus, but used generically for gastropods’ |
| Fijian and Rotuman | ||
| Bauan | siði | (1) ‘Trochus spp.,’ (2) ‘first element in various compounds naming kinds of gastropods and a few bivalves’ |
| Wayan | ðiði | (1) ‘Generic for (i) gastropod molluscs (Gastropoda) with external shells, i.e. univalve or snail-like shellfish and (ii) chitons (Amphineura) and rock oysters, which resemble typical gastropods in that they stick to rocks and/or that the shell is not hinged.’ (2) ‘In a limited range of contexts, a generic for all molluscs with external shells’ |
| Rotuman | sisi | ‘edible shell-fish, the periwinkle’ |
| Polynesian | ||
| Niuean | hihi | ‘snail (generic), usu. first element in compound names for certain gastropods’, e.g. hihi Afelika, ‘African snail’, hihi tea ‘a seasnail’ |
| Hawaiian | hiihii/wai | ‘grainy snail, Neritina, in both fresh and brackish water, eaten’ |
| Samoan | sisi | ‘name given to small snails in general’ (Milner); ‘fresh water molluscs’ (Pratt)89 |
| sisi-tai | ‘saltwater winkles’ (Pratt) | |
| sisi-vai | ‘freshwater molluscs’ (Pratt) | |
| sisi-vao | ‘landsnails’ | |
| Futunan | sisi | ‘small shellfish spp., Neritidae and Naticidae’ |
| Uvean | hihi | ‘kind of shellfish, Janthinidae and Neritidae’ |
| Tikopia | sisi | ‘marine and land snails incl. Melampus and Nerita spp.’ |
| Mere-Ifira | sisi | ‘small snail (nerites and others of similar shape)’ |
| Rennellese | sisi | ‘edible nerite shells’ |
| Nukuoro | sisi | ‘mollusc, Neritidae spp.’ |
The semantic range of these terms may be arranged in a scale as shown in Table 8.9. The semantic ranges of reflexes of *sisiq in better documented Oceanic languages are shown in Table 8.10.
Has *sisiq expanded its range of reference from domains (1) or (2), or did it already have wider scope in POc? Was it already polysemous in POc, with both a broad and a narrow application?10
About 50 families of gastropods or snail-like molluscs, comprising hundreds of genera, are found in the Indo-Pacific region. In almost all languages where the dictionary definitions specify particular families or genera, these refer to nerites (members of the family Neritidae) and to other fairly similar groups of small edible marine gastropods (grade 3). In many Oceanic-speaking communities such small shellfish are regularly collected in large quantities and often they supply a more significant part of the protein consumption than any other gastropods. This, then, may have been its primary sense in POc.
| (1) | one or two species of edible molluscs, sometimes a gastropod, sometimes a bivalve, sometimes unidentified |
| (2) | Nerites (Neritidae), small snail-like shells, with large body whorls) |
| (3) | two or more of several several groups of edible small globular gastropods (Neritidae and also Naticidae or moon shells (small, globular and smooth), Littorinidae or periwinkles (small, heavy shells, with only a few whorls, globular, spiral or turbinate in shape) |
| (4) | all edible gastropods |
| (5) | all molluscs with shells |
| a few spp. of molluscs | nerites | small globular gastropods: nerites, moon shells, periwinkles, etc. | all edible gastropods | all edible molluscs | Grade | |
|---|---|---|---|---|---|---|
| Takia | + | + | + | + | + | 5 |
| Wayan | + | + | + | + | (+) | 4-5 |
| Cemuhi | + | + | + | + | - | 4 |
| Samoan | + | + | + | + | - | 4 |
| ’Are’are | + | + | + | + | - | 4 |
| Fwai | + | + | + | + | - | 4 |
| Futuna | + | + | + | - | - | 3 |
| E Uvean | + | + | + | - | - | 3 |
| Tikopia | + | + | + | - | - | 3 |
| Niuean | + | + | ? | - | - | 3 ? |
| Rennellese | + | + | - | - | - | 2 |
| Nukuoro | + | + | - | - | - | 2 |
| NE Ambae | + | + | - | - | - | 2 |
| S Efate | + | + | - | - | - | 2 |
| SE Ambrym | + | - | - | - | - | 1 |
| Gela | + | - | - | - | - | 1 |
| Hawaiian | + | - | - | - | - | 1 |
POc *sisiq certainly referred to a class of snail-like (gastropod) molluscs but, given the very spotty data, it is difficult to determine the range of reference of this term. However, the fact that *sisiq reflexes are widely preserved as the first element in binominals, even in languages where a reflex of *sisiq no longer occurs by itself, indicates that it was a generic of some sort in Oceanic, participating in a number of binomials.
There is some evidence that *sisiq also served as a high-order generic, perhaps denoting all gastropods, or all edible gastropods. It was noted above that reflexes of *sisiq denoting edible marine gastropods in general occur in at least one language in each of the following subgroups: North New Guinea, SE Solomonic, New Caledonian, Fijian and Polynesian. It is likely that there are other cases - our sample is very limited. However, in only two languages in the sample do *sisiq reflexes also serve as a generic for shelled molluscs in general, including bivalves. Thus, there is no good evidence at present for attributing this last, much wider range of reference to POc *sisiq.
POc *manuk is a well attested form whose semantic range certainly included birds in general, but whose reflexes often have a broader scope. The implicational scale in Table 8.11, going from the most restricted domain to the most inclusive, predicts fairly well the kinds of variations in range seen among *manuk reflexes.
| (1) | birds and bats |
| (2) | birds, bats and flying insects |
| (3) | all land animals |
| (4) | all animals except humans |
| (5) | all beings |
| (6) | all objects |
| Tongan | manu | ‘animal, esp. bird but applied also to quadrupeds, reptiles, insects, and other land creatures but not to fish, shell-fish, and other sea creatures’. (Also manu-puna ‘bird or bat’; lit. ‘flying manti’) |
| Niuean | manu | ‘any living creature except humans and fish’ (Also manu-lele ‘bird and bat but not insects; lit. ‘flying manu’) |
| Maori | manu | ‘bird’ |
| Tahitian | manu | (1) ‘bird + flying insects’ (2) ‘land animal as opposed to iʔa’ |
| manu-manu | ‘insects’ [probably in vernacular sense, including arthropods] | |
| Hawaiian | manu | ‘bird, any winged creature’ |
| Rarotongan | manu | ‘creature, living thing, usually but not always bird; often used figuratively for a human being’ (also manu-rere, manu-reva ‘bird’, manu vaevaeʔā lit. ‘four legged animal’) and manumanu ‘insects, small flies’) |
| Samoan | manu | (1) ‘bird’ (also manu-lele ‘bird’) (Milner) (2) ‘animal, beast’ (Milner) (3) ‘any living thing’ (Utugagana) |
| Tokelauan | manu | (1) ‘animal, beast’ (2) ‘bird’ |
| Futunan | manu | ‘animal in general’; manu-lele ‘bird’ (lit. ‘flying animal’) |
| Uvea | manu | ‘animals in general’; manu-lele ‘bird’ |
| Ifira-Mele | manu | ‘bird, bat (and possibly butterflies)’ |
| Tikopia | manu | ‘terrestrial creatures: mammals, birds, arthropods, but ex cluding marine creatures’ |
| Rennellese | manu | ‘land fauna except human beings, turtles and fish (incl. birds and flying insects)’ (also manu-gege ‘birds and flying insects’) |
| Nukuoro | manu | ‘living things in general’ |
The semantic range of these terms may be arranged in a scale as shown in Table 8.12. This distribution of meanings suggests that PPn *manu had a narrow sense ‘birds and bats’ (with synonym *manu-lele), and an extended sense, roughly ‘all land animals other than humans’. Among the languages cited above, Niuean and Tongan (Tongic), Rarotongan and Tahitian (Eastern Polynesian), Rennellese, Samoan, Tokelauan and Tikopian (non-Eastern Nuclear Polynesian) have an extended sense.
In Futunan, Uvean, (and occasionally) Rarotongan, manu can be extended to all creatures other than humans and in Nukuoro that seems to be the standard meaning. This more extended use has a restricted distribution in contemporary Polynesian languages and was probably not present, or was rare, in PPn.
| 1 | 2 | 3 | 4 | 5 | 6 | ||||
|---|---|---|---|---|---|---|---|---|---|
| birds and bats | birds, flying insects | all land animals | all animals except humans | all beings except humans | all beings including humans | all things | Grade | ||
| Nukuoro | + | + | + | + | + | - | 5 | ||
| Rarotongan | + | + | + | + | + | - | 5 | ||
| Tikopia | + | + | + | + | - | - | 4 | ||
| Tongan | + | + | + | - | - | - | 3 | ||
| Niuean | + | + | + | - | - | - | 3 | ||
| Samoan | 1. | + | - | - | - | - | - | 1 | |
| 2. | + | ? | + | ? | - | - | 3 | ||
| Tokelauan | 1. | + | - | - | - | - | - | 1 | |
| 2. | + | ? | + | ? | - | - | 3 | ||
| Tahitian | 1. | + | + | - | - | - | - | 2 | |
| 2. | + | + | + | - | - | - | 3 | ||
| Rennellese | + | + | + | - | - | - | 3 | ||
| Ifira-Mele | + | + | (+) | - | - | - | 3 ? | ||
| Tahitian | + | + | + | - | - | - | 3 | ||
| Maori | + | - | - | - | - | - | 1 |
Cecil Brown, in a paper on the development of life form taxa in Polynesian languages, expresses doubts about the two-sense reconstruction. He concludes (Brown 1981:94) that ‘bird’ probably ‘constituted the primary, if not the only zoological reference’ of PPn *manu. If I understand him correctly, Brown prefers to interpret the fact that Tongan, Niuean, Samoan, Tikopia, Rennellese, Rarotongan, etc. agree in having a second, wider sense of *manu, as the result of semantic extensions made independently several times in the history of the Polynesian languages. He points out the importance of ecological factors. While birds must have been very prominent creatures in all the islands of the Pacific, most of the islands settled by Polynesians would have had very few land animals other than birds and small invertebrates. All had lizards, a very few had snakes, otherwise there were only those animals brought by humans: rats, dogs and (sometimes) pigs. When, in the post-European contact era, new quadrupeds (such as cattle, horses and goats) arrived it would have been natural to extend the reflex of *manu to these.
However, it seems to me more likely that *manu was already polysemous in PPn. The evidence indicating that PPn had a synonym for *manu in its narrower sense, namely the compound *manu-lele ‘birds, bats, flying insects’ (lit. ‘flying manu’) is consistent with the view that *manu had two senses in PPn. The evidence from Fijian and Rotuman (see below) also supports a polysemous reconstruction. Be that as it may, there can be little doubt that in the Polynesian languages ecological factors have motivated both extensions and contractions in the semantic range of reflexes of POc *manuk.
| 1 | 2 | 3 | 4 | 5 | 6 | ||||
|---|---|---|---|---|---|---|---|---|---|
| birds and bats | birds, flying insects | all land animals | all animals except humans | all beings except humans | all beings including humans | all things | Grade | ||
| Rotuman | + | + | + | + | - | - | 4 | ||
| Bauan | 1. | + | + | + | ? | - | - | 3 | |
| 2. | + | - | - | - | - | - | 1 | ||
| Wayan | 1. | + | + | + | + | - | - | 4 | |
| 2. | + | + | + | - | - | - | 3 |
Rotuman and the Fijian languages reflect only a reduplicated form *manu-manu. In Rotuman this term denotes land animals other than humans. In Wayan it denotes all creatures. In Bauan its narrower sense refers to birds and its wider sense includes all creatures. In all three languages a compound consisting of a reflex of *manu-manu plus a verb/adjective meaning ‘to fly, flying’ is used to distinguish birds and other flying animals from terrestrial animals.
| Rotuman | mɔn-mɔnu | ‘bird, insect, animal, including all land and air creatures but no sea creature’ (also mɔnmɔnu ferfere ‘bird, flying insect’) |
| Bauan | manu-manu | (1) ‘bird’ (2) ‘sometimes also animal, insect’ (also manumanu vuka ‘bird’ (lit. ‘flying animal)’ |
| Wayan | manu-manu | (1) ‘animal, creature, being, normally excluding human beings’ (also manumanu ðavu ‘bird’, lit. ‘flying animal’) (2) ‘sometimes used to denote land animals in contrast to ma rine animals, such as ika. fish and fish-like creatures’ |
| Kiribati | man | ‘animal (including fish), beast, insect, etc.’; ‘numeral classifier for people and animals’ |
| Puluwatese | maan | ‘creature, being, animal, insect’ |
| Carolinian | maal, maan | ‘generic term for any living creature except humans’, |
| -mal | ‘counting classifier for people and animals’ | |
| Chuukese | maan | ‘living creature of land or air, other than human’ |
| Woleaian | mal | (1) ‘animal, bird, animate object’; (2) ‘thing, inanimate object’ (these are listed as homonyms by Sohn and Tawerilmang 1976; note also mal ‘numeral classifier for animate beings including humans’) |
| Mokilese | mahn | ‘mammal, fruit bat, lizard, monitor, worm, bug, bird’ |
| Marshallese | men | ‘thing’ |
| men-in-mour | ‘animal, living thing’ (from men ‘thing’, -in- ‘of’, mour ‘life’) |
| 1 | 2 | 3 | 4 | 5 | 6 | |||
|---|---|---|---|---|---|---|---|---|
| birds and bats | birds, flying insects | all land animals | all animals except humans | all beings except humans | all beings including humans | all things | Grade | |
| Marshallese | + | + | + | + | + | + | 6 | |
| Woleiai | + | + | + | + | + | + | 6 | |
| Kiribati | + | + | + | + | + | - | 5 | |
| Puluwatese | + | + | + | + | + | - | 5 | |
| Chuukese | + | + | + | + | - | - | 4 | |
| Mokilese | + | + | + | + | - | - | 4 |
It can be seen that in this sample of Nuclear Micronesian languages most reflect POc *manuk with range of reference much broader than ‘bird’. In Kiribati and Mokilese the meaning is approximately ‘land animals other than humans’. In Carolinian and Woleaian the reflex refers to all animals other than humans (domains 1-4 on the implicational scale) but in Woleaian it can also denote any object (domains 1-6). In Puluwatese the nominal reflex of *manuk refers to living things in general, as is also the case for the numeral classifier reflex in most Nuclear Micronesian languages. In Marshallese and Woleaian the reflex of *manuk has become the term for ‘thing’.
It is likely that the primary sense of the PMic noun *man was at least as broad as ‘land creatures other than humans’ and that this form had already become a classifier for humans and animals.
Outside of the Nuclear Micronesian and Central Pacific groups, reflexes of *manuk generally have the narrower meaning ‘creature with wings: bird, bat, flying insect’. However, the paucity of good dictionaries for Oceanic languages in most regions of Melanesia limits our confidence in some of the glosses. The semantic range of reflexes of POc *manuk in some Oceanic languages other than Nuclear Micronesian and Central Pacific is shown in Table 8.15.
| Vanuatu | ||
| NE Ambae | manu | ‘birds and flying foxes’ |
| Mota | manu | ‘flying creature: bird, bat, beetle’ |
| Paamese | a-man | ‘bird’ |
| Sye | menoɣ | ‘bird’ |
| Lenakel | menuk | ‘bird’ |
| Anejom | in-man | ‘bird (includes flying-foxes but not insects)’ |
| New Caledonia | ||
| Cemuhi | meni | ‘birds (generic) and bats’ |
| Fwai | manik | ‘bird’ |
| Nelemwa | mālîc | ‘bird (generic)’ |
| Southeast Solomonic | ||
| Arosi | manu | ‘bird’ |
| ’Are’are | mānu | ‘bird, insect, anything that flies’ |
| Kwaio | manu | ‘bird’ |
| Lau | manu | ‘any creature that flies’ |
| Sa’a | menu | ‘bird, insect’ |
| Gela | manu | ‘flying creature, bird, insect’ |
| Meso-Melanesian | ||
| Nakanai | malu | ‘bird, insect (general term)’ |
| Tangga | man | ‘generic term for all species of bird’ |
| Simbo | maunuɣu | (1) ‘creature, being, including humans, birds, marine mammals, and possibly fish’ (2) ‘thing’ |
| cf. also | ||
| Marovo | oloko | (1) ‘all flying creatures including large and small bats’ (2) ‘animal, including all mobile land animals’ (Hviding 2005:175) |
| North New Guinea | ||
| Kairiru | mʸan | ‘bird’ |
| Manam | maŋ | ‘bird, fowl’ |
| Mumeng (Patep) | menaʔ | ‘bird, including cassowary’ |
| Yabem | mɔʔ | ‘birds (generic) and flying insects’ |
| Papuan Tip | ||
| Motu | manu | ‘birds and all things that fly including flying foxes’ (also manu-manu ‘flying insects’ (N. Oram field notes) |
| Molima | manu | ‘bird’ (also manu-manuwa ‘insects’) |
| Dobuan | manua | ‘birds (generic)’ (also manu-manua ‘insects and small creatures’) |
| Kilivila | mauna | ‘animal, bird, insect’ |
| Muyuw | man | ‘land creature, bird, animal, insect’ |
| Admiralties | ||
| Lou | mɔn-mɔn | ‘bird’ |
| Mussau | menua | ‘hawk, eagle’ |
| 1 | 2 | 3 | 4 | 5 | 6 | |||
|---|---|---|---|---|---|---|---|---|
| birds and bats | birds, flying insects | all land animals | all animals except humans | all beings except humans | all beings including humans | all things | Grade | |
| Simbo | + | + | + | + | + | + | 6 | |
| Muyuw | + | + | + | - | - | - | 3 | |
| Cemuhi | + | + | - | - | - | - | 2 | |
| Gela | + | + | - | - | - | - | 2 | |
| Lau | + | + | - | - | - | - | 2 | |
| Mota | + | + | - | - | - | - | 2 | |
| Motu | + | + | - | - | - | - | 2 | |
| Nakanai | + | + | - | - | - | - | 2 | |
| Manam | + | - | - | - | - | - | 1 | |
| Yabem | + | + | - | - | - | - | 2 | |
| Dobuan | + | - | - | - | - | - | 1 | |
| NE Ambae | + | - | - | - | - | - | 1 | |
| Arosi | + | - | - | - | - | - | 1 | |
| Nelemwa | + | - | - | - | - | - | 1 | |
| Lou | + | - | - | - | - | - | 1 |
What does the total range of comparisons tell us about the semantic range of POc *manuk? The variation among reflexes is striking. *manuk is most widely attested in the meaning ‘winged or flying creatures: birds, bats and flying insects’, i.e. grade 2 on the scale. However, the grade 3 meaning, encompassing all land animals, is common in Central Pacific and Nuclear Micronesian, and is attested in one Papuan Tip witness, Muyuw. Grade 4, 5 and 6 meanings occur but with more restricted distributions.
In terms of accounting for the distribution of meanings across subgroups, the most economical interpretation of the full range of comparative evidence in POc is to assume that *manuk referred only to winged or flying creatures: birds, bats and flying insects. This meaning dominates in subgroups other than Micronesian and Central Pacific. The wider meanings displayed by some daughter languages are best regarded as post-POc innovations.
This conclusion is consistent with the fact that outside of Oceanic, cognates of *manuk generally refer to chickens or to birds in general but not to quadrupeds or marine animals. Blust (2002) attributes to PMP *manuk the meaning ‘chicken (domestic fowl)’ on the basis of widespread reflexes in Western and Central MP languages. He also reconstructs a reduplicated form PMP *manu-manuk meaning ‘bird’, widely reflected with this sense in WMP languages. Blust argues that PAn had an unrelated term for ‘bird’ namely *ayam, which was replaced in this function by *manu-manuk in PMP. In PMP *ayam came to mean ‘domestic animal (of any sort)’.
Given that birds are the dominant indigenous land fauna in the island groups of Remote Oceania, and that there were no mammals native to this region, other than bats, it is not surprising that the generic term for birds and bats should be extended to other land animals (see discussion in §5.2.1). Here it seems is a case where a change in the meaning of a POc biotaxon can be attributed to changes in the range of animal life encountered by speakers of Oceanic daughter languages as they moved into new environments.
Osmond (this volume, ch.7) observes that reflexes of a form *manu-manuk, meaning approximately ‘insects and other small invertebrates’ (corresponding to the English vernacular category ‘insects’)11 are found in three widely separated subgroups of Oceanic: Papuan Tip, Southeast Solomonic and Polynesian. In certain languages in each of these groups the reflex of *manu-manuk stands beside and contrasts with a reflex of *manuk, meaning approximately ‘birds and other flying creatures’. Reduplications of *manuk, but with rather different meanings, are also found in Lou (Admiralties), Fijian and Rotuman (see §5.2).
In the definitions cited below from dictionary sources it is likely that the English gloss ‘insect’ is generally used in its vernacular sense.12
| Papuan Tip | ||
| Dobuan | manu-manua | ‘insects and small creatures’ |
| Motu | manu-manu | ‘beetles, insects’ |
| Sinaugoro | manu-manu | ‘insects’ |
| SE Solomonic | ||
| Bugotu | mau-manu | ‘insect’ |
| Gela | mau-manu | 1. ‘insect’ 2. ‘all living creatures: pigs, fish, birds, etc.’ |
| Polynesian | ||
| Tahitian | manu-manu | ‘insects’ |
| Rarotongan | manu-manu | ‘insects, small flies’ |
| Marquesan | manu-manu | ‘an insect’(pollex) |
| Mangarevan | manu-manu | ‘insect seen clinging to branches and fermentin: substances’ (pollex) |
| Pukapukan | manu-manu | ‘little insects’ |
| Tuamotuan | manu-manu | ‘bug, insect’ (pollex) |
| Nukuoro | manu-manu | ‘insectsp.’ |
| Rennellese | manu-manu hogau | ‘dragonfly’ |
| (1) | one or a few insect taxa |
| (2) | insects in general |
| (3) | all living creatures |
These data yield the rather simple implicational scale in Table 8.16. It can be seen that definitions of type (2) dominate. Just one language in the sample (Gela) has a greatly extended sense: ‘all living creatures’. Unfortunately, the sources give no evidence that would tell us whether forms with meanings of types (2) and (3) behave like true taxonomic generics or whether they are used only as collectives.
It might seem a straightforward matter to conclude that *manu-manuk was a generic or collective term for insects in general. However, two complicating factors must be addressed.
In many Oceanic languages a semi-productive use of nominal reduplication is to name a taxon which resembles that named by the simple form but is smaller or otherwise atypical — not quite the real thing (for discussion of this function of reduplication in Oceanic see Ross 2008a: 50-51). Given such a mechanism, we must consider the possibility that a reduplicated form of *manuk developed independently in Papuan Tip, Southeast Solomonic and Polynesian. On the formal level, such parallel development would not be surprising. However, it seems unlikely that the meaning of the reduplicated derivative would in both cases have been ‘insects, creepy crawlies’.
The other complicating factor is that a case can be made for attributing another sense to POc *manu-manuk, roughly ‘all creatures’ or ‘all creatures other than those that live in water’. It was noted above (§5.2.2) that reflexes of *manu-manuk have this broader range of reference in Lou, Rotuman and the Fijian languages. However, I prefer to regard the Lou, Rotuman and Fijian uses as post-POc innovations, in which the reflex of *manu-manuk replaced the *manuk reflex as the general term for creatures or land creatures. Thus the conclusion that *manu-manuk existed in POc as a general term for insects and other small land invertebrates still stands.13 Some languages, chiefly in Polynesia, have narrowed its application use to just one or a few insects but in one language its scope has been extended to encompass all living things.
We have looked into the semantic histories of five POc terms, *ikan, *pinaŋoda, *sisiq, *manuk and *manu-manuk, each a candidate for the status of a high-level generic or life form in the POc taxonomy of animals but each having reflexes with highly variable meanings. Approximate semantic reconstructions can be given for each POc term, as follows:
| *ikan | ‘high-level generic (life form) with two senses: 1. typical fish’; (2) ‘typical fish, sharks, rays, eels, cetaceans and dugongs, turtles and crocodiles, and possibly octopus and squid’ |
| *pinaŋoda | ‘collective term (not a life-form): seafood (other than fish) gathered on the reef’ |
| *sisiq | ‘folk generic: small globular gastropods (nerites, moon shells, periwinkles, etc.) which are an important part of the diet. Possibly also used as a collective or high-level generic for edible gastropods in general’ |
| *manuk | ‘high-level generic (life form): birds, bats and flying insects’ |
| *manu-manuk | ‘high-level collective or generic (life form): insects and other small land invertebrates’ |
Let us now return briefly to two questions asked at the outset. First, are the variations in the meanings of major generics across daughter languages random or do they follow an implicational scale, a natural sequence of semantic extension or contraction?
Variations in definitions given to reflexes of *ikan, *sisiq, *manuk and *manu-manuk do appear to conform to implication scales, roughly those sketched in Tables 8.1, 8.9, 8.11 and 8.16, respectively. (We leave aside *pinaŋoda, which was a collective rather than a true generic.) However, the nature of the data are such that we can only give approximate analyses. Undoubtedly, finer-grained data and analysis would show that this picture is too simple. While my main focus has been on the meanings of these five etyma, the foregoing question might also be asked of non-cognate terms that refer to ‘the same’ kinds of semantic categories. A wider comparative study that included non-cognate generic terms for kinds of animals would provide a broader base for testing hypotheses about implicational scales. However, such a study must be left for another occasion.
Finally, why have such marked differences arisen in the semantic scope of these five POc terms? Can some of the changes be attributed to differences in the range of animal life encountered by different communities? All I will say here is that Oceanic islands differ considerably in the richness of their land fauna, with atolls, and small, geographically remote high islands having extremely impoverished fauna and that this variation may have played a part in the semantic histories of two of the five terms considered here, *manuk and *manu-manuk. Brown (1981) has already pointed to the dominance of birds over other large land animals as a possible driver of semantic change in *manuk reflexes (see §5.2.1). But the other three terms all refer to aquatic (chiefly marine) animals and, leaving aside crocodiles and dugongs, the main types of marine fauna are fairly well represented throughout tropical Oceania. It follows that shifts in the referential range of reflexes of POc *ikan, *pinaŋoda and *sisiq must be explained in terms of other factors.