Some years ago, in a paper reconstructing Proto Polynesian (PPn) fish names, Robin Hooper suggested that ‘fishing terms and fish names are good candidates for inclusion in core vocabulary lists for Polynesian languages’ (Hooper 1994:187-8).1 She meant that names for kinds of fish and fishing methods may prove to be as stable as words in standard basic vocabulary lists, such as the Swadesh 100 and 200 word lists. Stability in the lexicon can be defined, operationally, in terms of retention rates of particular etyma in daughter languages. Her suggestion rested on the observations that Polynesian fish names ‘reveal a high degree of uniformity and semantic correspondence …, which is attributable to a comparative uniformity in the ichthyological fauna of the tropical South Pacific’ (Hooper 1994:186) and that in Polynesian communities fishing has great economic and cultural importance. Hooper did not undertake a close statistical analysis of retention rates for fish names or other fishing terms but her remarks invite such a study.
In this chapter I will investigate the stability of 52 of the 140 or so fish names that have been reconstructed to the level of Proto Oceanic (POc), a language spoken more than 3000 years ago. Examination of a sample of contemporary Oceanic languages shows that POc names for some fish taxa have been extremely persistent, others have been much less so, and yet others occupy the middle ground. This leads to the question: Why have some names been much more (or less) stable than others?
I will also draw on evidence from contemporary languages to estimate the total number of fish names likely to have been used by speakers of POc. Osmond (this volume, ch. 2) reconstructs some 145 POc names for fish taxa, of which 140 look to be well attested in terms of having reflexes distributed across more than one primary subgroup. Hooper (1994) reconstructed 115 fish names for PPn, and for Proto Nuclear Polynesian, a stage ancestral to all Polynesian languages except Tongan and Niuean, she reconstructed 147. Geraghty (1994) was able to recover some 150 secure terms for Proto Central Pacific, the immediate ancestor of Polynesian, Fijian and Rotuman.
Comparative analysis of contemporary lexicons for fishing communities speaking Oceanic languages and exploiting environments comparable to those of the Bismarck Archipelago, the probable location of POc, indicates that speakers of POc probably distinguished on the order of 400 names for kinds of fish. In that case the total of about 140 fish names so far reconstructed for POc represents about a third of the total. The question arises why so many names are missing from the reconstructed inventory.
R.H. Carcasson writes as follows of the fish fauna of the tropical Indo-Pacific region:
A surprisingly high proportion of Indo-Pacific species occur throughout the area: the Central Indo-Pacific. consisting roughly of the Malay Archipelago, the Philippines and New Guinea, has a richer fauna than any other part of the Region and has probably been the principal evolutionary centre from which the entire Indo-Pacific has been populated. As one moves away from this area, the fauna becomes progressively impoverished … (Carcasson 1977:13).
More than 150 families of fish occur in the Indo-Pacific region. Although the number of genera and species declines as one moves eastwards from New Guinea and the Bismarck Archipelago, the dominant families (those with most species, e.g. the wrasses, groupers, trevallies and jacks, damselfishes, cardinalfishes, surgeonfishes, gobies and moray eels) are present throughout the region.
The Bismarck Archipelago, the probable location of the POc speech community, is richer in marine life than any other region in the South Pacific apart from New Guinea. The coral reef systems of this region support several thousand species and several hundred genera.
As we are concerned with retention rates of POc lexical items it is relevant to ask what time span is involved. The breakup of POc can be dated with considerable confidence to between 3400 and 3100 BP. POc was almost certainly spoken in the Bismarck Archipelago, where the greatest concentration of primary subgroups of Oceanic is found today. Dating of its breakup to within a span of two or three centuries is possible because of the strong association between the appearance in the Bismarck Archipelago and subsequent spread across the SW Pacific of the highly distinctive archaeological culture known as Lapita and the initial diversification of Proto Oceanic within the Bismarcks and the subsequent spread of Oceanic languages across the SW Pacific (Green 2003, Kirch 1997, 2000, Pawley 2003c, 2007, 2008, Spriggs 1997a, Summerhayes 2001).
The latest critical reviews of C14 dates indicate that Lapita appeared in the Bismarcks around 3400-3450 BP (Specht 2007). By 3200-3100 BP bearers of this culture had settled the Reefs-Santa Cruz group, east of the main Solomons group (Green 2003, Green et al. 2008), and by about 3000 BP or earlier they had occupied Vanuatu (Bedford 2003, 2006), New Caledonia (Sand 2003) and Fiji (Nunn et al. 2004, Clark and Anderson 2009). By 2900 Lapita populations were present in the southern part of the Tongan group (Burley and Connaughton 2007) and by 2800-2700 they were in Samoa and the small islands between Tonga and Samoa: Futuna, Uvea (Wallis) and Niuatoputapu (Clark and Anderson 2009).
The carriers of Lapita material culture were, almost certainly, the first people to settle the Reefs-Santa Cruz group, Vanuatu, New Caledonia, Fiji and western Polynesia. Given that all these regions were occupied within 300-600 years of the first appearance of Lapita in the Bismarcks, we would expect that their first colonists spoke forms of Oceanic that were not greatly differentiated from their common ancestor, Proto Oceanic.2
Before proceeding to an analysis of the data on fish names we should consider some methodological issues.
The descriptions available on fish names and taxonomies in Oceanic languages are very uneven in quality and quantity. Although there are reasonably good general dictionaries for about 50 of the 450 or so languages in the Oceanic group only a handful provide something close to a comprehensive list of fish names together with scientific identifications and information about the indigenous taxonomy. In addition there are a few published papers and unpublished works that provide fairly extensive data. The statistical data cited in this paper are based on materials from a sample of 15 Oceanic languages plus a non-Oceanic language of Micronesia, Palauan, and a Malayo-Polynesian language of Sulawesi, Pendau (see Table 3.1).
The core of a folk taxonomy is a set of named categories (taxa) that form a semantic hierarchy in which all taxa are related by virtue of being included in a higher order category or by virtue of including lower order taxa. A term that includes two or more named subtaxa is a generic taxon. There may also be covert taxa, conceptual categories whose unity is indicated by other means than sharing a class name. For example, for many English speakers horses, zebras and donkeys form a covert category (horse-like animals), as do dogs and wolves. A distinction needs to be made between the core categories in the taxonomic hierarchy and other kinds of categories that crosscut these, e.g. ecological categories, like ‘fish of the reef’ vs ‘fish of the deep sea’, and functional categories, like ‘poisonous fish’.
| Name | Location | Subgroup | Main source |
|---|---|---|---|
| Arosi | Makira, eastern Solomons | SE Solomonic | Fox 1978 |
| Gela | Florida, central Solomons | SE Solomonic | Foale 1998 |
| K’marangi | atoll south of the Carolines | Polynesian | Lieber and Dikepa 1974 |
| Kiribati | atoll chain, Kiribati | Micronesian | Thaman & Tabano n.d. |
| Lau | Malaita, SE Solomons | SE Solomonic | Fox 1974. Akimichi 1978 |
| Marovo | New Georgia. W Solomons | NW Solomonic | Hviding 1990. 1996. 2005 |
| Marquesan | Eastern Polynesia | Polynesian | Lavondes 1977 |
| M’bunai Titan | Admiralty Islands. PNG | Admiralties | Akimichi & Sakiyama 1991 |
| Motu | Central Province, PNG | Papuan Tip | Oram n.d. |
| Mutu (Mandok) | Vitiaz Straits, PNG | North New Guinea | Pomponio n.d. |
| Niuatoputapu | small high island, W Polynesia | dialect of Tongan | Dye 1983 |
| Palauan | Western Carolines | Malayo-Polynesian | Helfman & Randall 1973 |
| Pendau | Central Sulawesi | Malayo-Polynesian | Quick 2005. 2010 |
| Satawalese | Central Carolines | Micronesian | Akimichi & Sauchomal 1982 |
| Teop | NE Bougainville, PNG | NW Solomonic | Shoffner 1976 |
| Uvean (Wallisian) | high island, W Polynesia | Polynesian | Rensch 1983 |
| Wayan | small high island, W Fiji | Fijian | Pawley & Sayaba 2003 |
An important distinction is between uninomial and binomial names, sometimes called primary and secondary terms. Uninomials are names that are either a single morpheme (oak, pine, apple) or are multimorphemic but with meanings that are idiomatic (she-oak, screw pine). A screw pine (pandanus) is not a kind of pine, and a she-oak (casuarina) is not a kind of oak. Binomials consist of a uninomial term plus a modifier, where the uninomial names the class and the modifier distinguishes members of the class (i.e. a white pine and redwood pine are kinds of pine, and white pointer shark and hammerhead shark are kinds of shark). Uninomials typically name taxa at the level of what Berlin (1992) calls ‘folk generics’. These are taxa that belong to a particular major generic or ‘life form’ such as ‘fish’, ‘bird’ and ‘tree’ and which themselves may contain subtaxa that are designated by binomials.
Binomials provided by native speakers need to be viewed with some caution. When informants are asked to name or discriminate between specimens they sometimes offer ad hoc descriptions, such as ‘spotted X’, or ‘red X’, which the investigator may mistake for genuine conventional names. Independent checks with a range of informants are desirable but not always possible.
Counting the number of lexical units in a folk taxonomy is not completely straightforward. There are several ways of counting members, which give different results. First, one can count the total number of terms that represent taxa. Each formally distinct term is counted separately even if it has synonyms, e.g. Australian English chook, fowl, and chicken ‘domestic fowl, Gallus gallus’ count as three terms. Second, one can count just the number of contrasting taxa, in the sense of distinct conceptual categories, treating as a single taxon any set of referential synonyms (i.e. two or more terms that have the same referential meaning). In this case, for some English speakers the three terms chicken, fowl and chook would represent one taxon. In this study totals generally refer to the number of distinct terms for fish taxa provided by each source, because it is not always possible to identify synonyms in a list of names.
Another variable is names for growth stages. For certain species in several groups (e.g. barracuda, eels, groupers, wrasses, mullet, trevally, sharks) Oceanic languages commonly distinguish two or more growth stages. In this study I follow the standard practice of treating each name for a growth stage as representing a distinct taxon.
Yet another variable is polysemy. Some terms have two senses, one referring to a more specific category, another to a more general one (e.g. cat can refer to the domestic cat, Felis felis, or more broadly to any member of the cat family (lion, tiger, cheetah, etc.) In this study generic and specific senses of fish names are not counted separately, chiefly because we lack reliable data on such sense differences for most languages.
Hooper (1994:189) points to the kinds of difficulties met in attributing meanings to reconstructed fish names. The comparative linguist is dependent on the quality of the definitions available in accounts of contemporary languages and on the measure of agreement between them. Usually it is possible to pin down the meaning of a reconstructed form to a family or a group of closely related families, e.g. ‘kind of rock-cod (Serranidae)’ or ‘kind of snapper (Lutjanidae) or emperor (Lethrinidae)’. In a minority of cases it can be said with considerable confidence that a name referred to a particular species or group of closely related species, or to a growth stage of a species or genus.
Hooper (1994) presents a brief statistical argument that Polynesian uninomial fish names, overall, show retention rates similar to those exhibited by etyma whose meanings belong to standard lists of basic vocabulary. She reconstructs 115 PPn fish names for generic and specific taxa, of which only three are binomials. She notes that if we compare the 112 PPn uninomials with lists for Uvean (Rensch 1988), Niuatoputapu (Dye 1983) and her own list for Tokelauan we get impressively high retention rates. The rates are about 66% for Uvean (74 retentions), 61% for Niuatoputapu (68 retentions) and 84% for Tokelauan (94 retentions). Hooper adds that ‘[if] we hypothesize that the Proto Polynesians, like their present-day descendants, used a vocabulary of about 130 monomial [uninominal] fish names, the retention rates are more plausible: 60% for Uvean, 52.2% for Niuatoputapu, and 72.3% for Tokelauan - close to the expected average retention rate of 70% for core vocabulary’ (Hooper 1994:187).
There are certain problems with this proposal. First, Hooper does not say how the ‘expected’ average retention rate of 70%’ for contemporary Polynesian languages was arrived at. It was presumably based on work in glottochronology indicating that, on average, languages will replace about 20 percent of words on the 200 item Swadesh list per 1000 years. An estimate of 30 percent replacement would make sense on the assumption that PPn broke up around 1400 years ago. This estimate is reasonable as a latest possible date of PPn breakup.
Second, Hooper’s assessment of average retention rates for fish names is based on just those 112 uninomials that have been reconstructed for PPn. These etyma will be a subset of the total number of fish names used in PPn (less stable names will not have been recovered). How small a subset is 112? For reasons indicated in §4, I estimate the number of uninomials in PPn to be more than 200. To take the case of Uvean, 74 retentions out of 200 would equate to a retention rate of 37 percent, still substantial but much less than 66 percent. Third, in some cases semantic matches between PPn fish names and their reflexes are not exact, e.g. the POc generic term for ‘shark’ is continued in some daughter languages by a term that refers to the Remora or sucking fish, which attaches itself to sharks. In standard glottochronological procedure, a form that continues a proto-language etymon but changes the meaning is not counted as a retention. In the case of PPn fish names that refer to particular families, genera or species, a reasonable compromise would be to count a reflex as retained in a daughter language provided that it refers to the same biological family as the etymon in the protolanguage.
There is, however, a more fundamental concern. Even a comparison of the average retention rates of PPn fish names with that of basic vocabulary will not leave us much the wiser. This is because the individual meanings in the basic vocabulary list do not share a constant retention rate. On the contrary, there is enormous variation in the rates at which forms representing different meanings are replaced. A comparative study of Austronesian languages (Dyen et al. 1967) indicates that, over a span of several millennia, the least persistent item on the Swadesh 200 item basic vocabulary list is many times more likely to be replaced than the most persistent item.
Using a 196 item meaning list Dyen et al. compared 89 Austronesian languages and calculated cognation rates over the total set of language pairs. The languages were drawn from diverse subgroups, most of them belonging to the Malayo-Polynesian subgroup, to which 98 percent of Austronesian languages belong. It was found that 10 meanings on the list show cognation rates of between 50 and 80 percent. Forty-two meanings have rates exceeding 20 percent. Only 70 meanings have rates of more than 10 percent. As many as 105 meanings show cognation rates of less than seven percent and 43 of these score below three percent.3
Admittedly, cognation rate and retention rate in basic vocabulary are not the same thing. The former is a measure of how often pairs of languages have cognate words for a given meaning on the list. The latter is a measure of how often a given reconstructed etymon is retained with the same meaning. But cognation rate in Dyen et al’s study will generally be a fairly accurate indicator of retention rate. For example, if a meaning, say ‘louse’, shows a high cognation rate it is almost certainly represented by an etymon (in this case, *kutu) that was present in a proto-language ancestral to most or all 89 languages in Dyen et al.’s sample and that has been continued by a high proportion of daughter languages.
Retention rates for what are approximately the most stable 60 items of POc basic vocabulary in a modified Swadesh list were presented in Pawley (2009). Retentions and replacements were recorded in a sample of 40 present-day Oceanic languages. Retention rates for individual items ranged between 97 and 23 percent. Each of the 24 most persistent POc etyma were retained by 70 percent or more languages, 44 etyma were retained by 50 percent or more languages, 54 by 40 percent or more, and 60 by 23 percent or more.
Given that individual basic vocabulary items vary so much in their retention rates, there is little point in asking whether POc fish names in general show retention rates similar to basic vocabulary in general. We need to ask questions sensitive to the variable retention rates of individual lexical items. First, retention rates needed to be determined for a sample of POc fish names Then it will become possible to make comparisons with the rates established for the 60 most POc etyma in the modified Swadesh basic vocabulary list. We can ask, for instance, how many fish names have a higher retention rate than the 20th, or 40th, or 60th most stable etymon in the basic vocabulary list.
Retention rates for 52 POc fish names, drawn from Osmond’s list of reconstructions, were investigated in 12 languages belonging to diverse subgroups: Titan (Admiralties group), Mutu (North New Guinea), Motu (Papuan Tip), Marovo and Teop (Meso-Melanesian), Gela (Guadalcanal-Gelic), Arosi and Lau (Makira-Malaitan), Kiribati and Satawalese (Micronesian), Wayan (Fijian) and Niuatoputapu (a dialect of Tongan, Polynesian).
Between them, the 52 POc names represent a wide range of fish families. In Osmond’s data each of the 140 secure reconstructions is well attested by cognate sets drawn from diverse subgroups. In my sample not all reconstructions are well attested: whereas Osmond was in some cases able to draw on cognates from scores of languages, my sample consists of just 12.
For convenience of presentation the 52 POc names are divided into four groups. Tables 3.2-3.5 show the presence (y) or absence (-) of reflexes of each etymon in the 12 contemporary languages. A question mark, accompanied by an explanatory footnote, indicates doubt as to whether the form should be counted as a reflex of the etymon. The columns showing the number of retentions (‘no. retentions’) and the percentage of languages that reflect the etymon then show alternate figures, one not counting, the other counting the doubtful reflex.
| *ikan | ‘generic for fish (and certain other free-swimming aquatic creatures)’ |
| *bakewa | ‘generic for sharks’ |
| *paRi | ‘generic for rays’ |
| *tuna | ‘generic for freshwater eels (Anguillidae)’ |
| *qawaq | ‘milkfish, Chanos chanos (Chanidae)’ |
| *saku | ‘needlefish and long toms (Belonidae)’ |
| *qonos | ’mature Sphyraena spp., possibly generic for all barracuda |
| *qalu | ‘taxon of barracuda (Sphyraenidae)’ |
| *taŋiRi | ‘Scombridae spp., probably incl. Spanish mackerel and wahoo’ |
| *qatun | ‘skipjack tuna, Katsuwonus pelamis, Scombridae’ |
| *walu | ‘Scombridae sp. or spp.’ |
| *saku-layaR | ‘sailfish (Istiophoridae)’ |
| *sabutu | ‘snapper (Lutjanus) or emperor (Lethrinus) sp. or spp.’ |
| Titan | Motu | Mutu | Teop | Marovo | Gela | Lau | Arosi | Wayan | Niuatoputapu | Kiribati | Satawalese | no. retentions | percentage | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| *ikan | y | - | y | - | y | y | y | y | y | y | y | y | 10 | 83 |
| *bakewa | y | y4 | - | y | - | y | y | y | - | - | y | y | 8 | 67 |
| *paRi | y | - | - | - | y5 | y | y | y | y | y | ?6 | y | 8-9 | 67/75 |
| *tuna | - | - | y | - | - | - | - | - | y | y | - | - | 3 | 25 |
| *qawaq | - | - | y | - | - | y | - | -7 | y | y | y | y | 6 | 50 |
| *saku | - | - | - | - | - | - | - | - | y | y | -8 | y | 3 | 25 |
| *qonos | - | y | - | - | - | y | y | y | -9 | y | - | - | 5 | 42 |
| *qalu | y | - | - | - | - | y | - | - | - | - | - | - | 2 | 17 |
| *taŋiRi | y10 | - | y | y | y | -11 | - | -12 | - | - | y | y | 6 | 50 |
| *qatun | - | - | - | y | - | y | y | y | y | y | y | y | 8 | 67 |
| *walu | - | y | - | - | - | -13 | - | - | y | y | - | - | 3 | 25 |
| *saku-layaR | y | - | - | - | - | - | - | - | y | y | ?14 | y | 4-5 | 33/42 |
| *sabutu | - | y | - | - | - | - | - | - | y | y | - | - | 3 | 25 |
| *kulabo | ‘snapper sp. or spp. (Lethrinidae)’ |
| *(ñ,n)opuq | ‘stonefish, Synanceia spp., incl. Synanceia verrucosa (Synanceiidae)’ |
| *taRaqan | ‘squirrelfish, Sargocentron sp. or spp. (Holocentridae)’ |
| *jumu | ‘triggerfish (Balistidae) and possibly leatherjackets (Monacanthidae)’ |
| *bebek | ‘generic for butterflyfish and coralfish (Chaetodontidae)’ |
| *kuRapu | ‘rock-cod, Epinephelus sp. or spp. including Epinephelus lanceolatus, Queensland grouper (Serranidae)’ |
| *palaja | ‘rabbitfish, Siganus spp. (Siganidae)’ |
| *laci | ‘Scomberoides spp., incl. Scomberoides lysan (Carangidae)’ |
| *qulua | ‘Caranx sp., possibly Caranx ignobilis. big-headed jack (Carangidae)’ |
| *pilu | ‘Caranx sp. or spp. (Carangidae)’ |
| *qatule | ‘scad spp. (Carangidae), incl. Trachurops crumenophthalmus, big-eyed scad’ |
| *taRutu(m,ŋ) | ’porcupinefish, Diodon spp. (_Diodon_tidae)’ |
| *kanase | ‘generic for certain mullet (Mugilidae)’ |
| *bunaR | ‘batfish, Platax sp. (Platacidae)’ |
| *tiqo | ‘goatfish sp. (Mullidae)’ |
| Titan | Motu | Mutu | Teop | Marovo | Gela | Lau | Arosi | Wayan | Niuatoputapu | Kiribati | Satawalese | no. retentions | percentage | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| *kulabo | y | - | - | - | - | - | - | - | -15 | y | - | - | 2 | 17 |
| *[ñ,n]opuq | y | y | y | y | y | y | - | ?16 | y | y | y | y | 10-11 | 83/91 |
| *taRaqan | - | y | - | - | - | y | - | - | - | y | y | y | 5 | 42 |
| *jumu | - | y | - | - | - | - | - | - | y | y | - | - | 3 | 25 |
| *bebek | - | y | - | - | - | - | y | - | - | - | y | - | 3 | 25 |
| *kuRapu | y | - | - | ?17 | - | y | y | y | y | - | y | y | 7-8 | 58/67 |
| *palaja | y | - | - | - | - | - | y | ?18 | y | - | - | - | 3-4 | 25/33 |
| *laci | y | y | y | - | y | y | - | - | - | y | y | - | 7 | 58 |
| *qulua | - | - | - | - | - | - | - | - | - | y | ?19 | - | 1-2 | 8/17 |
| *pilu | - | - | - | - | - | - | - | - | y | y | - | y | 3 | 25 |
| *qatule | y | - | - | - | - | - | - | - | y | y | - | - | 3 | 25 |
| *taRutu(m,ŋ) | - | - | - | - | - | - | y | - | - | y | y | - | 3 | 25 |
| *kanase | - | - | - | y | - | - | - | y | y | y | - | - | 4 | 33 |
| *bunaR | - | - | y | y | - | - | - | - | ?20 | - | - | - | 2-3 | 17/25 |
| *tiqo | - | y | - | y | - | y | - | - | - | - | y | y | 5 | 42 |
| *(k,q)ulapi | ‘parrotfish spp. (Carangidae) incl. Hipposcarus longiceps, longnosed parrotfish’ |
| *bala(ŋ,k)i | ‘Acanthurus and Naso spp. (Acanthuridae)’ |
| *qume | ‘unicornfish, Naso unicornis (Acanthuridae)’ |
| *koto(ŋ) | ‘large rock cod, Epiphenilus spp. (Serranidae)’ |
| *don(o,u) | ‘rock cod, possibly Cephalopholis (= Plectropomus) spp. (Serranidae)’ |
| *(k,q)umutuR | ‘sweetlips, Plectorhynchus spp. (Haemulidae)’ |
| *bʷa[p,w]a | ‘snapper, Lutjanus sp. (Lutjanidae)’ |
| *tasiwa | ‘snapper, Lutjanus sp. (Lutjanidae)’ |
| *kasika | ‘large emperor, Lethrinus sp. (Lethrinidae)’ |
| *palata | ‘dolphinfish (Coryphaenidae)’ |
| *bʷa(s,j)i | ‘moray eel (Moraenidae)’ |
| *muqu(ŋ) | ‘rabbitfish, Siganus sp., possibly Siganus spinus (Siganidae)’ |
| *tanipa | ‘sardine, Sardinella sp. (Clupeidae)’ |
| Titan | Motu | Mutu | Teop | Marovo | Gela | Lau | Arosi | Wayan | Niuatoputapu | Kiribati | Satawalese | no. retentions | percentage | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| *(k,q)ulapi | - | - | - | - | - | y | - | - | y | - | y | y | 4 | 33 |
| *bala(ŋ,k)i | y | - | - | - | - | - | - | - | y | y | - | - | 4 | 27 |
| *qume | - | - | - | - | - | - | y | y | - | y | - | y | 4 | 33 |
| *koto(ŋ) | y | - | - | - | - | - | - | - | y | - | - | - | 2 | 17 |
| *don(o,u) | - | y | - | - | - | - | - | - | y | y | - | - | 3 | 25 |
| *(k,q)umutuR | y | - | y | - | - | y | y | - | y | - | - | - | 5 | 42 |
| *bʷa[p,w]a | y | - | - | - | - | - | - | - | y | - | - | - | 2 | 17 |
| *tasiwa | - | y | - | - | - | - | y | - | - | y | - | - | 3 | 25 |
| *kasika | y | y | - | y | - | - | - | - | y | - | - | y | 5 | 42 |
| *palata | - | - | - | - | y | - | - | - | - | - | - | - | 1 | 8 |
| *bʷa(s,j)i | y | - | - | - | - | - | - | - | - | - | - | - | 1 | 8 |
| *muqu(ŋ) | - | y | - | - | y | - | y | - | - | - | - | - | 3 | 25 |
| *tanipa | - | - | - | - | - | - | - | - | y | -21 | - | - | 1 | 8 |
| *paya | ‘anchovy (Engraulidae)’ |
| *(k,q)aRua(s) | ‘generic for mullet (Mugilidae)’ |
| *bubu | ‘triggerfish, Balistes sp. (Balistidae)’ |
| *taRat | ‘unicornfish, Naso spp., probably incl. Naso brevirostris (Acanthuridae)’ |
| *qaroŋo | ‘surgeonfish, Acanthurus sp. (Acanthuridae)’ |
| *mamin | ‘Cheilinus spp. (Labridae), incl. double-headed Maori wrasse, Cheilinus undulatus’ |
| *taŋapa(R,r) | ‘large wrasse, probably Cheilinus sp. (Labridae)’ |
| *magaRut | ‘flying fish (Exocoetidae)’ |
| *panapa | ‘garfish, Hemiramphus sp. (Hemiramphidae)’ |
| *paRi-manuk | ‘eagle ray, Aetobatus narinari’ |
| *kaboRa | ‘catfish-eel, Plotosus sp. (Plotosidae)’ |
| Titan | Motu | Mutu | Teop | Marovo | Gela | Lau | Arosi | Wayan | Niuatoputapu | Kiribati | Satawalese | no. retentions | percentage | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| *paya | -22 | - | y | - | - | - | - | - | - | - | - | - | 1 | 8 |
| *(k,q)aRua(s) | y | - | - | - | - | y | y | - | - | y | y | y | 6 | 50 |
| *bubu | y | - | - | y | - | - | y | y | y | - | y | y | 7 | 58 |
| *taRat | - | y23 | - | - | y | - | - | - | y | - | - | y | 4 | 33 |
| *qaroŋo | y | - | - | - | - | - | - | - | - | y | y | - | 3 | 25 |
| *mamin | y | y | y | - | y | - | - | - | - | - | - | y | 5 | 42 |
| *taŋapa(R,r) | y | - | - | - | - | - | - | - | y | y | - | ?24 | 3-4 | 25/33 |
| *magaRut | y | - | - | - | - | - | - | y | - | - | - | y | 3 | 25 |
| *panapa | -25 | - | - | ?26 | - | - | - | - | - | - | y | - | 1-2 | 8/17 |
| *paRi-manuk | - | y | - | - | - | - | - | y | y | y | - | - | 4 | 33 |
| *kaboRa | - | - | - | - | - | y | - | — | y | - | - | - | 2 | 17 |
Table 3.6 shows the full range of retention rates for the 52 POc etyma. It can be seen that there is wide variation, with the most durable names being ten times more likely to be retained than the least durable.
| % retentions | Proto Oceanic fish name | |
|---|---|---|
| 83/91 | *(ñ,n)opuq | ‘stonefish, Synanceia spp.’ |
| 83 | *ikan | ‘fish (generic)’ |
| 67 | *bakewa | ‘shark (generic)’ |
| 67 | *paRi | ‘ray (generic)’ |
| 67 | *qalun | ‘skipjack tuna, Katsuwonus pelamis (Scombridae)’ |
| 58/67 | *kuRapu | ‘rock-cod, Epinephelus sp. or spp. (Serranidae)’ |
| 58 | *bubu | ‘triggerfish, Balistes sp. (Balistidae)’ |
| 58 | *laci | ‘Scomberoides spp., incl. Scomberoides lysan (Scrombridae)’ |
| 50 | *(k,q)aRua(s) | ‘mullet, generic (Mugilidae)’ |
| 50 | *qawaq | ‘milkfish, Chanos chanos (Chanidae)’ |
| 50 | *taŋiRi | ‘(Scombridae)_ spp., probably incl. Spanish mackerel and wahoo’ |
| 42 | *kasika | ‘large emperor, Lethrinus sp. (Lethrinidae)’ |
| 42 | *qonos | ‘mature barracuda (Sphyraenidae)’ |
| 42 | *(k,q)umutuR | ‘sweetlips, Plectorhynchus spp. (Haemulidae)’ |
| 42 | *mamin | ‘Cheilinus spp., incl. double-headed Maori wrasse, Cheilinus undulatus (Labridae)’ |
| 42 | *taRaqan | ‘squirrelfish, Sargocentron sp. or spp. (Holocentridae)’ |
| 42 | *tiqo | ‘goatfish sp. (Mullidae)’ |
| 33/42 | *saku-layaR | ‘sailfish (Istiophoridae)’ |
| 33 | *kanase | ‘generic for certain mullet spp. (Mugilidae)’ |
| 33 | *(k,q)ulapi | ‘parrotfish spp. (Scaridae), incl. Hipposcarus longiceps, longnosed parrotfish’ |
| 33 | *paRi-manuk | ‘eagle ray, Aetobatus narinari’ |
| 33 | *qume | ‘unicornfish, Naso unicornis (Acanthuridae)’ |
| 33 | *taRat | ‘unicornfish, Naso spp., probably incl. Naso brevirostris (Acanthuridae)’ |
| 25/33 | *palaja | ‘rabbitfish, Siganus spp. (Siganidae)’ |
| 25/33 | *taŋapa(R,r) | ‘large wrasse, probably Cheilinus sp. (Labridae)’ |
| 25 | *bala(ŋ,k)i | ’surgeonfish, prob, an Acanthurus wad Naso taxon |
| 25 | *bebek | ‘generic for Chaetodontidae, butterflyfish’ |
| 25 | *don(o,u) | ‘rock cod, possibly Cephalopholis spp. (Serranidae)’ |
| 25 | *jumu | ‘Balistidae, triggerfish, and possibly Monacanthidae, leatherjackets’ |
| 25 | *magaRut | ‘flying fish (Exocoetidae)’ |
| 25 | *muqu(ŋ) | ‘rabbitfish, Siganus sp., possibly Siganus spinus (Siganidae)’ |
| 25 | *pilu | ‘Caranx sp. or spp. (Carangidae)’ |
| 25 | *qaroŋo | ‘surgeonfish, Acanthurus sp. (Acanthuridae)’ |
| 25 | *qatule | ‘scad, incl. big-eyed scad, Trachurops crumenophthalmus (Carangidae)’ |
| 25 | *sabutu | ‘snapper (Lutjanidae) or emperor (Lethrinidae) sp. or spp.’ |
| 25 | *saku | ‘needlefish and long toms (Belonidae)’ |
| 25 | *taRutu(m,ŋ) | ’porcupinefish, Diodon spp. (_Diodon_tidae)’ |
| 25 | *tasiwa | ‘snapper, Lutjanus sp. (Lutjanidae)’ |
| 25 | *tuna | ‘generic for freshwater eels (Anguillidae)’ |
| 25 | *walu | ‘Scombridae sp. or spp.’ |
| 17/25 | *bunaR | ‘batfish, Platax sp. or spp., possibly Naso sp. or spp.’ |
| 17 | *bʷa[p,w]a | ‘snapper, Lutjanus spp. (Lutjanidae)’ |
| 17 | *kaboRa | ‘catfish-eel, Plotosus sp. (Plotosidae)’ |
| 17 | *koto(ŋ) | ‘large rock cod, Epiphenilus spp. (Serranidae)’ |
| 17 | *kulabo | ‘snapper sp. or spp. (Lethrinidae)’ |
| 17 | *qalu | ‘taxon of barracuda (Sphyraenidae)’ |
| 17 | *qulua | ‘Caranx sp., possibly Caranx ignobilis, big-headed jack (Carangidae)’ |
| 8/17 | *panapa | ‘garfish, Hemiramphus sp. (Hemiramphidae)’ |
| 8 | *bʷa(s,j)i | ‘moray eel (Moraenidae)’ |
| 8 | *palata | ‘dolphinfish (Coryphaenidae)’ |
| 8 | *paya | ‘anchovy (Engraulidae)’ |
| 8 | *tanipa | ‘sardine, Sardinella sp. (Clupeidae)’ |
We are now in a position to evaluate the hypothesis that some Proto Oceanic fish names have been as persistent as some basic vocabulary items.
It was noted in §2.5 that, in a sample of 40 daughter languages, the 60 most conservative POc basic vocabulary items in a modified 200 item Swadesh list show retention rates ranging between 97 and 23 percent. The 24 most persistent etyma were retained by 70 percent or more languages, 44 etyma were retained by 50 percent or more languages, 54 by 40 percent or more, and 60 by 23 percent or more.
No fewer than 38 of the 52 POc fish names examined show retention rates within the range exhibited by the 60 most stable basic vocabulary etyma. Eleven fish names fall within the range of the 44 most stable basic vocabulary etyma. The hypothesis is strongly supported.
Some small caveats must be attached to this conclusion. Methodological difficulties in comparing fish name retentions with basic vocabulary retentions were discussed in §2. In this study the fish name retention rates are based on just 12 languages, whereas the basic vocabulary retention rates are based on 40. (The same 40 languages could not be used in the fish names study for reasons stated in §2.1.) If the composition of the two samples, in terms of distribution across major subgroups, were markedly different this might bias the results, because some Oceanic subgroups are known to contain more languages that are lexically conservative than other subgroups. However, the composition is roughly similar so the effects of different sample sizes are probably small.
| family/group | POc no. | PCP % retained | PPm % retained |
|---|---|---|---|
| rays | 2 | 100 | 100 |
| damsels, sergeant-majors (Pomacentridae) | 2 | 100 | 100 |
| sweetlips (Plectorhinchinae) | 2 | 100 | 50 |
| boxfish (Ostraciidae) | 3 | 100 | 33 |
| eels | 7 | 86 | 14 |
| trevally, jacks etc (Carangidae) | 6 | 84 | 66 |
| scombrids | 6 | 84 | 84 |
| garfish (Heramphidae) | 5 | 80 | 80 |
| surgeons and unicorns (Acanthuridae) | 12 | 75 | 25 |
| snapper (Lutjanidae) | 4 | 75 | 50 |
| rock cod (Serranidae) | 4 | 75 | 50 |
| mullet (Mugilidae) | 3 | 75 | 75 |
| squirrelfish (Holocentridae) | 3 | 66 | 66 |
| goatfish (Mullidae) | 3 | 66 | 0 |
| parrotfish (Scaridae) | 7 | 57 | 57 |
| rabbitfish (Siganidae) | 7 | 57 | 43 |
| wrasses (Labridae) | 6 | 50 | 33 |
| emperors (Lethrinidae) | 5 | 40 | 20 |
| triggerfish (Balistidae) | 4 | 50 | 25 |
| marlin (Xiphiidae) | 2 | 50 | 50 |
| sharks | 7 | 43 | 43 |
| porcupinefish (_Diodon_tidae) | 5 | 40 | 40 |
Only two fish names have retention rates of more than 70 percent, ranking with the most stable 24 basic vocabulary items. One of these is the life form taxon *ikan ‘fish’. The other is the generic *(ñ,n)opuq ‘stonefish’. These repulsive-looking fish are not economically useful but are extremely dangerous denizens of the shallow sandy bottoms. The next three most stable forms consist of the respective generics for sharks and rays, *bakewa and *paRi, together with *qatun, which denotes one or more species of much-prized game fish in the bonito and tuna family. Sharks and rays are extremely salient fish for several reasons: they are highly distinctive in appearance and behaviour, very large, some kinds are dangerous, and they have economic importance. The next five names in the top ten all denote large or fairly large fish that are highly regarded as food.
Have names for taxa belonging to certain families of fish been more persistent than names for taxa belonging to other families? An answer to this question requires comparison of all secure POc reconstructions with a later language or languages. One set of comparisons that can readily be made is between POc, PCP and PPn. Table 3.7 specifies the number of retentions in PCP and PPn of reconstructed POc names for fish belonging to particular families or, in the case of sharks, rays and eels, high-order groups.
| language | no. retentions | percentage retained |
|---|---|---|
| Wayan | 29-30 | 56-58 |
| Niuatoputapu | 28 | 54 |
| Titan | 24 | 46 |
| Satawalese | 20 | 38 |
| Motu | 17 | 33 |
| Kiribati | 16-19 | 31-36 |
| Gela | 16 | 31 |
| Lau | 15 | 29 |
| Arosi | 12-13 | 23-25 |
| Marovo | 9 | 18 |
| Teop | 8-9 | 16-18 |
| Mutu | 8 | 16 |
Unfortunately, numbers in most groups are too low to make tests of statistical significance useful. Among those groups with four or more POc names it is noteworthy that carangids, scombrids, rock cod and mullet, all important food fish, show high retention rates in both PCP and PPn. But it is unclear why, of the four names in the sample representing carangids, one, *laci ‘Scomberoides spp. incl. Scomberoides lysan’, has been very durable (scoring 58%) while the other three have not (*qatule, probably denoting the big-eyed scad, *qulua and *pilu, both denoting Caranx sp. or spp., all score 25% or less). It is also striking that although most POc names for kinds of eels and for Acanthuridae (surgeonfish and unicornfish) are retained in PCP, very few are kept in PPn.
The 12 languages vary greatly in their retention rates for fish names (Table 3.8). This finding is consistent with what is known about retention rates for basic vocabulary among Austronesian languages in general (Blust 1981b, 1999b).
The percentages for certain languages are likely to have been deflated by the incompleteness of the fish name lists for these languages, particularly Kiribati and Mutu and to a lesser extent Arosi, Lau, Niuatoputapu and Teop.
| Wayan | 484 |
| Satawalese | 400 |
| Gela | 368 |
| Marovo | 316 |
| Titan | 287 |
| Uvean | 284 |
| K’marangi | 262 |
| Marquesan | 260 |
| Niua | 209 |
| Teop | 197 |
| Kiribati | 151 |
| Mutu | 145 |
Let us turn now to another set of questions. How many fish taxa were distinguished in POc? What proportion of the fish names were uninomials and binomials, respectively?
A fair idea can be gained by looking carefully at descriptions of a sample of present day languages. Investigators making fairly comprehensive studies of particular fishing communities living in rich marine environments in western Melanesia have consistently recorded more than 300 names for kinds of fish. Hviding (1996:193) says that speakers of Marovo distinguish by name over 400 kinds of fish; Hviding (1990) lists 316 Marovo fish names. For Gela, Foale (1998) lists 368 names (including 14 synonyms). During only 12-16 weeks fieldwork on M’bunai Titan, a language of the Admiralty Islands, Akimichi & Sakiyama (1991) recorded 287 fish taxa.
Figures for some communities in Remote Oceania are of a similar order. Helfman and Randall (1973) give 336 fish names for Palauan, a non-Oceanic Austronesian language. For Satawalese, spoken on an atoll in the Central Carolines, 400 names were recorded by Akimichi & Sauchomal (1982), based on seven months fieldwork. This number is high considering that Satawalese is a part of a chain of atolls, and the range of species and genera is likely to be fewer than in the vicinity of large high islands and especially, continental Melanesia. For the Western Fijian language of Wayan, 484 fish names were recorded of which 352 are uninomials and 132 are binomials (Pawley and Sayaba 2003). (This count excludes about 60 binomials that are probably descriptions rather than conventional names.)
Total counts for some isolated atolls and small high islands in the central Pacific are a bit lower. Lieber (1994) gives 262 names for Kapingamarangi, an isolated atoll south of the main Carolines group. Rensch (1983) gives 284 names for Uvean (aka Wallisian), in western Polynesia, and Lavondes (1977) gives 260 fish names for Marquesan. There are a number of other published lists which, although very useful, are plainly incomplete. For example, the list (Dye 1983) for Niuatoputapu, a language of western Polynesia, consists of 209 names (37 of these mark developmental stages of 12 species).
For Tobelo, a non-Austronesian language of the Bird’s Head family in North Halmahera, Taylor (1990) recorded 358 fish taxa. For Pendau, a Malayo-Polynesian language of central Sulawesi, Quick (2010) recorded 300 names, of which 73% were uninomials.
For reasons to be discussed below, it is likely that none of the inventories of fish names referred to above is exhaustive, not even those numbering over 400. There are various means of spotting the gaps and estimating their extent. We turn now to this matter.
| uninomials | binomials | % binomials | |
|---|---|---|---|
| Mutu | 104 | 41 | 28 |
| Gela | 252 | 100 | 27 |
| Wayan | 352 | 132 | 28 |
| Marovo | 194 | 122 | 38 |
| Satawalese | 278 | 122 | 30 |
| Titan | 279 | 8 | 3 |
| Uvean | 180 | 104 | 36 |
| Niua | 138 | 71 | 34 |
| K’marangi | 148 | 114 | 43 |
| Kiribati | 132 | 19 | 14 |
| Teop | 168 | 29 | 15 |
| POc | 140 | 1 | 1 |
| PCP | 150 | 3 | 2 |
| PPn | 112 | 2 | 2 |
One method of detecting gaps is simply to compare the total number of fish names recorded for each language in a representative sample. It can be seen that, among the 12 languages cited in Table 3.9, the total number of names varies greatly.27
Insofar as all these languages are spoken by fishing communities who use similar fishing techniques and have access to a pretty similar range of fish families and genera, and indeed species, we would expect the size of their lexicons to be more similar than they are. On these grounds alone it can be inferred that the lists numbering below 300 are probably incomplete, especially for the languages of Melanesia, and lists below 200 are seriously incomplete. However, it must be allowed that in some Pacific Island communities there has in recent times been a reduction in the range of fishing methods employed and a decline in knowledge of fish names and fish ecology.
Another method is to compare the ratio of uninomials and binomials that have been recorded, as well as the absolute numbers of each (Table 3.10).
Taking the five languages with the highest number of uninomials in Table 3.10 as representing the best described languages, we find an average of 271 uninomials. It is noteworthy that the three Polynesian languages in the sample range between 138 and 180 uninomials, well short of this figure.
In the best documented Oceanic fish taxonomies, binomials normally amount to between 27 and 38 percent of the total names (median 30 percent). If the proportion of binomials recorded for a language deviates markedly from this range it stands out as suspiciously anomalous. In three languages, Titan, Kiribati and Teop, the percentage of binomials is 15 percent or less. It is likely that in each case the list of binomials is seriously incomplete. Most striking is Titan, where binomials make up only three percent of the 287 fish names. We can infer that roughly 100 Titan binomials were not recorded. At the other extreme, Kapingama- rangi (43%) is considerably above the median; it may be that in this case the list of binomials includes some ad hoc descriptive forms and/or that the list of uninomials is far from complete.
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| Serranidae | 12 | 28 | 14 | 26 | 26 | 21 | 19 | 7 | 6 |
| Carangidae | 6 | 12 | 16 | 21 | 43 | 10 | 18 | 11 | 12 |
Yet another indicator of how complete the record of fish names for a particular language is, comes from cross-linguistic comparison of the number of taxa or names for each particular biological family (or sometimes a group of related families). Fairly consistent patterns emerge as to how many taxa are typically distinguished in each family or larger group. If the number of recorded taxa in a language is much lower than the average or median this is a sign that the list is incomplete.
Tables 3.11-3.22 give the number of names recorded for certain groups of fishes in eight contemporary languages plus PPn. The groups are usually a single family or a pair of closely related families, though in the case of sharks, rays and eels, each group consists of several families. In the case of PPn, it is to be expected that the totals for each family will generally be below the average for contemporary languages because the PPn reconstructions consist almost exclusively of uninomials and because the PPn list is probably more seriously incomplete than that of any of the contemporary languages.
The Serranidae and Carangidae (Table 3.11) both contain many species and many important food fish. Morphologically distinct species in each family are consistently classified into many taxa. (There seem to be no cases where a language has a generic for the diverse Serranidae family and the same is true for the Carangidae. Generics exist with much more restricted ranges.) For the Serranidae, the Teop list is anomalous in having only seven names. Of the 26 Gela terms for Serranidae, as many as eight are synonyms, so the number of distinct taxa is only 18. For the Carangidae, Niuatoputapu and Wayan are anomalous in having as few as six and 10 names, respectively, and Gela in having as many as 43. The Gela list includes four terms for growth stages (other than the unmarked, mature stage term) and five synonyms.
With hundreds of species each, the Labridae and Scaridae (Table 3.12) are two of the largest and morphologically most diverse families. Both families contain many important food fish. Kiribati and Niuatoputapu are anomalous in distinguishing only three and four terms for Labridae. Wayan stands out in distinguishing as many as 48. Twenty-seven of the Wayan terms are uninomials, so even if we discount some binomials as being merely descriptive the number of valid taxa remains high.
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| Labridae | 4 | 9 | 3 | 14 | 17 | 48 | 10 | 14 | 4/5 |
| Scaridae | 9 | 14 | 5 | 28 | 11 | 16 | 18 | 13 | 5/6 |
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| Mullidae | 8 | 10 | 3 | 15 | 8 | 10 | 9 | 6 | 6 |
| Mugilidae | 7 | 7 | 5 | 8 | 2 | 2 | 7 | 3 | 4 |
Mullidae, goatfish (aka grey mullet, Table 3.13) are a fairly numerous family of shallow water feeders, usually caught in nets or traps, and which are easily identified by the pair of barbels on the chin. There is usually a generic term for the family as well as a number of named kinds. Mugilidae, mullet (aka red mullet), are a small family of shallow water feeders, with just a few species present in any one locality, but are economically important and each species is often distinctively named. The Gela, Wayan and Teop lists are anomalous in distinguishing only two or three Mugilidae taxa.
The scombrids (Table 3.14) include several species much valued as game fish and usually finely discriminated in indigenous Oceanic taxonomies. However, the family is morphologically fairly diverse and it is not usual to find a generic for the whole family. Wayan is on the low side in number of scombrid taxa. Xiphiidae are a very small family of morphologically distinctive deep sea fish. The lists for Niuatoputapu and Gela lack names for two prominent kinds of Xiphiidae: swordfish and marlins, and Teop, unusually, subsumes all kinds in one taxon.
The Acanthuridae (Table 3.15) are a moderately us family of reef grazers which occur in schools. The larger species, especially, are valued as food and morphologically distinctive species are usually named separately. Satawalese has an exceptionally high number of Acanthuridae names, as many as 28 being uninomials. The Siganidae are a fairly small family of reef grazers with very sharp fin spines. They are good eating and are usually finely discriminated in the taxonomy. The Satawalese, Niuatoputapu, Uvean and Kiribati lists appear to be missing some terms.
The Holocentridae (Table 3.16) are small, brilliant shallow water coral fishes. There is often a generic for the family and invariably several taxa are named. The Balistidae, grazing fish of the coral reef and weedy shallows, number many species. Most are distinctively marked with bright colours and many are separately named. Some are poisonous. The Aluteridae comprise many species of small or medium sized grazing fish, similar in habits to the Balistidae. Some species are considered poisonous. Aluteridae and Balistidae are sometimes subsumed under a single generic in Oceanic languages. There is a marked gap between Satawalese and Wayan and the rest in number of names for Holocentridae and for Balistidae and Aluteridae.
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| scombrids | 8 | 12 | 7 | 10 | 9 | 6 | 13 | 6 | 4 |
| Xiphiidae | 1 | 1 | 3 | 2 | 1 | 3 | 2 | 1 | 1 |
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| Acanthuridae | 13 | 23 | 10 | 32 | 19 | 20 | 10 | 11 | 9 |
| Siganidae | 3 | 4 | 1 | 4 | 19 | 10 | 13 | 7 | 2 |
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| Holocentridae | 4 | 11 | 5 | 14 | 5 | 10 | 5 | 5 | 6 |
| Balistidae and Aluteridae | 5 | 11 | 3 | 22 | 11 | 25 | 9 | 11 | 1 |
The Lutjanidae (snappers, Table 3.17) are a large family of carnivorous fish, mostly bottom dwellers moving in schools in outer coral reefs and lagoons. The Lethrinidae (emperors) comprise about 20 species living mainly in shallow, coastal waters. Noteworthy here is the high number of terms for kinds of snapper in Gela, and the low numbers for Wayan, and also the high number of Titan terms for emperors. The fact that authors providing lists of fish names and identifications for Oceanic languages often follow conflicting scientific taxonomies creates difficulties in comparing group totals across languages. Taxonomists disagree, for example, over whether Lethrinidae and Plectorhinchinae (grunters, sweetlips) should be assigned to the family Lutjanidae or to separate families. Members of these closely related groups are carnivorous, usually have large heads, equal jaws and a forked tail-fin. They are an important food source and many taxa are distinguished but as a rule, Oceanic languages have no generic terms for any of the families (or sub-families) in question.
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| Lutjanidae | 13 | 16 | 13 | 8 | 23 | 7 | 17 | 11 | 10 |
| Lethrinidae | 3 | 10 | 7 | 13 | 15 | 10 | 23 | 7 | 3 |
| Plectorhinchinae | 0 | ? | 0 | 2 | 6 | 12 | 7 | 1 | ? |
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| sharks | 7 | 13 | 7 | 10 | 10 | 13 | 8 | 6 | 5 |
| rays | 5 | 7 | 5 | 7 | 4 | 9 | 5 | 10 | 3 |
Sharks and rays (Table 3.18) each fall into several families, some with several species. Individual species tend to be salient because of their size and distinctive morphology and behaviour and, in some cases, because they are eaten. In Oceanic languages there is almost always a generic term for sharks and another for rays, and the generics generally have several subtaxa. The Kiribati, Niuatoputapu, Titan and Teop totals for sharks are on the low side, as is the Gela total for rays.
Eels (Table 3.19) fall into several families, chiefly the Anguillidae (freshwater eels) and several families of sea or brackish water eels: Muraenidae (morays), Ophichthydae (snake eels), Congridae (conger eels) and Muraenesocidae (pike eels or pike congers). This highly salient group of fish is usually finely classified in Oceanic languages. There is often a generic term for eels as well as terms (often with subtaxa) for each of the families.
Sphyraenidae (barracuda, Table 3.19) are a small family of predatory fish, large fast and elongated, with very large sharp teeth that make them dangerous to humans. They are good eating. Generally, several growth stages and sometimes two or three mature kinds of barracuda are distinguished. It can be seen that several languages are on the low side in number of recorded eel and barracuda taxa.
The Clupeidae (Table 3.20) are small migratory shoaling fish of shallow coastal waters. They are good eating and are taken in nets. The lists for Gela, Niuatoputapu, Satawalese and Teop score notably lower than Titan and Wayan in number of Clupeidae, and it is likely that the lists are incomplete. The only member of the Chanidae is the milkfish, Chanos chanos, which is everywhere distinguished. The Atherinidae are a small family of silvery, sardinelike schooling fishes which are mainly used as bait.
The genus Synanceia (stonefish) is sometimes included by ichthyologists in the family Scorpaenidae (scorpionfish) and the two groups are lumped together here (Table 3.21). Species of this morphologically diverse family all have long poisonous pectoral and dorsal spines. The Synanceia have the most potent of all the fish venoms, secreted from glands at the base of their needle-like dorsal fin spines. They have large grotesquely shaped heads and mouths and conceal themselves by lying in mud, sand, coral or weeds, and are extremely dangerous. Wayan distinguishes many more scorpionfish and stonefish taxa than the other languages. The fact that most Wayan names are uninomials suggests that lists for some of the other languages are incomplete.
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| eels | 10 | 16 | 7 | 14 | 8 | 10 | 6 | 9 | 3 |
| barracuda | 2 | 5 | 2 | 5 | 3 | 5 | 5 | 4 | 3 |
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| Clupeidae | 1 | ? | 5 | 0 | 1 | 7 | 7 | 0 | 4 |
| Chanidae | 1 | 3 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| Atherinidae | 0 | 1 | 0 | 0 | 3 | 1 | 0 | 0 | 0 |
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| Scorpaenidae and Synanceiidae | 1 | 3 | 1 | 7 | 4 | 16 | 4 | 2 | 1 |
The Blennidae (blennies, Table 3.22) are a large family of agile, small elongated fish of shallow reef waters and tidal pools. The Gobiidae (gobies) are usually very small fish of tidal pools, reefs and estuaries. The Periopthalmidae (mudskippers) comprise a few small amphibious species of mangrove swamps and muddy shores. Figures for these three families are lumped together here both because lexicographical identifications are often vague, and because ichthyologists have trouble deciding how many families should be distinguished. The totals for Satawalese and Gela are spectacularly higher than the rest. Most Gela terms are binomials and it is possible that some represent ad hoc descriptive epithets rather than conventional names.
| Niua | Uvean | Kiribati | Satawalese | Gela | Wayan | Titan | Teop | PPn | |
|---|---|---|---|---|---|---|---|---|---|
| Blennidae, Gobidae and Periopthalmidae | 1 | ? | 0 | 23 | 28 | 5 | 2 | 1 | 1(?) |
This survey of family name totals suggests that even the larger inventories of fish names for Oceanic languages, those in the 350-480 range, are not complete. The Wayan list, for example, shows significantly below average numbers of names for Carangidae, Lutjanidae, Mugilidae and Scombridae.
POc speakers lived in the Bismarck Archipelago, a region with a very rich fish fauna. Lexical evidence (Osmond 1998) and archaeological evidence (Walter 1989, Kirch 1997) indicate that they exploited a wide range of fishing techniques. Given these considerations, we can infer that the number of fish taxa distinguished by POc speakers was on the same order as the largest numbers recorded for contemporary Oceanic languages, in the 400+ range, and that about 70 percent of fish names were uninomials.
The total number of fish names so far reconstructed for POc (145) falls far short of the number that we estimate were used by speakers of the language. Why are so many names missing from the reconstructed inventory? A similar question can be asked about PCP (150-160 reconstructions) and PPn (115).
One category of missing names can easily be identified: binomials. The only secure binomial reconstructed for POc is *paRi manuk ‘eagle ray, Aetobatus narinari (lit. ‘bird ray’). (*saku-layaR ‘swordfish’ is historically analysable into *saku ‘needlefish and long toms’ plus *layaR ‘sail’, but it was not a binomial: it is clear from contemporary evidence that *saku-layaR were not a kind of *saku) Of the 160 reconstructions in Geraghty’s (1994) list of PCP fish names, just two are binomials. Of the 115 fish names that Hooper (1994) reconstructs for PPn, just three are binomials. The lack of widespread fully cognate binomial terms confirms what many commentators (e.g. Rensch 1983) have said: binomials, or to be more exact, the secondary or modifying terms in binomials, are notoriously unstable. Binomials generally make up around 30 percent of the total taxa in Oceanic fish taxonomies (see §3). The chances are that most of this 30 percent of missing POc terms will never be recoverable.
The dramatic difference between uninomial and binomial retention rates, presumably, has something to do with the geographic distribution of species as opposed to genera. Folk generics, represented by uninomials, are typically applied to two or more species of a biological genus and are sometimes applied to whole families. The same genera and families of fish tend to occur throughout the tropical Indo-Pacific. By contrast, folk specifics, often represented by binomials, typically apply to a single biological species or a group of closely related species and the geographic distribution of species tends to be more localised. Thus, modifying terms for folk specifics do not ‘travel’ as well as generic terms. But perhaps the main reason is that while generic names are usually arbitrary, modifiers almost always describe a feature of the morphology, the behaviour or the ecological niche of particular species. Accordingly, a number of competing modifiers may suggest themselves as equally convenient ways of distinguishing a particular member of a folk genus, so that even when a species is ubiquitous the original modifier in a binomial will often be subject to competition.
So much for the binomials. This still leaves many missing uninomials unaccounted for. One might seek to explain variations in durability among uninomials in terms of (a) where names rank in the taxonomic hierarchy, (b) the relation of particular taxa to humans, e.g. their usefulness, importance in cosmology and ritual, or danger, (c) inherent properties of the referents, e.g. how striking or unusual they are in appearance or behaviour, and (d) the geographic distribution of particular species and genera.
While factor (d) very likely has had a considerable influence on the durability of binomials I do not think it has played much of a part with respect to uninomials. The latter generally refer to genera, families or high-order groups, and in those cases where they refer just to a single species it usually is one that is highly distinctive and widespread in the tropical Pacific. This is the case for the few POc reconstructions that can be identified as referring, or probably referring to a single species, e.g. *qawaq ‘milkfish, Chanos chanos’, *qume ‘unicornfish, Naso unicornis’, *mamin ‘double-headed Maori wrasse, Cheilinus undulatus’ and possibly *qalun ‘skipjack tuna, Katsuwonus pelamis’.
A reasonable hypothesis is that, other things being equal, the persistence of a name will correlate with its rank in the taxonomy: names of life-form taxa will be the most persistent, followed by major generics, minor generics, and specifics, in that order. There is some evidence in support of this hypothesis although what we can say is limited by the lack of precise semantic reconstructions for many POc names. The life form *ikan ‘fish’ and the major generics *bakewa ‘shark’, *paRi ‘ray’ and *kanase ‘mullet’ show high retention rates. However, the generics *bebek ‘butterflyfish and coralfish’, *jumu ‘triggerfish and leatherjackets’, *saku ‘needlefish and long toms’ and *tuna ‘freshwater eel’ all score quite low. In general minor generics have relatively low retention rates but the single most durable fish name, *(n,ri)opuq ‘stonefish’, was probably a minor generic, memorable because the stone fish is so dangerous.
I have not carefully scrutinised evidence bearing on the hypothesis that, at the level of folk generics and folk specifics, names for economically valuable fish will be more stable than other names (all other things being equal). There is some evidence that this is the case. For example, names for various members of the highly valued scombrid, carangid, rock-cod, snapper and mullet families tend to show high retention rates.
Further research should shed more light on these matters.