- 443

In spite of Fiji's relatively early entry into European consciousness (1643), and the central location and size of its two main islands, outsiders knew very little about the group or the languages there until the close of the eighteenth century. Much earlier, Tasman's report of a plethora of dangerous reefs kept Dutch ships from the area for a century, and the difficulties of later captains, such as Bligh and Wilson, bore out his report. For a time, these navigational difficulties, combined with Bligh's observations on the hostility of the Fijians, continued to keep European contact with the islands to a minimum. But even these obstacles were eventually surmounted by the all-powerful motive of profitable trade, based not on a European market for oriental spices and other such exotic commodities, but instead on an enthusiastic market in the orient itself for a parasitic tree: various species of Santalum, or, sandalwood. 2

Thus inspired by potential profits, two ships — the Marcia and the Fair American — purchased sandalwood in Fiji for the first time in 1804, - 444 and in the following year, Governor Gidley King of New South Wales wrote:

One of the earliest reports of just how advantageous that commerce could be is that of William Lockerby (1782-1853), born in Scotland but serving on the American ship Jenny. He wrote that the market value of the ship's cargo, acquired at a cost of approximately £50, was £20.000. 4 For such stakes many ship captains could no doubt have been persuaded to risk the reef-infested waters and a dubious reception from the Fijians. The captain of the Jenny was of such a mind and, according to Lockerby, wished to hold a monopoly on the trade as long as possible. It was this intention combined with a disagreement between Lockerby and the captain that led to what is most probably the first word and phrase list collected for a Fijian language. Lockerby explained his situation:

Lockerby made good his threat, and although it is difficult to know how effective it was in eliminating a monopoly, it was at least the means of preserving his conception of some Fijian words and phrases. In 1811, one - 445 of the copies, called “Directions for Fegee Islands by Wm Lockerby”, 6 fell into the hands of another sandalwood trader, Captain William Putman Richardson, of Salem. In addition to the sailing directions, the manuscript includes advice about carrying arms on shore, the most desirable articles for trade, personality sketches of various chiefs, and a “vocaboulary of their Tongue sufficnt [ninety-three items] to purchase Sandle Wood”.

Since the manuscript is not dated, we have little idea of how long Lockerby had been in Fiji when he compiled it, except for his statement that by the time the Jenny sailed, he “spoke the language in a tolerable manner”. 7 Like many documents written in a language whose orthography is as yet unfixed, the word list tells us more than the compiler intended. For instance:

1. The Fijian dialect area: Such a form as andonga (e duga) confirms the source as Vanua Levu, and specifically as Bua Province.

2. Lockerby's dialect of English: Including an r in his spellings iarse, ierse, and irse for yasi “sandalwood” and argasow for a gasau “the arrow” shows that Lockerby pronounced a postvocalic r as [∂], as in modern Received Pronunciation.

3. Difficulty in transcribing unstressed vowels: Lockerby's problems in identifying the unstressed vowels in Fijian pronunciation can be illustrated by some of the numerals he listed. In each case, the final vowel is unstressed. The correct forms appear in parentheses:

4. Features reflecting English spelling: English orthographic patterns are notorious for their inability to provide a consistent system for English itself, so it is no surprise to find its inconsistencies carried over to Lockerby's transcription. It is not difficult to interpret his -y as -e or -i, or his oo as u, but ng can represent two sounds in Fijian: [ŋ] and [ŋg] (written now as g and q). But then, Lockerby's spelling of English showed a degree of originality and variety not often encouraged by schoolmasters: “parshall” for “partial”, “grait” for “great”, and “gaird” for “guard”. 8

5. Grammatical recutting: Some of the forms cited are spelled with their article attached: Amboca for a buka “the firewood”, Antakie for a dakai “a bow, musket”. Lockerby did the same with some place names (as others did with Otahiti for 'O Tahiti): his Myimboo is actually mai Bua “from or at Bua”. 9

6. The character of the phrases: Lockerby's sandalwood-collecting sentences are rather pidginised, and are certainly without the niceties of expression that politeness might demand. Two examples are given below, - 446 first in Lockerby's transcription, then respelled according to the standard orthography, and finally “adjusted” to what might be a more reasonable translation.

As Lockerby was preparing to leave Fiji, in Tahiti a series of uprisings against the king and against Christianity and its bearers caused the missionaries there to ponder “their great unprofitableness in the work of the mission”. 11 Finally, in late October 1809, they boarded the Hibernia, originally bound for Canton via Fiji, where it was to stop long enough to pick up a cargo of sandalwood. But the reefs off the north coast of Vanua Levu damaged the ship so badly that it stopped at a small island off the Macuata coast while repairs were made. 12

One of the missionary party, the Rev. John Davies, was a careful diarist and later the chronicler of Tahitian mission history. Having arrived in Tahiti in 1801, and considered to be “something of a grammarian and, compared with his companions, a scholar”, 13 Davies was thus in a position to make the first professional statements about the Fijian language and culture. At first, he and his brethren had little contact with the Fijians, but as their stay stretched on to seven weeks, more trading was necessary to obtain supplies.

By the end of December, Davies was writing Fijian phrases, and after he left the Fiji Islands in late January, he thought it “might be expected we should make some observations on them and their inhabitants”. 14 Among lists of islands and descriptions of some customs appears a short word list (20 items), intended to show that “the main part of the language has no affinity with [Polynesian languages]”. The list, mistakenly called by im Thurn (the editor) a “somewhat confused vocabulary”, 15 shows a fairly sophisticated and systematic orthography, whose form is directly attributable to decisions made by the representatives of the London Missionary Society in Tahiti in March 1805. At that time, the five vowel letters (with something close to a “continental” pronunciation) and the consonants b, d, f, h, m, n, p, r, t, v, and w were chosen to make up the Tahitian alphabet. “Mr D[avies] and several others thought simplicity and - 447 uniformity (neither of which is characteristic of the English Alphabet) ought to be aimed at, so that the natives might easily learn to spell their own words . . .” 16 Incidentally, the b and d were included to represent sounds in loan words, although “it appeared a fact that the natives used b and p, d and t indifferently without observing the distinction”. 17

At a later meeting, the earlier (and more functional) five-vowel proposal was altered. In the new system, a was to represent the sound in “father”, Greek E that in “face”, and e that in “me”. One gets the impression that Davies did not favour this altered system, but he adopted it, and at least one convention — e representing [i] — is reflected in his spelling of Fijian words.

Since the Davies list 18 is short, it is reproduced here, with the now-standard spelling of the words. The numbers following the forms represent principles discussed following the list.

• 1. In many areas, the sound written in Bauan as y- is merely the fronting of a following a.
• 2. Here, the article a is written with the word.
• 3. The v, actually bilabial, was sometimes heard as a w, or p.
• 4. Here is illustrated the Tahitian mission's system of writing [i] as e. Davies also wrote Feje, which is consistent with the system.
• 5. Note in the discussion above what Davies wrote about Tahitian p and b, t and d. His k-g mishearing fits the same pattern.
• 6. For the forms Davies wrote with ng, four actually have [ŋ], and one

- 448

• [ŋg]. (Note that the English orthography has no way of distinguishing between these two sounds.)

Davies, with his superior training, and — more important — his experience with Tahitian, might have been able to present the scholarly world with a fairly accurate word list for Fiji. As it turned out, however, his manuscript remained unpublished for over a century, and he cited only a few Fijian words in his Tahitian grammar, printed in 1823. Even so, it is likely that his comments at that time are among the first published on Fijian.

Finally, and by a circuitous route, one of the sandalwood lists was successful in bringing the Fijian language to the attention of European and American scholars — without a century's wait for publication. When Captain Richardson deposited Lockerby's manuscript in the collection of the East India Marine Society, 19 he added his own and considerably more extensive list, collected sometime in 1811, at the wane of the sandalwood trade. The unlikely vehicle for its publication was the Journal of William Ellis, Narrative of a Tour of Hawaii, or Owhyhee; with Remarks on the History, Traditions, Manners, Customs and Language of the Inhabitants of the Sandwich Islands. Ellis (1794-1872) had been a missionary to Tahiti for a period long enough for him to have claimed later that he had devoted ten years to the “study of the uncultivated languages of the Pacific”. 20 It was this interest in Pacific languages that prompted him to follow his comments on the Hawaiian language with Davies' on Fijian, supplemented by a “Vocabulary of the Fejeean Language”. He justified its inclusion:

The immediate source mentioned was John Pickering, whose insightful principles outlined in his “Essay on a Uniform Orthography for the Indian Languages of North America” were in part responsible for the Hawaiian orthography. Those principles are not represented in his copy of Richardson's list, since he himself was not able to hear the language spoken. Thus Richardson's English-based orthography stands. Pickering was able, however, to alter the list to fit in with a pet hobby of his: Catherine the Great's “root words”, an ambitious project in comparative vocabulary. - 449 Some 55 words were printed in upper case, thus indicating their relationship to the Empress' list of some two to three hundred words. 23

Ellis alphabetised the list, deleted some 90 words and phrases, a seemingly arbitrary selection except for Richardson's last page. After concentrating on numerals, common words, and useful phrases for collecting sandalwood, the captain had elicited the Fijian translations for “piss”, “fart”, “a private vice”, “excrement”, and “a crooked c — t”. These the Rev. Ellis saw fit to delete in toto, even though they might have titilated the Empress, had she lived to read them.

Judging from Richardson's spelling of English, 24 he was much better educated than Lockerby. Still, his romanisation of some Fijian words shows possible influence from the latter's list. For instance, it is unlikely that both would have spelled waqa “boat” as wankey. For the most part, however, his vowel representations are more accurate than Lockerby's, especially his hearing of unstressed vowels. His spelling of the set of numerals, for example, is much improved. Still, the orthography remained essentially that of English, as Ellis remarked in his introduction to the list.

Grammatically, Richardson repeated Lockerby's error of attaching the article a to many nouns. Of a set of 18 words listed with initial a-, only two actually belong there. Semantically, Richardson went beyond his predecessors in explaining many forms with something more than a one-word gloss. For example:

One particularly interesting gloss is that for Pappelange (vavālagi): a white man or a ghost. The latter meaning is not widely known today and may explain what the protagonists of a minor controversy among amateur philologists in Fiji were unable to. 25

Eventually, the list (Pickering's copy, not the abbreviated Ellis version) made its way into the hands of Wilhelm von Humboldt. Oddly, Humboldt regretted that the words were rather too specialised — names of plants, herbs, and goods — to be of much use for general comparative linguistics. 26 It seems a curious opinion, because of the 55 words that coincided with the Empress's list. Humboldt compared a number of items with Polynesian cognates, but the list is by no means exhaustive. 27

Richardson's list seems the last contribution of the sandalwood traders, for, early in the second decade of the century, they were replaced by seekers of beche-de-mer, this too for an Asian market. Unlike their predecessors, these traders left (so far as we know) no records so esoteric as word lists. Finally, the practical motives of the early visitors were replaced by a more scientific one.

- 450

One such discovery voyage was that of the Astrolabe, commissioned by France to find out the fate of the missing ship La Perouse. Like many ships at that time (1827), the Astrolabe approached Fiji from Tonga, sailing north through the Koro Sea almost to Taveuni, then southward to Matuku, and finally across to Nukulau, where she anchored. 28 This last spot is a likely one for Gaimard, 29 the zoologist, to have collected a word list of almost 300 items, plus numerals, from an informant named Tomboua-Nakoro [Tabua-Nakoro?] 30 with the help of interpreters from Manila and Guam. 31 It is difficult to imagine communication resulting from this linguistic ménage, but the list is surprisingly accurate.

Even the transcription shows a high degree of consistency. Except for representing [u] by ou, a French habit that seems to die hard even in their work with Austronesian languages, Gaimard used the “continental” vowel system. 32 He neatly avoided the ng pitfall into which the first English transcribers stumbled, keeping [ŋg] and [ŋ] apart by writing them ng and nh. But other consonants provided expected difficulties for a speaker of French. The greatest of these was c. Gaimard transcribed ciwa “nine” as zioua, zivoua, tiva, and diva. These examples illustrate another predictable problem — w — but in many examples, it is transcribed correctly.

As in the former lists, the bilabial v was sometimes written as b (as in binaka for vinaka “good”), or as v (vizaï 33 for vicai “copulate”). It is probable that Gaimard was aware of his inconsistency here, for he gave alternate forms for some words, such as lebou, and levou for levu “large”.

It is the alternate forms of a different nature that give importance to Gaimard's list. First, differences in pronunciation between Bau and Lakeba are pointed out. For example:

This and other examples illustrate the palatalisation of d and t before i that occurs at Lakeba and other parts of the Lau group. Other pairs point out vocabulary differences between the two groups. But above all, the presence of the alternating forms illustrates the sensitivity of the Fijians to dialect variation, a phenomenon that was later to prove a barrier to translators and lexicographers.

Occasional individual words also stand out. For instance, the use of mazou (macu) “satiated” is now (and perhaps was then) definitely non-Bauan, but was common in Lau and a fairly small area near Nukulau, - 451 where the Astrolabe was anchored. The translation for “to smoke”, ouvou tabaka (uvu tabaka), contains the first recorded instance of a loan word from English. Literally, the form means “blow tobacco”, but the current idiom is kana tavako — “eat tobacco”, and within Fijian's gender system, the word is now classified in the “edible” category, along with literal foodstuffs. In this form, the b probably represents v; Gaimard's confusion between the two sounds is illustrated above. Then, too, the French spelling of the source could have influenced him to write b.

Grammatically, Gaimard duplicated his predecessors' difficulties with word division. For some forms, for example, Sampon ponhi (sa bogibogi) “it is morning”, the prenasalisation for the stop consonant appears as part of the preceding word. In the next decade, the eventual solidifiers of the Fijian orthography, Cargill and Cross, were to experiment briefly with the same solution to a troublesome phonemic problem. This example shows another common problem: writing the verbal particle sa as part of the same word. And although a still appears prefixed to some words, d'Urville wrote that it seemed only to indicate the article. 34

Because of the quality of their work, it is unfortunate, first, that Messrs d'Urville and Gaimard did not have the time to collect a more extensive vocabulary, and second, that the Wesleyans who followed them apparently never saw the word list. If they had, they might have avoided some problems with Fijian orthography, especially the Tonganised spelling Feejee, which persisted long enough to establish a pattern of European mispronunciation of the name. 35

Even though sandalwood traders and explorers helped open Fiji to the rest of the world, their stays there were fleeting and their influence minimal. Hindered by lack of time or inclination to study the language seriously, they gathered word lists that were products of perhaps curiosity, a few hours of leisure time, or a desire to purchase and load sandalwood more effectively.

The first word lists approaching the scope of a dictionary grew out of a different and non-marketable commodity — some hundreds of thousands of Fijian souls to be won for Christianity. The initial efforts at conversion can be traced to the Rev. Mr Davies, who, after his exile from Tahiti, returned there in 1812. His opportunity to introduce Christiantiy to Fiji arrived in 1825 in the person of Takai, a footloose Fijian, who, desirous of seeing the world, travelled as far as Port Jackson (Sydney) and thence to Tahiti. After he had observed the activities of the mission there, he requested that Christian teachers be sent to Fiji. To this end, two Tahitians — Hape and Tafeta — were appointed to go. To help prepare their way, Davies and Takai compiled a primer — Sa Alphabeta Na Vosa Faka Fiji — said to be the first book printed in the Fijian language. 36

- 452

After a two-year stay in Tonga and a return to Tahiti, the two teachers, joined by a third, finally reached Fiji in 1830. But once there they proved to be singularly ineffective in their efforts to convert the Fijians, or even to learn the language. 37 Nor were they particularly encouraged in their mission by Fijians with any influence or power. Tui Nayau, of Lakeba Island, by now conscious of the aura of prestige and riches surrounding Europeans, held out for vavālagi missionaries until in 1835, two representatives of the Wesleyan Methodist Mission Society landed at Lakeba. One of these, David Cargill (1809-1843), was largely responsible for the subsequent work on the language, for, like Davies, he was considerably better educated than his colleagues.

The approach of the Wesleyans to the problems of language analysis was orderly and explicit. Cargill's instructions read:

After his arrival at Lakeba, Cargill wasted little time. When only a year had passed, he reported that his dictionary was in progress: “I have inserted in it nearly 3000 words, exclusive of the names of persons & places: the number is being daily increased. The accent, — the pronunciation, meaning & derivation of the words are attempted.” 39

By July 1838, Cargill had collected “5000 or 6000 words with their signification, accentuation and probable derivation.” 40 A series of events, however — the death of his wife, his removal to England, his subsequent reassignment to Tonga, and his suicide there — prevented the publication of most of Cargill's work. But his linguistic studies, although often unacknowledged by his colleagues, served not only as pedagogical material for newly arrived missionaries, but also as a foundation for the published works that followed. His own manuscript survives only 41 in the form of a copy by the Rev. Thomas Jaggar, with additions to Cargill's Lakeba words from the areas of Bau, Rewa, and “Nandy”. 42

Because of its scope and attention to at least some grammatical details, Cargill's Lakeba glossary went far beyond its predecessors, approaching the status of a dictionary. 43 Features that are distinctive or at least much - 453 more developed than in earlier works are: (1) some diacritics to indicate pronunciation; (2) grammatical information; and (3) a large number of entries for other areas.

Since, by the late 1830s, the missionaries had developed the Fijian writing system to almost its present form, 44 Cargill's diacritics to aid pronunciation seem not to convey much reliable information. He used macrons and breves 45 but did not indicate whether they marked vowel quality or quantity. For some words, a lack of diacritics is especially confusing. Examples are two loan words from English — prisoni “prison” and palesi “palsy” — each of which should show some compensation of stress and length to match the English model. 46

It is Cargill's grammatical annotations that separate his work from the earlier lists. His first contribution, part-of-speech affiliation, followed the classical format in which he worked: a division into substantive or noun, adjective, verb, adverb, interjection, article, and personal pronoun. Bases with multiple functions were treated like the following: “Babalavu a[djective]. s[ubstantive]. long-length”. Or, “Funumalie n[oun]. peace, quietness”; Funumalie [a new entry] a[djective]. quiet, easy”.

In addition, Cargill wisely realised that it was necessary for a dictionary to identify certain nominal bases that required suffixed possessive forms (such as “Ba, na s. a branch”) and verbs that were capable of appearing with suffixes indicating a grammatical function somewhat like transitive (such as “Ba, takina, v. to deny”). Curiously, he chose to list derivatives separately, rather than under the common base. Thus, one finds as head words lakoni, lako, lakomai, lakova, and lakovakanadaku — all forms of lako “go”. Examples of the same kind of proliferation of entries are 16 pages of forms with the common prefix vaka- (causative, simulative) and four pages with vei- (reciprocal, distributive). Thus, Cargill's practice of separating derivitives as well as listing some bases twice, added greatly to the number of entries.

Because of the additions from Jaggar and Hazlewood, the dictionary shows clearly two kinds of sound correspondence among the Fijian languages. The first (for Lakeba) is a palatalisation of t and d before i, both (unfortunately) written as j. The following example shows the Lakeba and Bau forms: “Baji, -na s. a tooth. Bati, -na B[auan] dia[lect]”. The second, illustrated sparsely in Hazlewood's data from Vanua Levu, is a correspondence between t (in Bau and Lakeba) and glottal stop (in Nadi). The symbol for the latter is not written, for it was not considered a proper language sound: “In the Nandy Dialect the letter t is not sounded; it is always omitted . . .”

Cargill's Lakeba list also shows other characteristics of Lauan. Generally, all words in the list with f or p are from either Tongan or English. For example, funumalie “quiet” (Tongan has mālie) and papa - 454 “plank, board” are clearly Tongan. Words in the list from English are mainly the result of Bible translation: words for angel, Pharisee, fig, prison, prophesy, prophet, palsy, baptise, and Passover. But the secular world of the European had also begun to be mirrored in the vocabulary: farthing, lead pencil, slate pencil, and pussy.

Because of the difficulties presented by the multiplicity of languages in Fiji and complicated by the abandonment of Lakeba as the official language of the mission, Cargill's dictionary never reached publication directly. It did serve, however, as the basis for a work that achieved recognition in the scholarly world.

Just as Cargill was entering into the most difficult period of his stay in Fiji — his own debilitating illness, and the fatal illness of his wife and sixth child — the United States Exploring Expedition, under the command of Charles Wilkes, arrived in Fiji. Among its scientists was included Horatio Hale (1817-1896), a young Harvard graduate serving as ethnographer and philologist. During the brief time of their fortunate overlap, Cargill and Hale met, 47 and in July 1840, the month of Cargill's departure, Hale copied out his Lakeba grammar and vocabulary. Hale enumerated the sources for his final product:

As a result of supplementing Cargill's substantial contribution by the other sources mentioned, Hale's “Vitian Dictionary” is perhaps a modest beginning at a multidialectal glossary. But the majority of the 2,688 entries are for Lakeba, thus making the work principally a dictionary of that language, not the others.

Hale made other changes, not necessarily to improve on Cargill's work, but instead (as he indicated in his preface), to make the work conform to the other glossaries in the volume. For this reason, and as he wrote, “to facilitate the comparison of the different Oceanic languages”, 49 he - 455 changed the orthography in the direction of a more “phonetic” representation that could be more easily read by scholars: c to δ, g to ŋ, and the prenasalisation restored to the stops and trill: b to mb, d to nd, dr to ndr, and q to ŋq.

Also, evidently with a view toward comparison, he deleted all words that appeared to be borrowings. The only indication left that Lakeba uses some words with p is an occasional bracketed form such as “Vala[pala] rotten, rottenness”. Words with f were deleted entirely.

Hale dispensed with an overt part-of-speech classification, using instead such clues in the translation as that just illustrated, articles, and the sign of the infinitive.

Phonetically, Hale's experience and training put him far beyond Cargill. For instance, he was the first to mark vowel quantity (separating, for instance, such words as δava “what?” and δavā “hurricane”) 50 and the first to describe the phonetic nature of the so-called dropped k and t in Ra and some Vanua Levu languages: “In Somusomu and in many parts of Vanua Levu, the k is never pronounced, its place being supplied by a slight catching of the breath, as in the Samoan and Hawaiian.” 51

Hale was also the first to recognise the limited distribution of the sound his predecessors wrote as j: “In the dialect of Lakemba, the j is added, to express the sound of t before i, which is nearly that of tsh, — or such as is heard in the English words Christian, question.” 52

And although he continued writing [nr] as dr, he was aware that the d was in this instance excrescent: “The r is used both by itself, and preceded by n. In the latter case, the sound of d is generally, though not always, inserted between the n and r, merely, it would seem for euphony . . .” 53

The remainder of Hale's comments on the sound system are so astute that they bear quoting in full:

The v is one of the most remarkable elementary sounds in the language, on account of the wide range of its variations. Like the Spanish b, it is pronounced by closing the lips together, and according - 456 to the greater or less force of pronunciation, it is heard as v, f, p, or b, and occasionally even as m. Thus the word vanua, country, is sounded frequently fanua, banua, and panua; levu, great, is commonly pronounced nearly as leb'; δava; what, as δapa; and the name Viti levu has been written by different persons Feetee leb, Beetee lib, and, utterly corrupted, Metaleep. At the beginning of words, it is more often heard as f, and in the middle as b or p. In some few words, the sound of p is so distinct that the missionaries were induced, at first, to write it with this letter; but they find it impossible to keep up the distinction, and at present the sounds of p, f, v, and b, (not preceded by m,) wherever they occur, are expressed by the same letter, v.

Finally, whereas even a grammarian as late as Churchward (1941) arranged his discussion of Fijian sounds according to the order of the Roman alphabet, Hale attempted a strictly phonetic classification:

Although its categories are mixed (“labial”, “dental” and “gutteral” represent articulatory positions: “liquid” represents a manner of sorts), the scheme is innovative. For example, Hale realised that v, to a certain extent, occupied the position that p would usually take. The neat 4 x 4 chart shows Hale's attempt to fit the Fijian system into a symmetrical arrangement. Such symmetry of patterning is a feature often found in sound systems, but that of Fijian resists all attempts to fit it into such a scheme. Hale must have come to the same conclusion, for the question mark after the s indicates his uncertainty about its position.

With the publication of Hale's part of the Wilkes Expedition papers, reliable and detailed information was finally available for Fijian. But although Ethnography and Philology preceded the first full-scale dictionary by four years, still it was not as readily available as Hale might have wished. Fifty years (almost to the day) after he left Cargill in Rewa, he wrote, “Unfortunately the American Government had only a few hundres copies printed for distribution . . . and it is now a very rare book.” 56

- 457

While Hale was preparing his book for publication, Fijian politics were moving in a direction that was to have a profound effect not only on his and Cargill's work, but also on all the word lists collected earlier. Until then, European interests had begun in the east and had spread only slightly westward. But with the rise of Bau and its chief, Cakobau, came the selection of its language, rather than one from Lau or Vanua Levu, as the lingua franca. Thus, the mission efforts were redirected toward Bauan, and when Hazlewood's dictionary appeared in 1850, “Feejeean” was assumed to be Bauan. As a corollary, the other languages were relegated to the category of “dialects”, the early word lists were doomed to remain largely unknown, and the expectations of the first lexicographers were never fulfilled.

• CARGILL, DAVID, 1841. Memoirs of Mrs Margaret Cargill . . . including notices of the progress of Christianity in Tonga and Feejee. London, Mason.
• —— and others, 1839. Feejeean Dictionary. Mitchell Library Manuscript A 2065.
• CODRINGTON, Robert H., 1885. The Melanesian Languages. Oxford, Clarendon Press.

• DERRICK, R. A., 1957. A History of Fiji. Suva, Government Press. Third Edition.
• DODGE, Ernest S., 1972. “A William Lockerby Manuscript in the Peabody Museum of Salem,” The Journal of Pacific History, 7: 182-8.
• DUMONT D'URVILLE, J. S. C., 1834. Voyage de la Couverte l'Astrolabe. Paris, Ministère de la Marine.

• ELLIS, William, 1825. A Journal of a Tour Around Hawaii, the Largest of the Sandwich Islands, by a Deputation from the Mission on those Islands. Boston, Crocker & Brewster.
• —— 1827. Journal of William Ellis. Narrative of a Tour of Hawaii, or Owhyhee; with Remarks on the History, Traditions, Manners, Customs and Language of the Inhabitants of the Sandwich Islands. London, Fisher, Son, and Jackson.

• HALE, Horatio E., 1846. Ethnology and Philology, vol. vi of United States Exploring Expedition, 1838-42. Philadelphia, Lea and Blanchard.
• —— 1890. Letter to [?], Ontario, Canada, July 11. Manuscript in the Library of the School of Oriental and African Studies, University of London.
• HUMBOLDT, Wilhelm von, 1836-9. Über die Kawi-Sprache auf den Insel Java, 3 vols., Abhandlungen der der königlichen Akademie der Wissenschaften zu Berlin.

• IM THURN, Sir Everard Ferdinand and Leonard C. WHARTON (ed.), 1925. The Journal of William Lockerby, sandalwood trader in the Fijian Islands during the years 1808-9: with an introduction and other papers connected with the earliest European visitors to the islands, Hakluyt Society Series II, No. 52. London, Hakluyt Society.

• NEWBURY, C. W. (ed.), 1961. The History of the Tahitian Mission, 1799-1830 [by John Davies]. The Hakluyt Society, Second Series No. cxvi. Cambridge, The University Press.

• PICKERING, John, 1820. An Essay on a Uniform Orthography for the Indian Languages of North America. Cambridge (Mass.), The University Press.

- 458 Page is blank