^{page 620 note 1} See in particular Capell, A., A linguistic survey of the south-western Pacific (SPC, 136), Nouméa, New Caledonia, 1962, 186–8, 226–7Google Scholar; Anthropology and linguistics of Futuna-Aniwa, New Hebrides (Oceania Linguistic Monographs, 5), Sydney, 1960, 165 Google Scholar; ‘ Oceanic linguistics today’, Current Anthropology, III, 4, 1962, 380, 391–2Google Scholar.

^{page 620 note 2} The name of Polynesian outliers has been given to a number of small communities which are Polynesian in language and culture but living on islands situated in the geographical area of Melanesia. The outliers mentioned in the present article are Mae (or Emwae), Futuna, and Aniwa in the New Hebrides, and Tikopia due east of the Solomon Islands. In the geographical area of Polynesia a distinction is usually made on linguistic and other cultural grounds, between the western (or central) languages and cultures (e.g. Samoa, Tonga, and to some extent Fiji) and the eastern (or, more accurately the peripheral) languages and cultures (e.g. Tahiti, Marquesas, Cook Islands, New Zealand, and Hawaii).

^{page 620 note 3} I should like to express my acknowledgments and thanks to Professor Raymond Firth for his help in the preparation of this study.

^{page 620 note 4} Until, that is, the appearance of Elbert's, S. H. ‘Internal relationships of Polynesian languages and dialects’, Southwestern Journal of Anthropology, ix, 2, 1953, 147–73CrossRefGoogle Scholar. Even in that article, however, the evidence of Tikopian is not considered.

^{page 620 note 5} For a modern structural study of a Melanesian phonological system, see Scott, N. C., ‘A study in the phonetics of Fijian’, BSOAS, XII, 3–4 , 1948, 737–52Google Scholar, especially 740–3. Some of the languages of Melanesia have preserved reflexes of another liquid (reconstructed as γ or R) which seems to have left no trace in Polynesian).

^{page 620 note 6} The languages of western Polynesia have in general a richer phonemic and morphophonemic inventory than those of eastern Polynesia, from which they also differ in other characteristic ways.

^{page 621 note 1} Such a movement can only be described as reverse if one accepts the conventional view that Polynesia was settled originally from the mainland of Asia or from the islands of South East Asia or the Far East.

^{page 621 note 2} This point has been aptly made apropos of Rotuman by Goodenough in his comment to Capell's recent review article (‘ Oceanic linguistics today’, Current Anthropology, III, 4, 1962, 407–8Google Scholar).

^{page 621 note 3} That is Elbert's suggestion, put forward in his comment to the same review-article (pp. 405–6). That even a small community like that of Tikopia may have heterogeneous origins, is well illustrated in Firth's, Raymond History and traditions of Tikopia, Wellington, 1961 Google Scholar.

^{page 621 note 4} Dempwolff, O., ‘Die L-, R- und D-Laute in austronesischen Sprachen’, Zeitschr. f. Eingebvrenen-Sprachen, xv, 1924–1925, 19–50, 116–38, 223–38, 273–319 Google Scholar.

^{page 621 note 5} Brandes, J. L. A., Bijdrage tot de vergelijkende hlankker der westersche afdeeling van de Maleisch-Polynesische taalfamilie, Utrecht, 1884 Google Scholar.

^{page 621 note 6} Ibid 30.

^{page 621 note 7} Ibid 42, 138–40.

^{page 622 note 1} Biggs, B., ‘The structure of New Zealand Maaori’, Anthropological Linguistics, III, 3, 1961, 9 Google Scholar.

^{page 622 note 2} Pukui, M. K. and Elbert, S. H., Hawaiian-English dictionary, Honolulu, 1957 Google Scholar, p. xxix.

^{page 622 note 3} See the rubric for words beginning with l and n in Hawaiian, pp. 173 and 237.

^{page 622 note 4} C. Vernier, , Introduction à la langue tahitienne, Paris, 1959 Google Scholar.

^{page 622 note 5} Ibid., 14.

^{page 622 note 6} This is based on field-work carried out in Tonga in 1949. See also Churchward, C. M., Tongan grammar, London, 1953 Google Scholar, 1, ‘l: nearly as in led, but somewhat suggestive, at times, of r as in thread’.

^{page 622 note 7} Churchward, S., A Samoan grammar, Melbourne, 1951, 8 Google Scholar, states that ‘l is sounded as in English, except where it stands between a, e, o oru (not i) and i, as in aliali, appear; ‘eli, dig… (etc.). In these cases the l sound is less decided, and approaches to a rather indistinct r ord’; cf. Marsack, C. C., Teach Yourself Samoan, London, 1962, 16–17 Google Scholar, ‘(l) substantially the same as in English but pronounced more softly. At times the l is softened almost into an r. In fact l and r are often treated as interchangeable in Samoan, in that in introduced words containing an r …an l is often substituted for the r’.

^{page 623 note 1} Elbert, op. cit., 153.

^{page 623 note 2} Dempwolff, O., ‘Fiktion und Hypothese in der Sprachwissenschaft’, Annalen der Philosophie, 1922, 246–57Google Scholar.

^{page 623 note 3} A fiction being a scientific invention (Erdichtung) made for practical purposes and needing to be ‘justified’ whereas a hypothesis needs to be ‘verified’ and is concerned with reality; ‘Vaihinger grenzt die Fiktion gegenuber der Hypothese dahin ab, dass letztere entdecken will, Verifikation verlangt und auf die Wirklichkeit direkt ausgeht, erstere dagegen erfinden will, der Justifizierung bedarf und auf indirekte Berechnung der Wirklichkeit ausgeht’, Ibid., 246–7.

^{page 623 note 4} Ibid., 248.

^{page 623 note 5} Brandstetter, R., Die Lauterscheinungen in den indonesischen Sprachen, Luzern, 1915, 1917, 1921, 31–2, 40–2Google Scholar.

^{page 623 note 6} Dempwolff, ‘Fiktion und Hypothese’, 250.

^{page 623 note 7} Dempwolff, ‘Die L-, R- und D-Laute’, pub. 1924–5.

^{page 624 note 1} Ibid., 21.

^{page 624 note 2} Ibid., 34; see also 38: ‘Ausdrücklich sei nochmals hervorgehoben, dass dieses Lautsystem nur ein wissenschaftlich.es Mittel der Sprachvergleichung sein soil und nicht den Anspruch macht, in vollem Umfang eine historische Wirklichkeit von Menschen gesprochener Laute zu erschliessen’.

^{page 624 note 3} Ibid., 34–5.

^{page 624 note 4} Ibid., 36.

^{page 624 note 5} O. Dempwolff, Vergleichende Lautlehre des austronesischen Wortschatzes, 3 vols., Berlin-Hamburg. The first volume is entitled Induktiver Aufbau einer indonesischen Ursprache (I A) and appeared in 1934. The second volume is entitled Dedulctive Anwendung des urindonesischen auf austronesische Einzelsprachen (DA) and appeared in 1937. The third volume is entitled Austronesisches Wörterverzeichnis and appeared in 1938.

^{page 624 note 6} DA, 166.

^{page 624 note 7} Although Dempwolff does not equate Proto-Indonesian (PIn.) with Proto-Austronesian (PAn.) until he reaches the last section of his survey (DA, § 160, 193–1), it may be taken that he regards PIn. and PAn. as coterminous.

^{page 625 note 1} DA, 192.

^{page 625 note 2} The symbol ø indicates that this phoneme does not seem to have a reflex in the languages under consideration in this study. No further reference will therefore be made to it.

^{page 625 note 3} G. B. Milner in a comment to Capell, A., ‘Oceanic linguistics today’, Current Anthropology, III, 4, 1962, 416–17Google Scholar.

^{page 625 note 4} DA, 193.

^{page 625 note 5} ‘Es drängt sich bei rein linguistischer Betrachtung der Schluss auf, dass die beiden Gruppen nur dadurch zu einer weitgehend übereinstimmenden Entwicklung der Laute des UIN. gekommen sind, dass sie ursprünglich zum Sprachgut eines einheitlichen Volkes gehörten. Damit scheint die rassische Verschiedenheit der dunklen krausharigen Melanesier von den hellfarbigen, schlichtharigen Polynesiern in krassem Widerspruch zu stehen’ (DA, 193).

^{page 625 note 6} ‘Daraufhin lässt sich folgendes Bild entwerfen: Ein hellfarbiger Volksstamm von einheitlicher Sprache hat vor vielen Generationen die Inselwelt des stillen Ozeans kolonisiert…. Seine Muttersprache blieb bei diesen langen und weiten Fahrten nicht rein, sie behielt aber einen gemeinsamen Wortschatz und gemeinsame Lauterscheinungen sowohl in dem Sprachgut, das er an andere Völker abgab, als auch in dem, das er selbst hinüberrettete’ (DA, 193).

^{page 625 note 7} DA, 166 and 192.

^{page 626 note 1} Dyen, I., ‘The Tagalog reflexes of Malayo-Polynesian D’, Language, XXIII, 3, 1947, 227–38CrossRefGoogle Scholar.

^{page 626 note 2} At the time, it had not yet of course been suggested that certain outliers might have a phonemic distinction between l and r, the Tikopian evidence of Raymond Firth having apparently been overlooked.

^{page 626 note 3} Milke, W., ‘Zur inneren GHederung und geschichtlichen Stellung der ozeanisch-austronesischen Sprachen’, Zeitschr. f. Ethnologie, LXXXIII, 1, 1958, 58–62 Google Scholar.

^{page 626 note 4} See also Milke, W., ‘Beiträge zur ozeanischen Linguistik’, Zeitschr. f. Ethnologie, LXXXVI, 2, 1961, 162–82Google Scholar.

^{page 626 note 5} DA, 166 and 192.

^{page 626 note 6} See p. 630, n. 3, concerning Rotuman.

^{page 626 note 7} Grace, G. W., The position of the, Polynesian languages within the Austronesian (Malayo-Polynesian) language family (IJAL, Memoir 16), Baltimore, 1959 Google Scholar.

^{page 627 note 1} DA, 165.

^{page 627 note 2} Whereas this could be shown to have been the case in Malagasy (Hova); cf. DA, 85–6.

^{page 627 note 3} DA, 184.

^{page 627 note 4} See Grace, op. cit., 25–7 and 33–34.

^{page 627 note 5} DA, 7.

^{page 627 note 6} Optional prenasalization of the initial consonant of a reconstructed word in PIn. is not to be confused with the occurrence of homorganic prenasalization by accretion (nasaler Zusatz) or substitution (nasaler Ersatz) as a regular and still productive morphophonemic feature in modern Indonesian languages.

^{page 628 note 1} DA, 137 (for Melanesian) and 180 (for Polynesian).

^{page 628 note 2} DA, 136.

^{page 628 note 3} DA, 141.

^{page 629 note 1} Scott, op. cit., 743.

^{page 629 note 2} Hockett, C. F., A manual of phonology (IJAL, Memoir 11), Baltimore, 1955, 95, 124 Google Scholar.

^{page 629 note 3} Raymond Firth, in course of publication.

^{page 629 note 4} Capell, op. cit., 1960.

^{page 629 note 5} Capell, A., The Polynesian language of Mae (Emwae), New Hebrides, Auckland, Linguistic Society of New Zealand, 1962 Google Scholar.

^{page 630 note 1} Another Polynesian settlement in Melanesia, to the south of the Solomon Islands.

^{page 630 note 2} In addition to the information soon to be published by Firth concerning l and r in Tikopian, ample material is found scattered in his books and articles, particularly in ‘ The analysis of Mana: an empirical approach’, Journal of the Polynesian Society, XLIX, 1940, 483–510 Google Scholar, and in We, the Tikopia, London, 1936, passim.

^{page 630 note 3} The status of Rotuman itself has long been a matter for controversy. Owing to extensive borrowings from Melanesian and Polynesian sources, it has not yet been found possible to state with certainty which Rotuman linguistic stratum is more recent and which, more ancient. For that reason Rotuman has been deliberately excluded from the present study. It is mentioned in the context of Tikopian, however, because legendary and topographical evidence points to Rotuma as having been the original home of at least some of the ancestors of present-day Tikopians; cf. History and traditions, 158.

^{page 630 note 4} The present writer has put forward a similar explanation to account for certain pairs of doublets which occur in Samoan; see Milner, G. B., ‘Notes on the comparison of two languages (with and without a genetic hypothesis’) in Linguistic comparison in South East Asia and the Pacific, London, 1963, 28–3Google Scholar.

^{page 631 note 1} The Tikopian words in this section occur in the texts given in an appendix to Raymond Firth, History and traditions of Tikopia, or were obtained by personal communication with the author himself.