² Dempwolff‘s .

³ A Malagasy (Madagascar) dialect.

⁴ Melanesian.

⁵ Polynesian.

⁶ See R. J. Wilkinson, A Malay-English Dictionary (Romanised) s.v. tělu (Mytilene, Greece, 1932).

⁷ A Minangkabau informant (Emyt Dollan, a native of Fort de Kock, a tutor in Malay at Yale University in 1944) gave tigo. It is so recorded by R. van Eck, Een en Ander over het Minangkabausch-Maleisch, IG 2.2.967 (1880). It is recorded as tigŏ by J. L. van der Toorn, Het Minangkabausche ten opzichte van het Maleisch, IG 3.1.531 (1881).

⁸ Benkoeleesche Raadsels (Těkoq-Těki), Tijd. 37.99 (1894). He uses the transcription tigaw in Bijdrage tot de letterkunde van den Serawajer en Besemaher in de afdeeling Manna en P. O. Manna (Residentie Bengkoelen), Tijd. 37.65-97 (1894).

⁹ Seraway has other marked Minangkabau features besides final o for Malay a. For example, Helfrich transcribes sěpoelo'ah ‘IO’ and toedjo'ah ‘7’ (dj — my j and oe = my u), and my Minangkabau informant (see note 7) gave sapuluah and tujuah. (I cannot explain the meaning of Helfrich's apostrophe.) Nevertheless, Seraway is included with Malay by at least two isoglosses, the negatives taq and tidaq (Helfrich's q is defined as a ‘swallowed’ k, presumably a glottal stop). The corresponding Minangkabau negatives are da? and inda?.

¹⁰ In the work cited at the end of note 8.

¹¹ Raymond Kennedy's suggestion.

¹² Helfrich's material for the Besemah dialect indicates that that dialect is very similar to the Malay of the southern part of the Malay Peninsula. According to a note at the bottom of p. 66 of Helfrich's article, the Seraway and Besemah dialects were assigned to ‘Midden-Maleisch’ by Dr. J. Brandes (no further reference given). If the Besemah and Seraway dialects are typical of ‘Midden-Maleisch’, then this term has no foundation in linguistic fact despite its wide currency. Cf. Esser's map of the languages of Sumatra, Atlas van Tropisch Nederland 9 (uitgegeven door het Koninklijke Nederlandsch Aardrijkskundig Genootschap in Samenwerking met den Topographischen Dienst in Nederlandsch-Indië, 1938), in which ‘Midden-Maleisch’ is treated as a language different from Malay and Minangkabau; and also the map entitled ‘Linguistic Divisions of Sumatra’ in Loeb and Heine-Geldern, opposite p. 352, in which the same area and a large part of Esser's Malay area are called ‘Mediane language’.

¹³ Simaloersch woordenlijstje, Bijd. 56.304 (1904).

¹⁴ ‘7’ and ‘10’ are recorded as tujōh and sapulōh respectively, not showing the change of the back vowel in a closed syllable to ua.

¹⁵ Simaloereesche Texten, Bijd. 71.585 (1916).

¹⁶ Op.cit. 603.

¹⁷ 't Lampongsch en zijne tongvallen, Tijd. 18.123 (1872).

¹⁸ H. N. Kiliaan, Javaansche Spraakkunst 181 ('s-Gravenhagc, 1919).

¹⁹ K. F. Holle, De Batoe Toelis te Buitenzorg, Tijd. 17.484 (1869).

²⁰ De telwoorden in ‘t Balineesch, Tijd. 21.171 (1875). The divisions ‘high’ and ‘low’ can be assumed to correspond respectively to Javanese Kromo and Ngoko. It is to be noted, however, that the phrase hanak tigang diri ‘three people’ is quoted (173) as ‘low’. The reason for this incongruity is unclear to me.

²¹ It. A. Dr. Hoesein Djajadiningrat, Atjèhsch-Nederlandsch Woordenboek, 2 vols.; Batavia, 1934. The author lists all three forms mentioned, but uses Ihèë in the introduction (x) in discussing the numerals.

²² Final -èē is the regular Achínese (Koetaradja dialect) equivalent of Malay final -u; for example, Achínese malèē equates with Malay malu ‘be embarrassed’. Phonetically èē is something like [], according to C. Snouck Hurgronje, Studiën over Atjehsche klank- en schriftleer, Tijd. 35.422 (1893). lh- is the regular equivalent in the same dialect of Malay těl-; for example, Achínese Ihō‘ ‘deep, bay’ equates with Malay tělok ‘bay, pond’. Snouck Hurgronje says (op.cit. 381) that phonetically lh is l followed by h; but Raymond Kennedy has told me that he heard it as h followed by l. Some Achínese dialects retain těl-, romani zed as teul-.

²³ J. H. Neumann, Schets der Karo-Bataksche spraakkunst, VBGKW 43.120 (4e Stuk, 1922). No material is immediately available for Gayo, also in north Sumatra.

²⁴ H. von Rosenberg, Vergelijkende tafel der getallen von een tot tien in de talen der bewoners van Sumatra, Tijd. 1.434 (1853), quotes tellau, a form clearly related to *telu, for Redjang.

²⁵ H. N. van der Tuuk, op.cit.

²⁶ H. Sundermann, Kurze Formenlehre der Niassischen Sprache, Tijd. 28.119 (1883).

²⁷ N. Adriani, Spraakkunstige schets van de taal der Mentawei-eilanden, Bijd. 84.90 (1928).

²⁸ O. L. Helfrich, Nadere bijdrage tot de kennis van het Engganeesch, Bijd. 71.514 (1916).

²⁹ J. Rigg, A Dictionary of the Sunda Language (s.v. tolu), VBGKW, Vol. 29 (1862). There is no immediately available material on Madurese, spoken on the island of Madura and the adjacent part of Java. The presence or absence of tiga would, however, have no effect on the main points of this paper.

³⁰ A. G. Vordermann, Sasaksche woordenlijst, Tijd. 38.408 (1895). Also R. C. van den Bor, Nederlandsch-Sasaksche woordenlijst (Prajaasch dialect) 9, VGBKW, Vol. 56 (5e stuk, 1907). It is worth noting that Bor records dasa beside pulu (the inherited form) in the higher tens, but not for ‘10‘, where (s.v. tien) he gives only sěpulu.

³¹ The checking included Dayak (Borneo), the Toradja and other languages of Celebes, West-Flores (Lio dialect), Tettum (on Timor), and others. For Tettum (also called Teto, Tetum, Tettun; on Timor) the Diccionario Teto-Português of R. das Dores (Lisbon, 1907) cites tika ‘three. Malay term used only for the hours of dawn, but falling into disuse’. The letter g does not occur in the transcriptions. Since this dictionary would almost certainly have k for the g of a borrowing, the dictionary's indication that tika is a borrowing from Malay (and does not retain the k of the Sanskrit word cited later) seems to be entirely trustworthy. Tettum tolo ‘3‘ is cited without comment.

³² Madurese, of course, is excepted from this statement.

³³ R. Pischel, Grammatik der Prakrit Sprachen; Strassburg, 1900. On p. 320 he quotes Ardhamāgadhi, and Jaina-Śaurasēnī duga = Sanskrit dvika ‘a pair’, but fails to quote tiga = trika. The Prakrit dictionary of H. D. T. Sheth, (Calcutta, 1923-8), lists four occurrences of tiga.

³⁴ Loeb and Heine-Geldern 7 f.

³⁵ Cf. E. Hultsch, Inscriptions of Asoka, Corpus Inscriptionum Indicarum, Vol. 1 (Oxford, 1925). t for tr is general throughout all the inscriptions. Original medial k is generally represented by k, but sometimes by g, as in the Sanskrit paralokam ‘other world’. It has also dropped out (presumably after first changing to g) in diyadhiyam = Sanskrit *dvikārdhika 'one and a half in the Maski Brahmagiri, Śiddāpura, Rock-Inscriptions. In fact the loss of medial k is more frequent in the inscriptions than the change to g. But since medial k is also represented by g, it is likely that some dialects still retained medial g. In quoting from the Asoka inscriptions I have selected forms from the inscriptions in or near southeast India, since that is the area from which the colonization of the East Indies probably took place, according to B. H. M. Vlekke, Nusantara: a History of the East Indian Archipelago 16 (Harvard University Press, 1943).

³⁶ See note 6; susun tělu ‘in a triple row or layer’; buah kěras tělu ‘a candle-nut with three pips’; buah salak tělu ‘a salak nut with three pips’.

³⁷ Rulon S. Wells suggests a similar instance in English. The word kith (OE cȳþ ‘acquaintance, relationship, native land‘) survives only in the phrase kith and kin.

³⁸ Loeb and Heine-Geldern 97.

³⁹ The first element is a form meaning ‘one’. No emphasis is placed on the apparently closer equivalence of Prakrit s with Javanese s, for Javanese has only one sibilant and would presumably have replaced Sanskrit ś with s in borrowing.

⁴⁰ What evidence there is suggests a closer relationship with Javanese than with Malayan. Von Rosenberg's citation of rowa ‘2' (op.cit.) and v. d. Tuuk's citation of ruwa ‘2' (op.cit. 143, without indication of dialect) for Lampong seem to go more closely with Javanese ro ‘2’ than with Malayan *dua. Similarly Paminggir (v. d. Tuuk, op.cit. 124) lalor ‘fly’ and hulor ‘worm’ seem to go more closely with Javanese Idlěr and ulěr than with Malayan Halat and *ulat.

⁴¹ Loeb and Heine-Geldern 324.

⁴² Dempwolff's (op.cit.) duva'.

⁴³ Loeb and Heine-Geldern 307.