Baca PDF (OCR): The Reconstruction of Proto-Malayo-Javanic

VERHANDELINGEN

VAN HET KONINKLIJK INSTITUUT VOOR TAAL-, LAND- EN VOLKENKUNDE

BERND NOmOFER

THE RECONSTRUCTION OF PROTO-MALAYO-]AVANIC

'S-GRAVENHAGE - MARTINUS NIJHOFF 1975

THE RECONSTRUCTION OF PROTO-MALAYO-jAVANIC

VERHANDELINGEN

VAN HET KONINKLIJK INSTITUUT VOOR TAAL-, LAND- EN VOLKENKUNDE

BERND NOTHOFER

THE RECONSTRUCTION OF PROTO-MALAYO-)AVANIC

\\\,~\\UJK INs"'1: f·/ ~() "~J' ' .-f ~ , ~ VOOR '" il \~, 1~6 -f; ~~/ "~!?~~ 'S-GRAVENHAGE - MARTINUS NIJHOFF 1975

PREFACE

This book is a slightly revised and corrected version of my Ph.D. dissertation (Yale University, 1973). The work concerns the reconstruction of the phonemes of Proto-MalayoJavanic, the last proto-language which Sundanese, Javanese, Malay, and Madurese directly continue. Part 1 contains a lexicostatistical calculation of the degrees of relationship among the four languages under investigation and a brief description of the phonology and morphophonemics of each language. Part 2 is devoted to the reconstruction of the ProtoMalayo-Javanic phonemes. it is shown that the distinction between the two Sundanese vowels a and ; is not inherited and that Javanese loanwords are the main source of Sundanese words containing d. It is also shown that evidence from Malayo-Javanic languages requires the reconstruction of a number of Proto-Malayo-Javanic phonemes which hitherto have not been reconstructed for proto-languages of higher order or the proto-language of highest order, i.e. ProtoAustronesian. At the beginning of Part 2 we discuss the methodological principles applied in the determination of inheritance and borrowing. The appendix contains the basic vocabulary lists for the four languages, a map showing previously assumed language boundaries separating Sundanese, Jakarta Malay,

Javanese, and Madurese and a revised map showing language boundaries as revealed in the course of this research as well as Sundanese dialect maps. An index of the ProtoMalayo-Javanic reconstructions follows.

The word index and the section entitled "Additional notes on the *S : *b distinction and new hypotheses regarding the reflexes of *w and *S" were added to the original text in 1974.

The topic of the dissertation was suggested to me by Professor Isidore Dyen. I wish to thank him for his constant and patient guidance of my ana1ysis. I wou1d 1ike to thank Vale University, the Nationa1 Science Foundation, the Conci1ium on International and Area Studies of Ya1e University, and the project "The Genetic C1assification of Languages-Austronesian" directed by Professor Dyen and supported by the Nationa1 Science Foundation, for their support of my graduate work and research. I am grateful to my fellow-student Curtis D. McFar1and for all tangib1e and intangib1e support he has given me. I have profited from discussions with Shigeru Tsuchida. Fina11y, I wou1d 1ike to thank the Indonesian peop1e, particu1ar1y my friend and assistant Adang Affandi from Sandung, for making my field work in West Java such a fruitfu1 experience.

Page I. Abbrev;at;ons ••••••••••••••••••••••••••••••••••••• XI 11. Bib1iography ••.•••.•••••••••••••••••••••••••••••• XV 1. Introduction •••••••••••••••••••••••••••••••••••••• The four 1 anguages •••••••••••••••••••••••••••••• 4 Sundanese ••••••••••••••••••••••••••••••••••••• 4 Javanese •••••••••••••••••••••••••••••.•••••••• 8 01 d Javanese •••••••••••••••••••••••••••••••••• 16 Ma1ay ••••••••.•••••••••••••••••••••••••••••••• 21 Madurese •••••••••••••••••••••••••••••••••••••• 23 Convent i ons ••••••••••••••••••••••••••••••••••••• 28 2. Reconstruction •••••••••••••••••••••••••••••••••••• 35 Methodo1ogica1 princip1es ••••••••••••••••••••••• 36 The PMJ morpheme structure •••••••••••••••••••••• 45 Madurese double consonants •••.•••••••••••••••••• 47 The PMJ vowe1s ••••••.••••••••••••••••••••••••••• 50 PMJ a ••••••••••••••.•••••••••••••••••••••••••• 51 PMJ i ......................................... 52 PMJ u ••••••••••.••••••••••••••••••••••••••••.• 63 PMJ il ••••••••••••••••••••••••••••••••••••••••• 72 The PMJ semivowe1s •••••••••••••••••••.•••••••••• 82 PMJ w ••••••••••••••••••••••••••••••••••••••••• 83 PMJ y ••••••••••.•••••••••••••••••••••••••••••• 89 The PMJ consonants ••••••••••••••••••••••••.••••• 95 PMJ m, n, n, 1), 1, 1. r. and 5 •••••••••••••••• 95

Page PMJ m ••••••••••••••••••••••••••••••••••••••••• 95 PMJ n......................................... 96 PMJ n ..........•..•.....•....................• 98 PMJ !J ••••••••••••••••••••••••••••••••••••••••• 99 PMJ 1 ••••••••••••••••••••••••••••••••••••••••• 101 PMJ 1 ......................................... 102 PMJ r ......................................... 1 04 PMJ s ......................................... 105 PMJ d, j, z, and g •••••••••••••••••••••••••••••••• 107 PMJ d ••••••••••••••••••••••••••••••••••••••••• 1 07 PMJ j ......................................... 108 PMJ z ••••••••••••••••••••••••••••••••••••••••• 109 PMJ g .•.••.•.••...•...•.•....•....••.•••....•. 111 PMJ c, p, t, and k •••••••••••••••••••••••••••••••• 113 PMJ c ••••••••••••••••••••••••••••••••••••••••• 11 3 PMJ p, t, and k ••••••••••••••••••••••••••••••• 113 PMJ p ••••••••••••••••••••.••••••••••••••••• 116 PMJ t ...................................... 117 PMJ k •••••••••••••••••••••••••••••••••••••• 119 PMJ t ............................................. 121 PMJ 8 and b ••••••••••••••••••••••••••••••••••••••• 123 PMJ 8 •••••••••••••••••••••••••••••••••••••• 124 PMJ b •••••••••••••••••••••••••••••••••••••• 136 PMJ 0 and d ••••••••••••••••••••••••••••••••••••••• 145 . PMJ 0 •••••••••••••••••••••••••••••••••••••• 147 PMJ ~ •••••••••••••••••••••••••••••••••••••• 153

Page PMJ R .•••••••.•.•..•.••.•••..••••..•....•...•.•.• 160 The laryngeals ..••.•..•..•••..••...........•....• 166 PMJ h •.••••.••...••••••..•••.•.•.••.••.•.•.•• 168 PMJ q .....•....•.•••.....••.•••.•••••..••••.• 178 PMJ hiatus .•.....••...•..•.•••••.••••.•••••.. 181 PMJ double consonants af ter vowels other than ·a.184 The i nfi xed nasa 1 •..••.•.••..•.•••.•••••••••••••• 188 PMJ doubled monosyllables ...•........•••.....•..• 194 3. Conclusion ..••.••.•.•.........•.......•.••••.•••... 201 Footnotes ••••.•.•••••.•...•....••••••••••••••.•.•.•.•• 211 Appendix I: The 200-word lists ••.••..•...•...••....•.• 226 Appendix 11: Maps .••.••...........•..•••.•..••....•••. 235 Index of PMJ reconstructions ...•.......•.....••..•••.. 244 Word index .•.•..••..•..•.••••••..•••.•••..•...•.•.•••. 260 Additional notes on the ·S :·b distinction and new hypotheses regarding the reflexes of ·w and ·S •••.•••• 289 Addenda et Corri genda •.........••.••.••••••••.•••....• 314

I. Abbreviations

B. = Bangkalan (Mad. d.)

Bant. = Banten (Snd. d.)

Bad. = Baduy (Snd. d.)

Band. = Bandung (Snd. d.)

Baw. = Bawean (Mad. d.)

Bijd. = Bijdragen tot de Taal-, Land-, en Volkenkunde van Nederlandsch-Indië. The Hague.

Bog. = Bogor (Snd. d.)

Bug. = Buginese

Cian. = Cianjur (Snd. d.)

Cir. = Cirebon (Snd. d.)

Fi. = Fiji

Fu. = Futuna

(H) = High (SS) 10 = Isidore Dyen (only used in the citation of reconstructions) Ind. = Indramayu IPLS = Indo-Pacific Linguistic Studies, Part I (=Lingua 14) Jak. = Jakarta (Mal. d.) Kad. = Kadipaten (Snd. d.) Kang. = Kangean (Mad. d.) Ked. = Kedah (Mal. d.)

Kel. = Kelantan (Mal. d.)

(L) = Low (SS)

= Madurese Mad.

= Language. Baltimore. Lg. (M) = Mid (SS) Mad. d. = Madurese dialect = Makassarese Mak. Mal. = Malay Mal. d. = Malay dialect = Merina Mer. Mlc. = Malacca

NgO. = Ngaju-Oayak

= New Javanese NJv.

NJv. (B) = literary

NJv. (0) = dialectal

= archaic NJv. (0)

N.S. = Negri Sembilan (Mal. d.)

00 = Otto Oempwolff (only used in the citation of reconstructions)

OJv. = Old Javanese

P. = Pamekasan (Mad. d.)

Pal. = Palembang

PAN = Proto-Austronesian

Pen. = Penang (Mal. d.)

PHN = Proto-Hesperonesian

PMJ = Proto-Malayo-Javanic

Pur. = Purwakarta (Snd. d.)

PWI = Proto-West-Indonesian

S. = Sumenep (Mad. d.)

Sa. = Saa

S.-B. = Southern Banten (Snd. d.)

Sm. = Samoan

Snd. = Sundanese

Snd. d. = Sundanese dialect

S.P. = Southern Peninsula (Mal. d.)

Sp. = Singapore (Mal. d.)

SS = status-style

Su. = Sumatra (Mal. d.)

Sumo = Sumedang (Snd. d.)

Tag. = Tagalog

TBt. = Toba-Batak

Tijd. = Tijdschrift voor Indische Taal-, Land-, en Volkenkunde, iutgegeven door het Bataviaasch Genootschap. Batavia.

To. = Tonga

U. = Ulawa

(VH) = Very High (SS)

(VL) = Very Low (SS)

ZES = Zeitschrift für Eingeborenen-Sprachen. derlin.

11. Bibliography

All works consulted are listed below. Works referred to in the text are assigned abbreviations for convenience; the full bibliographical reference follows the abbreviation af ter the colon:

Bergsland and Vogt: Bergsland, Knut and Vogt, Hans. 1962. On the validity of glottochronology. Current Anthropology, vol. 3, 2.115-129.

Brandes, J.L.A. 1884. Bijdrage tot de vergelijkende klankleer der westersche afdeeling van de Maleisch-Polynesische taalfamilie. Utrecht.

Brandstetter, Renward. 1911. Gemeinindonesisch and Urindonesisch. Renward Brandstetter's Monographien zur Indonesischen Sprachforschung. Luzern.

Coolsma 1904: Coolsma, S. 1904. Soendeneesche spraakkunst. Leiden. Coolsma 1913: Coolsma, S. 1913. Soendaneesch-Hollandsch woordenboek. Leiden 2nd ed.

Dempwolff: Dempwolff, Otto. 1934-38. Vergleichende Lautlehre des austronesischen Wortschatzes: 1. Induktiver Aufbau einer indonesischen Ursprache, ZES 15 (1934); 2. Deduktive Anwendung des Urindonesischen auf austronesische Einzelsprachen. ZES 17 (1937); 3. Austronesisches Wörterverzeichnis. ZES 19 (1938). Barlin.

Dyen 1945: Dyen. Isidore. 1945. Spoken Malay. New Vork.

Dyen 1947: Dyen. Isidore. 1947. The Taga10g ref1exes of MalayoPo1ynesian D. Lg. 23.227-38.

Dyen 1949: Dyen, Isidore. 1949. On t~e history of the Trukese vowe1s.

Lg. 25.420-36. Dyen 1951: Dyen, Isidore. 1951. Proto-Malayo-Polynesian *Z. Lg. 27.534-40. Dyen 1953a: Dyen, IsiJore. 1953a. The Proto-Malayo-Polynesian laryngeals. Ba1timore. Dyen 1953b: Dyen, Isidore. 1953b. Dempwolff's *R. Lg. 29.359-66. Dyen 1956: Dyen, Isidore. 1956. The Ngaju-Dayak 'old speech stratum'. Lg. 32.83-7. Dyen 1965a: Dyen, Isidore. 1965a. Formosan evidence for some new Proto-Austronesian phonemes. IPLS 285-305. Dyen 1965b: Dyen, Isidore. 1965b. A lexicostatistical classification of the Austronesian languages. Indiana University Publications in Anthropology and Linguistics, Memoir 19, supplement to the International Journalof American Linguistics. Dyen 1967: Dyen, Isidore. 1967. A descriptive Indonesian grammar. New Haven. Dyen 1971: Dyen, Isidore. 1971. The Austronesian languages and ProtoAustronesian. In Sebeok, Thomas A. 1971. Current Trends in Linguistics, vol. 8.5-54. Dyen and McFarland: Dyen, Isidore and McFarland, Curtis D. 1970. ProtoAustronesian etyma constituting an Austronesian cognate finder list. Mimeographed. Fokker, A.A. 1953. Tatabunyi Sunda. Bahasa dan Budaya Sunda. 1,6. Haudricourt: Haudricourt, André G. 1965. Problems in Austronesian comparative phonology, IPLS 315-29. Hendon: Hendon, Rufus S. 1964. The reconstruction of *-aw in ProtoMalayo-Polynesian. Lg. 40.372-80.

Horne: Horne, Elinor C. 1961. Beginning Javanese. New Haven and London. Jansz: Jansz, P. 1932. Practisch Javaansch-Nederlandsch woordenboek. Semarang, Soerabaia, Bandung, 's-Gravenhage, 3d ed. Juynbo11: Juynbo11, R.J. 1932. Oudjavaans-Neder1andsche woordenlijst. Leiden. H. Kern: Kern, H. 1918. Verspreide geschriften 8. 's-Gravenhage. R.A. Kern 1906: Kern, R.A. 1906. 't 1emes in 't Soendaasch. Bijd. 59.385-562. R.A. Kern 1940: Kern, R.A. 1940. De Soendasche ö-k1ank. Bijd. 99.111-8. Ki1iaan 1897: Ki1iaan, H.N. 1897. Madoereesch spraakkunst. I. Inleiding en klankleer. 11. Woordleer en syntaxis. 2 vo1s. Batavia. Ki1iaan 1904: Ki1iaan, H.N. 1904. Madoereesch-Neder1andsch woordenboek. I, 11. Leiden. Ki1iaan 1919: Ki1iaan, H.N. 1919. Javaansche spraakkunst. 's-Gravenhage. Macdona1d, R. Ross and Soenjono Darjowidjojo. 1967. Indonesian reference grammar. Washington. Panganiban, José Villa. 1972. Diksyunaryo-Tesauro Pi1ipino-Ing1es. Lungsod Quezon. Pigeaud (n.d.): Pigeaud, Th. n.d. Javans-Neder1ands handwoordenboek. Batavia. Pigeaud 1967: Pigeaud, Th. 1967. The 1iterature of Java, vol. I. Ras 1968: Ras, J.J. 1968. Lange consonanten in enige Indonesische talen. Bijd. 124.521-41

Ras 1970: Ras, J.J. 1970. Lange consonanten in enige Indonesische talen 11. Bijd. 126.429-47. Sakiyama: Sakiyama, Osamu. 1970. Das *~ der austronesischen Ursprache. Journalof Osaka University of Foreign Studies 22.107-24. Sa1zner: Sa1zner, Richard. 1960. Sprachat1as des Indopazifischen Raumes. Wiesbaden. Stevens 1965: Stevens, A1an M. 1965. Language levels in Madurese. Lg. 41.294-302. Sumukti: Sumukti, Rukmantoro Hadi. 1971. Javanese morpho1ogy and morphophonemics. Corne11 University unpub1ished dissertation. Tuuk, H.N. van der. 1869. Vergelijkende woordenlijst van Lampongsche tongvallen. Tijd. 17.569-75. Tuuk, H.N. van der. 1870. Brieven betreffende het Lampongsch. Tijd. 19.362-410. Tuuk, H.N. van der. 1872. 't Lampongsch en zijne tongvallen. Tijd. 18.118-56. Tuuk, H.N. van der. 1897-1912. Kawi-Ba1ineesch-Neder1andsch woordenboek. 4 vo1s. Batavia. Uh1enbeck, E.M. 1949. De structuur van het Javaanese morfeem. Bandung.

Robins: Robins, R.H. 1953. The phono1ogy of the nasa1ized verbal forms in Sundanese. Bulletin of the School of Orienta1 and African Studies 15.138-45.

Stevens 1966: Stevens, A1an M. 1966. The Madurese ref1exes of ProtoMa1ayo-Po1ynesian. Journalof the American Orienta1 Society 86.147-56.

Stevens 1968: Stevens, A1an M. 1968. Madurese phono1ogy and morpho1- ogy. New Haven.

Van Syoc: Van Syoc. Wayland B. 1959. The phonology and morphology of the Sundanese language. University of Michfgan unpublished dissertation.

Walbeehm. A.H.J.G. 1897. De taal soorten in het Javaansch. Batavia. Wilkinson: Wilkinson. R.J. 1932. A Malay-English dictionary (romanized). Mytilene. Greece.

Zoetmulder: Zoetmulder. P.J. 1950. De taal van het Adiparwa. een grammaticale studie van het Oudjavaans. Verhandelingen lembaga Kebudajaan Indonesia K.B.G. vol. 79. Bandung.

1. INTRODUCTION

1.1. This work is a study of the historical relationships among Sundanese, Javanese, Malay, and Madures~ based on the methods of comparative linguistics. The cognate features of these four important languages are compared and investigated for their implications regarding their past history and in particular the last proto-language which all four languages directly continue. This last proto-language is referred to as Proto-Malayo-Javanic (PMJ).

1.2. The limitation to these four languages is based on a likely subgroup of which these four languages are the most important members. This subgroup was constituted in Dyen's lexicostatistical classification (1965b:26), in which some 250 languages we re classified into subgroups on the basis of shared cognates in a list of 196 words of basic meanings. He named this subgroup the 'Javo-Sumatra Hesion'. It fncludes as coordinate members, Sundanese, Javanese, and a further subgroup, the 'Malayic Hesion' which includes Malay and Madurese among its members. The following list is extracted from Dyen's classification and shows all members of the ~avo-Sumatra Hesion':

1- Ma1ayic Hesion

1- Ma1ayan Subfami1y

1- Ma1ay

2. Minangkabau

3. Kerintji

2. Madurese

4. Lampungic Subfami1y

3. Achinese

1- Lampung

2. Krol!

2. Sundanese

3. Javanese

Evidence from 1anguages other than Ma1ay. Madurese. Sundanese. and Javanese (Old Javanese and New Javanese) is not considered here because of the mass of data avai1ab1e for these four 1anguages alone. The sparsity of data avai1ab1e for Kerintji. Lampung1• and Kroë wou1d have made consideration of these 1anguages difficu1t in any case.

1.3. Dyen's 1exicostatistica1 ca1cu1ation of the re1ationships among Sundanese. Javanese. Ma1ay. and Madurese showed the fo11owing percentages of cognation for the 200-word 1ists2:

A new lexicostatistical calculation of the relationships among the four languages was made taking into account the regular correspondences established below and thus almost completely eliminating errors due to undetermined borrowings which can be detected by the presence of irregular correspondences. It tended to confirm Dyen's classification. In particular it supported subgrouping Malay with Madurese.

Dyen's percentage for the comparison of the Sundanese and Javanese 1 ists, 37.1 percent, is however too high' ~ Sundanese borrowed heavily from Javanese. Af ter eliminating all words which are determinably borrowings the new percentage is 33 percent.

The following table shows the percentages of cognation among the four languages for both the lOO-word lists3 (lower-left hand corner) and the 200-word lists4 (upper right-hand corner) as they were calculated af ter the determination of the regular correspondences. The percentages for the lOO-word lists support those made for the 200-word lists. Those for the lOO-word lists are proportionally higher because the vocabulary is less fragile:

A tree-configuration of the re1ationships among the four 1anguages as determined by 1exicostatistics is as fo11ows:

Sundanese Ma1ay Madurese Javanese The four 1anguages

1.4. Sundanese. Sundanese is the first (i.e. the native) language of about 15 mi11ion people in the Pasundan region, West Java. In the north, the Sundanese speaking area reaches the Java Sea on1y in the area of Pamanukan. Otherwise, it is separated from the sea main1y by a thin strip of Javanese and the much deeper wedge-shaped area of Jakarta-Ma1ay which inc1udes Jakarta and thins toward the south with its southernmosttip just north of Bogor.

Map 1 shows the 1anguage borders between Sundanese and Javanese, and Jakarta Ma1ay as they were found in a recent study.5

1.4.1. The Sundanese phoneme inventory.6 The Sundanese vowels constitute a seven-term system:

i • u e a 0 a

The Sundanese consonants constitute a nineteen-term system:

P t c k q7 b d j 9 m n n ~ 1 r s h w y

1.4.2. Distributional restrietions of individual phonemes. The phonemes c, j, n, and w do not occur preceding pause. The vowel u does not occur preceding wand the vowel i does not occur preceding y. The vowel a does not occur following y and preceding q, h, or y and furthermore does not occur in sequence with any other vowel. The vowel o does not occur preceding w. No vowel immediately follows or precedes pause. The vowel e and 0 do not occur in sequence with the vowels i and u. There are no sequences of identical vowels.

1.4.3. Sundanese morphophonemic alternations8

1.4.3.1. Initial alternations. A major type of alternation involves the initials of bases whose post-pausal form has an initial p, t, c, s, k, q, and a number of bases whose post-pausal form has an initial b. This alternation appears in combination with an active prefix which itself has many alternants. For convenience we will call all of the alternants of the active prefix the N-alternation. The alternations are as follows:

{a} The N-alternation appears as mbefore a base whose postpausal form has an initial pand a number of bases whose post-pausal form has an initial b. Af ter the active prefix alternant m-, the p or b of the post-pausal form is replaced by 0: e.g., pakeq, makeq (L) 'to use'; b~~k~t, m4uk4t 'to bind'.

(b) The N-alternationappears as nbefore a base whose postpausal form begins with t. Af ter the active prefix alternant n-, the t of the post-pausal form is replaced by 0: e.g., tutup, nutup (L) 'to close'.

Cc} The N-alternation appears as nbefore a base whose postpausal form begins with s or c. Af ter the active prefix alternant n-, the s or c of the post-pausal form is replaced by 0: e.g., susul, nusul 'to pursue'; cokot, iiokot (L) 'to take away'.

(d) The N-alternation appears as ubefore a base whose postpausal form begins with k or q. Af ter the active prefix alternant u-, the k or q of the post-pausal form is replaced by 0: e.g., kirim, ~irim (L) 'to send'; qinum, ninum (L) 'to drink'.

Ce} The N-alternation appears as ~abefore non-alternating initials. Examples of non-alternating initials are the following: bacir, nabacir 'to flee'; dapan, ~adapan (L) 'to lie on the belly'; jinjin, naji~jin 'to lift up with the hand'; g~in' nag~i~ 'to wake up'; minak, naminakan 'to oil s.t.'; nuhun, Uanuhunk4n 'to thank s.o. for', nawaq, nanawaqan 'to reach for, attempt'; naran, ~anarank~n 'to name s.t •• report'; 14094h. ua14ug4h 'to come together in order to begin to celebrate a feast'; raut. ~araut 'to bark. rind. split'; walJwan. !)awaUwalJ 'to do s.t. without using a model'; yaktiq. ~ayaktik4n 'to make become true'; hartiq. !)ahartiq 'to understand'.

1.4.3.2. Final alternations. The following final alternations appear before the suffixes -k4n and -naq:

(a) If the pre-pausal form of a base ends in k. the k of the prepausal form is replaced by ~ before -k4n: e.g .• balik. malik4n (l) 'to return s.t.'.

(b) If the pre-pausal form of a base ends in n. the n of the prepausal form is replaced by ~ before -naq: e.g •• qau4n, qa~4naq 'his soup'.

(c) If the pre-pausal form of a base ends in q, the q of the prepausal form is replaced by ~ before -k4n and before -naq: e.g., tojoq, nojok4n 'to direct towards'; gadeq, gadenaq 'its size' .

1.4.4. The Sunoanese materlal. The main source used for the Sundanese material was Coolsma 1913. It is useful to employ a set of symbols which is in part different from Coolsma's. In the following list of symbols which differ, the symbol used in this work precedes the colon and Coolsma's follows: j: dj; c: tj; Y :j; n : nj; IJ : ng; u : oe; 4 : eu; a : ~.

Furthermore in conformity with the phonemic analysis presented above I have used the following additional conventions in citing words from Coolsma. Coolsma's wand y have not been written af ter a homorganic vowe1 9: e.g., his boewah 'fruit' is written buah; his bijuk 'rotten, sme11ing' is written biuk. A q has been inserted in Coolsma's sequence of identica1 vowe1s: e.g. his buuk 'hair' is written buquk. A q has been p1aced before his initia1 vowe1: e.g., his isuk 'ear1y morning' is written qisuk. Fina11y a glotta1 stop has been p1aced af ter his fina1 vowe1: e.g., his duwa 'two' is written duaq.

1.5. Javanese. Javanese is spoken by about 50 mi11ion peop1e10 in Central and East Java. It is a1so spoken in the thin strip a10ng the north coast of West Java, with the exception of the areas around Pamanukan and Jakarta. There is a1so a significant number of speakers in Sumatra and in Dutch Guiana.

Map 2 shows the 1anguage borders separating Javanese, Madurese, Sundanese, and Jakarta Ma1ay as they appeared in Sa1zner's Atlas.

1.5.1. The Javanese phoneme inventory.11 Javanese has the fo11owing eight vowe1 phonemes:

Javanese has the fo11owing twenty-six consonant phonemes:

P t ~ c k q12 b d ~ j 9 m n n ij mb nd n~ nj ~g 1 r s h w y

1.5.2. Distributiona1 restrictions of indi~dual phonemes.

(a) The vowel a does not occur before pause or before q, except in interjections: e.g., haqa 'yes'.

(b) The phoneme h occurs initially only in interjections: e.g., ham 'yes'.

(c) The phoneme h occurs intervocalically only between identical vowels, usually a, rarely u, still more rarely between one of the other vowels. It occurs between different vowels only dialectally.

(d) The phoneme q occurs only in final position. In interjections it can also occur intervocalically.

(e) The phonemes t, d, c, j, n, y, and w do not occur in final . . position.

(f) The phonemes b, d, and 9 occur in final position only in the Western group of Javanese dialects (the dialects of Banjumas and Tegal) and correspond to final p, t, and k in the Eastern group. The latter can be subdivided into a Central subgroup (the dialects of Bagelen, Jogjakarta, Kedu, Surakarta, Semarang, and Djepara-Rembang) and an Eastern subgroup (the dialects spoken in the province of East Java). In the Central subgroup final p, t, and k alternate with b, d, ~d 9 when followed bya suffix: e.g., sabap 'cause', but (with the suffix -aké), sababaké 'is caused'. In the Eastern subgroup p, t, and k always appear even when followed by a suffix: e.g., sabapaké. For convenience Javanese forms are cited with final b, d, and 9 instead of final p, t, and k.

(g) The phonemes mb, nd, n~, nj, and ~g do not occur before pause.

(h) The penultimate vowels é and ~ in an open or closed syllable in the Eastern subgroup correspond to i and u in the Central subgroup: e.g., Eastern subgroup séséh, Central subgroup siséh 'side' ; Eastern subgroup sósóh, Central subgroup susó'h 'nest'. 13

(i) The phoneme k occurs before pause only af ter a. According to Kiliaan (1919:53) Javanese -ak and -aq vary freely: e.g. jdaq. idak 'to step on'. In the citation of Javanese forms we will list both alternants only if both are given in Pigeaud (n.d.).

1.5.3. Javanese morphophonemic alternations14•

1.5.3.1. Initial alternations. The major initial aTternation in Javanese involves the initials of bases whose post-pausal form begins with p, w, t, ~, s, c, or k. This alternation appears in combination with the N-alternation. The alternations are as follows:

(a) The N-alternation appea~s as mbefore a base whose postpausal form begins with p, w, or b, and a few bases whose post-pausal form begins with avowel. Af ter the active prefix alternant m-, the p or w of the post-pausal form is replaced by 0 and the b of the post-pausal form remains: e.g., pillh, miléh 'to choose'; woco, màc~ (L) 'to read'; bukaq, mbukaq (L) 'to open'; uléh, muléh (L) 'to return home'.

(b) The N-alternation appears as nbefore a base whose postpausal form begins with t, ~, d, ~, or j. Af ter the active prefix alternant n-, the t or ! of the post-pausal form is rep1aced by ~ and the d, 9, or j of the post-pausa1 form remains: e.g., tipés, nipés 'to become thin or flat', ~intéo' nin~é~ 'to strike a note'; dadi, ndadi (L) 'to become worse'; quwór, nçuwóraké (L) 'to make higher'; j3n3~, nj3na~aké (L) 'to name, ca11'.

(c) The N-a1ternation appears as nbefore a base whose postpausa1 form begins with s or c. Af ter the active prefix a1ternant n-, the s or c of the post-pausa1 form is rep1aced by ~: e.g., si1éh, ni1éh (L) 'to borrow'; campór, namp6r 'to mix'.

(d) The N-a1ternation appears as nbefore a base whose postpausa1 form begins with k or g, and before a base whose post-pausa1 form begins with avowel. Af ter the active prefix a1ternant u-' the k of the post-pausa1 form is rep1aced by ~ and the g of the post-pausa1 form remains: e.g., kumbah, Dumbah 'to wash'; górèO' ogórèn 'to fry in oi1'; 61ah, nó1ah 'to cook a dish'.

(e) The N-a1ternation appears as Oa-, U3-, m3-, or abefore monosy11abic bases. The naand D3- alternants on1y appear with transitives and the m3- and aalternants on1y appear with intransitives: e.g., kon, Uakon (L) 'to order'; lap, D31ap 'to wipe'; qón, m39ón 'to descend'; d6s, ad6s 'to take a bath'.

(f) The N-a1ternation appears as n3- or m3- before a base whose post-pausal form begins with 1 and r. The Oaalternant appears with transitives, the m3w alternant with intransitives: e.g., liwat, naliwat 'to boil rice'; laku, malaku (L)

'to walk'.

1.5.3.2. Final alternations. Many bases exhibit two forms, one that appears before pause and the other that appears before certain suffixes: the pronominal suffixes -ku (L) 'my', -mu (L) 'your, -né ~ -é (L), -nipón ~ -ipón (H) 'his'; the nominalizing suffix -an; the locative suffix -ni ~ -i, the locative imperative suffix -nono ~ -ono; the causative suffixes -qaké ~ -aké (L), -qakan ~ -akan (H); and the causative imperative suffix -qno ~ -no. The alternations are as follows:

(a) If the pre-pausal form of a base ends in h, the h of the pre-pausal form is replaced by 0 before a vowel-initial suffix: e.g., kumbah, kumbaan 'laundry'.

(b) Base-alternations before -ku, -mu, -né ~ -é:

(1) If the pre-pausal form of a base ends in 0, the 0 and any preceding consecutive 0 of the pre-pausal form is replaced by a: e.g., konco 'friend', kancaku 'my friend'.

(2) If the pre-pausal form of a base has é in the final closed syllable, this é of the pre-pausal form is replaced by i before the suffix -é (but not before -ku or -mu) for some speakers: e.g., muréd 'student', muridé (others murédé) 'his student'.

(3) If the pre-pausal form of a base has 6 in the final closed syllable, this ó of the pre-pausal form is replaced by u for all speakers: e.g., sapór 'train', sapuré 'the train'.

(c) Base-alternations before -an:

(1) If the pre-pausal form of a base ends in 0, the 0 of the pre-pausal form is replaced b,)' 21: e.g., tako 'to come', takan 'arrival'.

(2) If the pre-pausal form of a base has é or 6 in the final closed syllable, the é or ó of the pre-pausal form is replaced by i or u respectively: e.g., tulés (L) 'to write', tulisan (L) 'writing'; raóp 'to wash one's face', raupan 'water in which face has been washed'.

(3) If the pre-pausal form of a base ends in i, the sequence i + a is replaced by è: e.g., bali (L) 'to return', balèn (L) 'change'. In some instances the sequence of vowels remains: e.g., dadi (L) 'to become', dadian 'ghost' .

(4) If the pre-pausal form of a base ends in u, the sequence u + a is replaced by 0: e.g., turu (L) 'go to bed', turon (L) 'place to sleep'. In some instances the sequence of vowels remains: e.g., adu (L) 'to fight', aduan 'a fight'.

(5) If the pre-pausal form of a base ends in é, the sequence é + a is replaced by è: e.g., 6mbé 'to drink', ómbèn 'a drink'. In some instances the sequence of vowels remains: e.g., gawé (0) 'to work', gawéan (L) 'action, work'.

(6) If the pre-pausal form of a base ends in ó, the sequence 6 + a is replaced by 0: e.g., aog6 (L) 'to use, augon

(L) 'act of doing'. In some instanees the sequence of vowe1s remains: e.g., jaró (L) 'inside', jaróan 'intestines'.

(d) Base-a1ternations before -ni ~ -i, -nono ~ -onb, -nan ~ -an, and -nipón ~ -ipón:

(1) If the pre-pausa1 form of a base ends in 0, this 0 and any preceding consecutive 0 of the pre-pausa1 form is rep1aced by a: e.g., biso (L) 'can, be ab1e to', dibisani 'be ab1e to be done'.

(2) If the pre-pausa1 form of a base has é or ó in the fina1 c10sed sy11ab1e, the é or ó of the pre-pausa1 form is rep1aced by i or u respective1y: e.g., tu1és (L) 'to write', nu1isi 'to write on, to'; wis6h 'to wash one's hands', misui 'to wash hands (s.o. else's)'.

(3) If the pre-pausa1 form of a base ends in i or u, the or u of the pre-pausa1 form is rep1aced by è or 0 respective1y: e.g., ga ni (L) 'fire', oganèni (L) 'to cook'; 1aku (L) 'to walk', Ua1akoni (L) 'to undergo, endure'.

(4) If the pre-pausa1 form of a base ends in é or 6, the é or 6 of the pre-pausa1 form is rep1aced by è or 0 respective1y: e.g., 6mbé 'to drink', Dómbèni 'to give a drink to'; jaró (L) 'inside', jaroni (L) 'to make deeper'.

(e) Base-a1ternations before -qaké ~ -aké, -qakan ~ -akan, and -qno ~ -no:

(1) If the pre-pausal form of a base ends in O. this 0 and any preceding consecutive 0 of the pre-pausal form is replaced by a: e.g •• toto 'to arrange'. nataqaké 'to arrange for s.o.'.

(2) If the pre-pausal form of a base ends in i or u. the i or u of the pre-pausal form ·is replaced by è or 0 respectively: e.g .• lali (L) 'to forget'. Ualalèqaké (L) 'to try to forget'; tamu (L) 'to meet'. namoqaké 'to find for s.o.'.

(3) If the pre-pausal form of a base ends in é or 6. the é or ó of the pre-pausal form is replaced by è or 0 respectively: e.g .• suwé (L) 'long time'. nuwèqaké (L) 'to cause to take a long time'; b6d6 'stupid'. mbódoqaké 'to make a fool of'.

(4) If the pre-pausal form of a base ends in U or n. the U or n of the pre-pausal form is replaced by q in some instances: e.g .• takon (L) 'to ask'. nakoqaké (L) 'to ask for information'.

1.5.4. The Javanese material. The material for Javanese is mainly drawn from Pigeaud (n.d.). The orthography of the source has been modified in the citations. As in the case of Coolsma's symbols for Sundanese. Pigeaud's dj. tj. j. nj. ng. and oe have been replaced by j. c. y. n. n. and u; Pigeaud's e has been replaced by the symbol a. Furthermore the following modifications have been made to bring the material into a phonemic writing. Pigeaud's final k has been replaced by final q except when it occurs af ter a: e.g .• his pijak 'to separate'

is written piyaq, but his ibak 'fu11' is not modified. His i and u before a fina1 consonant have been rep1aced by é.and ó respective1y: e.g., his ba1ik 'on the contrary' is written ba1éq; his baku1 'tradeswoman' is written bak61. His 0 before a fina1 consonant has been rep1aced by 0: e.g., his ba10D 'low level' is written balon. His fina1 a and any preceding consecutive a have been rep1aced by 0: e.g., his waca 'to read' is written woco.

Some of the Javanese material has been drawn from Jansz, particu1ar1y the names of animals, p1ants, and trees. Pigeaud 1ists on1y a few zoologicalor botanical terms. Furthermore, Jansz has been consu1ted for the qua1ity of the penu1timate vowe1 in words that are 1isted with 0 in Pigeaud's dictionary. Jansz distinguishes ó and à in his transcription. In the quotation of words from Jansz the same substitutions as for Pigeaud are app1ied. In addition, n has been substituted for Jansz's ~ and single consonants are written instead of double consonants.

1.6. 01d Javanese. 01d Javanese is a term which is emp10yed to designate a 1anguage that is on1y known from texts. These texts are from different periods and of a rich variety in content and form. They have in common that all of them were written in the pre-Is1amic period of Javanese history. A list of the texts used by Juynbo11 for his dictionary - from which the material presented in this work is drawn - can be found in the preamb1e to the dictionary (pp. VIII-XI).

1.6.1. The 01d Javanese symbols. 01d Javanese has the fo110wing vowe1 symbo1s: a, a, i, 1, u, ü, r, f, 1, 1, e, ai, 0, a, and ö.

The consonant symbo1s are the fo110wing 34 symbo1s:

p t t c k ph th ch kh b d ~ j 9 bh dh jh gh m n ~ n D 1,r c,~,s h

w y

Anusw!ra: ~ Wisarga: h Anun!sika: •

1.6.2. The ~honetic characteristics of the 01d Javanese s~bo1s. The fo110wing presentation of the phonetic characteristics is based on H. Kern. The symbo1 a probab1y was pronounced 1ike Sanskrit or Tega1 Javanese a. lts pronunciation is that of a as it a1so occurs in Sundanese, Ma1ay, Makassarese, and other 1anguages of the area. Concerning the quantity of the vowe1 H. Kern points out that a probab1y was pronounced 1ike Makassarese a, i.e., a long a. Since the single author writes sometimes a and sometimes ä in the same words, e.g., mati and mati, Kern conc1udes that the quantitative distinction between a and ä was disappearing in 01d Javanese. According to him it is very doubtfu1 that the distinctions in quantity as they were made by the poets of the 12th and 13th century actua11y agreed with the pronunciation of the time. The symbo1 i in open sy11ab1es probab1y was pronounced as in modern Javanese. In c10sed sy11ab1es however it probab1y represents avowel whose pronunciation is c10ser to that of Sundanese or Madurese i in the same environment. The symbo1s u and ü probably we re pronounced as in modern Javanese, at least in open syllables. The symbol r can only be interpreted as a vowel in Sanskrit words. In inherited words it represented the sequence ra: e.g., rno, r3uö. The symbol r represented the sequence rö. The same holds for ! and I which represented the sequence la and lö. The symbol a probably was pronounced like modern Javanese pepet. Kern posits th at ö was pronounced like a long modern Javanese pepet. Short and long pepet are confounded in a number of instances, particularly wh en followed by a consonant: e.g., ahöb, ahab. As opposed to the symbols presented so far, the phonetic characteristics of the two symbols e and 0 are difficult to determine. In modern Javanese we have é and è, ó and 0 respectively. Kern believes that dialectal differences in the pronunciation of e and 0 probably already existed in the Old Javanese period. The symbol ~i probably was pronounced ai. It occurs in Indic words. Even this ai (as well as the inherited ai) must have been weakened to e at a very early stage of Javanese as is indicated by the occurrence of rake besides rakai in the documents.

Aspirated consonant symbols only occur in Sanskrit loan words except for a few misspellings of inherited words. The unaspirated voiceless consonant symbols and the semivowels could have differed only little from the Sanskrit sounds, for the latter are assigned values that do not differ appreciably from those of modern Javanese. Final k was pronounced Ek] as in Sundanese, and not [?] as in New Javanese or Malay. Kern's hypothesis is based on the fact that final k and final t are replaced by g and d respectively before w: e.g., tak + wruh appears as tag-wruh; tat + wruh appears as tad-wruh. [Here the writing with g and d respectively may follow the Sanskrit rules of sandhi.] Just as d is the voiced counterpart of t, g is the voiced counterpart of k, and not q. Of the three sibi1ants on1y s is Javanese. All three sibi1ants were pronounced 1ike modern Javanese s. According to Kern, the symbo1 h in initia1 position must have been a 'softer breath-expu1sion' than Dutch or Indic h. As a fina1 consonant h was pronounced clear1y just as in modern Javanese. The anunäsika on1y occurs in the sy11ab1e om. The wisarga written at the end of a meaning unit or before a consonant is pronounced 1ike a modern Javanese h.

1.6.3. 01d Javanese morphophonemics15

1.6.3.1. Initia1 a1ternations. We wi11 discuss base-a1ternations in combination with the N-a1ternation. We list on1y those alternations which are different from those of New Javanese:

(a) The N-a1ternation appears as an- ~ manbefore a base whose I post-pausa1 form begins with s. Af ter the active prefix a1ternant an- ~ man-, the s of the post-pausa1 form is rep1aced by ~: e.g., sämbut 'grasp', nämbut 'to grasp'.

(b) The N-a1ternation appears as an- ~ manbefore a base whose post-pausa1 form begins with d or j. Af ter the active prefix a1ternant aQ- ~ maU-' the d or j of the post-pausa1 form remains: e.g., doh 'distance', aUdoh 'to withdraw'; juru 'leader, head', anjuru 'to be at the head of, lead'.

1.6.3.2. Fina1 alternations. The foliowing alternations occur before the locative suffixes -i ~ -ani, the causative suffix -akan, and the nominalizing suffix -an:

(a) Before -i:

(1) If the pre-pausal form of the base ends in a, the sequence a + i is replaced by è: e.g., lara 'pain', u1are 'to wound'.

(2) If the pre-pausa1 form of the base ends in i, the sequence i + i is rep1aced by 1: e.g., pati 'dead', umatl 'to ki11'.

(b) Before -akan, -ani, and -an:

(1) If the pre-pausa1 form of the base ends in i or u, the i or u of the pre-pausa1 form is rep1aced by y or w respective1y: e.g., uni 'noise', mauunyakan 'recited'; susu 'mi1k', sinuswakan 'was put to the breast'.

(2) If the pre-pausa1 form of the base ends in ö or 0, the ö or 0 of the pre-pausa1 form is rep1aced by w: e.g., raUö 'to hear', paranwan 'fame, reputation'.

(3) If the pre-pausa1 form of the base ends in e, the sequence e + a is rep1aced bye: e.g., 1imbe, 1umimbekan 'tightened'.

In 01d Javanese manuscripts and in the traditional Javanese writing system we find double consonants before suffixes: e.g., OJv. ucappan (base: ucap), OJv. anuturrakan (base: tutur), OJv. manutussi (base: utus). For a simi1ar phenomenon in Madurese, see section 1.8.3.3.

1.6.4. The 01d Javanese material. As a1ready indicated in 1.6. the 01d Javanese material is taken from Juynbo11. The symbo1s U' n, and a have been substituted for Juynbo11's ng, n, and ~ respective1y.

1.7. Malay. Malay is the first language of about 10 million people on the Malay Peninsula. in certain areas of Sumatra. coastal Borneo. Jakarta and Ambon. anr other parts of the Indonesian Archipelago. Bahasa Indonesia or Indonesian is one of the standard dialects of the Malay language. the other being the Malay of Malaysia and Singapore.

1.7.1. The Malay phoneme inventory.16 The Malay of the Malay Peninsula has the following six vowel phonemes:

The consonant phonemes are:

P t c k q b d j 9 m - n n D 1 r s h w y

1.7.2. Distributional restrictions of individual phonemes. The phoneme a does not occur in the last syllable of a word. The phoneme q occurs optionally af ter and before pause. Finally. the phonemes b. d. c. j. g. and n do not occur in syllable final position.

1.7.3. Malay morphophonemic alternations17

1.7.3.1. Initial alternations. The major initial alternation in Malay involves the initials of bases whose post-pausal form begins with P. m. t. n. k. U. s. and n. This alternation appears in combination with the N-alternation. The alternations are as follows:

(a) The N-alternation appears as mambefore a base whose postpausal form begins with p or b. Af ter the active prefix alternant mam-, the p of the post-pausal form is replaced by' and the b of the post-pausal form remains: e.g., pinjam, maminjam 'to borrow'; buka, mambuka 'to open'.

(b) The N-alternation appears as manbefore a base whose postpausal form begins with t, d, c, or j. Af ter the active prefix alternant man-, the t of the post-pausal form is replaced by , and the d, c, or j of the post-pausal form remains: e.g., tules, manules 'to write'; dapat,mandapat 'to obtain'; cari, mancari 'to look for'; jual, manjual 'to sell' •

(c) The N-alternation appears as manbefore a base whose postpausal form begins with s. Af ter the active prefix alternant man-, the s of the post-pausal form is replaced by ,: e.g., sapu, manapu 'to brush'.

(d) The N-alternation appears as maU- before a base whose postpausal form begins with k, g, or h, and before a base whose post-pausal form begins with avowel. Af ter the active prefix alternant man-, the k of the post-pausal form is replaced by 0 and the 9 or h of the post-pausal form remains: e.g., kirem, manirem 'to send', gosoq, mangosoq 'to scrub'; hito~, manhitoD 'to count'; isi, manisi 'to fill up'.

(e) The N-alternation appears as mabefore a base whose postpausal form begins with m, n, n, !J, 1, r, or w. The initialof the post-pausal form remains: e.g., minta, maminta 'to request'; naeq, manaeki 'to go up against'; nani, manani 'to sing'; ~aUa. maUaUa 'to open wide'; lateh. ma1ateh 'to train'; rokoq. marokoq 'to smoke'; warna. mawarnakan 'to color'.

1.7.3.2. Fina1 a1ternations. If the pre-pausa1 form of a base ends in q. the q of the pre-pausa1 form is rep1aced by k before the suffixes -1 and -an for some speakers: e.g •• dudoq. mandudoqi or mandudoki 'to occuPY'. pandudoqan or pandudokan 'occupation'.

1.7.4. The Ma1ay material is quoted from Wilkinson. Severa1 orthographic changes have been introduced. Wi1kinson's ch. ny. ng. and ~ have been rep1aced by c. n. D' and a respective1y.

Furthermore. in conformity with the phonemic ana1ysis presented above I have used the fo110wing additiona1 conventions in the citation of words from Wilkinson. Wi1kinson's fina1 k and ' have been rep1aced by q: e.g. his buka' 'to open' is written bukaq and his dudok 'to sit' is written dudoq. Furthermore his i and u before a fina1 consonant have been rep1aced bye and 0 respective1y: e.g .• his rampin 'pretti1y slender' is written rampeU; his rambut 'hair' is written rambot. In severa1 instances Wi1kinson 1ists words with and without h. This fact is indicated by using parentheses around the h: e.g •• (h)atap 'roof'. The symbo1s wand y have been substituted for his fina1 u and i where other material shows that they are non-sy11abic: e.g .• bau 'to sme11'. but ba1ay 'hall'.

1.8. Madurese. Madurese is spoken by about 6 mi11ion peop1e in parts of East Java. the is1and of Madura. and on a large number of smaller is1ands in the vicinity. such as Kangean and Bawean (see map 2).

1.8.1. The Madurese phoneme inventory. Madurese has the follow1ng n1ne vowel phonemes:

a u é á ó è a o

Stevens (1968:18) constructs three vowel collections: (1) 'alternating vowels' containing the vowels i, u, a, and a; (2) 'nonalternating vowels' containing the vowels i, è, u, 0, a, á, and a; and (3) 'special vowels' containing the vowels é and ó. Non-alternating and special vowels only occur in loan words.

The 'alternating vowels' have the following allophones: i: [i], [è]; u: [u], [0]; a: [á], [a]; and a: [aA], [a].

Madurese has the following twenty-six consonant phonemes:

P t ~ c k b d ~ j 9 bh dh ~h jh m n n 1 r s h w y

1.8.2. Distributional restrictions of individual phonemes.

(a) The phoneme h only occurs in loan words except in West Madurese. In West Madurese final h occurs in contrast with final vowel, but only before pause. In the citation of Madurese material this contrast is indicated by writing a final h in parentheses: e.g., bu~i(h) 'back'.

(b) The phoneme w in inherited words only occurs af ter a homorganic vowel and a following different vowel.

(c) The phoneme y in inherited words only occurs af ter a homorganic vowel and a following different vowel and in morpheme final position.

(d) The phoneme a does not occur immediately before another vowel, w, y, q, or he.

(e) The phonemes i, è, á, and a do not occur immediately before w.

(f) The phoneme q does not occur in initial position.

(g) The phonemes ~, ~, ~h do not occur before r or 1.

(h) The sequence rd does not occur.

(i) Syllable-final c, j, jh, ~, ~, ~h, b, d, and w do not occur except in a geminate cluster.

1.8.3. Madurese morphophonemic alternations19

1.8.3.1. Vowel alternations. The following vowel alternations appear in Madurese. The vowels è, à, and a alternate with i, u, and á respectively af ter voiced and aspirated stops and af ter liquids, q, w, y, or no consonant, if the preceding syllable begins with a voiced or aspirated stop. The vowels i, u, and á alternate with è, à, and a respectively in all ot her environments: e.g., ~áq 'to' + rèya (L) 'this', ~áqiyá (L) 'this way' (with the loss of base-initial r); buwáq, màwaq 'to carry'.

1.8.3.2. Initial alternations. The major initial alternation is that which involves the initialof bases whose post-pausal form begins with p, t, ~, or k, and the initialof a few bases who se postpausal form begins with b, dh, ~, jh, or gh. This alternation appears in combination with the N-alternation. The alternations are as follows:

(a) The N-alternation appears as mbefore a base whose postpausal form begins with pand before a few bases whose post-pausal form begins with b. Af ter the active prefix alternant m-. the p or b of the post-pausal form is replaced bye: e.g .• pa99haO' maggha~ 'to hit with a sword'; báca(h). maca(h) 'to read'.

(b) The N-alternation appears as nbefore a base whose postpausal form begins with t or ~ and before a few bases whose post-pausal form begins with dh or 9. Af ter the active prefix alternant n-. the t. ~. dh. or 9 of the post-pausal form is replaced by 0: e.g •• totop. notop 'to close'; ~a~~an. na~~aO 'to be wide open'; dhuddhuq. noddhuq 'to point'; 9ápaq. napaq 'to reach'.

(c) The N-alternation appears as nbefore a base whose postpausal form begins with c or s and before a few bases whose post-pausal form begins with jh. Af ter the active prefix alternant n-. the c. s. or jh of the post-pausal form is replaced by 0: e.g •• coca. noco 'to stab'; soqon, noqon 'to request'; jhujjhu. nujjhu 'to stick. prick' (also: ajhujjhu 'to stick. prick').

(d) The N-alternation appears as ubefore a base whose postpausal form begins with k and before a few bases whose postpausal form begins with gh or h. Af ter the active prefix alternant n-. the k, gh. or h of the post-pausal form is replaced by 0: e.g .• kèrèm. nèrèm 'to send'; ghin9hun. OènghuD 'to carry on back or hip'; hormat. Dormadhi 'to honor'; ako(h), nakè(h) 'to confess'.

(e) The N-alternation appears as abefore a base whose postpausal form begins with a nasalor a liquid and before most bases whose post-pausal form begins with b, d, ~, j, g, bh, ~h, jh, gh, and h: e.g., bájhiqághi, abájhiqághi 'to disgust'. For further examples, see Stevens (1968:9lff.).

In some instances doublets appear, one showing the alternation described under (a) through (d) depending on the consonant that appears initially in the post-pausal form of the base, the other showing the alternation described in this paragraph: e.g., bássaè, massaè or abássaè 'to wet'.

(f) The N-alternation appears as ambefore a few bases whose post-pausal form begins with b or bh. Af ter the active prefix alternant am-, the b or bh of the post-pausal form remains. There is always a competitive form with the aalternant: e.g., bujá 'salt', ambujái saghárá(h) 'to salt the sea (i.e. to do s.t. useless)' or abujái 'to salt'; bhábhájái, ambhábhájái or abhábhájái 'to endanger'.

(g) The N-alternation appears as anbefore a few bases whose post-pausal form begins with dh or jh. Af ter the active prefix alternant an-, the dh or jh of the post-pausal form remains: e.g., dháddhi, andháddhiághi 'to bring about'; jhálá sottra(h), anjhálá sottra(h) 'to use a silk fishing net'.

(h) The N-alternation appears as anbefore a few bases whose post-pausal form begins with gh. Af ter the active prefix alternant aO-, the gh of the post-pausal form remains: e.g., ghá~hui, a~ghá9hui 'to own, possess'.

1.8.3.3. Fina1 a1ternations. Before the suffixes -án ~ -an and -i ~ -è the base-a1ternations are as fo11ows. If the pre-pausa1 form of the base ends in a stop, the stop of the pre-pausa1 form is rep1aced by its voiced aspirated counterpart: e.g., tètèp, nètèbhi 'to give in custody', tètèbhán 'what is given in custody'; kassot, uassodhi 'to wipe off', kassodhán 'broom'; arak, noroqhi 'to increase', oroghán 'an increase'.

Note that fina1 consonants except q are doub1ed before dissy11abic suffixes such as -ághi ~~aghi: e.g., tètèp, tètèppaghi 'to give s.t. into s.o.'s custody'; onta1, nanta11aghi 'to throw away'; but: bájhiq, abájhiqághi 'to disgust'. For double consonants before suffixes in Javanese, see section 1.6.3.2.

1.8.4. The Madurese material. The Madurese material is taken from Ki1iaan 1904. The fo110wing orthographic changes have been made: j, jh, c, y, n, ~ and a have been substituted for his dj, djh, tj, j, . d \I nJ, ng, an ~. Furthermore, the fo11owing changes have been made to bring the material into a phonemic writing: the diacritic over i and u has been ignored; á has been substituted for his ~; è and 0 have been substituted for his e and 0 respective1y.

Conventions

1.9. The asterisk (*) is used in the meaning 'unattested'. In most contexts it wi11 be c1ear1y used as the equivalent of 'ProtoMa1ayo-Javanic'. In a few instances it is used in the meaning 'unattested' for an ear1y Sundanese, Javanese, Ma1ay, or Madurese word where the context does not permit confusion.

1.10. Four levels of reconstruction are distinguished throughout this study:

(a) Proto-Austronesian (PAN): a reconstruction is ProtoAustronesian if it is based on at least one cognate in an Indonesian language and one in a Melanesian and/or Polynesian language.

(b) Proto-Hesperonesian (PHN): a reconstruction is ProtoHesperonesian if it is based on a Tagalog cognate beside a cognate in a West-Indonesian language, but no Melanesian or Polynesian cognates.

(c) Proto-West-Indonesian (PWI): a reconstruction is Proto-WestIndonesian if it is based on a Malayo-Javanic cognate and also a Toba-Batak, Ngaju-Dayak, or Merina cognate, but no cognate in Tagalog, Melanesian and Polynesian languages.

(d) Proto-Malayo-Javanic (PMJ): for the principles leading to a Proto-Malayo-Javanic reconstruction, see section 2.2.

1.11. Certain conventions are followed in the citation of reconstructions:

(a) If areconstruction attributed to PMJ is identical to a reconstruction made by Dempwolff and/or Dyen for PHN or PWI the particular level of reconstruction follows the PMJ reconstruction in parentheses. In case it is PHN the Tagalog cognate is listed and if it is PWI the other West-Indonesian cognate is listed: e.g. *kunin (PWI, NgD. kani9-an 'brass') indicates that areconstruction *kuni9 is made on the basis of Malayo-Javanic evidence which is to be associated with an

1dent1cal PWI reconstruction made by Dempwolff and/or Dyen; *suru~ (PHN, Tag. su:loij 'to go ahead, forward') ind1cates that areconstruction *surun is made on the basis of Malayo-Javanic evidence wh1ch is to be associated with an 1dentical PHN reconstruction made by Dempwolff and/or Dyen.

(b) If areconstruction attributed to PMJ differs from a PHN or PWI reconstruction made by Dempwolff and/or Dyen the part1cular level of reconstruction follows the PMJ reconstruct10n in parentheses. If the reconstruction is PHN, the Tagalog cognate is listed af ter the level of reconstruction, and 1f 1t is PWI, a West-Indonesian cognate is listed af ter the level of reconstruction. The cognate is followed by Dempwolff's and/or Dyen's reconstruction: e.g.,*Bunkuk (PWI, TBt. bukkuk 'bent' : bunkuk) indicates that areconstruction *Bunkuk is made on the basis of Malayo-Javanic evidence wh1ch is to be associated with a PWI reconstruction bu~kuk made by Dempwolff and/or Dyen; *ga~çiD (PHN, Tag. gá:rin '1vory' : gadiD) indicates that a reconstruction *ga~çiD is made on the basis of Malayo-Javanic evidence which is to be associated with a PHN reconstruction *gadi~ made by Dempwolff and/or Dyen.

(c) If areconstruction attributed to PMJ differs from one made by Dempwolff and Dyen for PHN or PWI, and if Dempwolff's and Dyen's reconstructions are non-identical, the PMJ reconstruct10n is given followed in parentheses by the particular level of reconstruction. As in (b), the Tagalog cognate is given if the level is PHN, or a West-Indonesian cognate is given if the level is PWI along with the reconstruction made by Oempwolff (preceded by the abbreviation 00) as well as the reconstruction made by Oyen (preceded by the abbreviation 10). Thus, the reconstruction *rakit (PWI, TBt. rakkit 'raft' OD Ra(U)kit, ID [1953b] R3akit) fndicates that *rakit is reconstructed on the basis of Malayo-Javanic evidence which is to be associated with a PWI reconstruction *ra(u)kit made by Oempwolff and a PWI reconstruction *R3akit made by Oyen 1953b. Similarly *BaR2aq (PHN, Tag. bá:ga 'charcoal': 00 baRa, 10 [1965a] baRaH) indicates that *BaR2aq is reconstructed on the basis of Malayo-Javanic evidence which is to be associated with a PHN reconstruction *baRa made by Oempwolff and a PHN reconstruction *baRaH made by Dyen 1965.

(d) If areconstruction attributed to PMJ can be associated witl, one made by Dempwolff and/or Dyen for PAN, the same conven~ tions as stated above apply with the exception that only Dyen's reconstructions are considered. Furthermore, an eastern (Melanesian or POlynesian) cognate is cited. Thus, the reconstruction *niuR (PAN, Fi., Sa. niu 'coconut palm') indicates that *n;uR is reconstructed on the basis of MalayoJavanic evidence which is to be associated with an identical reconstruction made by Dyen and McFarland. Similarly, *Balliq (PAN, Sa. holi 'to buy' : bali [?h]) indicates that areconstruction *Balliq is made on the basis of MalayoJavanic evidence which is to be associated with a PAN reconstruction ba1i[?h] made by Dyen and McFar1and.

1.12. Dempwolff's reconstructions are written in Dyen's symbo1ism (1971:23) with the exception of Dempwolff's ~. ~. and h which are retained. Dyen's reconstructions are written in Dyen's symbo1ism with the exception of his Tand 0 for which we write ~ and ~ respective1y. This symbo1ism is adhered to in new reconstructions. However. different correspondences are assigned to some of Dyen's symbo1s (such as j or z).

1.13. The alternative choices in an ambiguous reconstruction are enc10sed within brackets: e.g .• *pa[rr. RR]ut means indeterminab1y *parrut or *paRRut.

1.14. Tentative or 'prob1ematic' proto-phonemes are given. whenever we are not ab1e to separate c1ear1y the effects of secondary changes (ana1ogica1 change or borrowing) from regu1ar phonetic change. These tentative proto-segments are indicated by using subnumera1s: e.g •• *DaU~aR1. Both possib1e tentative proto-segments are usua11y cited by postposition of the subnumera1s in reconstructions which are uncertain in respect to one tentative proto-segment: e.g •• *caiR13 means *caiR1 or *caiR3•

1.15. Hyphens are inserted to separate affixes from a base: e.g .• um-iwö (= urn + base) 'to take care'.

1.16. Morphologically complex forms are for the most part cited only for Old Javanese. For the other languages only the base is cited.

1.17. Of the languages under investigation Sundanese, Javanese, and Madurese have for certain meanings choices of words which are socially determined and depend on the relations between the speaker and the person addressed, and also on the status of the referee, i.e. the subject of discourse. This phenomenon has been referred to by terms such as language levels (Stevens 1965) or language types (Dutch taalsorten, Walbeehm 1897). We prefer the term 'status-style'. Five status-styles are distinguished in the citation of words in this study. The following table shows the names of the various status-styles in the three languages and the abbreviations used for each of them.

Javanese is the language with the largest number of words which are marked with respect to a certain status-style. In all statusstyles fewer words are involved in Sundanese and Madurese. Semantic categories, such as numbers ot animals which are part of a household (e.g., chicken, dog, cat, etc.) and which have low-high pairs in Javanese, have only words which are unmarked as to status-style in

Sundanese and in Madurese. Sundanese has about 400 high-level words, Javanese has about 600, and Madurese has about 100 words.

It is generally inferred that the status-sty1e distinction is a Javanese innovation which was borrowed by Sundanese and Madurese. Coo1sma (1904:11) says:

"The usage (of the status-sty1e distinction - SN) by the Sundanese very probab1y originates from an imitation of the Javanese who according to history have had great inf1uence on West Java in the past ... The origin of this pecu1iar usage bj the Javanese, i.e. to produce a distinction between peop1e a1so on the 1anguage level, is exp1ained by the caste-system which was brought to Java by the Hindus ••. 75% of the approximate1y 400 Sundanese high-level words appear in Javanese. 72% of the approximate1y 400 10w-1eve1 words are a1so found in Javanese."

In this study all words, whether marked with respect to statussty1e or not, wi11 be considered cognate un1ess there is evidence to the contrary.

2. RECONSTRUCTION

2.1. The reconstruction of Proto-Malayo-Javanic phonemes is divided into three sections: (1) vowels, (2) semivowels, and (3) consonants. Within each of the sections the reflexes of each phoneme are discussed. Those proto-phonemes for which the evidence seems undisputable are presented first. Only a limited number of reconstructions containing the particu1ar phoneme under consideration are cited in these cases. Many more examples cou1d be given, but such a listing would not contribute to the purpose of this work. For other proto-phonemes, an attempt has been made to give as many examp1es as possib1e, although no claim of exhaustiveness is made.

2.1.1. The following chart shows in which sections the PMJ vowels, semivowels, and consonants are presented. In general, the order is determined by procedure from simple to complex:

Semivowels and consonants:

2 10 t /2.1 ~ f c k I EIJ h q ' p . L3J !20 9d EO j z 9 I ra m n i'I !I I 1 r s . 12.7 • y I~ R1-R3

Vowels:

12•6. a u a

Methodological principles

2.2. Since PMJ is a sub-proto-language or a meso-language, those elements which came from the proto-language, but survived in only one member-language (the so-called 'meso-language cases'), are reconstructed as PMJ on the basis of their comparison with cognates outside this subgroup: e.g., we reconstruct *qasuq (PAN, Sm. inu-asu-asu 'to sip, lap': Wasu[ ]), although only Javanese has a cognate: OJv. asu, NJv. asu (L) 'dog'. We are aware of the fact that any element which cannot be placed in the sub-proto-language is potentially a 'mesolanguage case', as cognates might be found outside this subgroup, but take this to be part of the unavoidable limitations on the application of the comparative method.

To reduce the effects of borrowing the following principles have been applied in determining PMJ reconstructions when no cogn~te is known to occur in any outside language. These principles are based on the historical relations which can reasonably be supposed to have existed among the peoples speaking the four languages under investigation. Malay and Javanese were the languages of empires which maintained political supremacy in the Indonesian Archipelago. Malay was the language of Srivijaya (approximately 7thcentury - 13th century) which dominated Sumatra, most of the Malay Peninsula, and West Java. Javanese was the language of the empire at Mataram, which is first mentioned about 890 A.D. and whose capital was later transferred to

Kediri in 929 where it remained until 1222. Javanese was also the language of the successor empires, Singasari (1222-1293) and Majapahit (1293-1520). The Kediri empire established commercial relations with Sumatra and the Malay Peninsula. At the turn of the first millenium the Javanese invaded Sumatra to attack Srivijaya which soon afterwards retaliated by attacking in Java. At the beginning of this millenium Srivijaya and Kediri (the latter under Airlangga) ca me to a rapprochement. According to the Nagarakertagama, a poem composed in 1365 by Prapanca, the kingdom of Singasari established suzerainty over Sumatra, parts of the Malay Peninsula, West Java, Bali, and Madura. It was probably at this time that Srivijaya lost its control over the Straits of Malacca and Sunda, both of which it had controlled for many centuries. Majapahit - according to the Nagarakertagama - comprised among ot hers all of Sumatra, parts of the Malay Peninsula, Mentawai, West Java, Brunei, Madura, Bali, and the Bandas. Aside from the political developments in which it played an important role, Malay became the language used by traders and therefore became widely current throughout the islands as a means of communication between the speakers of different local languages. These historical facts lead to the assumption that both Sundanese and Madurese, the languages of people whose territories had been dominated by these empires, probably borrowed heavily from the language of their masters. Furthermore, the historical facts lead to the assumption that Malay and Javanese borrowed from each other, both empires having been in close contact and having intruded into each other's domain. On the basis of these assumptions we reconstruct a PMJ form if comparab1e meaning-forms occur in two 1anguages and cou1d otherwise be exp1ained as borrowings, but on1y:

(a) ff one of the comparab1e meaning-forms occurs in 01d Javanese and the other occurs in one of the other three 1anguages: e.g., *timpa9' OJv. timpan 'crfpp1e', Mal. timpan, tempary 'limping, cripp1e'.

(b) if one of the comparab1e meaning-forms occurs in Sundanese and the other in Madurese: e.g., *tunOuh, Snd. tunduh, Mad. ton4u 'sleepy'.

We a1so reconstruct a PMJ form if comparab1e meaning-forms occur fn on1y two 1anguages if for any combination with New Javanese and/or Ma1ay [except for the combination Ma1ay-Madurese which is exc1uded, because of the strong probabi1ity that these two 1anguages formed a later subgroup (see section 1.3.)]

(a) the meaning is basic and there is no pecu1iar reflex which 1ends itse1f to a borrowing hypothesis: e.g., *huntuq, Snd. huntuq (L), OJv. huntu, NJv. untu 'tooth', or

(b) the meaning is non-basic and the words are not so simi1ar as to 1end themse1ves to a borrowing hypothesis: e.g., *[Bb]uncil, Snd. buncir 'to expand (of the wa11s of a sack, of a rice-barn, a1so of"the stomach)', Mal. buncet 'distended (of the stomach)'.

If comparab1e meaning-forms occur in three or all four 1anguages we reconstruct a PMJ form, if all the forms in the 1anguages fo110w the posited regu1ar deve10pment without showing any pecu1iar reflex: e.g., *kutan, Snd. kutau, NJv. kutaU (0), Mal. kutan, Mad. kotan 'undervest'. In instances of unexpected reflexes, we have recourse to the borrowing hypothesis (see below).

2.3. This section presents an example of the methodological principles used in the determination of borrowing and inheritance: (1) The correspondence system of a language is that which applies to its basic vocabulary. When many doublets are found whose differences cannot be explained as due to analogical changes, that member of the doublet is considered to be inherited if its elements correspond to the regular reflexes found in the rest of the basic vocabulary and that other member is considered to be borrowed if its elements do not correspond.

Since Sundanese has a large number of doublets whose differences cannot be explained as due to analogical change, it is a good example to demonstrate the application of this principle. In the following list of Sundanese doublets the entry following (A) indicates the inherited Sundanese reflex and the entry following (B) indicates the reflex which appears in words borrowed from Javanese. In a few instances we find a loan from Malay.

PMJ a (see section 2.6.4.).

(A) 4, (B) Cl: (A) Mn4r 'full and good (of a rice-grain), (B) banar (L) 'true, right'; cf. NJv. banar 'true, right'; (cf. *bannClr, p. 74)

(A) 4 (B) u: (A) lts4h, (B) lusuh 'worn out, used up, weak'; cf. NJv. lusóh (0) (0) 'rotten, worn out'; (cf. *lassah, p. 75).

(A) i before *u, (B) Cl: (A) simbur, (B) sambur; cf. NJv. samb6r

PMJ h- (see section 2.15.1.): (A) h-, (B) q-: (A) hastm 'sour', (B) qasam 'tamarind'; cf. NJv. asam 'tamarind', 'sour' (0); (cf. *hassam, p. 186)· PMJ -ay (see section 2.7.2.4.2.): (A) -eq, (B) -iq: (A) hateq (L), (B) qatiq 'liver, heart, feelings'; cf. NJv. at; (L) 'heart, feelings, soft inner parts (e.g. of bamboo), liver (e.g. as food)'; (cf. hatay, p. 93). PMJ 0- (see section 2.13.): (A) d-, (B) r- : (A) daudtr 'the wild cotton tree', (B) randuq 'the cotton tree'; cf. NJv. ranqu 'cotton tree'; (cf. *OaUOaR2' p. 100). PMJ -q- (see section 2.15.2.2.): (A) -q-, (B) loss of the consonant followed by the contraction of the vowels: (A) ruqum, (B) qa-rum; cf. NJv. a-róm 'fragrant'; (cf. *ruqum, p. 179) PMJ B- (see section 2.12.): (A) b-, (B) w-: (A) bituq 'to fire a weapon, explode, erupt', (B) watuq 'to come out'; cf. NJv. watu (L) 'to appear, come out'; (cf. *Battuq, p. 126). PMJ -R2- (see section 2.14.2.): (A) -r-, (B) loss of the consonant followed by the contraction of the two vowels: (A) taruh 'to bet', (B) toh 'to stake';

(0) 'to bespatter'; (cf. *sambur, p. 141). (A) t, (B) a: (A) tanth 'earth, ground' , (B) tanah 'land, area'; cf. Mal. tanah 'land'; (cf. *tanah, p. 73).

cf. NJv. t~h 'to bet. stake'; (cf. *taR2uh. p. 163). PMJ -aw (see section 2.7.1.3.3.):

(A) -oq. (B) -uq: (A) ba~oq 'water'. si-baffoq 'to wash one's hands'. (B) banuq 'water'; cf. NJv. banu (L) 'water; (cf. "baiiaw. p. 88).

In the 200-word Swadesh list for Sundanese 21 the fo11owing thirty-six words which have an etymo1ogy contain inherited reflexes: PMJ a > Snd. 4:

has4p 'smoke' (cf. *hasap. p. 171); bas4h 'wet' (cf. *Bassah. p. 185); b41ah 'to split' (cf. *Ballah. p. 125); b4rat (L) 'heavy' (cf. BaR2R2at. p. 131); bH4!) (L) 'belly' (cf. *Batta!). p. 126); d4kH (L) 'near' (cf. [D~]akkat. p. 157); h4r4t 'narrow' (cf. *haR2R2at. p. 162); k4r4t 'to cut' (cf. *karrat. p. 48); 1494n (L) 'hand' (cf. *laU9pn. p. 72); sa4tik 'few' (cf. *qattik. *qantik. p. 194); tan4h .. . 'earth' (cf. *tan~h. p. 73); s4s4h 'to wash' (cf. *sassah. p. 72).

PMJ a ) Snd. i before u: ti1uq 'three' (cf. ta11uq. p. 79). PMJ h- > Snd. h-: has4p 'smoke'. see above; hateq (L) 'liver'. see abovei hejoq 'green' (cf. *hijaw. p. 54); h4r4t. see above; hirup 'to live' (cf. *huDip. p. 151); hujan 'rain' (cf. *huzan. p.

lll}; huluq 'head' (cf. *huluq, p. l70); huntuq (L) 'tooth' (cf. *huntuq, p. 38). PMJ 0- > Snd. d-: datau (L) 'to come' (cf. *Datan, p. l49); daun 'leaf' (cf. *Daun, p. l48); deneq (L) 'to hear' (cf. *DaunaRl' p. 76); diq 'in, at' (cf. *Diq, p. l48); duaq 'two' (cf. *Duaq, p. l48). PMJ B- > Snd. b-: bas~h 'wet', see above; batuq 'stone' (cf. *Batuq, p. l28); b~lah 'to split', see above; bentaD 'star' (cf. *BintaD' p. l3l); b~rat (L) 'heavy', see above; bereq (L) 'to give', see above; b4t4D (L) 'belly', see above; binih 'seed (cf. *Binih, p. l26); bitis 'leg' (cf. *Battis, Bantis, p. 193); boroq 'to hunt' (cf. *Buraw, p. 88); buah 'fruit' (cf. *Buah, p. l27); buluq 'feather' (cf. *Buluq, p. l29); bun tut 'tail' (cf. *Buntut, p. l32). PMJ -R2- > Snd. -r-: b~rat (L) 'heavy', see above; bereq (L) 'to give', see above; h4r4t 'narrow', see above; quran ~arereaq 'we (cf. *quR2a~,

PMJ -ay > Snd. -eq: hateq (L) 'liver', see above; bereq (L) 'to give' (cf. *BaR2R2a~, p. l26).

p. 163).

PMJ -aw > Snd. -oq:

boroq 'to hunt', see above; garoq 'to scratch' (cf. *garaw, p. 88).

Of the other items in the 200-word list which we have an etymology only three exhibit reflexes associated with loan words: galut 'to fight'; cf. NJv. ga16t 'to fight'; (cf. *gallut, p. 119); kambaU 'flower'; cf. NJv. kamban (L) 'flower'; (cf. *kambaU' p. 140); banar (L) 'true, right', see p. 74.

There are ten more words which exhibit borrowed reflexes. All of them are identical in shape and meaning to Javanese words. It is interesting that none of them has a cognate in the other Malayo-Javanic languages or in languages outside the subgroup. They are labuq 'ashes, dust' [cf. NJv. labu '(thick) dust']; gadeq (L) 'big' [cf. NJv. ga4é (L) 'big']; gatih 'blood' [cf. NJv. gatéh (L) 'blood']; qandog 'egg' [cf. NJv. an~èg (L) 'egg']; hibar 'to fly' [cf. NJv. ibar 'to fly']; cakal (L) 'to hold' [cf. NJv. cakal (L) 'to hold']; tamba~ (L) 'to sing' [cf. NJv. tamban (L) 'to sing']; lamas 'smooth' [cf. NJv. lamas 'smooth, elegant']; lampaU 'straight' [cf. NJv. lampaU 'straight']; kandal 'thick' [NJv. kandal 'thick'].

The vast majority of the words in the list are composed of phonemes which never appear in instances of doublets. Thus, it is difficult to ascertain whether the form is truly inherited or has been borrowed: e.g., limaq'five' contains reflexes which occur in words together with both set (A) and set (B) reflexes. Such lexical items do not have any bearing here.

(2) If no representative of a proto-segment which exhibits doublets occurs in the basic vocabulary, that member of the doublet is considered to be inherited which clearly contains other elements which can be determined to be inherited reflexes. So, for example, although we find two reflexes for PMJ -ay in Sundanese, it can be shown that -ay is the inherited reflex and -eq is the reflex appearing in loanwords. Consider the following example where we get a Sundanese doublet for PMJ -ay (see section 2.7.2.4.1.):

(A) -ay, (B) -eq: (A) balay 'little wall made of stones',

(B) baleq 'bench, front-hall'; cf. NJv. balé (0) (0)

'bench, front-hall'; (cf. *balay, p. 90).

Since the segment -ay occurs in other words with inherited reflexes which cannot be considered borrowings, -ay must be considered the inherited form: e.g., Snd. h41ay 'numerical coefficient for tenuous objects such as garments' is the reflex of*hallay (p. 91). It has been determined that Snd. 4 is the inherited reflex of *a.

(3) In languages with status-style distinctions the correspondence system is that which applies to the majority of its low words. Again, Sundanese will serve to demonstrate the application of this principle.

A check through R.A. Kern 1906 revealed that about 75% of the Sundanese very high and high words which are unambiguous in respect to whether they contain inherited reflexes or reflexes associated with loan words have reflexes associated with loan words.

There are instances of status-style doublets. The low word contains inherited reflexes and the high word reflexes associated with loan words:

s's'p (L), sasap (H) 'to suck'; (cf. *sapsap, p. 200).

(4) The correspondence system of a language is that which applies to most of its bound morphemes. It is expp.cted that this will ag ree with the correspondence system of the basic vocabulary. For example, the following two Sundanese suffixes contain inherited reflexes:

*-kan, Snd. -ktn, OJv. -a-kan, NJv. -a-kan (H), Mal. -kan, Mad.

Kang. -a-ghan22 , Mad. Baw. -a-kan 'causative'.

*-an, Snd. -tn, OJv., NJv. -an, Mal. ka- -an, Mad. -an 'to be affected by'.

The PMJ morpheme structure

2.4. Three types of morphemes are distinguished in PMJ: monosyllables, dissy11ab1es (origina1 dissy11ab1es or doub1ed monosy11ab1es), and trisy11ab1es.

The PMJ monosy11ab1es consist of the sequence CVC where C is consonant (inc1uding semivowe1) and V is vowe1.

There are two types of dissy11ab1es in PMJ. If the two V's are identica1 the sequence is [a] CV1(C)CV1C, and if the two V's are nonidentica1 the sequence is eb] CV1(C)(C)V_1C. The parenthese~ mean optiona11y present.

The PMJ trisy11ab1es consist of the sequence CV1(C) + [a] or eb] or (presumab1y) CV_1(C)(C)V1(C)(C)V_1C or CV_1(C)(C)V1(C)CV1C.

The formu1as imp1y: (1) no initia1 and no fina1 vowe1s are permitted; (2) most vowe1 sequences are permitted if the vowe1s are nonidentica1, though there is no evidence for the occurrence of the fo1lowing PMJ vowel sequences: *aa, *ia, *ai, *au; (3) a cluster consists of no more than two consonants; (4) clusters occur unly between vowels.

The most common PMJ consonant clusters are nasal clusters (i.e. nasal + non-nasal consonant) and the clusters which appear in doubled monosyllables (see section 2.18.). PMJ also has a large number of double consonants (see sections 2.5.1. and 2.16.). In PMJ dissyllabic bases there is evidence for only one cluster which is not a nasal cluster or a double consonant, namely the cluster *-ks- (e.g.,*ruksak, p. 120).

The consonants *n and *R3 do not occur in preconsonantal, postconsonantal, and final position; *1 and *R2 do not occur in initial, preconsonantal, and postconsonantal position; *j, *z, *c, *~, *8, and *0 do not occur in preconsonantal and final position; *w, *Rl' and *q do not occur in preconsonantal and postconsonantal position; *D does not occur in postconsonantal position; and *y, *b, and *ç do not occur in preconsonantal position.

Thus, the following consonants occur in initial position: *w, *y, *m, *n, *n, *U' *1, *r, *s, *d, *j, *z, *g, *c, *p, *t, *k, *~, *B, *b, *0, *ç, *Rl' *R3, *h, and *q.

The following consonants occur in preconsonantal position: *m, *n, *u, *1, *r, *s, *d, *g, *p, *t, *k, and *h.

The following consonants occur in postconsonantal position: *y, *m, *n, *1, *r, *s, *d, *j, *z, *g, *c, *p, *t, *k, *~, *8, *b, *0, *ç, and *h.

The following consonants occur in final position: *w, *y, *m, *n, *D' *1, *1, *r, *s, *d, *g, *p, *t, *k. *b, *Rl' *R2, *h. and *q.

Madurese double consonants

2.5. Madurese is the only Ma1ayo-Javanic 1anguage which distinguishes single from double consonants. All consonants occur doubled except q. There are 1anguages outside the Ma1ayo-Javanic subgroup that a1so distinguish single from double consonants, e.g., I10cano, Buginese, and Makassarese. Most instances of double consonants in these 1anguages can be associated with the position af ter PAN a or PHN a, e.g., PAN panuq, I10cano punno, Buginese panno, Snd. pinuh, OJv. panuh, NJv. panóh (B), Mal. panoh 'ful1' (cf. PMJ pannuh, p. 79); PHN lapas, Bug. 1appaq, Mak. 1appasaq, Snd. 14pas, OJv., NJv. lapas, Mal. lapas, Mad.lappas 'free, loose'. In Madurese, as in ot her languages, there are a1so many instances of double consonants which do not occur af ter *a.

2.5.1. Double consonants af ter PMJ a. There are two possib1e hypotheses with respect to exp1aining the origin of double consonants af ter *a. Either PMJ had double consonants af ter *a and Madurese retained the double consonants, whi1e the other Ma1ayo-Javanic 1anguages reduced them to a single consonant; or Madurese double consonants were deve10ped af ter *a. The first hypothesis is supported by Malay manuscripts in which words such as modern Malay basar, karat, and taloq are spelled b.ss.r, k.rr.t, and t.11.q respective1y, i.e., with double consonants. A1so in an 01d Javanese document of the 10th century (Jayapattra) double consonants can be found af ter wh at appears as modern Javanese a: e.g., the name Gallam [for modern Javanese Galam], pajjah [for modern Javanese pajah (H) 'dead,].23

There is no reason at this time not to reconstruct PMJ double consonants af ter PMJ a exr.ept when the following consonant is *h or *q. There is no necessity to reconstruct*-hh-, or*-qq-, since evidence for these does not appear in Madurese, the only language with double consonants (c.f. *sahaR, p.174( and *kaqan, p. 73). Thus we reach the following reconstructions:

*karrat (PAN, Sa. qolo 'to cut off the ends': kaRat), Snd. k4r4t, OJv., NJv. karat, Mal. karat, Mad. karraq 'to cut'.

*lassuU (PHN, Tag. lus6n 'rice-mortar': lasun), Snd. lisuD, NJv. lasóU' Mal. lasoU' Mad. lassoD 'rice-mortar'.

We also reconstruct double consonants af ter *a in case a Madurese cognate is lacking: e.g., BaR1Rlas, p. 126.

Since languages such as Ilocano, Buginese, Makassarese, and Madurese which have double consonants af ter *a are geographically remote from one another, one might argue that not only PMJ but also proto-languages of higher order and perhaps the proto-language of highest order had double consonants af ter *a. These double consonants we re reduced to single consonants in all languages except for those mentioned above.

A possible argument against the reconstruction of double consonants af ter *a was presented by Dyen (1971:44). Ilocano has puno 'full' and a syncopated form na-pno af ter a vowel-final prefix. This alternation could argue for an independent development of double consonants af ter vowels which reflect *a, since "such an alternation is unlikely to have developed if it would have brought three consonants together." (Id.) However, since double consonants also occur in positions other than af ter *a and since such instances are not rare, one can still argue for areconstruction with double consonants also af ter *a.

For PMJ double consonants in other environments, see section 2.16.

2.5.2. Madurese double consonants by assimilation. In many instances Madurese double consonants can be attributed to the assimilation of the final consonant of the first syllable. This assimilation probably is the regular development in the B. and P. dialects:

*taptap, Snd. tatap, Mad. taptap, Mad. B. tattap 'to hit with the flat hand'.

*tastas (PHN, Tag. tistfs 'to tear up'), OJv. a-natas 'came out', NJV. tatas 'well-cutting' (0), 'cut off', Mal. tatas 'slit open, rip up. Implies that the force is applied from the inner side, e.g., wh en a chick breaks out of its egg', Mad. tastas, Mad. B. and Mad. P. tattas 'to hatch'.

2.5.3. A Madurese double consonant as reflex of PMJ -nh-. In a single instance Madurese has a double consonant where Sundanese and Old Javanese have -Uh-:

*taUhiq, Snd. tauhiq (H) 'to get up', OJv. a-taUhi 'to awake', NJv. taUi 'to get up', Mad. tau~è(h) 'to stay up all night'.

2.5.4. Madurese double consonants by analogy. The double consonants -ttin Mad. pèttö(h), pèttöq 'seven' (from *pituq) and -11- in Mad. bállu(h), bálluq 'eight' (from *w2aluq) are perhaps analogical to the double consonants -11- in Mad. tallö(h), telloq 'three' (from

*ta11uq) and -nnin Mad. annam 'six' (from *qannam).

The PMJ vowe1s

2.6. Dempwo1ff reconstructed four PAN vowe1s: PAN a, PAN i, PAN u, and PAN a. Four vowe1s are a1so reconstructed for PMJ and it is posited that the PAN vowe1s continued into PMJ. The PMJ vowe1s are denoted by the same symbo1s. Dempwo1ff reconstructed this PAN vowe1 distinction based on a four-vowe1 system, because he found it sufficient to account for the vowe1 systems of the 1anguages he investigated. He treated the origin of Ma1ay e and 0 in non-fina1 sy11ab1es (from *i and *u respective1y) as 'Tendenzen zur Lautverschiebung' (2.22) and the origin of Javanese é, è, ó, and 0 in non-fina1 sy11ab1es and è and o in fina1 sy11ab1es (for *i and *u respective1y) as 'unerk1ärte Ausnahmen' (1.86). Dyen (1953a:7 f.) treated them as different dia1ecta1 ref1exes of *i and *u respective1y. As wil1 appear bel ow, we have put forth hypotheses simi1ar to those of Dyen's by attributing the origin of these vowels to secondary deve10pments inc1uding dia1ecta1 and inter1inguistic borrowing. Dyen 1949 showed that a PAN four-vowe1 reconstruction was sufficient to exp1ain the origin of the Trukese nine-vowe1 system. In this section it wi11 be shown that a PMJ four-vowe1 system probab1y is sufficient to account for the seven-vowel system of Sundanese (i, e, u, 0, a, a), the eight-vowel system of Javanese (i, é, è, u, Ó, 0, a, a), the six-vowe1 system of Ma1ay (i, e, u, 0, a, a), and the nine-vowe1 system of Madurese (i, é, è, u, Ó, 0, a, f, a). For a tabulation of the ref1exes of the PMJ vowe1s in the four 1anguages, see section 3.

2.6.1. PMJ a

2.6.1.1. PMJ penultimate and ultimate a. PMJ penultimate and ultimate a appears as Sundanese, Old Javanese, and Malay a. PMJ ultimate a before PMJ -q (which disappearp.d in Javanese, Malay, and in Central and East Madurese, see section 2.15.2.3.) became NJv. -0. PMJ penultimate a before PMJ ultimate a before PMJ -q also became NJv. O. Otherwise PMJ penultimate and ultimate a appear as NJv. a. The Madurese reflexes are á and a (distributed according to the Madurese vowel alternation rule given in section 1.8.3.1.):

*qapaq (PAN, Sa. t-aha 'what': ?apa?), OJv. apa, NJv. öpO (L), Mal. apa, Mad. apa(h) 'what'.

*kitaq (PAN, To., Fu. kita 'I': kita?), Snd. kitaq, OJv. kita 'you', NJv. kito 'we, us' (B), 'I, my, mine' (0), Mal. kita , 'we; you and I', Mad. kèta 'I'.

*lambUQ(PWI, TBt. lambun 'side'), Snd. lambuU 'the hollow space under the ribs on the side', OJv. lambuU' NJv. lamb6U' Mal. lamboD 'side, flank', Mad. lambhuD 'the sides of the body above the hip'.

*rampas (PHN, Tag. gá:pas 'cutting': Ra[m]pas), Snd. rampas 'to loot', OJv. r-in-ampas 'looted', NJv., Mal., Mad. rampas 'to loot'.

*siUgah, Snd. siUgah 'to go out of the way', OJv. s-um-iUgah 'went out', NJv. siUgah (0) 'to avoid', Mad. sèo9há 'to withdraw'.

2.6.1.2. PMJ antepenultimate a. PMJ antepenultimate a appears as Sundanese and Old Javanese a, and became New Javenese and Malay a, and Madurese á or a (distributed according to the vowe1 alternation ru1e given in section 1.8.3.1.):

*ga1ugaq, OJv. ga1uga, NJv. g(a)lugo, Mal. ga1uga, Mad. ghá1ughá(h) 'red dye'.

*tamiao, Snd. tamiao, OJv. tamyaU, Mal. tamian 'kind of bamboo'. *tarimaq (PWI, NgD. tärima 'to receive': ta[r]ima), Snd. tarimaq (L), OJv. a-narima, NJv. t(a)rimo (L), Mal. tarima, Mad. tarèma(h) 'to receive'.

A1though Sundanese and 01d Javanese cognates are 1acking, a reconstruction with antepenu1timate a has a1so been made in the fo110wing instance:

*Sa1irau24 (PHN, Tag. ma11:1aU 'su1phur': baliraU), NJv. waliran, Mal. ba1eran, Mad. bá1iráq 'su1phur'.

In the fo110wing comparison PMJ antepenu1timate a and PMJ penu1timate a contracted in earl ier Javanese af ter the 10ss of *R:

*SaRaniq25 (PAN, Fi. tan-ane 'man': baRani[?h]~ OJv. wäni, NJv. wani (L), Mal. barani 'brave'.

The following comparison is an instance of the contractior, of a penu1timate vowe1 with u1timate a in Javanese. PMJ antepenu1timate a thus became a Javanese penu1timate vowe1 and appears as a (see section 2.6.1.1.):

*pariaq, Snd. pariaq, NJv. paré, Mal. paria, Mad. parèya(h) 'a gourd' •

2.6.2. PMJ i.

2.6.2.1. PMJ u1timate i. PMJ u1timate i appears as Sundanese and 01d Javanese i. and became Madurese i or è (distributed according to the vowe1 alternation ru1e given in section 1.8.3.1). It appears as

New Javanese and Malay i when originally followed by PMJ -q (which was lost, see section 2.15.2.3.). Otherwise it became New Javanese é and Malay e:

*gantiq (PHN, Tag. gantf 'requital': ganti), Snd. gantiq (L) 'to change', OJv. g-um-anti 'to follow', NJv. ganti (DJ 'in place of', Mal. ganti, Mad. ghántè(h) 'to replace'. *jahit (PAN, Sm. sai-sai 'to bind together': zaqit), OJv. j-inahit 'cut apart (7)', NJv. jaát, Mal. jahet, Mad. jháiq 'to sew'. *kamunin, Snd., OJv. kamunin, NJv. kamunén, Mal. kamuneU' Mad. kamonèo 'kind of plant'. *qujiq26 (PHN, Tag. qu:riq 'carat': uzi), Snd. qujiq 'to examine metals', NJv. uji (0), Mal. uji 'to test silver or gold'. New Javanese exhibits è as a dialectal reflex of PMJ i before consonants other than PMJ -q: *gapit (PWI, TBt. gappit 'to pinch': ga[m]pit), Snd. Bad. gapit, NJv. gapét, gapèt (0) 'nipper, clamp', Mad. S. and Mad. Kang. ghápèq 'nipper'. *guliU (PWI, TBt. guliU 'rolled'), Snd. gu~guliU 'roll-cushion', gu-guliu-an 'to roll, swing back and forth', OJv. an pa-gulin-an 'revolving', NJv. guléu 'roll-cushion', O-g6léU' o-gólèn (0) 'to swing back and forth, incline (of a ship), slanting', Mal. guléU,goleU 'to rollover', Mad. ghulin, Mad. S. ghu-ghuliU 'roll-cushion'. *rapih, Snd. rapih 'to be good with one another, of one opinion', NJv. rapéh (0) (0), rapèh (0) 'completely free (of pain), clear, in order', Mad. rap~ 'attached, devoted to'.

In instances 1ike the fo11owing Sundanese exhibits e as reflex of u1timate *i:

*kuniO (PWI, NgD. k5niu-an 'brass'), Snd. konen, OJv. kunin,

NJv. kunéQ, Mal. kuniU' Mad. kön~Ü 'yellow'.

*tajin, Snd. tajen, NJv. tajén, Mal. tajen 'stareh, made of ricef1our', Mad. tajhin 'mash, pap'.

2.6.2.2. PMJ penu1timate i. PMJ penu1timate appears as Sundanese, 01d and New Javanese, and Ma1ay i. It became Madurese i or è (distributed according to the vowe1 alternation ru1e given in section 1.8.3.1.):

*gi1aq (PAN, Fi. ki1a 'wild': gi1a[?h]), Snd. gi1aq 'shy, mad, scared', OJv. gi1a 'to shudder', NJv. gi1b 'not wise, be mad, to abhor', Mal. gi1a, Mad. ghi1á(h) 'mad, menta11y 111'.

*h1riq (PHN, Tag. h{:liq 'stimulus': hiri), Snd. hiriq 'bad, jea1ous', OJv. aU-iri, ma-hiry-y-a 'jealous', NJv. irèn (=iri+an) 'jealous' (0), Mal. (h)iri hati 'spite, malice', Mad. ~rè(h) 'jealous' .

*ki1a~, OJv., NJv. ki1aQ 'sugar-syrup', Mal. Sumo ki1aQ 'a fermented drink obtained from cane-sugar or pa1m-sugar', Mad. kè1aQ 'thick syrup out of which sugar is made'.

PMJ penultimate i before PMJ -aw, and PMJ -ay (which became Snd. oq, eq respective1y, see sections 2.7.1.3.1. and 2.7.2.4.2.) became Snd. e. Let us first consider examp1es of PMJ i before PMJ -aw:

*hijaw (PHN, Tag. h{:raw 'cock with green feathers': hizaw), Snd. hejoq, OJv. a-hijo, NJv. ijó (L), Mal. (h)ijaw 'green', Mad. èjhu(h) in ffèyor èjhu(h) 'kind of coconut with a green skin'. *limaw (PAN, Fi. mo1i 'citrus-fruit, orange'), Snd. 1emoq, OJv. lima, Mal. 1imaw, Mad. 1êmà(h) 'citrus-fruit'. Snd. 1imoq 'citrus-fruit' probab1y is a loan from a dialect or 1anguage such as Javanese in which *i before *-aw became i. A simi1ar instance of Snd. i before *-aw in a presumed loanword is found in the fo11owing comparison: *qi1aw (PAN, Fi. qi1o-i10 'mirror': ?i1aw), Snd. qi1oq, OJv. man-ilo, NJv. i16 'to mirror' , Mal. i1aw 'shimmering (as sun1ight in water)'. PMJ penu1timate i before *- y: *[t~]i[rR2]ay, Snd. tereq (L) in qindun tereq 'step-mother', Mal. tiri 'step-' in maq tiri 'step-mother'. In same instanees Sundanese exhibits e as reflex of PMJ penu1timate i a1though not fo11owed by a PMJ diphthong. In the fo11owing four comparisons the u1timate vowe1 is a. Perhaps earl ier Sundanese i underwent partia1 assimi1ation to this vowe1 in a dialect of Sundanese: *Ba1i1aq27 (PHN, Tag. ba1i1á 'weaver's sword': balija), Snd. bareraq (with assimi1ation of 1 to r), NJv. w(a)lir3, Mad. bá1i1~(h) 'weaver's sword'. *lintah (PHN, Tag. 1intáq 'leech': 00 lintah, 10 1intaq), Snd. 1entah, NJv., Mal. lintah, Mad. l~nta 'leeeh'. *pintaq (PWI, TBt. pitta 'to desire': pi[n]ta), Snd. pentaq (L), OJv. ma-pinta, NJv. pinto (B), Mal. pinta, Mad. pènta(h) 'to ask'. In the fo11owing instanee there is a related Javanese word which can be considered the source for the Sundanese form with e:

*kihuij, Snd. keoU' NJv. kéyóU' Mal. kioU 'land-snail'.

PMJ penultimate i became New Javanese é in the Eastern subgroup of Javanese dialects in open or closed syllables, if the word-final syllable contained a like vowel (see section 1.5.2.):

*pilih (PAN, Fi. vili 'to pick up': piliq), Snd. pilih, OJv. a-pilih, NJv. mil~h, méléh (0), Mal. pileh, Mad. pèlè 'to choose' .

*pitik, Snd. pitik 'chick', OJv. pitik 'chicken', NJv. pitéq, pétéq (0) 'chick', pitéq (L) , hen, chi cken ' , Mad. pètèq 'chick'.

PMJ penultimate i became New Javanese è presumably in a northeastern dialect in instances like the following:

*minnak (PWI, NgO. minak 'oil': minak), Snd., OJv. minak, NJv. minaq, mènaq (0), Mad. mènnaq 'oil'.

*liruq, Snd. liruq 'confounded', NJv. liru (L) 'in place of',

k-liru (L) (0) 'wrong, mistaken', k-lèru 'confounded', Mal.

ka-liru 'bewilderment', Mad. lèro(h) 'in place of'.

Wh en the ultimate vowel appears in New Javanese as è (see section 2.6.2.1.) PMJ penultimate i likewise appears as è in the following comparisons:

*hiriU' Snd. m-iriu, OJv. a-hiriU' NJv. iréu, m-iréu, m-èrèu (0) 'inclining', Mal. m-ereU 'to lean over', 'incline to one side', Mad. m-èrèu 'to lie on one side, not horizontal.' *licik, Snd. licik 'deceitful, cunning', NJv. lic~q 'insipid, cowardly', lécéq (0), lècèq 'to sit loosely, to take to one's heels', Mad. lècèk 'no,t true, lying'.

*titis (PAN, Fi. titi 'to flow, run'), Snd. titis 'to pour out', OJv. titis 'to drip down', NJv. tités 'to pour out', tètès 'drop, drip', Mal. tites 'a gentle drop', Mad. tètès 'to drip'.

In some instances Malay exhibits competing forms, one with a penultimate i, the other with a penultimate e. Thus, we can posit two dialects, one in which PMJ penultimate i appears as i and another in which this penultimate i became e. In the following instances it is best to explain the existence of competing forms as borrowings from the e-dialect into the i-dialect, or conversely:

*ligar, NJv. ligar (0) 'billof exchange', Mal. ligar, legar 'to make the circuit of the rice-mill', Mad. P. lèghar 'to go round, circle'.

*liwat, Snd., NJv. liwat (L), Mal. liwat, lewat, Mad. lèbát 'past, af ter'.

*pisah, Snd. pisah, OJv. a-pisah, NJv. pisah, Mal. pisah, pesah, Mad. pèsa 'to separate'.

*rim[Bb]as (PWI, TBt. rimbas 'adze': rimbas), Snd. rimbas 'adze', OJv. rimbas 'plane, shredder', NJv. rimbas 'chopped' (0), Mal. rimbas, rembas 'adze'.

*simpan (PAN, Sm. sipa 'to lean to one side'), Snd. simpan 'to branch off', OJv. a-nimpan 'to take a side-road', NJv. simpan-an 'sideroad', Mal. simpan, sempan 'to branch off', Mad. sèmpan 'to give way'.

*siram (PWI, NgD. siram 'to spray': si[r]am), Snd. siram 'to water, spray', OJv. s-in-iram 'watered', NJv. siram, Mal. siram, seram, Mad. sèram 'to water'.

In the following instanee there is a related Javanese word which can be considered the souree for the Malay form with e: *qir1d, Snd. qirid 'to drive before one, take along', OJv. aU-irid-akan 'to take s.o. along as a slave', NJv. iréd 'pulled away' (0), 'lead' (B), èrèd (0) 'to drag along', Mal. eret 'to drag at', e.g. mau-eret kuda 'to lead a horse by the snaffle', Mad. èrèt 'to drag, tow s.o. or s.t.'. In some comparisons the Malay cognate with penultimate e occurs without a by-form w1th penultimate i: *ci~ak, Snd. citak 'form, cast, press', NJv. ci~aq 'cast, form', Mal. cetaq 'to cast or shape by the use of a mould' , Mad. S. P. cè~aq 'cast, form'. *g1wan. Snd. giwaU. NJv. giwaU (0). Mal. gewant Mad. ghibáu 'mother-of-pearl'. *qikuR (PAN. To. iku 'tail. end': w4ikuRl ), OJv. iku, Mal. ekor 'ta1l', Mad. èkor in soroy èkor ' a comb the handle of which 1s shaped like a tail '. *11nsir, Snd. liusir 'to be in a slanting position (of the sun)', OJv. linsir 'to go toward sunset', NJv. linsér 'afternoon', Mal.le!)ser 'side-slipping', Mad. l~os~r 'afternoon'. *t1was (PAN, To., Fu. ma-tsiva 'poor'), Snd. OJv., NJv. tiwas 'to perish', Mal. tewas 'being worsted', Mad. tèb~s 'accident, disaster'. In the following instanee there is a related Javanese word which can be considered the souree for the Malay form with e: *p1cak, OJv. picak 'one-eyed', NJv. picaq, picak (0), pecaq (0) 'blind'; Mal. pecaq 'blind', 'crushed or driven in at one pOint,28, Mad. B. pècak 'blind'.

2.6.2.3. In a number of instances Sundanese, Malay, and Old Javanese exhibit a penultimate e matching a New Javanese é in words which have a different ultimate vowel. We also find instances of Sundanese, Malay, and Old Javanese penultimate and ultimate e matching New Javanese penultimate and ultimate~. FinaJly we find instances in which Sundanese and Malay and rarely Old Javanese exhibit an ultimate e af ter a different vowel matching a New Javanese ultimate è. Madurese exhibits i or è in all these instances (distributed according to the vowel alternation rule given in section 1.8.3.1.).

The question is whether the reconstruction of a phoneme *e for these matchings is necessary or whether the Sundanese and Malay forms containing e can be attributed to borrowings from Javanese. Madurese has no bearing on this problem, since it exhibits in these words the same reflexes as for those for which PMJ i was reconstructed. We know from our preceding discussion that PMJ penultimate i dialectally became NJv. é or é and that PMJ ultimate i dialectally became NJv. ~. Presumably dialectal forms containing é or è spread over the whole of the Javanese speaking area and were subsequently borrowed by Malay and Sundanese. This is all the more likely to have happened, since all the words in question belong to that part of the vocabulary which is considered non-basic. There is therefore no compelling reason at this point to assign a distinction between *i and *e to PMJ, and we leave any decision in this matter to further research.

For comparative purposes a list of all words which have e in Sundanese, Old Javanese, and Malay, and é or ~ in New Javanese is given bel ow. No PMJ reconstruction is. made unless a cognate occurs in a language outside this subgroup. It should be kept in mind that the Sundanese and Malay forms are considered likely to be borrowings and that the Madurese forms are ambiguous as to whether they are inherited or borrowed:

Snd. qaneh '~trange, curious', NJv. anèh 'unused to, peculiar', Mad. anè 'strange, curious'.

*ban~iU (PWI, NgO. bantiu 'wild ox'), Snd. banteU. OJv. banten, NJv. bantèn, Mal. banteu, Mad. bh'n~èu 'wild ox'.

Snd. bebas 'finished, completely gone', bebas (L) 'paid (of debt);, NJv. bèbas (0) (0) 'gone', b6bas-an 'without extras', Mal. bebas 'freedom of restraint', Mad. bhibhás 'paid (of debt)'.

*bibik (PHN, Tag. bib( 'duck': bibi), Snd., OJv. bebek, NJv. bèbèq (L) 'duck'.

Snd. begal, NJv. bégal, Mal. begal, Mad. bhighál 'highwayrobbery'.

*biUkuk (PAN, Sm. piqo 'bent'), Snd. beUkok, NJv. b6Ukoq, Mal. benkoq 'bent'.

Snd. belaq 'to help, be at one's side', OJv. bela 'to follow s;o. into death', NJv. bélo (B) 'partner, sharer', Mal. bela 'blood-offering in the way of self-immolation', Mad. bhillá(h) 'to endanger O.S. for s.o. else'.

Snd. belek 'to have red eyes', OJv. belek 's.o. who is suffering from an eye-disease', NJv. bèlèq-an 'to suffer from inflamed eyes'.

Snd. berak 'dung', NJv. béraq (0) (VL) 'dirt', Mal. beraq (vulgar) 'to defecate'.

Snd. celen, NJv. cè1èo (L), Mad. cèlèo 'swine'.

Snd. comel, NJv. cómèl (L) 'to talk a lot', Mal. comel 'to murmur, grumble' , Mad. c~èl 'to ask a lot'.

Snd. qepok (H), NJv. ~pèq, Mal. epoq, Mad. B. èpoq 'receptacle for sireh-requisites'.

Snd. gooseo, NJv. góosèO' Mad. ghuOsèo 'rattle, bell '.

Snd. jejer, NJv. jéjèr, Mal. jejer, Mad. B. jhijhir 'one af ter another, orderly line'.

Snd. jeoke1 'unhappy, impatient', NJv. jèokè1 'itching (not to be able to hear or see anymore)', Mal. jevkel 'peevish'.

Snd. jenkol, NJv. jéokol, Mal. jeokol 'kind of tree which delivers malodorous fruit'.

Snd. joged, NJv. jógèd, Mal. joget 'dancing-girl'.

NJv. l~lèh 'soft (melted in the sun)', Mal. leleh 'to trickle gently', Mad. lèllé 'to drip off (e.g. melted sugar)'.

Snd. sered, NJv. sèrèd, Mal. seret 'to bail behind'.

Snd. tembok, NJv. témboq, Mal. temboq, Mad.t~buq 'wall '.

Snd. tempel 'to hang at, against, stick', NJv. t~pèl 'to stick on', Mal. tempel 'to plaster or stick on'.

NJv. tèpaq 'sireh-box' (0), 'salver', Mal. tepaq, Mad. tèpaq 'sireh-box' •

In ot her instances non-final New Javanese é or è are the reflexes of an early Javanese or PMJ vowel sequence -aior -iawhich is attested in Old Javanese data or data from other MalayoJavanic languages. 29 An example is the following comparison:

*~aiO, Snd. deqeO 'raw meat, cut into slices, spiced, and then ,~ " dried in the sun', OJv. gen'dried meat', NJv. ~enqen, geugen

(0) (0) 'dried meat'. Mal. daeU 'slice into thin strips and drying in the sun'.

The Sundanese cognate in the preceding instance probably is the result of the following development: (1) PMJ i partially assimilated to the preceding vowe1 resulting in a form *daeU; (2) the initial vowel assimilated to the final vowel and a -qwas inserted between the two identical vowels. since Sundanese does not have sequences of identical vowels. 30

Sundanese deUdeO'dried meat (sold in a warong)'. Mal. dendeU. and Mad. ~hiU~hio 'jerked meat' are treated as loans from Javanese. TobaBatak. a non-Malayo-Javanic language. also has dendeu'jerked meat'; this form is probably also a borrowing. either directly from Javanese or. more likely. indirectly via Malay.

2.6.2.4. PMJ antepenultimate i. PMJ antepenultimate i appears as Sundanese and Old Javanese i. and became New Javanese and Malay a. and Madurese á or a (distributed according to the vowel alternation rule given in section 1.8.3.1.):

*tiUgaluU. Snd .• OJv. tingaluu. NJv. t-r-aOgalón. Mal. tangaloO. Mad. taugháluU 'civet-cat'.

*ti09ilio. OJv. ti99ilio. NJv. taTIgilé~. Mal. taOgileo 'anteater'.

In the following comparison Javanese exhibits the contraction of the penultimate and ultimate vowel. PMJ antepenultimate i thus became a Javanese penultimate vowel and therefore appears as i (see section 2.6.2.2.):

*binuaO (PHN. Tag. banuaD31 'octomelis': banu[w]an). Snd. binuaU' OJv. winoD' NJv. winoU' Mal. banuaO 'kind of tree'.

2.6.3. PMJ u

2.6.3.1. Different dia1ecta1 ref1exes a1so,occur for PMJ u, as they do for PMJ i, in Sundanese, Javanese, and Malay. The question arises whether a distinction between two vowe1s, *u and *0, shou1d be assigned to PMJ or whether certain forms can be attributed to borrowings. This prob1em is 1ike that of determining whether PMJ distinguishes *i and *e (see section 2.6.2.3.). We conc1ude here too that such a distinction does not need to be assigned to PMJ at this time.

2.6.3.2. PMJ u1timate u. PMJ u1timate u appears as Sundanese and 01d Javanese u, but became Madurese u or 0 (distributed according to the vowe1 alternation rule given in section 1.8.3.1.). It became New Javanese and Malay u when originally followed by PMJ -q (which was lost, see section 2.15.2.3.). Otherwise it became New Javanese Ó and Malay 0:

*jujuq, Snd. jujuq 'to keep a loose hand over, give lots of food', NJv. juju, Mad. jhujhu(h) 'to feed'.

*timun (PAN, Fi. timo 'melon'), Snd. bont~ timun, NJv. ka-timón, Ma1.timon, Mad. B. P. tèmàn, S. an-tèmon 'cucumber'.

*hunus (PAN, Sm. unus-i 'to pull out': hlunus), OJv. um-unus, NJv. unós, Mal. (h)unos, Mad. onos 'to pull out, draw off'. New Javanese exhibits ~ as a dialectal reflex of PMJ u before consonants other than PMJ -q:

*cankul (PWI, NgO. sankul 'hoe'), NJv. caukól, cankol (0), Mal. canko1 'hoe'.

*palluh,OJv. a-pal uh 'not strong', NJv. palóh, pal~h (0) 'impo- , tent', Mal. paloh, Mad. pallo 'to sweat'.

*rapuh (PAN, Fi. ravu 'smash up'), Snd. rapuh 'brittle, frail',

OJv. rapü(h) 'tired', NJv. rapJh, rap~h (B) 'tired, exhausted' , Mal. rapoh 'brittle, fragile'.

In the following comparison Sundanese exhibits a form with 0 as reflex of PMJ ultimate u. The penultimate vowel is a. Perhaps earl ier Sundanese u underwent partial assimilation to the preceding vowel in a dialect of Sundanese:

*Bauk, Snd. baok, OJv. wok, NJv. woq32 'beard', Mal. baoq 'hair along the line of the jaw-bone'.

2.6.3.3. PMJ penultimate u. PMJ penultimate u appears as u in Sundanese, Old and New Javanese, and Malay. It became Madurese u or ~ (distributed according to the vowel alternation rule given in section 1.8.3.1.):

*jujur, Snd. jujur, NJv. jUjór, Mad. jhujhur 'to go straight through'.

*lurah, Snd., NJv., Mal. Java lurah, Mad. lbra 'headman'.

*tumpau (PAN, Fi. tu va 'to pile up': tupa~, tumpa9)' Snd. tumpau 'to be on it, lie on it, go along', OJv. tumpau 'to sit', NJv. tumpan 'on top, lie on another' (0), Mal. tumpan 'join in with others', Mad. tompau 'to put on s.t., go along.'

PMJ penultimate u before PMJ -aw, PMJ -aw (which both became Snd. -oq, see sections 2.7.1.3.1. and 2.7.1.3.3.), and PMJ -ay (which became Snd. -eq, see section 2.7.2.4.2.) became Snd. o. The following comparison is an instance of *u before *-aw:

*pukaw (PHN, Tag. pU:kaw 'inspiration'), Snd. pokoq, Mal. pukaw 'narcotic'.

The following comparison is an instance of Snd. u betore *-aw in a presumed loanword: *pulaw (PWI, TBt. pulo 'island,)33, Snd. puloq, NJv. pUló, Mal. pulaw, Mad. pol~(h) 'island'. PMJ u before *-aw: *tujjaw (PAN, Sa. i_it034 'to stand on the course': tuzu[?h]), Snd. tojoq 'to conform, go towards, rely on', OJv., NJv. tuju 'to aim', Mal. tuju 'to point at, make straight for', Mad. t~jjhu(h) 'exact'. Snd. tujuq 'to conform to' is probably a loan, perhaps from Javanese or Malay. *[Bb]u[dç]aw, Snd. S.-B. bodoq 'an additional dish, eaten mainly by the Baduys, consisting of rotten fish', Mal. Ked., Kel. budu 'anchovies pickled in brine af ter being dried and partially decayed'. PMJ u before *-ay: *huay (PAN, Sa. ue 'rattan': quaq3aYi)35, Snd. hoeq, OJv. hwi 'rattan'. In one instanee Sundanese exhibits 0 as reflex of PMJ penultimate u before *-ay: *[rR2]uay, Snd. roay in kacan roay, Mal. ruay in kacan ruay 'a small variety of the lalab'. *Bunkuk (PWI, TBt. bukkuk 'bent': bunkuk), Snd. bunkuk, bOnkok, OJv. wuUkuk, NJv. wUUkóq, m-bunkóq, Mal. bonkoq, Mad. bunkàq 'bent, bowed'.

In a few comparisons Sundanese exhibits competing forms, one with a penultimate (and ultimate) u, the other with a penultimate (and ultimate) o. In the following two instanees there is a related Malay word which can be considered the source for the Sundanese form with 0:

*surun (PHN, Tag. s~:lon 'go ahead, forward'), Snd. surun' soroD' OJv. ma-nuru9' NJv. suróU' Mal. soroD,.Mad. S. sàro~ 'to push a10ng'.

In one comparison the Sundanese competitive form with penultimate 0 can be regarded as a borrowing from Javanese or Ma1ay:

*bunkar (PHN, Tag. bunká1 'to root up the ground'), Snd. bunkar, bonkar 'to lift, raise, break open', NJv. bunkar 'to un10ad' , 'break open' (0), bÓrykar (0) 'broken open', Mal. bUnkar, boUkar 'to heave up', Mad. bhuOkar 'to uproot'.

In the fo110wing instance Sundanese has a cognate containing 0 without a by-form containing u. This form can be regarded as a borrowing from an unidentified Sundanese dialect in which *u became 0:

*~uk~uk (PAN, Fi. tutu 'to beat the drum'), Snd. toktok, NJv. tutóq . 'to knock', Mal. tutoq 'to crush rattans into soft . fibrous pulp', Mad. ~~k~àk 'to give a tap on the hand or the knee'.

PMJ penu1timate u became New Javanese 0 in the Eastern subgroup of Javanese dia1ects in open and c10sed sy11ab1es if the word-fina1 sy11ab1e contained a 1ike vowe1 (see section 1.5.2.):

*pucuk (PHN, Tag. pusók 'ardor, impetuosity'), Snd. pucuk I " 'top-branch1et', OJv. pucuk 'point', NJv. pucoq, pocoq (0) 'top,point', Mal. pucoq 'shoot. top-branch1et'. Mad. B. P. p~c~k, Mad. B. P. S. pÓ-p~c~k 'the sharp point of a sharp weapon'.

According to data in the Co11ection Pigeaud36 penu1timate u became 6 a1so before non-1ike vowe1s in the dia1ects of Pasuruan and Koeta-Arja (both are 10cated in the province of East Java). Wh ether this development is regular or not is not clear: *sugih, Snd. sugih, NJv. sugéh, sógéh (0), Mad. s~ghi 'rich'. Wh en the ultimate vowel appears in New Javanese as 0 (see above), PMJ penultimate u likewise appears as ~ in some words, such as:

*cuDur, Snd. cu]ur 'upper lip of an animal, mouth', NJv. cun6r (0), cÓ~ór (0), co~or 'nose, mouth of an animal', Mad. B. càO~r 'to eat'.

*sun[D9]uO' Snd. sunduO 'bamboo-basket in which the grasscutters store their grass', NJv. sun260 'bamboo-basket', s~n9~D (0) 'pole with a basket fixed to its top (to pick fruit)'.

In some instances Malay exhibits competing forms, one with a penultimate u, the other with penultimate o. Thus, we can posit two dialects, one in which PMJ penultimate'u appears as u and another in which this penultimate u became o. In the following instances it is best to explain the existence of competing forms as borrowings from the o-dialect into the u-dialect, or conversely:

*lutuO' Snd., OJv., lutu~, NJv. lutó~, Mal. luto~, loto] 'kind of monkey'.

*tulu~, (PHN, Tag. tu:lo~ 'to help'), Snd. tulu~ (L), OJv. an-ulu~, NJv. tulÖ~, Mal. tul09' tolo~, Mad. t~lo~ 'to help'.

In the following comparison there is a related Javanese word which can be considered the source for the Malay form with 0:

*qupah (PHN, Tag. qu:pa 'reward': 00 upah, 10 [1953a] upaq), Snd., OJv. qupah, NJv. upah (0), ópah, Mal. upah, opah, Mad. ~pa 'reward, payment'.

In a number of comparisons the Malay cognate occurs without a by-form with u:

, *rumpun, Snd. rumpun, NJv. rumpon (0) 'off', Mal. rompon 'mutilated, cut off'.

*su~gin, Snd. sUngin 'to burn figures into wood or bamboe with a red-hot iron, to tattoo', NJv. sungén, Mal. sOngen' Mad. sOnghin 'to paint'.

*[t~]uhur (PAN, Sm. tu 'to stand': [t~]uquq), Snd. tuhur 'dry', Mal. tohor 'shallow'.

*tumBun' Snd. tumbun 'seed-bud in a coconut, female privy parts', Mal. tombon 'spherical lump of any sort, seed-bud in a coconut', Mad. tombun 'rectum'.

\ , *tumpuk, Snd. tumpuk, NJv. tumpoq, Mal. tompoq, Mad. tompoq 'small heap'.

In an instance like the following the Malay form with 0 can be attributed to a borrowing from Javanese, since a Javanese word of the same shape is available:

*Bunsuq (PHN, Tag. bunsóq 'youngest': bu[n]su), Snd. bunsuq, OJv. wUnsu, NJv. bUnsu (0) (B), bdnSU (0) (B), Mal. bOnsu, Mad. bUns~(h) 'youngest-born'.

2.6.3.4. In some comparisons Sundanese and Malay exhibit a penultimate 0 matching New Javanese J in words which have a different ultimate vowel. We also find comparisons in which Sundanese, Malay, and rarely Old Javanese exhibit penultimate and ultimate 0 matching New Javanese penultimate and ultimate O. Finally, we find comparisons in which Sundanese and Malay exhibit an ultimate 0 af ter a different penultimate vowel matching New Javanese ultimate à. Madurese exhibits u or à in all instances (distributed according to the vowel alternation rule given in section 1.8.3.1.).

Again, we are faced with the question of whether to reconstruct an additional proto-phoneme 0 or whether to resort to the principle of borrowing as an explanation. The matchings can be explained in the same way as was done in the case of Sundanese, Malay, and Old Javanese e matching New Javanese é or è. All words in question have non-basic meanings and can therefore be explained as borrowings from Javanese dialects.

The following list gives some comparisons which exhibit the matchings of Sundanese, Malay, Old Javanese 0, New Javanese Ó or 0:

*bucur (PWI, NgO. busor 'leak,:37 bucu[rJ), Snd. bocor, NJv. bocor, Mal. bocor, Mad. bhucor 'leaking'.

NJv. bokar, Mal. bokor, Mad. S. P. bhukor 'rimmed plate, shallow bowl;.

Snd. bolon, NJv. b~lan, Mad. bhulu~ 'opening, hole'.

*buru~ (PWI, NgO. bä-buro~8 'wholesale': bu[rJu~), Snd. borot) 'to make a good buy', also: 'to buy everything', NJv. , , , boron, Mal. boron, Mad. malle bhurun-an 'wholesale, in a lump' •

Snd. cobaq (L), NJv. cóbO (L), Mal. coba, Mad.cobhá(h) 'to test, try'.

Snd. NJv. ' , , , cocog, cocog, Mad. cocok 'to agree'.

Snd. gand0!l' NJv. , gan~o~, I Mad. ghan~hun 'to carry on the back.'

NJv. , gan~oo, Mal. ganton' Mad. ghant~n '1 arge earthenware vesse l'

Snd. godeg, NJv. gó~ég 'side-whiskers', Mal. godek 'short and pendulous, whisker'.

Snd. gonjak, NJv. gónjaq (0), Mal. Pen. gonjaq 'to make fun of, ridicule' .

Snd. goren. NJv. gör~D' Mal. goren. Mad. ghurin 'to cook in a pan'. Snd. gorok 'to insert a kris or another weapon into the throat through the mouth and then turn it. to cut the throat'. NJv. gàr~q. Mal. goroq 'to cut the throat'. Mad. ghuruk 'to cut the throat'.

*gusuk (PWI. Mer. kusukä 'to rub'). Snd. gosok. NJv. gàsoq. Mal. gosoq 'to clean by friction. scrub'.

Snd. goyan 'motion. sound. noise'. NJv. g~yaO 'move back and forth. noise'. Mal. goyan 'oscillation'. Mad. ghuylD 'to rollover. roll about'.

*kan~un (PAN. Fi. kata 'basket'). Snd. kantOD' NJv. kan~oD' Mal. Jak. kanton. Mad. kan~oD 'pouch. pocket'.

Snd. kolon. NJv. k~làD (0). Mal. koloD 'space under anything'.

*kun~ul (PAN, Sa. quu 'scrotum'). Snd. kontol 'the testicles, scrotum', NJv. kàn~èl 'scrotum', Mal. kontol 'short, thick, pendulous object', Mad. B. P. kon~ol. S. k~ncè kon~~l 'padlock'.

Snd. koson 'empty. without content', NJv. kosoD 'empty' (0). 'vacant'. Mal.koson 'empty. hollow. 39

\ , Snd. rampog. NJv. rampog. Mal. Java rampoq. Mad. rampok 'to go robbing and plundering'.

Snd. rombak40 'to take away. clear away. NJv. r6mbaq 'to tear down. up'. Mal. rombaq 'to unravel. untying'.

Snd. rompaD' NJv. r~mpan 'with pieces missing (e.g •• a missing tooth. also the teeth of a comb)'. Mal. rompaD 'tattered and torn (of thatch and matwork). holed in many places.

, \ ( ) .... Snd. sero9' NJv. sero~ -an , Mal. seron, Mad. sero9 'askew, aslant'.

Snd. tonkol, to-tonkol, NJv. tonkal, Mal. to~kol, Mal. Java ta-tonkol, Mad. to~kol 'tunny-fish'.

Snd. qomo~ (L), NJv. omo~ (D), Mal. omon 'to talk, gossip'.

In some instances non-final New Javanese Ó and 0 are the reflexes of an early Javanese or PMJ vowel sequence ua or au. 41 This case is analogous to the explanation of é and ~ as continuing the PMJ vowel sequences -iaand -ai-. In the following example Old Javanese data indicate that New Javanese penultimate and ultimate ~ are the result of the contraction of an early Javanese sequence -ua-. NJv. dadal 'a sweetmeat of rice-flour' is the continuation of an Old Javanese form dwadwal 'flour-pastry'. Sundanese and Malay dodol and Madurese dhudhul 'sweetmeat of rice-flour' seem thus clearly to be loans from Javanese.

2.6.3.5. PMJ antepenultimate u. PMJ antepenultimate u appears as Sundanese and Old Javanese u. It became New Javanese and Malay a, and Madurese á or a distributed according to the vowel alternation rule given in section 1.8.3.1.

*gulimpan, Snd., OJv. gulimpan, NJv. g(a)limpan' Mal. galimpan, galempan, Mad. ghálimpa~ 'to lie astretch'.

*kuranjiq, Snd. kuranjiq, Mal. karanji, Mad. karanjhi(h) 'kind of tree'.

PMJ antepenultimate u appears as u in an unidentified dialect of Javanese.

*suligiq (PHN, Tag. sulf:giq 'spear' : suligi), Snd. suligiq, OJv.

suligi 'spear, NJv. s(a)ligi, suligi 'bar', Mal. saligi 'wooden dart or javelin', Mad. B. salèghi(h) 'club, cudgel'. *surambiq (PHN, Tag. sulamb{q 'eaves [of a house]': sur-ambi), Snd. surambiq, NJv. s(a)rambi, surambi, Mal. sarambi, Mad. sarambhi(h) 'a long low closed verandah made by prolonging the eaves of the main building'.

2.6.4. PMJ a

2.6.4.1. The Malay reflexes. PMJ a appears as Malay a in the penult and a in the final syllable:

*qannam (PAN, Sa. ono 'six': ?anam), OJv. nam 'six', ka-nam 'the sixth', NJv. nam, Mal. anam 'six', Mad. (an)nam 'six. *sassah, Snd. s+s4h 'to wash', Mal. sasah 'switching, beat the clothes in the wash'.

PMJ penultimate a became Malay a if followed by PMJ h or another vowel:

*paar, NJv. pbr, Mal. pahar 'pedestal tray'.

*Danak (PHN, Tag. dá:hak 'to spit out': OD daak, 1D [1953a] [d9]ahak), OJv. rahak, NJv. Tagal ruwaq, Mal. dahaq 'to cough up phlegm'.

2.6.4.2. The Madurese reflexes. PMJ a appears as Madurese a in the penult and the final syllable:

*la9Uan, Snd. l+O+n (L), OJv., NJv. la9an, Mal. lanan, Mad.

laUUan 'lower arm, hand'.

PMJ a became Madurese a before *-p, *-t, and *-1 wh en they in turn became Madurese -q. The same development is also seen before *-h, which was then lost in Madurese (see section 2.15.1.5.).

*hatap (PAN, To. qato 'to cover a roof': qatap), Snd. hat4p, OJv. hatap, NJv. atap, Mal. (h)atap, Mad. ataq 'roof'. *karrat (PAN, Sa. qo10 'cut off': kaRat), Snd. k4r4t 'to cut (off)', OJv. karat, krat 'was cut off', NJv. karat, Mal. karat, Mad. karraq 'to cut off'.

*la1a1 (PAN, Fi. 1a10 'f1y': 1a1aj), Snd. 1a14r, OJv., NJv. lalar, Mal. lalat, Mad. 1a1aq 'fly'.

*tanah42 (PAN, Sa. ano 'land, earth': tanaq), Snd. tan4h, Mal. tanah, Mad. ta na 'earth, ground'.

PMJ a before PMJ -qand PMJ hiatus became Mad. b~

*kaqa~, Snd. k4q4~ 'to be 10ne1y', OJv. kü~ 'pain of love', NJv. k6U 'desire, grief, pain of love', Mad. koqà~ 'to be 10ne1y, orphan'.

*haay, Snd. h4ay, Mad. bway 'to yawn'.

2.6.4.3. The Javanese reflexes. Old Javanese 1ike Sundanese, has two vowe1s cal led pepets; of these ö is the conventiona1 transcription of the 01d Javanese vowe1 sign which represents the long vowe1 corresponding to the short vowe1 transcribed ~ by Juynbo11 (in this study a).

R. A. Kern 1940 ob serves that OJv. ö became NJv. u and that in Juynbo11's 01d Javanese-Dutch dictionary some words have two spe11ings, one with short pepet, the other with long pepet, e.g. lahaO' 1ahöO. He conc1udes that if the pronunciation of 01d Javanese words in fact agreed with their spelling, then OJv. ö must have been an unstab1e sound, even more than is indicated by the dictionary. Kern a1so states that 01d Javanese forms with ö must have existed for every 1exical item which contains u in New Javanese, because - as he argues - on1y this sound became NJv. u. In some instanees the change ö-+u is a1ready represented in 01d Javanese according to Kern: e.g., ranö, ruOu 'to hear'.

This study proposes the fo110wing hypotheses in respect to the Javanese ref1exes of PMJ a:

(a) PMJ 3 became 01d Javanese a or ö and New Javanese a in the penu1timate and in the fina1 sy11ab1e. The 01d Javanese symbo1 ö does not occur in initia1 position and the 01d Javanese symbo1 a does not occur in fina1 position. In other positions the two symbo1s a and ö were apparent1y used interchangeab1y43:

*b3nnar (PHN, Tag. bani1 'pious': banar), Snd. btn4r 'fu11 and good (of rice-grains)', OJv. banar, NJv. ban ar (L), Mal. banar 'right, good', Mad. Kang. bhannar in di-ghinl bhannar 'to be led on the right track'.

(b) PMJ a before PMJ h became 01d Javanese a, Ö, or u and New Javanese u in the penu1timate and Ö in the fina1 sy11ab1e. Oia1ecta11y, it a1so became New Javanese penu1timate and u1timate à (see the section on PMJ u).

Three ways of exp1aining the two 01d Javanese ref1exes suggest themse1ves: (1) the spelling with a or ö might be an archaism and u represents the pronunciation in 01d Javanese; (2) a and örepresent the pronunciation in earl ier 01d Javanese and urepresents the pronunciation in later 01d Javanese; and (3) the two spe11ings might represent two 01d Javanese dia1ects, where one dialect has a or ö as reflex of *a before h, and the ot her has u, where New Javanese continues the u-dialect:

*dahaq, Snd. matak ti-d4haq 'to push s.o. against the chest', OJv. ma~-daha 'to push with the hand', ka-döha 'pushed aside', NJv. duwo 'to push away'.

*qampah, Snd. qamp4h, OJv. mämpah (=ma + ampah), NJv. ampóh, ampoh (0) 'to check, restrain'.

(c) Two PMJ a's in a word which contains PMJ intervoca1ic or fina1 h became 01d Javanese a .. a, a .. ö, or u •. u and New Javanese u •. ó (for the dia1ecta1 reflexes, see the section on PMJ u). A1so in this case three ways of exp1aining the 01d Javanese ref1exes suggest themse1ves: (1) the spe11ings a .• a and a .. ö might be an archaism and the spelling u .. u might represent the pronunciation in 01d Javanese; (2) a •• a and a •. ö represent the pronunciation in earl ier 01d Javanese and u •• u represents the spelling in later 01d Javanese; and (3) the two spe11ings might represent two 01d Javanese dia1ects, where New Javanese continues the u-dialect.

, 44 , *laha!), Snd. Hh4!), OJv. 1aha!), 1ahö~, NJv. 1uwo!) • 1uho!} (B). 1ówón 'to prefer'.

*gallah. Snd. g4Hh. OJv. galah. ga1öh. NJv. gu16h (0) '.dirty'.

*lassah. Snd. 14s4h 'b1eached. dirty'. OJv. 1asah 'rotten'.

1usuh 'worn out'. NJv. 1usóh (0) (0). losoh 'rotten. worn out'. Mad. 1assa 'rotten. worn out'.

*pappah45 • Snd. p4p4h. OJv. pupuh. NJv. pupÓh (0). Mal. papah 'to strike. beat'.

In one comparison 01d Javanese exhibits a penu1timate and ultimate a from a PMJ base with a penu1timate *i and an ultimate *a:

*qiah46 (PHN. Tag. q{:hiq 'urine': 00 ia[h]. 10 [1965a] [ ]iS1aq). Snd. k-iqih. OJv. ayah. ayöh. uyuh 'urine'. p-ayöh. p-öyah. p-uyuh 'to have to urinate'. NJv. uyóh CL) 'urine'. puyuan (= p + uyuh + an) 'to pass urine'. n-uyóh (L) 'to urinate' Mal. k-am-eh 'to pass uri"~'.

The following developments are posHed for Javanese. The anticipated contraction of the two vowels (see the section on PMJ hiatus) did not occur because of the prior development of initial *i to y. This could have appeared first in a form like OJv. payah with prefix *pa-. The initial a of OJv. ayah may be analogical in origin.

In Sundanese the following developments probably took place. The ultimate vowel assimilated to the penultimate vowel. A glottal stop was inserted between the like vowels since no sequence of like vowels occurs in Sundanese.

In Malay the following developments probably took place. The ultimate vowel assimilated to the penultimate vowel. Subsequently the two like vowels contracted to *i (see the section on PMJ hiatus).

(d) PMJ a before PMJ final *R became OJv. -ö or-u and New Javanese u:

*qiBaR (PAN. To. ifo 'saliva': ?ibaR). OJv. iwö 'to take care. do one's utmost'. Mad. èbar 'saliva'.

(e) If the penultimate vowel was PMJ a and the ultimate vowel was *a before *R. the Old Javanese reflex is a .• ö or u .. u and the New Javanese reflex is u •• u:

*oau~a~ (PAN. Fu. Sm. 10Do 'to hear': ~aDaR23)' Snd. deneq (L). OJv. maD-ranö 'to hear'. n-~aOö. runu 'hearing. listening'. NJv. k-ru~u (L). n-~u~u (0). Mal. danar. Mad. Kang. ~aODar 'to hear'.

(f) PMJ a before PMJ -BB- became Old Javanese a and New Javanese u:

*laBBih (PHN. Tag. lab{ 'more': 00 labih. IO [1953a] labiq), Snd. l4wih (L), OJv. lawih. NJv. luweh (L). Mal. labeh,

Mad. 1 abbi 'more'. *taBBas, OJv. tawas, NJv. tuwas (0), Mad. tabbás 'advantage, prosperity' .

*taBBak47 (PAN, Fi. teve 'circumcise': tabak), OJv. tawak, twak, NJv. tuwak 'to stab, pierce through', Mal. tabaq 'bar used for stone breaking'.

(g) Early Javanese hiatus of *a's which is the result of the loss of *R or *q contracts to Old Javanese a, Ö, or u and New Javanese 6 [for the possible ways of explaining the spellings, see under (b)]:

*gaqam, Snd. g~q~m-an, OJv. ka-göm-an, NJv. ka-góm 'scared, shocked' .

*kaqam, Snd. k~q~m 'to be under water, sit in water', OJv. ma-kam 'to sit in water for a long time', (yan) pa-köm 'which were in the water for a long time', k-in-um 'bathed', NJv. k~m 'to sit in water (in order to soften)'.

*paR2R2ah48 (PHN, Tag. pig'q 'to squeeze out': paRah), Snd. p~r~h (L) 'eye-drops, eye-water', OJv. pöh-i 'squeezed out', pa-pah 'to drip in', NJv. póh (0) 'squeezed out', pupÓh (0) 'eye-water', Mal. parah 'to express', Mad. parra 'eye-drops'.

(h) PMJ penultimate a before *RR and a following different vowel became OJv. a or waf ter the loss of the consonants in early Javanese. Old Javanese wand the following vowel contracted at a later stage of Javanese:

*taR2R2ab (PHN, Tag. tigáb 'to yawn': taRab), Snd. t~rab, OJv. twab, NJv. a-tob, Mad. darrap49 'to beleh'.

(i) The PMJ sequence -aabecame New Javanese o. Presumably Old

Javanese had a form *pwar which contracted at a later stage of Javanese. The New Javanese form is por 'pedestal tray': *poar. p. 1~.

2.6.4.4. The Sandanese reflexes. Sundanese has two different vowels. 0 and ~,which can be found to match reflexes of PMJ o. In the literature they are usually referred to respectively as the 'long and short pepets'. So far the existence of these two Sundanese vowels has constituted a major problem in Austronesian comparative studies. Most researchers ignored this problem. even those who included Sundanese as a criterion language in their comparative investigations. R. A. Kern 1940 is the only schol ar who attempted an analysis of the origin of the two Sundanese pepets. He reconstructed an Indonesian pepet and arrived at the following conclusions with respect to its reflexes in Sundanese:

1. The pepet is lengthened in open syllables and in the final syllable wh en followed by h (primary lengthening).

2. Secondary lengthening takes place when short and long pepet occur in the same word. All pepets are lengthened.

3. No further rules can be given for the transition from a short to a long pepet; it is arbitrary but in every instance fixed. (p.1l8)

R.A. Kern thus suggested sporadic change by using the phrase 'its transition is arbitrary'.

As has been pointed out in section 2.3.1. this study proposes that ~ is the regular reflex of PMJ 0 and that the distinction between ~ and 0 is not inherited. Javanese loan words. and in a few instances Malay loan words. are thus the main source of Sundanese words containing o.

(a) The following is an example of PMJ penultimate and ultimate 0:

*kappal (PHN, Tag. kf;pil 'pressed down tight as cooked rice within a fist'; kapal), Snd. k~p~l 'grasp, handful', OJv. kapal-kapal 'clotted mass of rice', NJv. sa-kapal 'clenched fist, handful', Mal. kapal 'lump, clod', Mad. kappal 'handful'.

(b) PMJ a became Sundanese when the final syllable contained *u: *[Bb]a[tt, ~~]u~, Snd. bitu~, Mal. bato~ 'large bamboo-tree'. *cakkur, Snd. cikur, Mal. cakor, Mad. cakkor 'creeping weed'. *ka[rr, R2R2]uh, Snd. kiruh. Mal. karoh 'turbid, muddy'. *saggun, Snd. sigun, NJv. sagóU 'badger'.

*talluq (PAN, Sa. olu 'three'; talu?), Snd. tiluq, OJv. talu, NJv. talu (L), Mad. tallo(h), talloq 'three'.

*tammuq (PWI, TBt. tomu 'to meet'; tamu), Snd. timuq (L), OJv., NJv., Mal. tamu. Mad. tammo(h) 'to find, meet'.

*tannun (PWI, TBt. tonun 'to weave'; t nun), Snd. tinun, OJv. tanun, NJv. tan6n, Mal. tanon, Mad. tannon 'to weave'.

*pannuh (PAN, Sa. honu 'full'; panuq), Snd. pinuh. OJv. panuh, NJv. panóh (B), Mal. panoh 'full'.

*tappu9 (PWI, NgD. tepoU 'baker's ware'; tapu~), Snd. tipuD' Mal. tapon, Mad. tappen 'flour'.

In the western parts of the Sundanese speaking area and in the northeastern enclaves PMJ a before PMJ u has the reflexes ~, a, and 0 according to the dialect. The Baduy (Cibeo, Cipiit), Saketi, Mandalawangi, Gurusul, Lebak, and Cijaralang dialects are~- dialects. The Cipacung, Cikeusal, Cirende, Cipaku, Cilayang, and ~ibeusi dialects are a-dialects and the Dukuh, Baros, Jiput, Mancak, Parean, and Lelea dialects are o-dialects.

Since Sundanese merges *i [which appears as i in most instances (see the section on PMJ i)] and *a before *u, all standard Sundanese words of the shape .• i(C)(C)u •. which have no known cognates in other Austronesian languages are ambiguous in respect to the origin of the penultimate vowel. Only the 4-, a-, andodialects do not merge these two vowels. A word like Snd. lintuh 'fat' has no cognates in any other Austronesian language as far as I know. Only dialectal information resolves the ambiguity as to the origin of Sundanese i: the 4- dialects have the form l4ntuh, the adialects the form lantuh, and the 0- dialects the form 10ntuh. 50 The dialectal forms indicate that the earl ier Sundanese form must have been *lantuh.

The same holds for Snd. tihul 'tree stump thrown into a fire to keep the fire smouldering on', Parean toul, Baduy (only Cipiit) t~hul. Earlier Sundanese must have had a form *tahul.

There are two instances in which the Sundanese dialectal reflexes indicate that PMJ had a penultimate a. In both instances Malay exhibits an inexplicable i. A possible solution is that PMJ had a doublet. one with penultimate a, the other with penultimate i:

*hazzuk5l (PHN, Tag. irok 'sugar-palm': izuk), OJv. haduk, NJv. dóq, Mal. ijok 'black sugar-palm fiber'; and *hanzuk, Snd. qinjuk 'black sugar-palm fiber'. The Sundanese dialectal forms are q~njuk, qanjuk, and qonjuk 'black sugar-palm fiber'.

*na[rr, R2R2]uq, Snd. niruq, Mal. niru, niru 'winnowing basket'. The Sundanese dialectal reflexes are n~ruq, naruq, and noruq 'winnowing basket'.

(c) There are instances of *a > Snd. e wh en before eq < early Snd. -ay < *-ay, *-aRl : e.g., Snd. deDeq (L) 'to hear' (cf. *oaDDaRl, p. 76); Snd. bereq (L) 'to give' (cf. *BaR2R2ay, p. 126).

(d) In one instance an early Sundane~e *-ayfrom PMJ -aRlassimilated to an early Sundanese *4 in the following syllable before the contraction of early Sundanese *ay to e could take place. Sundanese exhibits 4y:

*paR1Rlam, Snd. p4y4m 'katan which has been sweetened by ragi', Mal. param 'to store fruit to let it ripen'.

(e) There are a few instances in which inherited Sundanese words contain a. Examples are qaDg4s 'already' and hant4q 'not, no'. For some speakers the initial two elements are dropped: ng4s and nt4q. These are taken to be fast speech forms of the careful and/or slow speech forms qa994s and hant4q. Both occur frequently at relatively great di stances from the end of phrases.

(f) There are instances of Sundanese loan words that contain Snd. 4. For the meaning 'to hear' Sundanese has two words, r4n4q and rUDuq, that are part of the high status-style and correspond to the low status-style deneq. The high words are probably borrowings from Javanese. Considering the three ways of explaining the Old Javanese spellings a .. ö (raUö) and u .. u (ruDuq), there are three possible ways in which Sundanese might have borrowed these two words: (1) if we consider the possibility that the spelling a .• ö is an archaism and that the spelling u .. u represents the pronunciation in Old Javanese, then Snd. r4D4q probably was borrowed in the pre-Old Javanese period, while Snd. rUDuq was borrowed during the Old Javanese period; (2) lf we consider the possibility that a .. ö represents an earlier Old

Javanese pronunciation and u •. u represents a later 01d Javanese pronunciation, then Snd. rt~tq presumab1y was borrowed before a .. ö became u •• u, whi1e Snd. ru~uq was borrowed af ter this change had occurred; and (3) if we consider the possibi1ity that the two spe11ings represent two 01d Javanese dia1ects, then Snd. rtntq can be exp1ained as a borrowing from the a .. ö dialect and Snd. runuq as a borrowing from the u •. u dialect.

Snd. rthak 'mucus, ph1egm' was perhaps borrowed from a dialect with a form 1ike OJv. rahak 'mucus, ph1egm', represented by Tega1 Javanese ruwaq 'mucus, ph1egm'. The expected Sundanese form is *dthak and not rthak. Ma1ay has dahak 'coughing up ph1egm'. The Javanese and Ma1ay evidence implies *Dahak. PMJ 0- however became regu1ar1y Snd. d- (see the section on PMJ 0, p. 147)

The PMJ semivowe1s

2.7. Dempwo1ff reconstructed two PAN semivowe1s, *w and *y. Both are accepted as part of the PMJ reconstructed phoneme inventory. Fo11owing Dyen (1971:26), we make an additiona1 distinction for *w in initia1 position. Thus the two tentative proto-phonemes *w1- and *w2- are reconstructed.

PAN wand PAN b between *a's merge in PMJ w: e.g., *qawak (PAN, Fi. 1-èwe 'f1esh, contents': ?awak), Snd. qawak (L), OJv. awak, NJv. awaq (L), Mal. awaq, Mad. abáq (L) 'body, trunk of body'; *sawah (PAN, Fi. ~ava 'harvesting season': sabaq), Snd., OJv. sawah, NJv. sawah (L), Mal. sawah, Mad. sab~ 'wet rice-fie1d'.

PMJ w occurs in three positions: initia1, intervoca1ic, and final.

and PMJ y in four positions: initia1, intervoca1ic, postconsonantal, and final.

2.7.1. Pf-tJ w.

2.7.1.1. PMJ w1 and PMJ w2. PMJ w1 disappeared in Sundanese and Madurese. It became h in Malay and 01d and New Javanese w. PMJ w2 disappeared in Sundanese. 01d Javanese has ww as its reflex before a vowe1 other than Before *u its reflex is w. The second member of the 01d Javanese geminate contracted with the fo110wing vowe1. 01d Javanese w before u disappèared in New Javanese (for a simi1ar deve10pment, see PMJ B before *u, p. 132).

2.7.1.1.1. PMJ w1. Simple instances are 1acking. The following comparison is worth considering despite its comp1ications:

*w1aR3iq (PAN, Sa. w~li 'for a whi1e': waR12 i?), Snd. poeq (?<*pa-w1aR3iq) 'time-space between sunrise and sun-set', poek4n (=poeq+k4n) 'to dry in the sun'. OJv. we 'sun, day', ta~a~e (=ta~ah ~ we) 'noon' , NJv. p~ 'to dry in the sun (by hanging outside)', taUa1Jé (0) 'noon' , Mal. (h)ari, Mad. arè(h) 'day'.

Snd. poeq and NJv. pé are considered morphologica11y complex forms, exhibiting the prefix *pa-. PMJ pa-w1aR3iq became Snd. poeq by the fol10wing development: (1) PMJ R3 disappeared (see the section on PMJ R), and (2) perhaps the sequence *aw became 0 and the sequence aiq became eq in this single instance. NJv. pé is probab1y due to the fo110wing deve10pment: PMJ R3 disappeared in early Javanese and the resulting vowel sequence ai contracted to New Javanese I. Earlier Javanese waf ter a consonant and before a front vowe1 disappeared (as in *OuR2iq, OJv. rwi, NJv. ri 'thorn', see p. 148).

2.7.1.1.2. PMJ w2. *w2ayaq, Snd. qayaq, OJv. wwaya 'there is'. *w2aDaq (PAN, Fi. wara-i 'not to be': waqa?), OJv. wwara 'there is', NJv. óra 'there is not', Mal. ada, Mad. bá~á(h) 'there is'. *w2a1uq (PAN, Fi. wa1u 'eight': w2a1u?), OJv. wwa1u, NJv. wè1u 52 , Mad. bá11u(h) 'eight'. *w2uyah (PWI, NgD. guyah 'salt': uyah), Snd. quyah, OJv. wuyah, NJv. uyah (L), Mad. buy~ 'salt'.

2.7.1.2. PMJ w in intervoealie position. PMJ w in intervoea1ie position appears as w in all 1anguages exeept Madurese where it beeame b:

*kawah (PHN, Tag. ká:wa 'saueepan': 00 kawah, 10 [1953a] kawaq), Snd. kawah 'erater, big saueepan', OJv. kawah 'a big saueepan', NJv. kawah 'erater', 'eau1dron' (B), 'eooking pan' (0), Mal. kawah 'vat, eau1dron, erater'. *tawaR2I (PHN, Tag. tambág 'wedding-gift': ta[m]baR), Snd. tawar 'to bargain, make a bid', OJv. a-tawa 'tendering', NJv. t~w~ (L) 'to ask for, ask for the priee', Mal. tawar, Mad. tabár 'to bargain, bid'. *tawaR2II ,(PHN, Tag tá:bag 'tasteless': ta[m]baR), Snd. tawar 'braekish', NJv. towo 'fresh water', Mal. tawar 'tasteless', Mad. tabár 'sa1t1ess'.

2.7.1.3. PMJ w in fina1 position.

2.7.1.3.1. PMJ waf ter PMJ a appears as w in Malay, but eontraeted with this vowe1 in the other languages. The eontraeted resu1t is Sundanese -oq, 01d Javanese -0, New Javanese -ó, and Madurese u or ~, and West Madurese -uh or -oh (distributed according to the vowe1 alternation ru1e given in section 1.8.3.1.):

*[Bb]akaw (PAN, To. pako 'kind of tree': bakhaw), Snd. bakoq, Mal. bakaw 'mangrow'.

*pisaw (PHN, Tag. p{:saw 'knife'), Snd. pesoq, NJv. pisó (0), Mal. pisaw 'knife'.

*tinjaw (PWI, Mer. tsindzu 'to ob serve from the distance': tinjaw), Snd. tenjoq 'to see', OJv. tinjo, NJv. tinj~ (0) (0) 'to visit', Mal. tinjaw 'to crane the neck, be on the wateh'.

*tu~aw, (PHN, Tag. tu~áw 'insect'), Snd. tonoq, Mal. tunaw, Mad. B. t~~~(h) 'sand-f1ea'.

In one instanee the Javanese cognates inexp1icab1y exhibit fina1 u:

*panaw (PAN, Sa. hano 'white spots on skin'), OJv., NJv. panu, Mal. panaw, Mad. panb-pan~ 'white spots on the skin'.

In a few instanees the Madurese cognate inexp1icab1y exhibits fina1 -~Y or -uy:

*ba~aw (PWI, NgO. banao 'stork'), Snd. ba~oq, OJv. bano, NJv. J I bano, Mal. baUaw. Mad. bhaDuy 'stork'.

*ga1aw, Snd. ga1oq, NJv. ga1ó (0), Mad. gh~luy 'mixed'.

*kabbaw53 (PAN, Fi. karavau 'ox, cow': kabaw), OJv. kabo, NJv. kab6, Mal. ka-r-baw, Mad. ka-r-bhuy 'buffa1o'.

*jaraw, Snd. jaroq, NJv. jar~, Mad. jh~ruy 'a bamboe with holes a10ng its length to al10w crosspieces to be put in for fence-making'.

In t:he verbs this additional y perhaps is the suffix *-iq. Let us consfder the following comparison:

*tamuq (PQI, TBt. tamu-e 'guest': tamu), Snd. ta-tamuq, OJv. tamu, NJv. tamu (H) 'guest', Mal. tamu 'to regale, feed up', Mad. tam~y 'guest'.

We also find OJv. tamuy 'guest' which is perhaps the morphologfcally complex form of the simple form *tamuq: *tamuq+iq became OJv. tamuy. The suffix *-iq did not contract with *u af ter the loss of the final glottal stop of *tamuq but became its homorganic semivowel (for the contraction of vowels af ter the loss of *-qin Javanese, see the section on the laryngeals). The New Javanese continuation of OJv. tamuy is an expected tami (H) (0) 'guest' (for the development of OJv. -uy to NJv. -i, see bélow). TtÀ's latter form already occurs in an Old Javanese text: a-na-nami 'to treat, regale'. It was probably wrftten bya scribe in whose dialect -uy had already contracted to -i. Madurese has tam~y, presumably like OJv. tamuy and NJv. tami from *tamuq+iq.54

2.7.1.3.2. PMJ final waf ter PMJ i contracted with this vowel in Sundanese and Malay. The contracted result is Sundanese -iq and Malay -i. In Old and New Javanese an interchange of syllabicity took place between the two vocoids. In Madurese the order and the syllabicity of the vocoids was reversed:

*BaR2iw (PWI, TBt. bari 'to taste rotten': baRiw), Snd. bariq55 'old, stiff, cold (of food which has been stored for a long time), sour, fermented, rotten', NJv. wayu 'to have been standing for a long time, too old, rotten', Mad. b~ruy 'rotten, musty, stale (of food)'.

*laR2iw (PHN, Tag. 1agy6q 'soul ': 1aRiw), Snd. 1ariq 'to fo110w a trace', OJv. pa-1ayw-ana 'to be run af ter', NJv. ka-1ayu (0), k-1ayu 'to want to go a10ng (with the mother, master, etc.)', Mal. 1ari 'to run'.

In one instance a PMJ doublet exhibiting an interchange of sy1- 1abicity and non-sy11abicity might have to be reconstructed: Snd. kaiq 'wood' probab1y is from *kaiw, whi1e West and East Sundanese kayuq56, OJv., NJv., Mal. kayu, and Mad. kaju(h) 'wood' probab1yare from *kayuq. These two forms can be associated with Dyen's (1971:43) reconstructions *kaS2iw and *kaS2uy respective1y.

2.7.1.3.3. PMJ fina1 waf ter PMJ a contracted with this vowe1 to Sundanese -oq, Ma1ay, 01d and New Javanese -u, and Madurese -u or 0 and West Madurese -uh or -bh (distributed according to the vowe1 a1ternation ru1e given in section 1.8.3.1.).

The first to reconstruct this fina1 sequence was Dyen (1953:363, fn. 18). His on1y examp1e was *buR3aw. The reconstruction was an attempt to combine two of Dempwolff's reconstructions *buRaw and *buru. The ref1exes assigned to Dyen's *-aw were Tag. -aw, TBt. -0 (?), NJv. -u, Mal. -u, NgD. -aw, Sa. -0. Hendon (1964:372 ff.) reconstructs *-aw instead to account for Tag. -oq, NgD. -aw, Fi. -0, and OJv. -ö. He suggests to reassign *-aw from the correspondence to which Dyen had assigned it. Howeve~Dyen's correspondence is c10se1y patterned with that of *-ay. A further 1anguage in which Dyen's *-aw and *-ay are parallel is Sundanese, the reflex of *-ay being Snd. -eq (see section 2.7.2.4.2.). Furthermore there are other comparisons to support Dyen's assignment. Hendon's material shows some non-conforming evidence. In two of Hendon's basic correspondences Old Javanese has an -u instead of an anticipated -ö. Hendon suggests that these Old Javanese words can be considered loans from Malay or that a historical correct form in -ö appears as -u in the manuscripts due to scribal corruptions. On the other hand it is also possible, as Hendon admits, "that archaiz1ng scribes may have substituted unhistorical -ö for an etymologically correct -u"(op. cit.: 379, fn. 34).

We reconstruct PMJ -aw in the following comparisons:

*Buraw (PAN, Sa. huru 'to run': 00 buRaw and buru, IO [1953b] buR3aw), Snd. boroq (L) 'to hunt, chase, pursue', OJv. a-buru 'to hunt', NJv. buru (L) 'to hunt' (B) (0), 'to try to catch', Mal. buru 'to hunt, drive game', Mad. buru(h) 'to run away'. Mad. bhuru 'to hunt' is treated as a borrowing from Malay or Javanese. Snd. bu-buru (L) 'to hunt' is also treated as a loan from one of these two languages. The reduplication occurred at a later stage of Sundanese. There is also a PMJ reconstruction *buruq, see p. 138. *[dqO]agaw or *ga[dq]aw, Snd. gadoq (L), Mal. dagu 'chin'. *garaw, Snd. garoq 'to scratch (with nails), OJv., NJv. garu 'to harrow', Mal. garu 'scraping with a blunt point, scratching', pan-garu 'harro~/', Mad. gh~ru(h) 'to scratch'. See also *[Bb]u[dq]aw, p. 65 ; *tujjaw, p. 65.

*banaw (PAN, Sm. panu-panu 'smeared': banu[?h]), Snd. banoq 'water', si-banoq 'to wash one's hands', OJv. ba~u, NJv. banu (L) 'water', Mal. banu 'fermented coconut-water used in dyeing', Mad. bh~no(h) (H) 'to urinate'.

2.7.1.3.4. Ambiguous reconstructions. An ambiguous reconstruction with [uq, aw] is employed when a Sundanese cognate is lacking and when Old and/or New Javanese and/or Malay exhibit -u, and/or Madurese exhibits -u(h) or ~(h):

*semm[uq, aw]57, OJv. samu 'to look like, appearing', NJv. samu 'meaning, to let know', 'as if, outward appearance' (0), Mal. samu 'to deceive by means of lying tales, humbug', Mad. sammo(h) 'what can be drawn up from outer appearance'.

2.7.2. ~. PMJ Y in initial position appears as y in Sundanese and Old and New Javanese, and became j in Madurese. The Malay reflex is indeterminable because of the absence of a cognate set with a Malay member. In intervocalic position it appears as y in all languages but Madurese where it became j. In postconsonantal position it appears as y in Sundanese and became j in Madurese. The Javanese and Malay reflexes are indeterminable because of the absence of cognates in these languages. For PMJ y in final position, see the section with that heading.

2.7.2.1. PMJ yin initial position. There are two pr-1J reconstructions with initial y:

*yuQYun, Snd. ka-yu~un 'lovely, beautiful', Mad. ka-junjun 'lovely, beautiful, alluring'.

*yuyuq (PAN, To., Fu. ququ 'kind of big land-crab': yuyu), OJv.

a-yuyu 'lobster, crayfish', NJv. yuyu 'crayfish'.

2.7.2.2. PMJ Y in intervocalic position

*payuQ (PHN, Tag. pá:yoQ 'umbrella'), Snd. payu~ (L), OJv. payu9' NJv. paY&3' Mal. payo~, Mad. paju~ 'umbrella'.

*payaD' Snd. payan 'drag-net', OJv. pa-mayan-an 'fisher-boat', NJv. payau 'drag-net', Mal. payau 'fishing', Mad. pajin 'drag-net' •

2.7.2.3. PMJ y in postconsonanta1 position. The on1y reconstruction with postconsonanta1 y is *yu~un, see above.

2.7.2.4. PMJ y in fina1 position.

2.7.2.4.1. PMJ fina1 y af ter PMJ a appears as Sundanese and Malay -ay, and became ~ladurese -6.y or -ay (distributed according to the vowe1 alternation ru1e given in section 1.8.3.1.), and contracted to New Javanese I. 01d Javanese exhibits doub1ets in some instances, one with fina1 ai or ay, the other with contraction to -eo As in the case of the 01d Javanese ö and u, there are three possib1e ways of exp1aining the two spe11inqs: (1) the spe1lings with -ay is an archaism and -e might represent the pronunciation in Old Javanese, (2) -ay represents the pronunciation in earl ier 01d Javanese and -e the pronunciation in later Old Javanese, and (3) the two spellings might represent two dia1ects, one in which *-ay became -ay, and one in which *-ay became -e, where New Javanese is from the e-dialect. The spelling -ai is considered a scriba1 corruption:

*qanay (PAN, Sa. s-ane 'termite': ?anay), Snd. Kad. qanay, Mal. anay-anay 'termite'.

, é ~ *qangay, Snd. qangay-qaugay, NJv. auge-aryg , Mad. aUghay 'mo1e cricket' .

*ba1ay (PAN, Fi. vale 'house'), Snd. ba1ay 'little wall made of stones', OJv. balay 'hall', bale in bale nyäsa 'annex', NJv. balI (0) (0) 'front-hall, bench', Mal. ba1ay 'pub1ic building', Mad. bh~láy 'place for reception'.

*Ba~kay (PHN, Tag. bankáy 'corpse': ba~kay), Snd. bankay in ~ , sawan ba~kay , 'apop1exy', OJv. wankay, wanke, NJv. wanke (0) (B), bauké 'carcass', sawan waUké 'apop1exy', Mal. bankay 'corpse', sawan ba~kay 'apop1exy'.

*ga[09]ay60 (PWI, TBt. gade 'pledge, pawn': ga~ay), OJv. ga~e 'pledge, pawn', NJv. ga~é (L), Mal. gaday 'to p1edge, pawn'. *gaway6l (PHN, Tag. gá:way 'witchcraft'), OJv. gawai, gawe, NJv. gawé (L) 'work' , Mal. gaway 'tool', Mad. ghibáy 'work'. *qintay (PHN, Tag. qintáy, hintáy 'wait': 00 hi[n]~ay, 10 [1953a] [qh]in[t~]ay), OJv. a~-intay, man-inte, NJv. int~ (B), Mal. intay 'to pry, watch'. *kasay (PWI, NgD. kasai 'ointment'), Snd. kasay (H) 'object used in washing', OJv. kasay 'cosmetic used as a hairwash'.

*bu1ay59 (PAN, Sa. hu-hu1e 'cataract'), OJv. bu1e, NJv. bu1é, Mal. bu1ay 'albino', Mad. B. bhu1áy 'dirty white'.

*ha11ay, Snd. h41ay, OJv. h1ai, Mal. (ha)lay 'a numeral coefficient for tenuous objects such as garments'.

*lantay (PAN, Sa. ladre 'stones of the oven'), Snd. lantay-an 'bamboo-stic~s fastened horizontally to sticks standing upright in the ground, serving to hand rice at. The rice is to dry in the sun'. OJv. l-in-antay 'wrapped in a mat (?)', NJv. lanté (0) (H) 'mat', Mal. lantay 'flooring', Mad. lantay 'rattan-mat to sit on'.

*laway62 (PHN, Tag. lá:bay 'skein': labay), Snd. laway (H) 'to prepare yarn for weaving', OJv. lawe, NJv. lawé, Mad. labJy 'yarn'.

*11mBay, OJv. an-limbe 'to extend', Mal. limbay, Mad. lèmb~y 'to sway the arms outwards'. *pakay63 (PHN, Tag. pá:kay 'mission'), NJv. pak~an 'clothes, horse-equipment' , Mal. pakay 'to use'. *pa[rr, R2R2]ay, Snd. p~ray 'to open up (of a lump of flour), corrode', Mal. bar-paray-paray 'to disperse in all directions, breaking up, crumbling as a clod of earth or fungus'. *pulay, OJv. pule, NJv. pulé, Mal. pulay, Mad. pOlay 'kind of tree'. *[rR2]amBay, Snd. ra-rambay, Mal. rambay, Mad. ramb~y 'kind of tree that gives fruit'. *rantay64 (PWI, TBt. ratte 'chain': [r]antay), Snd. rantay 'linked together', OJv. rantay, rante 'band', NJv. rant~, Mal., Mad. rantay 'chain'. *[rR2]am[Bb]ay, Snd. r~mbay, Mal. rambay 'to flow (of tears)'. *salay65 (PWI, TBt. sale 'to smoke'), Snd. salay 'to hang bundles of rice on ~rancatan (to dry)', NJv. sal~, salèh (0) 'fruit biscuit (dried banana) ' , Mal. salay 'slow roasting', salay-an 'the rack nearest the kitchen-fire', panalay 'a sort of grid or platform for sun-drying'. *sampay (PAN, Fu. safe 'to carry on the shoulder'), Snd., Mal., Mad. sampay 'to hang loosely'. *simpay, Snd. simpay 'band, hoop', NJv. simp~ (0) 'iron-wheelband', Mal. simpay, Mad. s~mpay 'band, hoop'. *[t~]aDkay (PAN, Sa. ake 'twig': takay, tankay), Snd., Mal. tankay'stalk, stem, haulm'.

*tapay (PHN. Tag. tá:pay 'dough'). Snd. Bant. tapay. NJv. tapé. Mal.,Mad. tapay 'yeast. leavened cake'.

In one comparison the Sundanese cognate exhibits final -~y instead of an expected final -ay. perhaps through assimilation of the final vowel to the initial vowel:

*pattay.Snd. p~t~y. NJv. patl. Mal. patay 'stink-bean'.

2.7.2.4.2. PMJ final y af ter PMJ a contracted with this vowel in all four languages and became Sundanese -eq. Javanese and Malay -i. and Madurese -i or -~ and West Madurese -ih or -~h (distributed according to the vowel alternation ruie given in section 1.8.3.1.).

Oyen (1949:421 fn. 5) was the first to reconstruct *-ay to explain the correspondence Tag. -ay. TBt. -et NJv. -i. Mal. -i. NgD. -ay. Mer. -i. Further instances exemplifying *~yappear among the following:

*hanay (PHN. Tag. há:nay 'to put in a row': hanay). Snd. pi-haneq. NJv. p-ani. Mal. ani 'to arrange the warp on the loom'. *hatay (PAN. To.qate 'liver': qaCay?i). Snd. hateq (L) 'heart. feelings'; anatomically: 'liver'. OJv. hati 'heart'. h-in-ati-ati 'in what one trusts'. NJv. ati (L) 'heart. feelings. soft inner parts (e.g. of bamboo). liver (e.g. as food) , • Mal. (h)ati 'heart. feelings'; anatomically: 'part of body made up of the liver. gall. and heart'. Mad. at~(h) 'heart. feelings'; anatomically: 'liver'. *guramay66. Snd. gurameq. NJv .• Mal. Java gurami 'kind of fish'. *[jz]unay. Snd. joneq. Mal. juni 'kind of plant'. *kan~ay. Snd. Bant. kandeq 'a bag of katun or other material'. NJv. kandi (0) (0) 'bag. pouch'. Mal. kandi-kandi 'basket'.

Mad. kanqhi 'bag. pouch'.

*palay 67 (PHN. Tag. pá:lay 'rice-plant': pajay). Snd. pareq (L).

OJv. pari. NJv. pari (L). Mal. padi 'rice'.

See also *[t~]i[rR2]dY. p. 55; *huay. p. 65.

2.7.2.4.3. Ambiguous reconstructions. An ambiguous reconstruction with *[iq. ay] is employed when a Sundanese cognate is lacking and when Old and/or New Javanese and/or Malay exhibit -i. and or Madurese exhibits -i or -~ and West Madurese -ih or -~h (distributed according to the vowel alternation rule given in section 1.8.3.1.):

·k3nqBq, ay]68 (PWI. TBt. hond i 'water-goglet': kanq[i]). NJv.

kan~i. Mal. kandi. Mad. kanqhi(h) 'water-goglet'.

2.7.2.4.4. PMJ final y af ter PMJ u appears as Sundanese -uy and became Madurese -uy or -~y (distributed according to the vowel alternation rule). It contracted with the preceding vowel to Malay and New Javanese i. As in the case of PMJ final y af ter PMJ a. Old Javanese has doublets. We either find an uncontracted form with -uy or 'a contracted form with -i. For the three possibilities of explaining these doublets. see section 2.7.2.4.1. Instances exemplifyina *-uy appear among the following:

*qapuy69 (PAN, To. afi 'fire': xlapuya). Old Snd. apuy. OJv. apuy, api 'fire'. NJv. api-api 'to warm o.s. at a fire', Mal. api Mad. apby 'fire,70.

*Baluy71 (PWI, Mer. valu 'to return. repent': baluy ). OJv. 'every time again', Mal. kam-bali 'to return'.

waluy 'return', NJv. wali (0) (0), bali (L). wali-wali (0)

*su~guy,OJv. s-in-u~gi 'be carried'. NJv. sungi 'carried on the head', Mad. sà~ghuy 'to carry on the shou1ders'. *suruy72, OJv. suruy, su ri 'comb' , NJv. suri ',weaving-comb', Mal. suri 'the comb of a Ma1ay loom', Mad. sbroy 'comb'. *tu1uy73 (PHN, Tag. tu1dy 'go on'), Snd. tu1uy 'to go forth', OJv. a-nu1uy-i 'to continue', a-nu1i-nu1i 'quick1y fo110w each other', NJv. tu1i 'immediate'. *turuy74, Snd. Cir. turuy, Parean tuyur, OJv., NJv., Mal. turi, Mad. t~r~y 'kind of tree'.

The consonants

2.8. In this first section on the reconstruction of the PMJ consonants we sha11 deal with the fo110wing proto-phonemes: *m, *n, *n, *~, *1, *1, *r, and *s, and their reflexes.

2.8.1. PMJ m. PMJ m is reconstructed in all five positions: initial, intervocalic, preconsonantal, postconsonantal, and final. It appears as m in all four languages in all positions except in certain environments in PMJ doubled monosyllables (see the section on PMJ doubled monosyllab1es, p. 194).

2.8.1.1. PMJ m in initial position:

*mataa (PAN, To., Fu., Sm. mata 'eye': maCa?), Snd. mataq (l), OJv. mata, NJv. moto (l), Mal. mata, Mad. mata(h) 'eye'.

*manis (PWI, TBt. manis 'sweet'), Snd. manis75 , 'sweet, nice', OJv. manis 'love1iness, sweetness, sweet', NJv. manes 'sweet' (0), 'nice, lovely', Mal. manes 'sweet', Mad. man~s 'sweet, lovely'.

2.8.1.2. PMJ m in intervoca1ic position:

*tumaq (PAN, To. Fu., Sm. tuma 'louse': tuma[?h]), Snd. tumaq, OJv. tuma, NJv. tum~, Mal. tuma, Mad. t~ma(h) 'louse'.

*taman (PAN, Sa. ama-a 'yard'), Snd., OJv., NJv., Mal., Mad. taman 'garden, p1easure-ground'.

2.8.1.3. PMJ m in preconsonanta1 position:

*kumpu1 (PHN, Tag. kump&l 'bunch'), Snd. kumpu1, NJV. kump61 (L), Mal. kumpo1, Mad. k~mp~l 'to gather, group').

*lumpuh (PWI, TBt. 1uppu 'lame'), Snd. 1umpuh, NJv. 1umpóh 'lame, weak (legs)', Mal. 1ayoh 1umpoh 'weak, 1imp, tender (of sores)', Mad. 10mpö 'weak, exhausted'.

2.8.1.4. PMJ m in postconsonanta1 position:

See the section on PMJ doub1ed monosy11ab1es, p.194 •

2.8.1.5. PMJ m in fina1 position:

*kirim, Snd. kirim (L) 'to send', OJv. kirim-an 'was sent', NJv. kirlm (L), Mal. kirem, Mad. kerèm 'to send'.

*su1am (PAN, Fi. qi-zu1a 'need1e'), Snd. su1am 'embroidery', OJv. su1am 'close together, p1aited', NJv. su1am (0) 'inter1aced', Mal. sulam, Mad. s~lam-an 'embroidery'.

2.8.2. PMJ n. PMJ n is reconstructed in all possib1e positions: initia1, intervocalic, preconsonantal, postconsonantal, and final. It appears as n in all positions in all languages.

2.8.2.1. PMJ n in initial position:

*nanah (PAN, Fi. nana 'pus, matter': naqnaq), Snd., OJv., NJv., Mal. nanah, Mad. nana 'pus, matter'.

*na~kaq (PHN, Tag. na~káq 'a fruit': na~ka), Snd. na~kaq, OJv. , , na~ka, NJv. no~ko, Mal. na~ka, Mad. na~ka(h) 'a fruit, specifically the jack'.

2.8.2.2. PMJ n in intervocalic position:

*qanuq (PHN, Tag. qanó 'the thing whatever you may call it': anu), Snd. qanuq, nuq 'who, which, so-and-so, such-andsuch', OJv., NJv., Mal. anu, Mad. an~(h} 'so-and'so, such-and-such' .

*taniq, Snd. taniq 'agriculture', OJv. tani 'land', NJv. tani 'peasant, farmer', Mal. Jak. tani 'farming', Mad. tanè(h} 'farmer' •

2.8.2.3. PMJ n in preconsonantal position:

*qundur (PWI, NgD. qundur 'to go backwards': u(n)du[rJ}, Snd. qundur, OJv. undur, NJv. undór (O), Mal. undor, Mad. ondhur 'to retreat, go back, retire'.

*kancin (PHN, Tag. kans{n 'gold-brooch': ka(U)ci~}, Snd. kanci~, NJv. kancé~, Mal. kanceU' Mad. kancèn 'button, rivet, bolt, stud, clasp'.

2.8.2.4. PMJ n in postconsonantal position:

For PMJ n in this position, see the section on PMJ doubled monosyllables (p. 194).

2.8.2.5. PMJ n in final position:

*salin (PHN, Tag. sá:lin 'translation'), Snd. salin 'to interchange, change', OJv. a-salin 'interchangeable', NJv. salén (L) 'to change, interchange', Mal. salen 'change of dress or food', Mad. salèn 'to change, interchange'.

*lamun, Snd. lamun (L), OJv. lamun, NJv. lamón, laman (D) (O), Mal. lamon, Mad. lamon 'provided it be'.

2.8.3. PMJ n. PMJ n is reconstructed in initial and intervocalic position. It appears as n in all languages in these two positions except when it occurred before *i. In this case it became Sundanese and Old and New Javanese n. Malay of ten has competing forms, one with n before i, the other with n before i. Thus, we can posit two dialects, one in which *n before i appears as n and another in which it became n. In the following instances it is best to explain the existence of competing forms as borrowings from the dialect in which *n before *i became n into the dialect in which *n before *i appears as n, or conversely.

2.8.3.1. PMJ n in initial position before avowel other than *i:

*ffawaq (PAN, To., Fu., Sm. ma-nava 'to breathe': nawa?), Snd. ffawaq, NJv. ~owo, Mal. nawa, Mad. nabl(h) 'soul'.

*naman, NJv. naman 'tasty', Mal. naman 'healthy feeling, tasty', Mad. naman 'sense of well-being'.

2.8.3.2. PMJ n in initial position before *i:

*niuR (PAN, Fi., Sa. niu 'coconut-palm'), OJv. nyu, Mal. nior, nior, Mad. n~yör, n~y~u76, 'coconut'.

*ffiluq, Snd. linuq (metathesis) 'smarting', OJv. a-linu 'sharp toothed', NJ~ linu 'to have a gnawing feeling (in the teeth, bones)', Mal. nilu, nilu77 'nerve pain', Mad. nèlO(h) 'to have a gnawing feeling in the teeth or in the bones'~

2.8.3.3. PMJ n in intervocalic position before avowel other than *i:

*qinaq (PAN, Fi. t-ina 'mother': ?ina?a), OJv. ina, NJv. in~, Mal. ina, Mad. éna(h) 'foster-mother, nurse'.

*tanaq, Snd. tanaq (L), OJv. tana, NJv. toffo (B), Mal. tana, Mad. tana (h) 'to i nterroga te' •

In one instanee Sundanese and Ma1ay exhibit an intervoca1ic n, whi1e 01d and New Javanese exhibit an intervoca1ic n:

*qanam (PAN, Fi. yana-yana 'to p1ait loose1y': ?anam), Snd. qanam, OJv., NJv. anam, Mal. anam 'to p1ait'. Mad. anam 'to p1ait' probab1y is a loan from Javanese.

2.8.3.4. PMJ ~ in intervoca1ic position before *i:

*haniq78 (PHN, Tag. qá:ni), NJv. ani-ani, Mad. a~~(h) 'to harvest rice, harvesting knife'.

*Buniq (PAN, Fi. vuni 'hidden': buffi[q?h]), Snd. buniq, OJv.

w-in-uni 'hidden', Mal. sam-buni 'concea1ment'.

*[Bb]-uniq, Snd. quniq (L), OJv. uni, NJv. uni (L), Mal. b-uni, Mad. m-Ön~(h) 'sound'.

*[Bb]annir,Snd. b4n4r79 'rice-dust', bran', OJv. a-manir-manir80 'in grains', NJv. manér80 , 'rice-dust, bran'.

2.8.4. ~. PMJ ~ is reconstructed in four positions: initia1, intervocali,c, preconsonantal, and final. It appears as D in all four 1anguages in initia1, intervoca1ic, and fina1 position. In preconsonanta1 position it was retained in Sundanese, but assimi1ated to the fo11owing stop in 01d and New Javanese, Ma1ay, and Madurese. For the ref1exes of PMJ preconsonantalo in doub1ed monosy11ab1es, see p. 194.

2.8.4.1. PMJ g in initia1 position:

*~a~aq (PAN, Fi. DaDa 'to drink without swa11owing': DaOa[?h]).

Mal. DaUa, Mad. Dana(h) 'to open wide the mouth'.

*Ua]an (PAN. Sa. s-ata 'name': DajaL]), Snd. Uaran (L), OJv. !Jaran 'name', NJv. !Jaran-i (1) 'to name', aran81 (L) 'name'.

2.8.4.2. PMJ n in intervocalic position:

*su~ut (PHN, Tag. su~6t 'proboscis [of insects, ants]'), Snd. su~ut (L) 'mouth', OJv. suuut 'fee1er, tentac1e', Mal. sU90t 'long sparse hairs on the upper lip, cat's whiskers, antennae of an insect'.

*ta!)is (PAN, Fi. taUi 'to weep': Ca!Jiso), Snd. ta!Jis (H), OJv. / 1 .... ta!Jis, NJv. ta!)es, Ma . tanes, Mad. ta!)es 'to weep, cry'.

2.8.4.3. PMJ ~ in preconsonanta1 position: *Oa!JOoR2' Snd. dand4r, OJv. ran~ö, ra!jrö, NJv. ran~u 'cotton tree' •

*janjiq (PWI, TBt. janji 'agreement': zanzi), Snd. j~jiq (L), NJv. janji, Mal. janji, Mad. jhánjhi 'agreement'. New Javanese a1so has an archaic form ja!Jji 'agreement'.

*ta!JguD (PWI, NgD. ta!JgoD 'to take responsibi1ity'), Snd. ta!Jgu~ 'to carry on the shou1der', OJv. t-in-angun-an 'be carried', NJv. tangon (L) 'to guarantee for, take care of', Mal. tangon 'to support on the shou1der, give security', Mad. taUghuO 'to give custody'.

2.8.4.4. PMJ ~ in fina1 position:

*qiriU (PWI, NgD. m-irin 'to fo110w': i[r]i~), Snd. qiri~ (H), OJv. um-irin-i, NJv. irén (0) 'to fo110w', Mal. iren 'Indian file', Mad. èrèn 'to fo1low'.

*cankiu, Snd. ca~kin (H) 'to ho1d in one's hand', OJv. cankiu, NJv. cankéD 'to carry with the hand which is hanging down', Mal. canken 'to take in both hands and lift up', Mad. canké~ 'to carry in the hand'.

2.8.5. PMJ 1. PMJ 1 is reconstructed in all five positions: initia1, intervoca1ic, preconsonantal, postconsonantal, and final. It appears as 1 in all 1anguages in all positions except in preconsonanta1 position where it disappeared in all languages but Madurese.

2.8.5.1. PMJ 1 in initial position:

*la9kah (PAN, Sa. laka 'to jump': la9kaq), Snd. lankah 'step', OJv. sa-lankah-lankah 'with every step', NJv., Mal. la~kah, Mad. la!)ka 'step'.

*lakiq (PHN, Tag. la-lá:ki 'man, male': lak;), Snd. lakiq, OJv., NJv., Mal. laki, Mad. laké(h) 'man, husband'.

In the following comparisons Sundanese exhibits doublets, one in which 1 appears as 1, the other showing assimilation to the fol10wing consonant:

*laruq (PWI, Mer. laru 'a potion used to stupefy fish': laru), Snd. laruq, raruq 'sap of the kawaoq which prevents or retards the fermentation of laha!)', OJv. den laroni (laru+ani) 'mixed with a sleeping potion', NJv., Mal. laru, Mad. lar~(h) 'ingredient to prevent or retard fermentation in toddy'.

*laris, Snd. laris, raris, NJv. larés, Mal. lares, Mad. B. lares 'in demand, selling well '.

In one instance the Sundanese cognate occurs without a by-form with initial 1:

I , \ *lurug, Snd. rurug, OJv. lurug, NJv. lurog, Mad. lorok 'to go on an expedition, go to war'.

2.8.5.2. PMJ 1 in intervocalic position:

*paluq (PAN, Fi. valu 'to fight': palu?), Snd. paluq, OJv., NJv.

pa1u 'hammer', Mal. pa1u 'to hit hard with a rigid weapon', Mad. pa10{h) 'hammer'.

*gi1io (PAN. Fi. 09i1i 'to grind in the hands'), Snd. gi1iO' NJv. gi1éo 'to grind, pulverize', Mal. gi1en 'to ro11 out, flatten with a roller', Mad. ghi1iD 'to turn around an axis', ghi1io-an 'mi11'.

2.8.5.3. PMJ 1 in preconsonanta1 position:

An examp1e of *1 in preconsonanta1 position can be found in the section on doub1ed monosyllables CP. 194).

2.8.5.4. PMJ in postconsonanta1 position:

*sa01iO' Snd. san1io 'to po1ish', OJv. s-in-a~lin 'po1ished',

NJv. san1én, Mal. san1eo, Mad. san1~n 'to po1ish'.

2.8.5.5. PMJ 1 in fina1 position:

*tata1 (PAN, Fi. qi-tata 'chip'), Snd .• OJv., NJv., Mal., Mad. tata1 'chip, wood-shaving'.

*tunga1 (PWI, Mer. tukan3 'unique'), Snd., OJv. tunga1, NJv.

tuoga1 (L), Mal. tunga1, Mad. tooghá1 'sole. unique'.

2.8.6. PMJ 1. PMJ 1 is reconstructed in intervoca1ic and fina1 --. . position. In intervoca1ic position it became Sundanese and 01d and New Javanese r, Ma1ay d, and Madurese 1. In fina1 position it became Sundanese and 01d and New Javanese r, Ma1ay t, and Madurese q.

PMJ 1 corresponds to Dyen's symbo1 j.

2.8.6.1. PMJ! in intervoca1ic position:

*hiluD (PAN, To. ihu 'nose': q2iju~), Snd. qiruD (L), OJv. irun. NJv. ir60' Mal. {h)idon' Mad. ~lÖD 'nose'.

*pal1as (PAN. Fi. vo~ota = vosota 'to endure': pajas, panjas), Snd. ptrts 'pain 1ike the pain which is fe1t when a hair is pulled out, smarting', Mad. pallas 'smarting like when hit with thin rattan'.

2.8.6.2. PMJ 1 in final position: *qanu182 (PAN, Sa. m-~nu 'to float': qanud), Mal. (h)anot, Mad. an~q 'to float'. *Buki!83 (PAN, Sm. puqe 'heap': bukid), OJv. wukir, NJv. wuk'r (B), Mal. buket 'hill'. *lalal (PAN, Fi. lala 'fly': lalaj), Snd. la14r, OJv., NJv. lalar, Mal. lalat, Mad. lalaq 'fly'. *pusa184 (PAN, Fu., Sm. uso 'navel-string': puslaja), Snd. pus4r 'the piece of navel-string which remains at the navel, navel, center', NJv. pusar (0) (0) 'navel-string, center', Mal. pusat 'center of circle, focus, navel'. *hulal (PAN, Fi. qulo 'maggot': ?ulaj), NJv. ular, Mal. (h)ulat, Mad. olaq 'maggot'.

In the following comparison Old Javanese exhibits competing forms, one with final r, the other with final d, and new Javanese exhibits one form with final rand the ot her with medial d. The occurrence of this d is inexplicable:

*lau185 (PAN, Fi. lau 'islands facing towards the winds': laud), OJv. lod 'sea', lor 'north', NJv. iwaq lód-an (? fish of the sea) 'whale', l~r (L) 'north', Mal. laot 'sea, ocean', barat laot 'north-west', Mad. laoq 'south'.

In the following comparison Old Javanese exhibits a doublet, one with a final r, the ot her with a final d. New Javanese only has a form with final r:

*lu1u1, Snd. 1u1ur, OJv. 1u1ud, 1u1ur 'ointment', NJv. 1u1Ór (0) 'ye110w cosmetic made of herbs', Mal. 1u10t 'to rub or massage with scents and cosmetics'.

2.8.7. PMJ r. PMJ r is reconstructed in initia1, intervoca1ic, preconsonantal, postconsonantal, and fina1 position. It appears as r in initia1, intervoca1ic, postconsonantal, and fina1 position in all 1anguages. For the ref1exes of PMJ preconsonanta1 r, see the section on doub1ed monosy11ab1es.

PAN ~ and PAN r in preconsonanta1 position and fina1 position and PAN R3 and PAN R4 merge in PMJ r: e.g., *kurkur (PAN, Fi. kukuva 'to scratch': kudkud) p. 199; *qa1ur (PHN, Tag. qá:10g 'pool of standing water': OD a1ur, ID [1957b] a1uR3]~ p.105; *Barat (PAN, Sm. afa 'storm': ha-baR4ata), p. 132.

2.8.7.1 PMJ r in initia1 position:

*rampu~, Snd. rampuD 'cut off (of a hand, foot, ear)', OJv. den rampun 'was devastated', NJv. ramp6D 'off, c1ear', Mal. rampoD 'bored, ho1ed, muti1ated', Mad. Kang. rampoD 'to break'.

/ *ra9ku1, Snd. raUku1, OJv. ma~-raDku1, NJv. ranko1 'to embrace' , Mal. ranko1 'to grasp and lift between both hands', Mad. raDk~l 'to finger (out of 1asciviousness)'.

PMJ initia1 r before PMJ -rbecame New Javanese 1:

*ririh, Snd. ririh 'ca1m, a11eviation', OJv. ririh 'carefu1, ab1e, understanding', NJv. 1iréh 'soft, ca1m', Mad. rêr~ 'comfortab1e, without haste'. Madurese a1so has a competing form here in which the same dissimi1ation as in New Javanese took p1ace: l~r~ 'comfortab1e, without haste'. *rurub, Snd. rurub 'to cover with a sheet', 'OJv. rurub, NJv. 1ur6b, Mad. rorop 'shroud, winding-sheet'.

2.8.7.2. PMJ r in intervoca1ic position:

*kuran (PHN, Tag. kU:1a~), Snd. kura~ (L), OJv. kuran, NJv. kuran (L), Mal. kura9' Mad. koran 'reduction, fa11 short, 1ess, not quite'.

*kara~ (PAN, Sm. qa1a 'a reef'), Snd., OJv., NJv., Mal., Mad. karan 'coral reef'.

2.8.7.3. PMJ r in postconsonanta1 position:

For examp1es of postconsonanta1 r, see the section on PMJ doub1ed monosy11ab1es.

2.8.7.4. PMJ r in fina1 position:

*qa1ur (PHN, Tag. qá:10g 'pool of standing water': 00 a1uR, JO [1953b] a1uR3), Snd. qa1ur 'track, trace of a rhinoceros, etc.', NJv. a1ur-an 'beaten buffa10 path, morass', Mal. a10r 'groove, cutting. furrow', Mad. a1~r-an 'small ditch'. *hatur (PAN, Fi. yatu 'row': h4atur), Snd. qatur, NJv. atör (0), Mal. (h)ator, Mad. ator 'to arrange'.

2.8.8. PMJ s. PMJ s is reconstructed in all five positions: initia1, intervoca1ic, preconsonantal, postconsonantal, and final. Jt appears as s in all 1anguages except in preconsonantal position. For the reflexes of *s in this position, see the section on PMJ doub1ed monosy11ab1es.

2.8.8.1. PMJ s in initial position:

*saput (PAN, Sm. aput-i 'to cover': s2aput), Snd., OJv. saput,

/ , NJv. sapot, Mal. sapot, Mad. sapoq 'to cover over slight1y'. *sappah (PHN, Tag. sapá 'bagasse of s.t. masticated': sa[m]pah), Snd. sipah (L) 'to chew sireh', OJv., NJv. sapah 'chewed-up quid of sireh', Mal. sapah 'quid of sireh', Mad. sappa 'what has been chewed, e.g., a1ready chewed quid of sireh'. 2.8.8.2. PMJ s in intervoca1ic position: 2.8.8.3. PMJ s in postconsonanta1 position: *su!)sa9 (PWI, TBt. sunsao 'inverted'), Snd., OJv., NJv., Mal. sunsan, Mad. s~osao 'to invert'. 2.8.8.4. PMJ s in final position:

*pusar (PHN, Tag. puS6d 'hairknot, coiffure': pusad), Snd. caiq pusir'spinning water', NJv. pusar-an 'circ1e', Mal. pusar 'spiral', Mad. pasar 'to turn, circ1e'.

*pasan (PAN, Sa. hata-i), Snd. pasan 'to put in order', OJv. am-asan 'to app1y', NJv. pasan 'to erect', Mal. pasan 'to put in order, in use', Mad. pasan 'to erect'.

*linsir, Snd. 1iOsir 'to be in a slanting position (esp. of the sun)', OJv. li9sir 'to give way, be past the noon-high, noon', NJv. li!Jsér (L) 'afternoon (about 3 p.m.)', Mal. linser, leDser 'side-slipping, sliding', Mad. le!)sèr 'af ternoon' •

*turus (PAN, Fi. nduru 'short house-pillar'), Snd. turus, NJv. turos 'stick, pole', Mal. turos 'massive post, pillar', Mad. t6r6s 'a cut-off stick'.

*pipis (PHN, Tag. pip{s 'pressed down flat'), Snd. pipis-an 'rubbing-stone' , OJv. pipis-an 'pressed flat', NJv. pipé's 'to rub down, pulverize', Mal. pipes 'to mash between two

, , surfaces', Mad. pepes 'to pulverize'; cf. OJv. um-ipis 'rub down, pulverize'.

2.9. In this section we shall deal with the proto-phonemes *d, *j, *z, and *g and their reflexes.

2.9.1. PMJ d. PMJ d is reconstructed in all positions: initial, intervocalic, preconsonantal, postconsonantal, and final. In initial, intervocalic, and postconsonantal position it appears as Sundanese, Old and New Javanese, and Malay d, and became Madurese dh. For its preconsonantal reflexes, see the section on PMJ doubled monosyllables. In final position it appears as d in Sundanese and in Old Javanese. New Javanese has d in its Western dialects; t before pause, but d before a suffix in the Central dialects; and t before pause and before a suffix in the Eastern dialects (see section 1.5.2.).

2.9.1.1. PMJ d in initial position:

*duqum, Snd. duqum, OJv. a-dum, NJv. d6m, Mad. dhuqum 'to distribute'.

*damar (PAN, Fu. lama 'toreh': damaRa), Snd. damar 'resin, toreh', OJv. damar, NJv. damar (L), damar sé16 'kind of resin', Mal. damar 'resin, toreh, resinous tree', Mad. dhámar 'lamp, light', dhámar báto 'kind of resin'.

2.9.1.2. PMJ d in intervocalic position:

*sidik, Snd. sidik, NJv. sidéq, Mal. sideq 'to examine'.

*kadan, Snd., OJv. kadan 'related, kindred', NJv. kadaD (0)

'brother, sister', Mad. kadhán in kadháD bárghá(h) 'kin'.

2.9.1.3. PMJ d in postconsonantal position:

*gandar, Snd. gandar 'the wooden shaft of an arrow', NJv. gandar

\aris-shaft', Mal. gandar cincen 'claw for gun in ring', / / Mad. ghandhar 'wooden shaft of a karis'.

*canduq (PWI, TBt. sandu 'opium'), Snd. canduq, NJv., Mal. candu, Mad. candhu(h) 'prepared opium'.

2.9.1.4. PMJ d in final position:

*parud86 (PAN, Fi. qi-varo 'rasp': parut), Snd. parud, NJv. par6d, , Mal. parot, Mad. parot 'rasp'.

*cacad, Snd., OJv., NJv. cacad, Mal., Mad. cacat 'defect, flaw, shortcoming'.

2.9.2. PMJ j. PMJ j is reconstructed in initial, intervocalic, and postconsonantal position. It appears as Sundanese, Old and New Javanese, and Malay j, and became Madurese jh in all positions.

PMJ j corresponds to Dyen's symbol z.

2.9.2.1. PMJ j in initial position:

*jarah, Snd. jarah, OJv. a~-jarah-jarah, NJv. jarah-rayah, Mal. jarah, Mad. jhárt 'to raid'.

*juruq (PHN, Tag. du:lo 'end, terminal extremity': zuru), Snd. juruq 's.o. who works at a low rank', i.e. juruq tulis 'copying clerk', OJv. juru 'leader, head, village-head', NJv. juru (0) 'man (woman) of, for', Mal. juru 'trained worker', Mad. jhuru(h) only in compounds such as jhuru kÖncè 'keykeeper'.

2.9.2.2. PMJ j in intervocalic position:

*tajiq (PHN, Tag. tá:riq 'cockspur': tazi), Snd. tajiq, OJv., NJv., Mal. taji, Mad. tajhi(h) 'artificial spur for fighting cock'.

*kajan (PHN, Tag. ká:rau 'nipa awning or shed': kazan), Snd., OJv., NJv., Mal. kajan, Mad. kajhán 'mat-protection against rain' .

2.9.2.3. PMJ j in postconsonanta1 position:

*ganji1 (PHN, Tag. gar{l 'defective in pronunciation': ga[n]zi1), Snd. ganjil 'too 1itt1e, too small for', NJv. ganjé'l '(one) more than the ri ght number', 'uneven' (0), Mal. ganj el, Mad. ghá'njhi 1 'uneven'.

*tanjuo, Snd., OJv. tanjun, NJv. tanj6n, Mal. tanjoD' Mad. tanjhu9 'name of a flower'.

2.9.3. PMJ z. PMJ z is reconstructed in initia1, intervoca1ic, and postconsonanta1 position. It became"Sundanese and Ma1ay j, 01d and New Javanese d, and Madurese jh in all positions.

PMJ z corresponds to Dyen's symbo1 Z (see Dyen 1951). Dyen 1ists eight comparisons, three in which his *Z appears in initia1 position, and five in which it appears in intervoca1ic position.

In the fo110wing I have 1isted all comparisons in which PMJ z occurs, in order to present further evidence for this reconstruction.

2.9.3.1. PMJ z in initia1 position:

*za1uhan87, OJv. da1uwan 'bark', NJv. d1uwan (L) 'paper', Mal. ja1uao 'vegetab1e parchment, native paper'.

*za1an (PAN, Sa. ta1a 'road': Za1ana), Snd. ja1an 'road, path', OJv. ma-da1an 'to go through', NJv. da1an (L), Mal. ja1an, Mad. jhá1~n 'way, road, path'.

*za1ujur, Snd. ja1ujur, NJv. d1uj6r88 (0), Mal. ja1ujor, Mad. jh{lujhur 'tack, bast, stitch'.

*zaR2um (PAN, To. hau 'need1e': Z2aR123um), Snd. jarum, NJv. dom, Mal. jarom, Mad. jh'rum 'need1e'.

*zauh (PAN, Sa. haqa-tau 'far': Zauq), Snd. jauh (L), OJv. doh, NJv. a-doh (L), Mal. jaoh, Mad. ~h{u 'far, distant'.

*za11iO' OJv. da1i9 'to open the eyes', Mal. ja1en 'to give a side-g1ance', Mad. B.P. jha11in 'to see s.o.'.

*zi1at (PWI, TBt. di1at 'to lick': 00 dilat, 10 [1951] Zi1at), OJv. a-di1at, NJv. dilat, Mal. ji1at, Mad. jhi1át 'to lick'.

*zua1 (PWI, TBt. jua1 'a measure of rice': zua1), Snd. jua1 (L) 'to se11', OJv. ma-dwa1, d-um-01 'sold', NJv. dol (L), Ma1. jua1, Mad. jhuw'l 'to se11'.

*zuR3uh (PAN, Fu., Sm. su 'wet, watery': ZuR1uq), Snd. juquh 'to flow out p1entifu11y (mi1k, pa1m-wine), OJv. duh 'sap, mi1k', du-duh 'sap, sauce', NJv. du-d6h 'sap', Mal. juroh 'syrup', Mad. B. jhuru 'sap, syrup'.

*za11as,NJv. da1as 'genuine, true', Mal. ja1as 'made p1ain, wound up, sett1ed'.

Prob1ematic is the fo110wing comparison inwhich 01d and New Javanese exhibit a doublet, one with initia1 d, the other with initia1 ~, corresponding to Sundanese initia1 j:

*zukut, Snd. jukut, OJv. dukut, ~ukut, NJv. dUk6t, guk6t (0) (H) 'grass'.

2.9.3.2. PMJ z in intervoca1ic position:

*Buz[aa]l, Snd. buja1 (L), NJv. wuda1 'navel', Ma1. buja1 'marked1y prominent (of navel)', Mad. bujha1 'navel'. *hazzan (PHN, Tag. h{:rin 'stick in throat': 00 hadan, 10 [1951] aZan) Snd. (h)4j:in, NJv. (a)dan, Mal. ajan, Mad. (aj)jhan

'to squeeze, press'. *[hq]izap89, NJv. idap, Mad. k-èjháq in bulu k-èjháq 'eye-lash'. *kiza090, OJv., NJv. kidaU' Mal. kijaD 'barking deer'. *kiZUD91 , OJv. kiduU' NJv. kidÓO (0) (0) 'singing, chanting, song', Mad. kèjhuQ 'song with rhyme, ballad'. *pizak (PWI, NgO. pijak 'to step on': pizak), NJv. pidaq, Mal. pijaq 'to set foot upon'. *huzan (PAN, Fi. quza 'rain': quZlaLa), Snd. hujan, OJv. (h)udan, NJv. udan (L), Mal. (h)ujan, Mad. ~jh'n 'rain'. 2.9.3.4. PMJ z in postconsonantal position: *hanzuaO' Snd. hanjuau, OJv. andoD' NJv. and~n, Mal. s-anjua~, l-anjuau 'dracaena'. *hunzan92, OJv. undan-undan-an, man-undan 'to invite', NJv. undaU 'to call', to invite' (0), Mad. onjháo 'to invite'.

2.9.4. ~. PMJ 9 is reconstructed in all five positions: initial, intervocalic, preconsonantal, postconsonantal, and final. In initial, intervocalic, and postconsonantal position it appears as Sundanese, Old and New Javanese, and Malay g, and became Madurese gh. For its preconsonantal reflexes, see the section on PMJ doubled monosyllables, p.194. In final position PMJ 9 appears as Sundanese and Old Javanese g, and became Malay q, and Madurese k. New Javanese has 9 in its Western dialects; k before pause, but 9 before a suffix in the Central dialects; and t before pause and before a suffix in the Eastern dialects (see section 1.5.2.).

2.9.4.1. PMJ 9 in initial position:

*galih, Snd. galih (VH) 'heart, mind', S.-B. 'the heartwood of a tree', OJv. galih 'aorta', NJv. gal~h 'heart, kernel (of wood)'. 'liver' (0). 'mind. heart' (B). Mal. Java galeh 'core and hard portion of a tree'. Mad. gháli 'the inner hard and heavy wood'.

*gilfr (PWI. NgD. girir 'order of succession'~ gili[r]). Snd .• OJv. gilir. NJv. gil€r. Mal. giler. Mad. ghilir 'succession. in rotation'.

2.9.4.2. PMJ 9 in intervocalic position:

*saguq (PAN. Fi. sako-sako 'pudding': sagu[?h]. sangu[?h]). Snd. saguq. NJv .• Mal. sagu. Mad. saghu(h) 'mealy pith'. *tagih (PWI. Mer. taki 'to dun': ta[~]gih), Snd. tagih. OJv. a-nagfh. NJv. tageh. tagéh (0). Mal. tageh. Mad. taghi 'to dun a man for debt'.

2.9.4.3. PMJ 9 in postconsonantal position:

*paogun (PWI. TBt. pangun 'hill '). Snd. panguu 'dome, outlook'. OJv. paugun 'dome. platform'. NJv. pang6n 'tower. platform'. Mal. pangon 'raised platform', Mad. panghun 'dome'.

*tangap (PHN, Tag. taUgáp 'accepted, admitted'). Snd. tangap 'to look for. listen for, to make s.o. perform, watch a play'. OJv. tangap-3n 'accepted', NJv. tangap 'to accept willingly' (0). 'to make people perform'. Mad. tangh'q 'to make people perform' .

2.9.4.4. PMJ 9 in final position:

*katta§3. OJv. kata-katag. NJv. (ka-)katag 'pulse', Mal. kataq 'rap. tap'. Mad. kattak 'pulse'.

*qurug. Snd. qurug-an 'to pour over. cover with'. OJv. in-urug-an 'covered'. NJv. urÓg 'to get seconds, cover with earth', Mad. ~r~k 'to fill up'.

2.10. In this section we will present the reflexes of PMJ c, p, t, and k.

2.10.1. PMJ c. PMJ c is reconstructed in initial, intervocalic, and postconsonantal position. It appears as c in all positions in all four languages.

2.10.1.1. PMJ c in initial position:

*cacin, Snd. caciD 'worm', OJv. cacin-an 'hit by worms', NJv. / , cace~, Mal. cace~, Mad. cacen 'worm'.

*cukur (PWI, TBt. sukkur 'to shave': cU[9]ku[r]), Snd. cukur

(L), NJv. cUk6r, Mal. cukor, Mad. cok~r 'to shave'.

2.10.1.2. PMJ c in intervocalic position:

*kacan (PWI, TBt. hatsaD 'bean': ka[~]ca~), Snd., OJv., NJv., Mal.

kacan 'pea, bean' , Mad. B.P. kaca~ 'peanut'.

*licin, Snd. licin 'hairless, smooth', OJv. licin 'fine', NJv.

licin, Mal. licen, Mad. l~c~n 'smooth, slippery'.

2.10.1.3. PMJ c in postconsonantal position:

*lancar (PWI, TBt. ratsar 'quick': lanca[r]), Snd. lancar 'clean, without trees', NJv. lancar 'quick', Mal. lancar 'to slip along smoothly', Mad. lancar 'quick and good (of .reading)'. *pancur (PWI, Mer. fantsuna 'outlet-pipe': pa9cu[r]), Snd. pancur 'glare (of light)', pancur-an 'bamboo-gutter, water-pipet, OJv. pancur-an 'fountain, artificial water-fall', NJv. panc6r 'to flow', 'to glare (of light)' (B), Mal. pancor 'to flow along a conduit', Mad. pancor-an 'pipe or gutter for the flow of water'.

2.10.2. PMJ p, t, and k. PMJ p, t, and k are reconstructed in all positions: initial, intervocalic, preconsonantal, postconsonantal, and final. They appear as p, t, and k in all languages in initial, intervocalic, and postconsonantal position. In preconsonantal position they disappeared in all languages except Sundanese and Madurese. In final position they appear as p, t, and k in Sundanese and in Old Javanese. In Malay *-p and *-t appear as pand t respectively, but *-k became -q. In New Javanese *-p and *-t appear as pand t respectively, but *-k became q except when it occurred af ter *a. There are two dialectal reflexes for *-ak in New Javanese, one in which *-ak became -ak and another in which it became -aq (see section 1.5.2.). In Madurese *-p, *-t, and *-k became Mad. -q ex cept for *-k af ter *a which appears as Mad. k.

Madurese has a number of doublets, one exhibiting a final glottal stop, the other exhibiting a final p, t, or k as continuations of PMJ forms with final *p, *t, or *k: e.g., *kulit (PAN, Fi. kuli- 'skin'), Mad. kÖlèq 'skin', ~Ól~t (= N + kol~t) 'to skin'. There are two possible ways of explaining doublets of this kind. They could be due to borrowings from Javanese or Malay or they could be due to certain analogical changes. In case no other evidence supports a hypothesis that an item with final p, t, or k from final *p, *t, or *k is a borrowing, it will be considered inherited and the appearance of final p, t, or k is best explained by the following analogical changes. As an example let us consider the Madurese doublet buwáq 'to hold, com- / / , / / prise, load': buwaq-an 'load' and buwat 'be loaded': buwadh-an 'the load'. In early Madurese there probably existed a base buw{q and a suffixed form *buw(t-an, since PMJ -t became Mad. q and PMJ -tbecame Mad. t., q thus alternating with t. By analogy to forms such as

Mad. tÖl~s 'to write': t~l~s-an 'writing' or Mad.pèk6l 'to carry on a carrying pole": p~kÓl-an 'carrying-pole', where s alternated with s and 1 alternated with 1 respectively, early Madurese could have formed a new alternation resulting in buwáq : buwáq-án or buwát : *buwát-an. The modern Madurese alternation t ~ dh found in the forms buw{t buw~dh-~ is the result of an analogical change on the model of the alternation which occurs in Madurese forms containing a PMJ final voiced consonant before monosyllabic suffixes. For example, PMJ -d became Mad. t, but PMJ -dbecame Mad. dh: e.g., *parud, Mad. par~t , / 'rasp': parodh-an 'rasped, grated'. Sy analogy to this model modern Madurese acquired the alternation t '" dh in buw~t : buwá'dh-á'n.

The following list shows some of the doublets. In some instances the form exhibiting a stop only occurs before a suffix and in some instances it only occurs before pause:

When only a Madurese form with a final stop can be found to continue a PMJ form with fina1 *p, *t, or *k, it wi11 be assumed that this form has rep1aced an older form with a fina1 q if there is no evidence that this form is borrowed: e.g., Mad. erop 'to interchange, confound' from *qurup, p. 117, is treated as a cognate.

2.10.2.1. ~

2.10.2.1.1. PMJ P in initia1 position:

*panah (PAN, Fi. vana 'to shoot': panaqa), Snd. OJv. panah, NJv. panah (L), Mal. panah, Mad. pana 'archery, bow'.

*panas (PAN, To., Fu. ma-fana 'warmth'), Snd., OJv., NJv., Mal.,

Mad. panas 'warm'.

2.10.2.1.2. PMJ P in intervoca1ic position: *pipiq (PWI, Mer. fifi 'cheek': pipi), Snd. pipiq, OJv., NJv., Mal. pipi, Mad. pèp~(h) 'cheek'.

*papan (PAN, Fu., Sm. papa 'plank'), Snd. papan 'plank, board', OJv. papan 'shield', NJv. pa pan 'plank' (0), Mal., Mad. papan 'plank, board'.

2.10.2.1.3. PMJ P in preconsonanta1 position:

See the section on PMJ doub1ed monosy11ables, p. 194 .

2.10.2.1.4. PMJ P in postconsonanta1 position:

*hampas, Snd. hampas 'what remains af ter squeezing out a fruit'. OJv. hampas 'sap pressed from a fruit', NJv. ampas, Mal. (h)ampas, Mad. ampas 'what remains af ter squeezing out a fru it, refuse'.

*kampi1 (PWI, NgO. kampil 'bag'), Snd., OJv. kampil, NJv. kampé'l, Mal. kampel, Mad. kampel 'large enve10pe-shaped matwork-bag'.

2.10.2.1.5. PMJ P in fina1 position:

*qucap (PAN, Fi. v-osa 'to speak, word': ?ucap, ?uncap), Snd. qucap, OJv. aV-ucap, NJv., Mal. ucap 'to utter, speak', Mad. pan-ócaq, pan-~cap 'saying'.

*hiRup (PAN, U. i1u 'to sup (yam soup)': S4iRuP), Mal. (h)irop, Mad. ~rÓq 'to drink'.

*qurup (PWI, TBt. qurup 'to help': u[r]up), Snd. qurup 'to barter', OJv. korup (ka+urup) 'confounded', NJv. uróp 'to barter ' , Mal. urop 'money-changing', Mad. ~rop 'to interchange, confound'.

2.10.2.2. PMJ t

2.10.2.2.1. PMJ t in initia1 position:

*tapak (PAN, Fi. tambak-a 'to put the hands'0n, s.t.': tapak, tampak), Snd., OJv. tapak, NJv. tapaq 'trace (foot, finger, etc.), print', Mal. tapaq 'palm (of hand), sole (of foot)', Mad. tapaq 'palm, sole'.

*timah (PWI, TBt. tima 'tin'), Snd., OJv., NJv., Mal. timah, Mad. térna 'tin'.

2.10.2.2.2. PMJ t in intervoca1ic position:

*kutuq (PAN, Fi. kutu 'louse': kuCu?a), Snd. kutuq 'parasite' , NJv. kutu 'wonns', Mal. kutu, Mad. k~t~(h) 'louse'. *putus (PAN, Sm. motu 'to break off'), Snd. putus 'done with', OJv. putus 'sett1ement, payment', NJv. put6s 'decision', Mal. putos 'severance, of the settlement or end of a dispute', Mad. p~t~s 'decided, done with'.

2.10.2.2.3. PMJ t in preconsonanta1 position:

See the section on PMJ doub1ed monosy11ab1es, p. 194.

2.10.2.2.4. PMJ t in postconsonanta1 position:

*qantiq, Snd. qantiq (l) 'to wait for', OJv. man-anty-a 'in order to wait', NJv. anti-anti (l), Mal. n-anti, Mad. B. ant~ 'to wait for'.

*runtuh (PWI, NgD. runto 'to crash': [r]untuh), Snd., OJv. runtuh, NJv. runt6h (l), Mal. runtoh, Mad. ronto 'to crash, topple down' •

2.10.2.2.5. PMJ t in fina1 position.

*hapit (PAN, Sa. ~i 'to carry under the ann': h3apit, h3ampit), Snd. hapit 'part of a loom, to pressure between two disconnected surfaces' , OJv. aa-hapit, apit 'pinched' , NJv. apét

'to have in between, part of a loom', Mal. (h)apet , 'pressure between two surfaces', Mad. apeq 'part of a loom', ap~t 'to have in between'.

*Buat (PAN, Fi. vuat-a 'harvest': buhat), Snd. di-buat 'to harvest'; OJv. wwat 'what is offered', NJv. wdt 'load', / Mal. buat 'to perform, make', Mad. buwaq 'to be able to hold (a load), buwádh-án 'the load'.

*lihat, OJv. lihat 'look, see!', NJv. liy~t mukaq 'to look straight ahead', Mal. lihat, Mal. Kel. liat, Mad. a-l~yaq-lèyaq 'to see, look'. M , *gallut ,NJv. galot, Mal. galot, Mad. ghalluq 'to fight'.

2.10.2.3. PMJ k

2.10.2.3.1. PMJ k in initial position: *kikir (PHN, Tag. k{:kil 'file'), Snd., OJv. kikir, NJv. kikér, Mal. kiker, Mad. k~k~r 'file, grate'.

*kiraq (PWI, TBt. hira 'approximate': ki[r]a), Snd. kiraq (L), OJv. a-kira-kira, NJv. kir~ (L), Mal. kira, Mad. k~ra(h) 'to estimate, calculate, think out'.

2.10.2.3.2. PMJ k in intervocalic position:

*pukiq (PAN, Fi. mata-vuki 'ulcer at the sole of the foot': puki[?h]), NJv. puki (0) (VL). Mal. puki (vulgar), Mad. p~ké(h) 'female genitals'.

*kukuq (PAN. Fi. kuku- 'claw, nail': kuS2kuS2)' Snd. kukuq 'claw, nail'. OJv., NJv. kuku 'nail'. Mal. kuku 'claw', Mad. k~k~(h) 'nail'.

2.10.2.3.3. PMJ k in preconsonantal position:

*raksuk, Snd. raksuk 'to put on clothes', OJv. ma-rasuk

'armoured', NJv. rasÓq 'to c10the'. *ruksak (PAN, Fi. rusa 'spoi1ed, destroyed': rusak), Snd. ruksak, OJv. rusak, NJv. rusaq, Mal. rosaq 'spoi1ing, ruining'. See a1so the section on PMJ doub1ed monosy11ab1es, p. 194. 2.10.2.3.4. PMJ k in postconsonanta1 position: *pankuq (PHN, Tag. paUkó 'carried 1ying on two arms in front': pa[n]ku), Snd. pankuq 'to ho1d with the hands, carry in the arms', OJv. p-in-ankw-akan 'held in the lap, NJv., Mal. pa~ku, Mad. pa~kd(h) 'to take in the lap'. *ti~kah (PAN, Fi. teUge 'to walk on one's toes': tikaq, tiUkaq), Snd. OJv., NJv. tinkah 'manner, way', Mal. tinkah 'musical ~ode, character, ways', Mad. tènka 'manner, ways'. 2.10.2.3.5. PMJ k in fina1 position: a) PMJ fina1 kaf ter PMJ 3: *hapak, Snd. hap4k, OJv. (h)apak, NJv. apak, Mal. (h)apaq, Mad. apak 'musty, frowny'. *qutak (PAN, Fi. quto 'marrow': q3utak, q3untak), Snd. qut4k, OJv. utak, NJv. utak, utaq (0), Mal. utaq, otaq, Mad. ~tak 'brains'. b) PMJ fina1 kaf ter vowe1s other than PMJ a: *qanak (PHN, Tag. qanák 'offspring': anak), Snd. qanak (L), OJv. anak, NJv., Mal., Mad. anaq 'chi1d', par-anagh-'n 'of mixed race'. *tasik (PAN, Fi. tazi 'ocean': [t~]asik), OJv. tasik, NJv. tas~q (B) 'sea', Mal. taseq 'lake', Mad. tas~q 'sea'.

2.11. ~. PMJ ~ is reconstructed in initia1, intervoca1ic, and postconsonanta1 position. It became Sundaneseand Malay t, and appears as t in 01d and New Javanese and in Madurese.

Haudricourt (p. 312f.) states that Oempwo1ff's PAN ~ on1y exists in Javanese and that it occurs se1dom and only in expressive words. He therefore suggests to ignore the distinction between *t and *~ for Proto-Austronesian which was suggested by Oempwo1ff. However, the existence of at: t distinction is not 1imited to Javanese, but it a1so exists in Madurese. Furthermore, a1though ~ occurs se1dom it does not only occur in expressive words as the comparisons 1isted be10w wi11 show. Oyen (1971:30) argues against Haudricourt's conc1usions when he says "although the evidence for *~ (Oyen writes *T) is extreme1y 1imited, it shou1d not be ignored ... "

2.11.1. PMJ ~ in initia1 position:

*~uku1 (PHN, Tag. tu:ko1 'overripe rice-grains': tU(!l)ku1), NJv. ~u~l 'to germinate, grow', Mal. tuko1 'sprout'.

2.11.2. PMJ t in intervoca1ic position:

*ba~~ak, OJv. ba!ak 'to cook', NJv. ba~ak, ba~aq 'to cook' (0), to prepare rice' (0) (H), Mal. bataq in juru bataq 'caretaker, cook to distinguished peop1e'.

NJv. ba~a!l' Mal. batau, Mad. bhá!aV 'dead body'.

*ba!a~ (PWI, TBt. batan 'coffin'), Snd. ba-batan, OJv. ba-ba~an'

*bi~iq, Snd. bitiq 'to hit s.o.'s 10wer leg with the foot',

NJv. bi~i 'to punch', Mad. bhi!è(h) 'to hit'.

*ki!aq (PAN, Fu. fe-kite 'to see each other again': kita), NJv. was-ki!a 'clear insight, foresight'.

*la~ak, NJv. 1a~aq (0), Mal. 1ataq 'dregs, lees, oi1 refuse' ,

Mad. la~ak 'mixture of oil and lye'.

*pa~iq (PAN, To. matsi 'essence': pa~i), Snd. ci-patiq, NJv. pati 'coconut-milk', Mal. pati 'cream or pick of finest portion of anything', pati santan 'cream of the coconut', Mad. pat~(h) 'coconut-milk'.

2.11.3. ~ in postconsonantal position:

*can!in' Snd. canti~, NJv. can!é9, Mal. Java canten, Mad. can!~~ 'scoop' .

*can!u~ (PWI, TBt. par-sottin-an 'temples'), NJv. can~ó~ 'curled hairlocks at the temples', Mal. canto~ 'bird's erectile crest or tuft of feathers on neck', Mad. can~oD " 'forehead of animals'.

*qin~il, Snd. S.-B. qintil, NJv. f " in!el, Mad. en~el 'kind of pastry' .

*kcln~al, NJv. kan~al, Mal. kantal, Mad. kan~al 'thick (of liquid)'.

*pan~aq, Snd. pantaq 'step, grade, rank, order', OJv. ma-pan~a

'grouped' , NJv. p~ntb, Mad. panta(h) 'group'. . . *pan~u~, NJv. pan~ó9' Mal. pant09' Mad. ·pan~on 'to club'.

*pantil . 95 , NJv. panté'l 'litt1e button, knob', Mal. pantel susu, . Mad. pant~l 'nipp1e, teat'.

2.11.4. Ambiguous reconstructions.

2.11.4.1. PMJ [t~J in initia1 position. An ambiguous reconstruction with initia1 *[ttJ is emp10yed when a Javanese and a tiadurese cognate are 1acking and both Sundanese and Ma1ay exhibit t: e.g., *[t~Juhur, p. 68.

2.11.4.2. PMJ [t~] in intervoca1ic position. The conditions for the reconstruction with *-[t~]- are the same as those stated in the preceding section: e.g., *[Bb]a[tt, . tt]un, p. 79. . ;,)

2.12. PMJ Band PMJ b. PMJ B is reconstructed in initia1, intervoca1ic, and postconsonanta1 position. PMJ b is reconstructed in initia1, intervoca1ic, postconsonantal, and fina1 position. The essentia1s of the correspondence 1eading to the reconstruction of *B and *b were first noted by Ki1iaan (1897:1.61) who suggested that Javanese w corresponds to Madurese band that Javanese b corresponds to Madurese bh. Since this matter has not been treated e1sewhere, the evidence is presented in extenso.

PMJ B in initial position became Sundanese, Ma1ay, and Madurese b, and 01d and New Javanese w. In intervocalic position it became Sundanese w, except when it occurred af ter u and before a different vowe1, in which case it disappeared. It became Old and New Javanese w, and Malay and Madurese b. In postconsonantal position it became b in all four languages.

PMJ b in initial, intervocalic, and postconsonanta1 position appears as Sundanese, Malay, and Old and New Javanese b. lts Madurese reflex is bh. In final position it appears as Sundanese and Old Javanese band Malay and Madurese p. New Javanese has different dialectal reflexes as in the case of final *d and *g. In the Western dialects it appears as b; in the Central dialects it appears as p before pause, otherwise as b. In the Eastern dialects it a1ways appears as p (see section 1.5.2.).

The phoneme inventory of PAN as proposed by Oempwolff contains only *b. Oempwolff explains the fact that his *b sometimes became Javanese band sometimes Javanese wand the fact that doublets occur as a 'Tendenz zur Lautverschiebung' (1934:90). In his study of the Madurese reflexes of PAN Stevens (1966: 156) comes to the conclusion that Madurese bh as reflex of *b probably Is the result of borrowing, Madurese b being the regular reflex.

2.12.1. PMJ B

2.12.1.1. PMJ B in initial position: *BaliBis96 , OJv. waliwis, NJv. m{a)liwés, Mal. balibes, Mad. bálibis 'teal'. *BaR2aq (PHN, Tag. bá:ga 'charcoal': 00 baRa, 10 [1965a] baRaH), Snd. baraq, OJv. wä, NJv. w~-w~ (O) CO), Mal. bara 'live coal'. *Barun97 (PHN, Tag. bá:lon-bá:lon 'hut':barun), OJv. warun-warun 'houses built by the army', NJv. war6~ (L) 'small shop', Mal. baron 'booth, stall', Mad. bárun 'small shop'. *Batas98 (PHN, Tag. bá:tas, 'short, direct course', b~:tis I 'brook': batas), NJv. watas, Mal. batas, Mad. batas 'boundary, limit, frontier'. *Ba!ah99 (PHN, Tag. bá:laq 'threat': bajah), OJv. ma-warah, NJv. warah, Mad. bál' 'to announce, say'. *Batuk (PWI, TBt. batuk 'to cough': batuk), Snd. batuk (L) 'ta cough', OJv. api-watuk 'ta do as if ane was caughing', NJv. watóq, Mal. batoq, Mad. bátoq 'to caugh'. *Baur, Snd. baur, OJv. ma-wor, NJv. w~r (B), Mal. baor, Mad. báur 'ta mix'. *Bawaq (PAN, Sa. haha 'to carry on the back': baba[?h]), Snd.

bawaq (L), OJv. a-mawa, NJv. wowo (B), m~w~, Mal. bawa 'to bring a1ong' •

*Baq~ak, OJv. a-wa9ak-waqak 'to powder', NJv.wa9aq, Mal. badaq, Mad. baqqhiq 'face-powder'. *Banna0102 (PWI, TBt. mona~ 'to win': bana~, manan), Snd. b4na~ (L) 'to want, get, win, can', OJv. wanaD 'authorized, win, can', NJv. wana~, wa-wanaD 'power and right over, have the power and the right to', Mal. manaU 'to prevail, win', Mad. banna~ 'authorized, permitted, allowed'.

*Ba11ah (PAN, Fi. mbo1a 'to split': ba1aq), Snd. b41ah 'to split'., OJv., NJv. wa1ah 'oar', Mal. balah, Mad. ba11á 'to split'. *Ba11as100, OJv. walas, w1as, NJv. walas, Mal. balas, Mad. ba11ás 'pity, mercy'.

*Balliq (PAN, Sa. holi 'to buy': ba1i[?h]), Snd. b41iq (L) 'to buy', OJv. wa1i, w1i in w1i hapu 'lime merchant', wa1i hara~ 'charcoa1 merchantJ , ma-mali 'to buy', Mal. bali, Mad. ba 11 i (h) 'to buy'.

*Ba11ut101 (PHN, Tag. bf:1ot 'small ro11': balut), OJv. walut, NJv. wa16t, Mal. balot, Mad. ba11uq 'ee1'.

*Ba11it (PAN, Fi. ve1i 'coi1ed': ba1it), Snd. b41it 'to wind around', OJv. pa-wa1it 'the man who makes roofs', a-ma1it 'a bound roof', NJv. wa1ét 'roof-cover (palm-leaves, straw, bound together with laths to form long bars', Mal. balet 'coil, turn, hitch (of rope), Mad. balliq 'roof-cover', ballit 'to wind around'. OJv. a-wilat 'to curl, wind around', NJv. wilat (0) 'to turn' exhibits metathesis.

*BaR1Rlasl03 (PHN, Tag. big{s 'hulled rice': baRas), Snd. beas (L). OJv. waas. wwas. NJv. w~s (H). Mal baras 'husked rice'. *BaR2R2ay (PAN. Sm. fo-aqi 'to give': baRlay), Snd. bereq (L) 'to give'. OJv. w-in-eh 'be given'. NJv. w~_w~hl04 (L), Mal. bari. Mad. barriql04 'to give'. *Bilahl05 (PWI, NgO. bila 'strip of bamboo': bilah), OJv. wilah 'floor of bamboo-laths (?)', NJv. wilah, Mal. bilah, Mad. bilt 'lath, strip of bamboo'. *BiRah (PAN, Fi. via 'alocasia': biRaq), OJv. wyah, NJv. wèw~h-an, Mal. birah, Mad. birá 'kind of plant'.

*Batta~ (PAN, Fi. mboto ni kete 'lower belly': batan, bantan), Snd. btttn (L), OJv., NJv. watan 'belly'.

*Battuq (PAN, Fi. votu 'to appear': batu), Snd. bituq 'to fire (of a weapon), explode, erupt (of a volcano)', OJv. watu, NJv. watu (L) 'to come out, appear', Mad. batt~(h) 'product, result' .

*Binih (PHN, Tag. binh{q 'seed': binih). Snd. binih, OJv. winih 'seed'. NJv. winéh 'seed' (0), winèh (0), Mad. binè 'seed'. A competitive form with penultimate *a is *Bannih (PWI, TBt. boni 'semen': banih), Mal. baneh 'seed'.

*Birin. NJv. wiréD, wir~D (0) or wiré9, wirèO kuné3 'red cock with yellow legs', Mal. bire!J 'bright red and yellow; fiercelooking (of the color of a fighting cock, especially about the legs)', Mad. birin 'red (of chicken), birin konèn 'with yellow and light-red feathers, a yellow beak and yellow legs' .

*Bassiq (PAN, Fi. vesi 'kind of spear' : basi, bansi, basi), Snd. btsiq, OJv. wasi, NJv. was i (L), Mal. basi, Mad. basse(h) 'iron'.

*BU~UR2107, Snd. bu~ur, OJv. wu~ü, NJv. wu~u, Mal. bU90r 'kind of tree'. *BusuR (PAN, Fi. vu~u 'the ends of a bow': busl uR12 d), OJv. wusü, NJv. wusu, Mal. busor 'cotton cleaner's bow'. *Buti[rR] (PAN, Sa. uhi 'wart': bu[t~]iR123)' Mal. buter 'grain, particle', Mad. butèr 'spilled crumbs of rice'. See also *Balilaq, p. 55; *BaR2iw, p. 86; *Bauk, p. 64; *Battis and *Bantis, p. 193; *BuBu~ and *BumBu.'J' p. 190; and *Buz[aa]l, p. 110. Javanese of ten exhibits doublets, one beginning with w, the other

*Buah (PAN, Fi. vua 'fruit': buaq), Snd. buah, OJv. wwah, NJv. w6h, Mal. buah, Mad. buwá 'fruit'.

*Bukil (PAN, Sm. puqe 'heap'), OJv. wukir, NJv. wukér (B), Mal. buket 'hill'.

*Bukuq (PAN, Fi. mbuku 'corner': buku[?h]), Snd. bukuq 'knuckle, knot, joint', OJv. wuku 'grain, joint', NJv. wuku (0) 'small pith', wuku 'joint' (0), Mal. buku 'knuckle, knot'; metaphorically: 'pith', Mad. buk6(h) 'joint, grain'.

*BuliR (PHN, Tag. bu{g 'bunch, cluster': buliR), OJv. wulihl06 , NJv. wuli in sa-wuli, Mal., Mad. bulir 'ear (of grain)'.

*Buluh (PAN, Sm. polo 'knife of bamboo': buluQ12a), Snd. buluh, OJv. wuluh, NJv. wulóh, wulOh (0), Mal. buloh, Mad. bulu in parr~D bulu 'kind of bamboo'.

beginning with b. The existence of these competing forms can be attributed to 10ans from a dialect or 1anguage in which *B became b or to a preceding nasa1. When initia1 *B was preceded by a word the fina1 consonants of which was a nasa1, *B became Javanese b; e.g., OJv. wa1as 'requita1, repay', kam-ba1as 'what we can do to repay it'. The fo110wing examp1es are instances in which New Javanese exhibits doub1ets:

*Baku1 108 (PHN, Tag. bá:ko1 'bamboo-basket': baku1), Snd. bakul, OJv. wakul, NJv. wak6l, bak6l (0), Mal. bakel 'basket'.

*BalaD109 (PHN, Tag. bá:laO 'locust': bala~), OJv. walaD' balaD' NJv. wala~, Mal. b-al-alaU' Mad. bala~ " " 'grasshopper'.

*Balik (PHN, Tag. bal{k 'return': balik)llO, Snd. balik 'to return', sa-balik-na 'on the contrary, on the other hand', OJv. walik 'reversed', balik 'wrong, again, instead of', NJv. waléq 'upside down, inside out, turned', baléq 'on the contrary, on the other hand', Mal. baleq 'reversal, going back', Mad. báliq 'to return, turn over'.

*Balaslll (PWI, NgO. baleh 'requital': balas), OJv. walas, NJv. (wa-)walas (0), walas-an 'requital', walas-an (0), balas-an (0) "reply', Mal. bal as, Mad. btl as 'requital, reply, repayment'.

*Banti9ll2, Snd. bantiD 'to dash against', OJv. ka-wantiu 'thrown down', am-bantiD 'to throw down', NJv. wantén 'to beat out and wash (clothes), bante9 " 'to throw down, dash against', Mal. banteD 'to dash against'.

*Batuq (PAN, Sa. h~u 'stone': batu? ), Snd. batuq, OJv. watu, NJv. watu (L) 'stone', batu in batu rai 'flat-roofed stone', Mal. batu, Mad. b{tO(h) 'stone'.

*Baya~113 (PWI, NgO. baya~ 'a top': bayan), OJv. wayan' NJv. waya~ (l) 'shadow-play', baya~-an (0) 'shadow (-picture)', Mal. bayaD' Mad. já9-blj1~-an 'shadow, vague outline'. *Ba~Uisl14 (PAN, Sm. pODi 'sullen': baUis), OJv. a-wanis 'furious', I ( / NJv. wa~es B) 'upset' , ba~es 'angry, malicious', Mal. ba~es 'cruel, heartless'. *Bisikl15 (PWI, Mer. bitsik~116 'towhisper': bisik), OJv. ma-wisikwisik 'to whisper', bisik-bisik 'whispering', NJv. wis6q 'secret instruction, biséq-biséq, ba-biséq, Mal. biseq, Mad. bis~q 'to whisper'. *Buitl17 , Snd. buit in bibit buit 'to originate from', OJv. wwit, wit-an 'beginning' , NJv. wet 'stem, stalk, origin, reason', NJv. bi-bet 'to originate from'. *Bulan (PAN, Fi. vula 'moon': bulala), Snd. bulan (l) 'moon, month' , OJv. wulan 'moon', NJv. wulan 'moon' (0), wulan (H) 'month', 'date' (0), bulan (0) 'moon', Mal. bulan, Mad. / bulan 'moon, month'. *Buluq (PAN, To., Fu., Sm. fulu 'hair': bulu[?h]), Snd. buluq (l), OJv. wulu, NJv. wulu 'body-hair', bulu-bulu 'feather', Mal. bulu, Mad. bulu(h) 'hair, plumage'.

In a few instances Sundanese exhibits doublets, one beginning with b, the other beginning with w. The member of the doublet which exhibits initial w is treated as a loan word from another language, probably Javanese:

*Balikat (PHN, Tag. bal{:kat 'shoulder':balikat), Snd. balikat, walikat 'shoulder-blade', OJv. walikat-an 'stiff', NJv.

walikat (0), w1ikat 'shou1der-b1ade', 1ikat-an 'to have a musc1e cramp', Mal. balikat in tu1a~ balikat 'scapu1a, shou1der-b1ade', Mad. bá1ikat 'shou1der-b1ade'.

~Bi1aD (PHN, Tag. b{:laO 'to count': bi1aO)' Snd. bi1aO (L), wi1a~ 'to enumerate', OJv. w-in-i1aO 'enumerated', NJv. wi1aO (L), Mal. bi1a~, Mad. bi1{n 'to enumerate'.

*Butaq (PWI, NgD. but~ 'blind': buta), Snd. butaq, wutaq, OJv. wuta, NJv. Wüto (B), Mal. buta, Mad. buta(h) 'blind'.

When both Javanese and Madurese exhibit competing forms, i.e. Javanese W-, band Madurese b-, bh-, PMJ initia1 B is reconstructed. As stated above, the existence of the Javanese competing forms can be attributed to borrowings from a dialect or 1anguage in which PMJ initia1 B became b or to a preceding nasa1. Presumab1y, the Madurese by-forms with initia1 bh's are borrowings from a 1anguage such as Ma1ay in which PMJ initia1 B became b. Madurese of ten borrows initial //. " // b from another 1anguage as bh. Examples are bharat 1n aDen bharat 'west-monsoon' , but see Mad. bár{q 'west' under *Barat, p. 132; or bhunt6t in ap6s bhunt6t 'tai1-strip', but see Mad. bunt6q 'tai1' under *Buntut, p. 132:

*Banunl18 (PAN, To., Fu., Sm. fa-fanu 'to make up': baOun), Snd. baDun 'occurrence, to erect', OJv. waOun 'was erected', / ba~un 'occurrence' , NJv. wa~on 'model, form, figure', / banon 'to bui1d', Mal. baDon 'risen posture, usual bui1d', / \ / , Mad. bhaOon, Mad. B. ba~on 'form, figure'. *Ba11aOl19 (PWI, TBt. bo1a~ 'spotted': balan), OJv. wa1aD 'spotted', NJv. walan in u16 (L) wa1a~ 'kind of snake', bala~ 'spot on the skin, scar', balau-buntal 'spotted (particularly said of animals)', Mal. balaO 'banded, brightly marked in contrasting colors', Mal. Java ular balaO 'kind of snake', Mad. B. ballán in Ólar balliD 'kind , / of snake', Mad. S.P. olar bhalla~ 'kind of snake'.

In a few instanees Old Javanese exhibits forms with initial w with or without a competing form with initial b, whereas New Javanese exhibits only a form with initial b. For most of these instanees there is a Malay word of the same shape available as a souree for explaining the Javanese form with b as being a borrowing:

*Binay (PAN, To. fine 'wife': binay), Snd. be-beneq tfiancee', OJv. wini, bini, NJv. bini (B), Mal. bini, Mad. binè(h) 'wife', binéq, bi-binéq 'of female gender'.

*Bintau (PWI, TBt. bittau 'star': bintan), Snd. bentau, OJv. wintan, NJv. bintan' Mal. bintau, Mad. binta~ 'star'.

There are two instanees in which such an explanation is not available. In the first there is a difference in meaning, in the second there is a difference in form:

*Batak120 (PHN, Tag. bá:tak 'pull': batak), OJv. watak, NJv. batak 'to pull(out)', Mal. bataq 'to rob, plunder'. *BaR2R2at (PAN, To., Fu., Sm. mama-fa 'heavy': baR2qat), Snd. b4rat (L), OJv. ma-wwat 'become heavy', bwat 'very, weight' a_bot121 , NJv. a-b~t (L) 'heavy, difficult', Mal. barat, Mad. barr{q 'weight, heaviness'.

When both Old and New Javanese exhibit an initial b without a by-form with initial wand Madurese exhibits an initial band a sound change which indicates that the word is inherited, Pfol·J initia1 B is reconstructed: *Ba1anak (PWI, NgO. b~lanak 'kind of fish': ba1anak), Snd. balanak, OJv. ba1anak, NJv. b1anaq, Mal. ba1anaq, Mad. bá1ánaq 'kind of fish'. *Barat (PAN, Sm. afä 'storm': ha-baR4ata), Snd. Bant. barat 'west', OJv. barat, NJv. barat (0) (0) 'west-monsoon', Mal. barat, / / Mad. baraq 'west'. *Buntut (PHN, Tag. buntót 'tai1 ': buntut), Snd., OJv. buntut, NJv. buntót, Mal. buntot, Mad. buntöq 'tai1'. See a1so *Bannak, p. 184. New Javanese exhibits the 10ss of an initia1 w from PMJ B when it occurs before u in the fo110wing comparisons: *BaR3haO (PHN, Tag. bagán122 'mo1ar tooth': 00 baRan' 10 [1953b] baR2qan), OJv. wahan' NJv. (u)waU 'molar'. *Bunka1, OJv. wuDka1 'stone', NJv. wunka1 'flat grindstone', uDka1 'to grind', Mal. boOka1 'lump, measure of weight, a stone', Mad. buOkal in bátö bu~kal 'grindstone'. 2.12.1.2. PMJ B in intervoca1ic position: *haBis, OJv. (h)awis 'exterminated, up, finished', Mal. (h)abis 'done with, all 'used up', Mad. Kang. ma-q-abis 'to use up, consume'. *haBuk (PAN, Sa. ehu-ora 'dusty': ?abuk), Snd. hawuk, OJv. hawuk-hawuk 'ash-gray', Mal. (h)aboq 'dust'. *haBuq (PAN, To. efu 'ashes': q2abu?a), Snd. hawuq 'furnace', OJv. hawü, NJv. awu, Mal. (h)abu, Mad. abu(h). *guBal, Snd. gual, NJv. kuwa1, Mal. gubal 'sapwood'.

*qiBaR (PAN, To. ifo 'sa1iva': ?ibaR), OJv. um-iwö 'to take care, do one's utmost', Mad. ébar 'sa1iva'.

*kaBu~, Snd. kawuD 'sugar-pa1m', NJv. kaw6U 'pa1m-1eaf' (0), Mal. kabon 'sugar-pa1m'.

*kuBuq (PHN, Tag. ku:bo 'hut'; kubu), Snd. Cir., Ind., Sumo kuwuq, OJv. a-kuwu, NJv. kuwu (0) 'head of a village', Mal. kubu 'stokade, semi-permanent fortification'.

*luBa~ (PHN, Tag. 1u:ba~ 'to plant root crops': 1uba~), Snd. 1ua~, OJv., NJv. luwaU' Mal. lubaU' loba~, Mad. 10bá~ 'ho1d'.

*raBun, Snd. rawun 'co11ection of things which are burnt', NJv. raw6n (0), Mal. rabon 'fumigated'.

*saBu~ (PHN, Tag. sá:boD 'cockfight': sabu~), OJv. sawu~ 'fighting cock', man-awu~ 'make fight', NJv. saw6~ (H) 'fighting cock', 'chicken' (0), Mal. sabo~ 'to fly at each other', ayam saboD' Mad. ajtm sabuD 'fighting cock'.

*saBut (PWI, NgO. sawut 'coconut-fiber': sabut), Snd. sawut 'cover of a fruit-pit', Mal. sabot 'fibrous she1l, husk, coir', Mad. sabuq 'fibrous she1l'.

*siBur, Snd. siwur, OJv. siwur-an, NJv. siw6r, Mal. sibor 'a sha110w scoop of coconut-shel1'.

*suBa~, OJv. s-in-uwdU 'rolled up', NJv. suwa9 (L), Mal. suban' Mad. sbba~ 'ear-stud'.

*taBur (PWI, NgO. tawor 'to strew': tabuR), Snd. tawur 'to strew', OJv. tawur 'offering', NJv. taw6r 'to strew money (offer)' (0) (0), 'offering, sacrifice', Mal. tabor, Mad. tabur 'to scatter, sow'.

*huBan (PHN, Tag. qu:ban 'white hair': 00 uban, 10 [1953a] quban), NJv. uwan, Mal. (h)uban, Mad. Ob'n 'gray (of hair)'.

In a few instances New Javanese exhibits competing forms, one with intervocalic w, the other with intervocalic b. These competing forms can be attributed to loans from a dialect or language (such as Malay) in which PMJ intervocalic B became b:

*riBut123 (PAN, To. faka-lifu-lifu 'to have a shivering fit': ribut), OJv. riwut, NJv. riwót (0) (0), ribót 'storm, stormy', Mal. ribot 'strong and, 3torm', Mad. B. rèbut 'storm'. *quBah124 (PWI, Mer. uva 'changed': ubah), Snd. qowah 'subject to change', OJv. ma -uwah-i 'to bring in order again', mowah (=ma+uwah) 'again', NJv. 6wah (=a+uwah) (L) 'other, ~ changed', obah (L) 'to move, movement' , Mal. ubah, obah 'to change, alter, modify', Mad. oba125 'to exchange, interchange'.

In three instances Javanese exhibits only b as reflex of PMJ B. These forms probably are borrowings from a dialect or language in which *B became b. In the first two instances a Malay word of the same shape is available as a source for explaining the Javanese forms as borrowings:

*taBBuq126 (PAN, Sa. e-ohu 'sugar-cane': tabuSa), Snd. tiwuq, OJv. tabü, NJv. tabu (L), Mal. tabu 'sugar-cane'.

*tuBaq127 (PAN, Sa. uhe 'fish-poison': tuba?), Snd. tuaq 'name of a root with stupefying effects (much used to stupefy fish)', NJv. tuba, Mal. tuba, Mad. tÖbá(h) 'fish-poison'.

In the following comparison it is not possible to explain the Javanese form as a loan from Malay:

*R13BBUn128 (PAN, Fi. rovu 'shoot': rabun), Snd. qiwuD' OJv. bu~, / NJv. bOD' Mal. raboD 'bamboo-shoot when young, soft and edible'.

2.12.1.3. PMJ B in postconsonantal position: *qamBat (PHN, Tag. qabát 'waylaying': a[m]bat), Mal. ambat129 'to obstruct', Mad. ambáq in báq~ambáq-~ 'to lie in wait'. *qamBuk, NJv. ambèq (0), Mal. amboq, Mad. ambuq 'mother'. *k3mBuD130 (PHN, Tag. kubÖn 'covering blanket, shawl': ka[m]bun), / NJv. kambon (0), Mal. kamboD' Mad. kambuD 'puffed up, 'swollen'. *kumBan' Snd., OJv. kumbaD' NJv. kumban (0), kÖmbaD' Mal. kumban' , / Mad. komban 'a large bee'. *lamBat (PWI, TBt. lambat 'be slow': lambat), Snd. lambat (L) 'to last long, long', NJv. lambat (L) (0) 'old, from long ago', Mal. lambat 'slow, take time', Mad. lambtq (L) 'earlier, former, a long time'. *lamBah, NJv., Mal. lambah, Mad. lambá 'meadow-land, low-lying land'. *rimBit, OJv. rimbit 'obstructed, hindered', NJv. sa-rimbét (0) 'man and wife', Mal. rimbet, rembet 'hampered, obstructed, having encumbrances; of a person with many dependents or impediments', Mad. rémbiq 'to expect a baby, have a baby', Mad. B. 'obstructed by what one takes along'. *sumBiq, Snd. sumbiq, NJv., Mal. sumbi, Mad. s~mbi(h) 'a rod that holds the cloth taut at the end of the loom'. *tamBir (PAN, Fi. qi-tambi 'flat basket': tabi[rR4]), Snd. tambir, NJv. tambér, Mad. S.P. tambir 'to hold close to the edge'.

*timBaq (PHN, Tag. timbáq 'pai1, bucket'; timba), Snd. timbaq, NJv. timb6, Mal. timba, Mad. tèmbá'(h) 'hand-bucket, dipper'. See a1so *BarBar, p. 200; *Ba1Ba1, p. 196; *limBay, p. 92;

*[rR2]amBay, p. 92; *tumBu~, p. 68.

2.12.2. PMJ b. The correspondences are Snd. b- -b- -b, OJv. b- -b- -b, NJv. b- -b- -b, Mal. b- -b- -p, and Mad. bh- -bh- -p (see section 2.12.):

2.12.2.1. PMJ b in initia1 position: *bakal I (PAN, Fi. vaDga 'stock': bakal, baDkal), Snd., NJv. bakal (L), Mal. bakal, Mad. bhákal 'raw material'. *bakal II , Snd. bakal (L) 'to be designed for, intended for', ba-bakal-an 'to be engaged', OJv. bakal 'fiancee', NJv. bakal-an 'fiancee', Mal. Java, Pal. bakal 'future, intended for', Mad. bhákal 'intended for'. *bakuD (PHN, Tag. bá:koD 'spider lily'), Snd., OJv. bakuD' NJv. bakó'~, Mal. bakoD' Mad. bhákÖD 'kind of plant'. *baluD' Snd. ba1uO 'bone (of animal)', OJv. baluD' NJv. balóD / (L), Mad. B. bhaluD 'bone'.

131 / . *banda9 ,OJv., NJv. bandélD' Mal. banda~, Mad. bhandha9 'klnd of fish'.

*ban~iD' Snd. bandiO' Mal. banden' Mad. bhán~hiO 'to compare, match' .

*banta 1 (PWI, TBt. battal' pill ow' ), Snd. bantal (H) 'pill ow' , ba-bantal 'small cushion, cushion to sit on', OJv. bantal 'cushion', NJv. bantal (H) 'pillow', bantal 'cushion', Mal. bantal 'cushion', Mad. bhántal 'pillow'.

*baR1ah (PHN, Tag. bá:gaq 'lung,)133, Snd. bayah, Mad. bhárá 'lung' • *baran (PWI, TBt. bara~ 'goods': ba[r]a~), Snd., NJv., Mal. bara~, Mad. bhárá'!) 'thing, stuff, goods'. *baris (PWI, TBt. baris 'row': ba[r]is), Snd. baris, NJv. bar{s, Mal. bares, Mad. bh{ris 'straight line, row'. *basiq (PWI, TBt. basi 's.t. extra': basi), NJv., Mal. basi, Mad. S. bhásè 's.t. thrown in, s.t. extra (for unseen expenses)' . *bataq, Snd. bataq, OJv. bata, NJv. boto (L), Mal. bata, Mad. bh{ta(h) 'brick'. *bawan (PHN, Tag. bá:wan 'garlic'), Snd., NJv., Mal. bawa~, Mad. bhel'b{!) 'on i on' . *bakkal 134 (PAN, Fi. mbakola 'people killed to be eaten'), OJv. b-in-akal 135 'be given as provisions', Mal. bakal 'stores for a journey, viaticum'. *b~ndun136, ".J NJ v. b~ d M 1 b~ d M d B bh~ dh 'd "n u!J-an, a. "n 0!J' a. . "n u!J am, dyke'.

*bantar132 , OJv. a-bantar 'quick', NJv. bantar 'quick, speeding', Mal. bantar 'to speed up', Mad. bhántar 'strict, hard, fierce'.

*ba!)sa1, OJv. b-in-a!)sa1 'fenced in', NJv. ba!)sa1 'hall, building', I Mal. ba~sa1 'shed, shelter', Mad. bha~sal 'a house with a roof of four slanting planes on a square as basis'.

*batur, Snd. batur 'p1inth, upright course of bricks', OJv. batur, I NJv. bator 'stoop, pavement', Mal. bator 'passage, corridor', Mad. bh{t6r 'lower part of the house'.

*ban9a~, OJv. baUaU 'to open the mouth', NJv. baUaU (B) 'spread apart', Mal. baUaU' Mad. B. bhauna9 'to en1arge'.

*biruq (PWI, NgD. biro 'b1ue': bi[r]u), Snd. biruq, OJv., NJv., Mal. biru 'b1ue', Mad. bhiru{h) 'green'.

*bi9u9' Snd. biOu~ 'disheartened', OJv. binu9 'disarranged, untidy', NJv. bin6n 'not to know what to do', Mal. biDon 'mudd1e-headed', Mad. bhiDoU 'desperate, at one's end'.

*bubu1, NJv. bubó1 'swelling, swollen', Mal. bubo1 'a cracked, fissured or u1cerated condition of the sole of the human foot or in the hoofs of horses', Mad. bhubhu1 'swe11ing at the foot'.

*bu1ad137 (PAN, Sm. pu1a 'eyes': buq1at, buklat), Snd. bu14d, Mal. bulat, Mad. bhulat 'round'.

*bu1us, Snd. bulus, NJv. bUlós, Mal. bo10s-bolos, Mad. bhulus 'sea-turt1e' .

*bunta1, Snd. buntal 'a white spot or strip over the head, blaze', NJv. buntal 'variegated, mu1ti-colored', Mad. bhuntal 'with a white tail (of cow or dove)'.

*buruq, Snd. buruq 'to hurry', NJv. ka-buru-buru, Mad. ka-bhuru 'in a hurry'.

*butuh, Snd. butuh, NJv. butóh, Mad. bhut~{h) 'to be short of s. t.'.

*buyar, Snd. buyar 'to spread out, disperse', NJv. buyar 'to dash apart', Mad. bhujár 'separated, dispersed'.

2.12.2.2. PMJ b in intervocalic position: *gabusI , Snd. gabus in kayuq gabus, NJv. gab~ 'cork', Mal. gabos 'rub on a soft surface; stropping or cleaning', kayu gabos, kayu pa-gabos 'soft spongy wood', Mad. gh{bhus 'the soft, cork-like inner parts of some kind of bamboo'.

*gabusII , Snd. gabus, NJv. gabós (0), Mal. gabos in ikan gabos 'kind of fish'.

*kabar, NJv. kabar (0) (0) 'weak, thin (second or third decoction in particular of coconut milk)', Mal. Mlc. kabar, Mad. kabhär 'coconut-milk obtained by a second pressing of the flesh'.

*kabbas138, NJv. kabas (0), Mal. kabas, Mad. ghabbhás139 , Mad. S. kabbh's 'to flick, jerk away, dust with a cloth'.

*kuba~, Snd. kuban 'a water-place or pond in the flat (not in the mountains), NJv. kuban (0) 'pond, lake', Mal. kuba~ 'wallow (buffalo), .

*labur (PAN, Fi. lambor-aka 'to pulverize earth with a stick'), Snd. labur 'to pour out, flow out', NJv. labór 'whitewash', Mal. labor 'to besmear', Mad. B. P. labhur140 'lime mixed with water'.

*labbat, Snd. l4b4t 'full with fruit (of a tree)', Mal. labat 'set densely together, dense', Mad. labbháq 'full with fruit'. *rabbut14l , OJv. ma-rabut, NJv. rabót 'to fight about s.t.', Mal. rabot 'to snatch, tear at', Mad. rabbhuq 'to fight about s.t.'.

*sabbah, Snd. s4b4h 'satiated, full', NJv. sabah 'to lie heavily in one's stomaeh, boring, have enough of'.

*tabbah, OJv. t-in-abah 'be hit', NJv. tabah 'to beat out (of a mat)', 'to hit, knock' (0) (0), Mal. tabah 'to beat a flat surface with a flat object', Mad. S.P. tabbhá 'to hit S.t.

or s.o. with a flat surface'.

*tabbas (PHN, Tag. tabás 'cutting off': [t]abas), OJv. t-in-abas NJv. tabóq, Mal. taboq 'to bore a hole into s.t.'. *rabbah143 (PAN, Fi. qova 'fallen in': R3abaq, Rambaq), OJv.

'cut down', Mal. tabas 'to cut down small plants'.

*tabbuk (PAN, Fi. tombu 'hole in a river-bed': tabuk, tambuk),

*qubin142 , Snd. batuq qubin, NJv. ubén (0), Mal. batu uben 'flagstone, paving stone'.

*rubun, Snd. rubun (H) 'to encircle, be round about', OJv.

r-in-ubun 'be flown around by', NJv. rubón, Mal. ru bon 'be round about, encircle, circle'.

rabah, NJv. rabah (H), Mal. rabah 'to fall in'.

2.12.2.3. PMJ b in postconsonantal position:

*gambar (PWI, NgO. gambar 'image, portrait': ga[m]ba[r]), Snd., NJv.; Mal. gambar, Mad. ghámbhár 'image, sketch, picture'.

*gambir (PHN, Tag. gamb{l 'fresh in one's memory'), Snd. gambir, NJv. gambér, Mal. gamber, Mad. ghtmbhir 'decoction from the leaves of a plant consumed with betel'.

*gimbal, Snd., NJv. gimbal, Mad. ghimbh~l 'to stick together (of hair)'.

*kamba~ (PAN, Fu. kopa 'curly hair'), OJv. kamba~, NJv. kamba~ (L), Mal. kamban 'expansion, blossoming out', Mad. kambhán 'flower'.

144 ( /. *kambar PHN, Tag. kambal 'twln': ka[m]bar), OJv., NJv. kambar 'twin', Mal. kambar 'form a match or a pair', Mad. kambh{r 'identical in shape, twins'.

*lambiO' NJv. lambéil 'spear' (0), Mal. lambe~, Mad. lambhi9 'spear'. *lambut145 (PAN, Fi. lombo-lombo 'soft'), OJv. lambut, NJv. lambót (L), Mal. lambot, Mad. lambhuq, lambhut 'soft, fine, small' . *sambur, Snd. simbur 'to bespatter' , OJv. ka-sambur-an 'bespatted', NJv. sambór (0) 'to blow, spit', Mal. sambor 'to bespatter from the mouth', Mad. sambhur 'to bespatter, spit'. *sumbaD' Snd. sumbaO 'not true (of a eopy), Mal. sumbaD 'to revolt, ineestuous, abomination', Mad. sombhä9 'not fitting, not belonging to eaeh other'. *sumbul, Snd., OJv. sumbul, NJv. sumból, Mad. sómbhul 'lidded basket for riee'. *sumbuq (PHN, Tag. sumb6q 'light': sumbu), Snd. sumbuq, NJv., Mal. sumbu, Mad. sombhu(h) 'wiek, fuse, slowmateh'. 1 *tambak (PHN, Tag. tambak 'embankment'), Snd., OJv. tambak, NJv. tambaq 'dam, dyke', Mal. tambaq 'banking, filling, leveling up', Mad. tambhäq 'dyke, dam'. *tambaD (PAN, Fi. tamba 'side': tabaD' tamban), Snd. tambaD' OJv. man-amba~-i, NJv. tamba~-ak{, Mal. tamba~, Mad. tambhá~ 'to ferry' . *timba~ (PHN, Tag. timbá~ 'weight'), Snd. timbaD 'to weigh, eompare', OJv. t-um-imban 'to eompete' , NJv., Mal. timba~, Mad. témbháD 'to weigh out'.

*qumbal, Snd. qumbal in jalma qumbal 'freight-porter, day-laborer', NJv. umbal (0) (0) 'to hire', Ombal-an (0) (0) 'reward', Mal. Java umbal 'freight-shipment', Mad. ombh~l 'to give s.o. work as a day-1aborer'.

2.12.2.4. PMJ b in fina1 position:

*si~kab. Snd. si~kab 'to draw open'. OJv. a-ni~kab. NJv. si~kab (0) 'to open'. Mal. si~kap 'to draw open'. Mad. sèDkap 'to ro11 up (of sleeves)'.

*ta~kub (PAN. Fi. taku 'crust of a turt1e': takub. taDkub). Snd. ta~kub 'to 1ie upside down'. taDkub-an 'to 1ie on one's be11y. put over s.t. with the ho110w side'. Mal. ta~kop 'to capture under a ho110w. hudd1ed up face downwards on the carpet or burying the face in the pi110w'. Mad. S. taDkOp 'bird-cage'. *qurab. Snd. qurab 'mixture of mea1s'. NJv. urab 'to mix'. Mal. urap. 'condiment of vegetables'. Mad. ~rap 'to mix'.

2.12.3. Ambiguous reconstructions

2.12.3.1. ·PMJ [Bb] in initia1 position. An ambiguous reconstruction with PMJ initia1 [Bb] is emp10yed when both Javanese and Madurese exhibit an initia1 band there is no means to determine whether the Javanese and/or the Madurese form(s) is (are) borrowed:

*[Bb]awah (PAN. Sa. haha 'lower part': babaq). Snd. bawah 'below'. NJv. bawah 'subordinate. be10nging to'. Mal. bawah. Mad. b'b' 'be10w,146.

*[Bb]annan147 (PWI. TBt. bona~ 'thread': banaa). NJv •• Mal bana3' Mad. banna~ 'thread'.

See a1so *[Bb]ukkaq; and *[Bb]u~kaq. p. 193.

An ambiguous reconstrll~tion with PMJ initia1 [Bb] is a1so emp10yed when Javanese exhibits an initia1 band Madurese exhibits competing forms. one with initia1 b. the other with initial bh:

/ *[Bb]ayar (PHN, Tag. ba:yad 'payment': bayad), Snd., NJv., Mal. bayar, Mad. báj{r, Mad. B. bhájár 'to pay'.

Fina11y, an ambiguous reconstruction with initia1 *[Bb] is emp10yed, wh en a Javanese and a Madurese cognate are 1acking: *[Bb]akaw, p. 85; [Bb]u[dq]aw, p. 65.

2.12.3.2. PMJ [Bb] in postconsonanta1 position. An ambiguous reconstruction with postconsonanta1 *[Bb] is emp10yed, when Madurese exhibits a doublet, one with postconsonanta1 b, the other with postconsonanta1 bh:

*kam[Bb]a~ (PHN, Tag. kambá~ 'spread, sweep, f10ating': ka[m]ba~), Snd. kamba~, OJv. k-um-amba~ 'f10ating', NJv. kamba~, kambaV-kamba~, kamba~-an 'f10ating (0), 'float of a fishing 1ine', Mal. kamban 'to float', Mad. kamb{~ 'float of a fishing 1ine, to float', kambhá~ in bháláy kambhá~ 'pavi11ion on a 1ake', dhámar kambhá9 'a lamp with f10ating wiek'.

*sam[Bb]ar148, OJv. man-ambar, NJv. sambar, Mal. sambar, Mad. S.P. sambar, Mad. Kang. sambhar 'to pounce to seize and carry off'.

*tam[Bb]a~, OJv. t-in-amban 'left behind', NJv. tamba~ (0) 'neg1ected', Mal. Jak. tamban 'separation from one's wife without divorce', Mad. tambt~ 'to be indifferent towards s.o., neg1ect', tambhá~ 'to live in divorce'.

2.12.4. Morphophonemic evidence for the distinction between *B and *b. Sundanese offers morphophonemic evidence for the distinction between PMJ initial Band b. As was pointed out in the section on

Sundanese morphophonemics (p. 5 ) the N-alternation appears as m or as ~a before bases with initial b. Whtn the N-alternation is m, the initial b of the base alternates with~. The bases whose initial b alternates with ~ of ten are continuations of PMJ forms with an initial *B: e.g., bereq, mereq (L) 'to give'; b41it, m41it 'to wind around'; baur, maur 'to mix'; b41iq, m41iq 'to buy'; bawaq, mawaq 'to bring along'; balik, malik 'to return'; banun, maDun 'to erect'.

In two cases verb bases with an initial b from *B have the Nalternation ~a: batuk (L) 'to cough', nabatuk4n149 'to cough up'; bantin' ~abantin150 'to dash against'. It is not clear to what this exceptional behavior, if that is what it is, is to be attributed.

Verb bases with initial b from *b have the N-alternation ~a: e.g., baris, ~abaris 'to sit in a row'; bandin' nabandi~ 'to compare, match'.

2.12.5. The evidence presented above shows that a phoneme *B existed in addition to *b in PMJ.

In initial and postconsonantal position *B and *b fell together in Malay and Sundanese. They also fell together in postconsonantal position in Javanese. In intervocalic position they only fell together in Malay. Morphophonemically initial *B and *b are distinguished in Sundanese.

The following table recapitulates the reflexes of *B and *b in the four Malayo-Javanic languages:

2.13 PMJ D and PMJ d. The phoneme inventory as proposed by Dempwo1ff contains a phoneme *d. He reconstructed it in all five positions: initia1, intervoca1ic, preconsonantal, postconsonantal, and final. According to Dempwo1ff the Javanese reflex of PAN ~ is ~ in initia1, intervoca1ic, and postconsonanta1 position, and d in fina1 position. Dempwo1ff considered the occurrence of Javanese r instead of ~ as indicative of a 'Tendenz zur Lautverschiebung' (1.90). Dyen 1947 ag rees with Dempwolff's hypothesis with respect to the Javanese ref1exes of PAN 9 except for fina1 which according to Dyen appears as Javanese r. Another comparativist who reconstructed PAN ~ was Sakiyama. Two of his criterion 1anguages were Ma1ayo-Javanic 1anguages, name1y Javanese and Madurese. He fai1ed to account for the correspondences which are described next. Stevens (1966:156) conc1udes - as in the case of Mad. bh - that loanwords probab1y are the the main source of Madurese words containing ~h.

We have been compelled to reconstruct PMJ 0 and PMJ ~ as part of the PMJ phoneme inventory. The two proto-phonemes exhibit the following correspondences. PMJ 0 in initial position became Sundanese and Malay d, Old Javanese and New Javanese r, and Madurese ~. Javanese forms with initial ~ can be attributed to borrowings from a dialect in which *0 became ~ or to borrowings from a language in which *0 became d (such as Malay). Also Oempwolff and Oyen 1947 assigned the appearance of a New Javanese 9 in some instances to borrowings. Another possible explanation of the appearance of an initial ~ as reflex of 0 is that it is the result of an analogical change. Wh en postconsonantal *0 became Javanese ~ (see the next page), the combination of a preceding element with a final nasal (e.g., the Nalternation, see p. 10) and a base with initial *0 would be common. But in the sequence * ... N+base, the 9 is postconsonantal and would be superficially indistinguishable from a base with an inhe~ited q (from *9, see below). Thus analogy could operate * ... N+~base : *~base as * .. N+Obase : *x. A possible result of such a development are doublets such as OJv. ra~ö and OJv. n~anö 'to hear', NJv. rUDu (L), nqunu (0) 'to hear' (from *oa33ClRl'p. 76). OJv. ra~ö and NJv. rU9u are the result of a phonetic change, OJv. nqa~ö and NJv. nqu~u the result of an analogical change.

PMJ 0 in intervocalic position became Sundanese and Old and New Javanese r, Malay d, and Madurese~. In some instances Old and/or New Javanese exhibit competing forms, one with intervocalic r, the other with intervocalic~. The forms with intervocalic g can be attributed to a dialect or a language (such as Malay) in which intervocalic *0 became d.

PMJ postconsonantal *0 became Sundanese and Malay d, Old and New Javanese and Madurese g. The occurrence of Old Javanese doublets, one exhibiting postconsonantal r, the other exhibiting postconsonantal d is taken to represent two Old Javanese diaTectal reflexes: e.g., OJv. ra~rö, ran~ö 'cotton tree' (from *Oa~aR2' p. 100).

PMJ ~ became Sundanese and Malay d, Old and New Javanese ~, and Madurese gh in all positions.

2.13.1. PMJ 0

2.13.1.1. PMJ 0 in initial position. The correspondences are as follows: Snd. d-, OJv., NJv. r- (analogically ~-), Mal. d-, Mad. 9-: *Oahay151 (PAN, To. laqe 'forehead': gahay), OJv. rahi 'forehead, visage', NJv. rai 'face, front', gai (0) 'forehead', Mal. dahi, Mad. ~ái(h) 'forehead'. *OaRah (PAN, Fi. ndra 'blood': [d9]aRlaQ2a), OJv. räh, NJv. rah (H)151~ Mal. darah, Mad. qárá 'blood'. *Oaraq (PHN, Tag. da-l~a 'maiden': gaRa), Snd. daraq 'a young woman who has her first child', hayam daraq 'a chicken which has laid eggs for the first time', OJv. rarä, gara 'maid', NJv. ror~ (8), l~ro (8)152 'maid', qoro 'half grown-up (of a chicken, plant), average, medium, Mal. dara 'maiden, virgin', Mad. qárá in ajám gárá 'a hen that has not laid any eggs yet, but will soon be doing 50'. 153 ( , *OaR2at PHN, Tag. da:gat 'sea, ocean': daRat), Snd. darat 'dry land, firm ground' , OJv. rät 'world, empire, earth', NJv. rat (8) 'world', 14al. darat 'dry land, upland'.

*DataR2l54 (PAN, Sm. lau-lata 'mountain-plain': 9ataRl)' Snd. datar, OJv. rata, NJv. roto (L), Mal. datar 'smooth, level, flat' . *Datuq155 (PAN, Fi. ratu 'master': Qatu?), OJv. ratu 'king, prince', an-ratu-n-9atu 'to acknowledge as king', ka-9atw-an 'empire, royal court', NJv. ratu 'prince' (B), 'princess' (B), 'king', Qatu, qatuq (B) 'prince, master', Mal. datuq156 'head of the family, elder', datu 'ruler, chief'. *Dahuq, Snd. dahuq, OJv. rahu, NJv. rau 'kind of tree'. *Oaun (PAN, To., Fu. Sm. lau 'leaf, foliage': cJahwan), Snd. daun, OJv. rwan (with metathesis), ron, ron-Qon, NJv. ron (H), ron-90n (0) 'leaf', ron-ron-an (H) 'foliage, leaves', Mal. daon, Mad. gáun 'leaf,157. *oa~~uq158, OJv. raDu 'stubborn', NJv. raDu 'disturbed, sad, displeased', Mal. daDu 'gasp'. *Oiq (PHN, Tag. d-itó 'here': di), Snd. diq, OJv. ri, ri-~, NJv. ré-!J (0) (0), Mal. di, Mad. 9i 'locative'. *Ouaq (PAN, Fi. rua 'two': gawS3a?]), Snd. duaq159, OJv. rwa 'two', an-pin-9wa-n, pi~-rwa, piD-ro 'for the second time', NJv. ló-r6 (L)160, rö-rö (B) 'two', pin-cjÓ 'second, for the second time', Mal. dua, Mad. guwá(h), 9uwáq 'two'. *OuR2iq (PWI, TBt. duri 'thorn': [dcJ]uRi), Snd. duriq, OJv . . 161 NJ . Mld . M d d . (h) 'th ' rWl , v. rl, a. Url, a .• url orn. See also *OaDDaR2, p. 100; *Oahak, p. 72; and *Da~~aRl' p. 76 . In the following instances Old and/or New Javanese are taken to show g as the result of an analogical change:

*Oalam (PAN, To., Fu., Sm. lala 'depth, bottom' : galam), Snd. Bad. dalim 'inside', OJv. ~alam-an, galam-an 'intestines', Mal. dalam 'palace, prince's home, inside, inner', di-dalam 'in', Mad. 9álam 'deep', è-qál~m 'in, inside'. OJv. dalam 'inside, deepest, palace, court, deep' and NJv. iD dalam 'in, inside' and dalam (H) 'house, hall' are taken to be loans from Malay. Mad. dhilam (H) 'house' is taken to be a loan from Malay or from Javanese.

*Oapur (PAN, To. lafu 'kitchen for feasts':gapuR3), Snd. dapur (L) 'kitchen', OJv. dapur 'cooking-place, kitchen', Mad. gápor 'kitchen'.

*Oatan (PHN, Tag. dat{n 'arrival ': datan), Snd. datan162 (L), OJv. 9ata9' NJv. gata~{H), Mal. datan, Mad. 9átan 'to come~. *Oayan, Snd. dayan 'maid at court, whore', OJv. qayan 'female of an animal, nun, cloister-servants', NJv. gayan 'servants' (0). 'whore' (0), Mal. dayan 'maid at court, girl, damsel', Mad. gijin 'whore'.

*oappaq (PAN, Fu. lofa 'fathom': gapa[?h]), Snd. dipaq, OJv. gapa, , NJv. ~apo, Mal. dapa, Mad. 9appa(h) 'fathom'.

*Oulan (PHN, Tag. du:lan 'low dining table': dula9)' Snd. dula9' I OJv. 9ulan, NJV. qulan (0), Mal. dulaD' Mad. gulan 'wooden tray'.

See also *OanOa9' p. 200 and *oagOag, p. 194.

2.13.1.2. PMJ 0 in intervocalic position. The correspondences are as follows: Snd. -r-, OJv., NJv. -r-, Mal. -d-, Mad. -~-:

*haOap (PAN, Sa. s-aro 'front': h3agap), Snd. harip163 (L) 'in front of', OJv. harap 'what is placed in front', m-arap-akana

'to offer the point (pinnacle)', haQap 'prevented', NJv. di-arap-i (0) 'one is in front', agap-agap-an (L) 'to be in front of the other', Mal. (h)adap, Mad. ag(q 'in front of,16~ *BaOak16 (PWJ, TBt. badak 'rhinoceros': ba[dd]ak), OJv. warak, NJv. waraq, baqaq (0), Mal. badaq 'rhinoceros'. *BuOiq (PAN, Fi. m-uri 'back': ?ugahf[?h]), Snd. Bant. and other areas buriq166 'behind', OJv. wuri 'back', i wuri 'from behind', NJv. wuri (B) 'behind', wuri-wuri (B) 'afterwards, later on', buri (L) 'behind', Mal. ka-mudi-an 'position behind, af ter', Mad. bu~i(h) 'behind'. *ciOuk (PHN, Tag. s{:lok 'spoon': ci[n]Quk), NJv. ciroq (0), ciqóq 'big scoop' , Mal. cedoq, cidoq 'to spoon up'. / *kaOOut (PHN, Tag. kulot 'curly, wavy': kadut), OJv. k-um-rut 'frowned', Mal. kadot 'wrinkle'. *maOuq, Snd. maruq, OJv., NJv. maru, Mal. madu, Mad. magu 'fellowwife' . *muOah (PHN, Tag. mu:ra 'cheap': 00 mudah, JO [1953a] mudaq), Snd. murah 'to get much at a low price, cheap', NJv. murah (L) 'cheap', Mal. mudah 'easy167, Mad. mog' 'cheap'. *siOaq (PHN, Tag. sil( 'they': siqa), OJv. sira '2nd, 3rd person', NJv. sirÓ '3rd person' (B), '2nd person (0) (0), Mad. dialectal sèqá'(h) '2nd person'. *suOuq (PHN, Tag. s~:roq 'spoon': su[n]du), NJv. suru (0); Mal. sudu, Mad. soqu(h) 'spoon'. *tiOuR (PAN, Sm. me-uru 'to sleep': tiquR2, tuquR2), OJv. turü, NJv. turu (L), Mal. tidor, Mad. tèqun168 'to sleep'. *huOaD (PAN, Fi. qura 'crab' : quça~, qunqaD)' Snd., OJv. huran,

NJv. ura!), u~a!) (D) 'shrimp', Mal. (h)uda!l 'crayfish, prawn', Mad. o~á!) 'shrimp'.

*huDip (PAN, Fu. ma-u1i 'to live': quqip, qunqip), Snd. hirup (L) (metathesis) 'to live', hurip 'to revive', OJv. hurip, NJv. urép (L),uq{P (D), Mal. (h)idop (metathesis), Mad. ~diq 'to live', par-~gibh-án 'kind1ing'.

2.13.1.3. PMJ D in postconsonanta1 position. The correspondences are as fo110ws: Snd. -d-, OJv., NJv. -q-, Mal. -d-, Mad. -q-:

*canDik, Snd. candik, NJv. canqéq, Mad. canqiq 'a bund1e of sireh 1eaves' .

*ganDu1, Snd. gandu1 in gada~ gandu1 'kind of papaya with a long sta1k', NJv. ganq61, ganqo1 (D) 'to hang down 100se1y, hanger, kind of papaya', Mad. ghándu1 in tandhu gh'ndu1 'a ki nd of . . . tanqhu, the hanging tanqhu (?)'.

/ *kanDu~ (PHN, Tag. kandoD 'held in lap': ka[n]qu~), Snd. kandu~, OJv. ma~-anqu~, Mal. kandoD' Mad. kan~u~ 'pregnant'.

*kanDur169 (PWI, TBt. hondur = hendur 'slack': kanquR), OJv. ka-kanqon (kangu+an) 'get 100se', NJv. kanqó170, kanqu (0), Mal. kandor, Mad. kan~ur 'slack, not taut'.

*kunDur (PHN, Tag. kund61 'wax-gourd': kun[q]ur), Snd. kundur, OJv. \ kunqur, Mal. kundor, Mad. konqur 'wax-gourd'.

*lanDak (PWI, TBt. si-gu-1andak 'porcupine': 1angak), Snd. 1andak, OJv. 1angak, NJv. 1angaq, Mal. 1andaq, Mad. 1angáq 'porcupine' .

*panDak (PHN, Tag. pandák 'short': pandak), Snd. pandak 'short', OJv. panqak, NJv. pangaq (0) 'dward', Mal. pandaq, Mad. I panqaq 'short, not long'.

*panDan171 (PAN, Fi. vandra 'screwpine': paD~an), Snd. pandan, / OJv., NJv. pangan, Mal. pandan, Mad. pangan 'smaller screwpine'. *panDam172 (PWI, TBt. pondom 'grave': pa[n]gam), OJv. a-man~am 'to bury o.s.', NJv. pangam 'to bury', Mal. tar-pandam 'hidden underneath', Mad. bhan~am173 'to bury'. *sanDaU (PWI, TBt. sandan 'carried on the shoulder': sançaU)' Snd. sandan 'clothes', OJv. sançaU-an 'was worn', NJv. san~aD-an, sandaU-paUan (L) 'clothes', Mal. sandaU 'to wear', Mad. sangáD 'what is worn as a cloth on the shoulder'. / *sanDar (PHN, Tag. sandal 'act of reclining': sançar), Mal. sandar, Mad. san9ár 'to recline, rest on a support'. *sanDuk (PWI, TBt. sonduk 'spoon': sanguk), Snd. sinduk, Mal. sandoq, Mad. sanguq 'spoon, ladle'. *sinDiR2 (PWI, NgD. sindir 'jest' : singiR), Snd. sindir 'veiled manner of speaking, veiled expression', OJv. ma-sinqy-an 'sing', NJv. sing~n (sindi+an)174 'song, singing', Mal. sinder 'tease, chaff by innuendo', Mad. B.P. sèngir 'to allude to one's love for a woman~. *tanDak, OJv. a-tangak, NJv. nangaq (D), Mal. Java tandaq, Mad. / tangaq 'to dance'. *tanDuk (PHN, Tag. tandók 'dry-cupping, cupping sucker made of horn': tanduk), Snd. tanduk, Mal. tandoq, Mad. tanquq 'horn'. *tinDas175 (PAN, Sa. irohi 'to crack a louse': tiças, tin~as), OJv. ka-tingas 'squeezed', NJv. tin~as, Mal. tindas, Mad. , tengas 'to crack a flea or louse on the finger-nail'. , *tunDuh, Snd. tunduh, Mad. tongu 'sleepy'.

*tunDuk (PW!, TBt. tunduk 'to submit to': [t]un[g]uk), Snd. / , tunduk, NJv. tunqoq, Mal. tundoq, Mad. tonquq 'to bow'.

2.13.2. PMJ q

2.13.2.1. PMJ 9 in initial position. The correspondences are as follows: Snd. d-, OJv., NJv. 9-, Mal. d-, Mad. 9h-: *gaguV' Snd. dadu~, OJv. gaqu~, NJv. gaq6~, Mad. 9hághuU 'thick rope'. *gampar (PAN, To. lafa 'flat': gapaR, çampaR), NJv. gampar 'low table, bench, throne', Mal. Java dampar 'low seat, stool', Mad. B.P. gh~mpar, Mad. S. 9há-qhámpar 'table with short legs'. *~asar (PW!, TBt. dasor 'level': gas[aa][r]), Snd. dasar, OJv., / NJv. gasar, Mal. dasar, Mad. ghasar 'level'. *gayuo (PW!, NgD. dayoo 'oar'), Snd. dayuD' NJv. gayoD' Mal. dayov, Mad. ghájuD 'oar'. *gampul 176 , NJv. dampól, Mal. dampol, Mad. ghampol 'a composition used for caulking boats'. *çassak, NJv. gasak, Mal. dasaq, Mad. ghassak 'to push together'. *guçaq, Snd. dudaq, NJv. ~ug~, Mad. ghugha(h) 'widower'. !gukuh, Snd. dukuh, OJv. gukuh, NJv. quk6h (L), Mad. ghuko 'hamlet, village'. *gukun, Snd. dukun 'name of songs the content of which isconnected with medicine, medicine-man', NJv. guk6n, Mal. Java dukon, Mad. ghukon 'medicine-man'. *gusun, Snd. dusun 'boorish, uncivilized', Mad. ghusbn 'peevish, hot-tempered'.

2.13.2.2. PMJ q in intervocalic position. The correspondences are as follows: Snd. -d-, OJv., NJv. -~-, Mal. -d-, Mad. -qh-: / *Buqug, Snd. budug, OJv. wu~ug 'scabies', NJv. bu~og, Mal. budoq, Mad. bu~hu k '1 eprosy' • *ca~a~, Snd. cada~, OJv. pa-ca~a~-a, NJv. ca~a~ (0), Mal. cada!) , / Mad. ca~ha~ 'ready'. / I *gaqa~, Snd. gadan, NJv. gaqan-an, Mad. ghaqhaU 'designed for'. /. I *gaquh, Snd. gaduh, NJv. gaqoh, Mad. gha~hu 'to loan'. 1]7 *guqau ,Snd. guda~, NJv. guqa!) , Mal. gudaD 'store, warehouse'. I *kacjaO, Snd. kada!) , NJv. kaqa!), Mal. kadalJ-kada!j, Mad. kaqhaD 'somet imes ' . / I *kaqawu~, NJv. kaqawon, Mal. kadaoD' Mad. kaqhabu~ 'kind of tree'. *kuquq, Snd. kuduq, OJv. kuqu, Mal. kudu 'bad'. 178 / *kuqun ,Snd. kudu!), OJv. a-kuqury-kuqulJ' NJv. kuqo~, Mal. Jak. ka-kudol) 'to wear over one's head'. *puqak, Snd. pudak, OJv. puçak, NJv. puqaq, Mal. pudaq, podaq, Mad. pÖdhák 'flower of the screwpine'. *pa~çaU179 (PWI, TBt. poda~ 'sword': pacja9)' NJv. paqa~, Mal. padan, Mad. paq9há!J 'sword'. *saq~a!J180, OJv., NJv. saqa!), Mal. sadaU' Mad. saqqhary 'intermediate' . *saqqap18l, OJv. saqap 'tasty', NJv. saqap 'spicy', 'bitter' (0), 'fine' (B), Mal. sadap 'pleasant, nif:e, tasty, agreeable', Mad. sa~~háq 'agreeable to the senses'. *saqqih182 , NJv. soqéh, Mal. sadeh 'sad, sorrowful', Mad. saQQhi 'furious'.

See also *~apqap, p. 198.

See also *[D~]aqaq, p. 157.

2.13.2.3. PMJ ~ in postconsonantal position. The correspondences are as follows: Snd. -d-, OJv., NJv. -Q-, Mal. -d-, Mad. -2h-:

*ban2a~, Snd. banda~ in taliq bandaD 'a cord which is worn around the king's neck or chest during a war (symbol to indicate that one wants to defeat and bind one's enemies)', OJv. ban2a~-an 'war-prisoners', NJv. banQa~-an 'captured and led away as a prisoner', Mad.bhán~há~ 'to take away by force'.

*kanga~ (PWI, TBt. handaD 'fenced enclosure'), Snd. kandan' OJv., / NJv. kan~an' Mal. kandaD' Mad. kan~han 'fenced enclosure for animals'.

/ *kan~as, Snd. kandas, NJv. kan~as, Mad. kan~has 'to run aground (of a ship)' .

*kangut, OJv. k-in-anQut 'carrying secretly', NJv. kanQ6t 'to carry, place between the belly-band and the belly', Mad. ghánqhuq183, Mad. B. kanqhuq 'to carry with one'.

*kun~an' Snd. kunda~, OJv. a-kunqan 'to have s.t. on or about one', NJv. kun~a9 'extra, partner', Mal. kunda~ in kundan-an raja .... I 'court-pages, attendants', Mad. kon~ha~ 'always together, inseparable' .

*lan~u~, Snd. landun 'long (of what is hanging down)', OJv. I lan~un' NJv. lanQoD 'stretched out long', Mal. landoD 'long (of a rope)', Mad. lan~hun 'long (of body)'.

*linQih (PWI, TBt. lindi 'added'), Snd. lindih 'to lie on s.t., subdue, defeat', OJv. l-in-inQih-an 'to be superseded', NJv. lin~{h 'to lie on s.t., defeat', Mad. lén~hi 'to defeat'.

*mun~uq. Snd. ta9kal munduq. NJv. mun~u. Mal. mundu. Mad. monghu(h) 'kind of tree'.

*pinga~ (PHN. Tag. pindtn 'jerked beef'). Snd. pindan. OJv. pinga~ 'fish or meat cooked in salt water'. NJv. pin~a9 'fishor meat-soup'. Mal. pindaD 'to cook (fish. prawns. etc.) in salted and spiced bouillon'. Mad. B.P. pénghá~ 'a dish of salted kabán or bhulusfish'.

*san9i!) (PWI. TBt. sandi!) :\board on the wa 11 , ). Snd. sandin 'against s.t .• close to one another. at one's side'. OJv. sangi9 'side. closeness. near'. NJv. sangé9 (0) 'close t~. against'. Mal. sande!) 'to sit side by side'. Mad. sanghin 'to have at one's side'.

, / *sunga9' Snd. sundaD' OJv .• NJv. sun~a9' Mad. songha9 'supported'.

*tangaq (PAN. Sa. adra-lo~a 'pattern on the back of turtles': taga? tanga?). Snd. tandaq (L). OJv. tanga. NJv. tÖngà. Mal. / tanda. Mad. tangha(h) 'sign'.

*tangi9' Snd. tandin. OJv. tangiD' NJv. ta-tangin-an 'competition'. Mal. tanden 'division into equal parts. proportion. share'. Mad. tanghi9 'competition'.

*tanguq (PWI. TBt. tandu 'hammock-litter': tangu). Snd. tanduq. NJv. tangu. Mal. tandu. Mad. tang hu 'hammock-litter'.

*tun~uD' Snd. tunduD' OJv. ma-tun9u~. NJv. tungd3' Mal. tond09' , Mad. ton~huD 'to chase away. oust'.

*qungaq. Snd. qundaq 'to be put on a stick or held on it. (of a cricket) put on hand (consequently one hits the hand from bèlow so that the cricket jumps up)'. OJv. an-unça 'to swing \ unwards'. NJv. ungo 'to swing upwards' (0). 'to fly (of a bird at a wire, kite)', Mad. ön~ht(h) 'ball thrown into the air', Mad. B. n-~n~há ght1téq 'to let a rice-bird fly at a rope'.

*qunga~ (PWI, TBt. qundan-qundan 'riddle'), Snd. qunda~, NJv. un9an (0), Mal. undaD' Mad. ón~h{n 'order, ordinance'.

*qun9uh, Snd. qunduh 'to pick fruit', OJv. kon~uh (=ka+unquh) 'be /. , shaken off', NJv. unqoh, un~oh (0) 'picked, collected (of fruit, honey)', Mad. önghu 'to shake (of a tree)'.

2.13.3. Ambiguous reconstructions

2.13.3.1. PMJ [O~J in initial position. An ambiguous reconstruction with PMJ initial [Dg] is employed when Old and/or New Javanese exhibit a cognate with initial 9 and a Madurese cognate is lacking:

*[0~]agaq184 (PAN, Fi. vaka-rara-kombi 'to cross the arms': 9aga[?h]), Snd. dadaq 'chest, breast, bosom', OJv. ~a~a, NJv. g~gd, Mal. dada 'breast, chest'.

*[Og]akkat (PHN, Tag. dik{t 'to stick': qakat), Snd. d4k4t (L)

'near', NJv. gakat (0) 'close together', Mal. dakat 'near'. *[Og]UYU~185 (PWI, Mer. truzun~ 'whale': quyun), Snd. duyun' OJv. guyu~, NJv. gUY6n, Mal. duyon 'sea-cow'.

*[O~]ugal, Snd. dugal, NJv. n-9ugal, Mal. Java dugal 'pert, bold, impudent'.

2.13.3.2. PMJ [O~J in postconsonantal position. An ambiguous reconstruction with postconsonantal *[09] is employed when Old and/or New Javanese exhibit postconsonantal g, Sundanese and/or Malay postconsonantal d, and a Madurese cognate is lacking:

*qan[Og]a~, Snd. S.-B. qandan, NJv. an9a~, Mal. andaj 'spar to which a sail is attached. yard. boom'. *gun[0~]ul186. Snd. gundul. NJv. gun~61. Mal. gondol 'bare. bald'. *gun[0~]il187. Snd. gundil 'without leaves. bald'. NJv. gun~él. gÓn~é'l (0) (0). Mal. gundel 'bare. bald'.

*qin[O~]aD' Snd. qinda9' OJv. in~a9' Mal. inda~ 'nun'. *pin[O~]ah (PWI. Mer. findra 'change of location': pinçah). Snd. pindah (l). OJv. pin~ah. NJv. pin~ah (H). Mal. pindah 'to move'.

*tun[O~]aq (PWI. Mer. tundra 'freight': [t]un[g]a). Snd. tundaq 'to put down to pick it up again quickly'. NJv. tun~ó 'to be given from one to another'. Mal. tunda 'drawn or following in the wake'. sa-tunda 'one in the wake of the other'.

An ambiguous reconstruction with postconsonantal *[Og] is also employed when Madurese exhibits competing forms. one with postconsonantal g. the other with postconsonantal gh:

/ *ran[Og]ah. NJv. ran~ah (0). Mal. randah. Mad. B. rangha. Mad. B.P.S. ran~( 'low'.

*tin[O~]ih (PHN. Tag. tind{ 'weight'. til{ 'to lie on top': ti[n]gih). Snd. tindih 'to fall upon'. OJv. tinçih 'what lies on top of s.t.'. NJv. ting~h 'to put weight on s.t .• Mal. tindeh 'lying one on another'. Mad. tèngi. tènghi 'to occupy a place'.

2.13.3.3. PMJ [dg] in intervocalic position. As was pointed out in section 2.9.1. P~'J intervocalic d became Sundanese and Malay d. Thus. PMJ intervocalic d and PMJ intervocalic 9 fell together in these two languages. When there is no Javanese and Madurese cognate and

Sundanese and Malay both have intervocalic d, an ambiguous reconstruction with *[d~] is therefore employed:

*sa[dd, ~~]uq (PAN, Fi. ma-~endra 'hiccup': sa[d~]u[?h]), Snd. siduq 'hiccup', Mal. sadu 'hiccup, sob'.

2.13.3.4. PMJ [D1J in intervocalic position. As was pointed out in section 2.8.6. PMJ intervocalic 1 became Sundanese and Old and New Javanese r, and Malay d. Thus, PMJ intervocalic 1 and PMJ intervocalic D fall together in these three languages. When there is no Madurese cognate and Sundanese, Ol d and/or New Javanese, and Mal ay exhibit rand d respectively, an ambiguous reconstruction with intervocalic [Dl] is employed:

*pa[DD, 1!]ih (PWI, Mer. feri 'injury': pa[d@h), Snd. p4rih, NJv. / pareh, Mal.padeh 'smart, ache'.

*sa[Dl]ay, NJv. sart (Hl 'to sleep', Mal. saday 'to lie prone, as of a crocodile on a mid-bank'.

*pu[Dl]i~, Snd., OJv. puri9' NJv. pur{~, Mal. pude~ 'gardencroton' .

2.13.4. The evidence presented in the preceding sections points to the necessity of reconstructing a phoneme *D in addition to *q in PMJ.

These two proto-phonemes D and 9 have different reflexes in initial position in Javanese [but initial *D merging with initial *d when *D is preceded by a nasal (see section 2.13.)] and in Madurese. In intervocalic position these two phonemes have different reflexes in Sundanese, Javanese, and Madurese; and in postconsonantal position in Madurese:

2.14. PMJ R. Three tentative *R's are reconstructed as part of the PMJ phoneme inventory. In each of the three, the Ma1ay and Madurese ref1exes are r. All three *R's disappeared in 01d and New Javanese and the vowe1s contracted wh en *R occurred in intervoca1ic position. PMJ Rl became Sundanese y, PMJ R2 became Sundanese r, and PMJ R3 disappeared.

Oyen 1953b distinguished four tentative PAN R's. His criterion 1anguages were Taga1og, Ma1ay, Javanese, Ngaju-Oayak, and Merina. The fo11owing tab1e shows the members of each of his correspondences:

The Ma1ay reflex for all four *R's is r; the Taga10g reflex is g.

Dyen's *R3 and *R4, and *r merge in PMJ (for an example, see the section on PMJ r) having *r as their reflex. Dyen's *Rl and *R2, however, need to be further subdivided to account for the evidence of PMJ language~. Dyen's *Rl corresponds to PMJ R2 (e.g., BaR2R2ay, p. 126) or PMJ R3 (e.g., *zuR3uh, p. 110). Dyen's *R2 corresponds to PMJ Rl (e.g., *BuR1uk, p. 161), PMJ R2 (e.g., *BaR2R2at, p. 137), or PMJ R3 (e.g., taR3R3as, p. 164).

2.14.1. PMJ Rl' PMJ Rl became Sundanese y, Ma1ay and Madurese r, and disappeared in Javanese.

2.14.1.1. PMJ Rl in initia1 position:

*paR1R1am, Snd. ptytm 'ketan which has been sweetened by ragi',

Snd. S.-B. ptytm 'to ripen fruit artificia11y', Mal. param 'to store fruit to let it ripen'.

*saR1a~ (PWI, TBt. sara~ 'next': sa[r]a3)' Snd. sayaD' Mal. sara~ 'nest', Mad. sara~ 'edib1e bird's nest'.

The sequence *-uR1- before *a became Snd. -uy-, just as *-uy (see section 2.7.2.4.4.). Before *a and *u, however, it became i, perhaps through -uy-:

*kuRaq (PWI, TBt. hura-hura 'tortoise': ku[r]a), Snd. kuyaq, Mal. kura-kura, Mad. kora(h) 'tortoise'.

*huR1am, Snd. hitm 'shade, shadow' , Mal. m-uram 'sombre, overcast'.

*BuRluk (PAN, Sa. mata-hu1u 'last ripeness': buR2uk), Snd. biuk, OJv. wük, NJv. w6q 'sme11ing, fou1 odor', Mal. buroq 'decayed' .

The sequence *- R1- before avowel other than *a became Sundanese e, perhaps through -ay-. In the single case before e < *i; *-aR1- >

*BaR1Rlas (PHN, Tag. big(s 'hulled grains': baRas), Snd. beas (L), OJv. waas, wwas, NJv. wos (H), Mal. baras 'husked rice'. *kaR1Rli~, Snd. keqeD' Mal. kare~, Mad. karr~9 'dry'.

2.14.1.2. PMJ Rl in final position:

*lapaRl (PAN, To. qa-lefa 'skinny becaüse of hunger': lapaR, lampa ), Snd. palay (H) (metathesis) 'to have a desire for, hungry, thirsty', OJv. lapä, NJv. lop~, Mal., Mad. lapar 'hungry' .

*qulaRl (PWI, Mer. ulatra 'serpent': ulaR), Snd. qoray188, OJv. ulä, NJv. ulO, Mal. ular, Mad. olar 'serpent'.

The sequence *-aRl perhaps became earl ier Sundanese *-ay and like PMJ -ay contracted to Sundanese e: *Da9~aRl' p. 76.

2.14.1.3. PMJ Rl in initial position. In the single instanee the sequence *Rlobecame Sundanese i, perhaps through *yo-: *RloBBun' p. 135.

2.14.2. PMJ R2. PMJ R2 became Sundanese, Malay, and Madurese r, and disappeared in Javanese.

2.14.2.1. PMJ R2 in intervocalic position:

*BaRah (PAN, Fi. mbo 'ulcer': baRoq), Snd. bar4h 'swollen', NJv. woh 'pig pimple', Mal. barah 'abscess', Mad. bárá 'swollen, swe 11 i ng , .189

*diR2iq, Snd. diriq (L), OJv. di-di, Mal. diri 'oneself, self', Mad. Kang. dhirèq190 'I'.

*hoR2R2ot, Snd. h4r4t, OJv. a-höt-a, NJv. a-ot (D) (H) 'narrow', Mal. {{h)a)rat 'constriction' .

*paR2iq (PAN, Fi. vai 'ray-fish': paRli[), Snd. pariq, NJv. pé', Mal. pari, Mad. pare(h) 'ray-fish'.

*taR2uh (PAN, Sa. elu 'to put (up)': taRluq), Snd. taruh, OJv. a-toh, NJv. tOh, Mal. taroh, Mad. tar~ 'to bet'.

*tuR2ut (PWI, TBt. turut 'consequently': tuRut), Snd. turut, OJv. t-um-üt-akan, NJv. t6t, Mal. turot, Mad. tor~q 'to follow', tórot 'to be passive, agree to everything'.

*quR2an (PAN, U. ule- 'kinship': ?uRan), Snd. quran, OJv. uaD' 191 , wwa~ ,NJv. won (L), Mal. oraD 'person, human being'.

*quR2at (PAN, Fi. ua 'vein, muscle': ?uR12aCa), Snd. qurat192 , OJv. ot-wat 'vein', NJv. ot-ot 'muscle, sinew', Mal. urat 'nerve', Mad. oraq 'muscle, vein'.

*taR2um (PHN, Tag. tá:yom 'indigo-plant,193: taRum), Snd. tarum, NJv. töm, Mal. tarom, Mad. tarÖrn 'indigo-plant'.

*zuR2uq (PHN, Tag. du: Y0194 'pulpit': zuRu, duRu), Snd. juruq, OJv. pa-du, pa-pa-don (du+an), NJv. pa-don, Mal. panjuru 'corner, angle'.

See also *BaR2aq, p. 124; *BaR2iw, p. 86; *BaR2R2ay, p. 126; *BaR2R2at, p. 131; *OuR2 iq, p. 148; *OaR2at, p. 147; *laR2 iw, p. 87; *paRl2ah, p. 77; *taR2R2ab, p. 77; and *zaR2um, p. 110.

2.14.2.2. PMJ R2 in final position: *BiBiR2 (PAN, Fi. mbembe 'vulva': bibiR)195, Snd. biwir, NJv. miwi 'to grin', Mal. biber, Mad. bibir 'lip'.

*hiliR2 (PHN, Tag. h{:lig 'inclined': 00 hiliR, 10 [1953a] qiliR), Snd. hilir 'downstream', OJv. (h)il; 'stream, to flow', NJv. ili 'stream', Mal. (h)iler 'downstream', Mad. Kang.

, , " e1er-e1er-an 'gutter'. *lanzaR2, Snd. 1tnjtr 'long and straight', OJv. ma~-landö 'at full length, stretched out'. *sinaR2 (PAN, Fi. ~ina 'torch': s2inaR), Snd. sinar 'light', OJv. s-um-ina 'shining, sparkling', Mal. sinar 'beam of light'. *tukaR2 (PWI, TBt. tuhor 'purchase-money': tuka[r]),Snd. tuktr (L) 'to exchange', OJv. tuku-nan 'is supposed to be bought', ma-nakw-a196 'wants to buy', NJv. tuku (L) 'to buy', Mal. tukar 'interchange of goods'. See a1so *BuDuR2, p. 127; *OaUOaR2' p. 100; *OataR2, p. 148; *sinOiR2, p. 152; *tawaR2I, p. 84; *tawaR2II , p. 84.

2.14.3. PMJ RJ. PMJ R3 became Ma1ay and Madurese r, and disappeared in Sundanese and Javanese. This is the weakest correspondence of PMJ R's. The evidence for *R3 is very 1imited:

2.14.3.1. PMJ R3 in intervocalic position:

*taR3R3as (PAN, To., Fu., Sm. toa 'iron-wood': taRas), Snd. ttas 'hard, not soft', OJv. a-twas, taas, NJv. a-tos 'hard', Mal. taras 'heart-wood' (if hard) in a tree'. See a1so *w1aR3iq, p. 83, and *zuR3uh, p. 110.

As for Snd. ttas it must be suspected of being a borrowing, since an Old Javanese word of the same shape is availab1e.

2.14.3.2. PMJ R3 in initial position. There is on1y one comparison exhibiting *R3 in initia1 position. As in the case of *taR3R3as the Sundanese word must be suspected of being a borrowing from Javanese. However, it appears that no c1ear-cut decision can be reached:

*R3usuk (PAN, Sa. lusu 'rib of a canoe': R12usuk), Snd. Cian. qusuk, NJv. uS6q 'rafter', Mal. rusoq, Mad. rosÓq197 'rib'.

2.14.4. Ambiguous reconstructions

2.14.4.1. PMJ R13. In one instanee Sundanese exhibits a final -iq corresponding to a final -er in Malay. It is not clear whether Sundanese final -iq is from PMJ *-iRl or *-îR3, since an earlier Sundanese *-iy (from *-iRl ) might have resulted in a final -iq:

*caiR13 , Snd. caiq 'water' Mal. caer 'watery'; (cf. wlaiR, p. 165).

2.14.4.2. PMJ R. Wh en a Sundanese cognate is 1acking and Old and/or New Javanese exhibit the lack of an r, while Malay and/or Madurese exhibit an r, an ambiguous reconstruction with *R is employed:

*diRus198 (PWI, TBt. duris 'to spray, water'), OJv. a-dyus 'to bathe', NJv. d6s 'to wash', Mal. diros 'to wet, water, irrigate' .

*kuncaR, NJv. kunco, Mad. B.P. koncar 'loose end of garment'. *talluR199 (PAN, Sa. sa-olu 'egg': CaluR123a), OJv. hantalü, Mal. talor, Mad. tallor 'egg'.

In the following instanee Madurese exhibits -D as reflex of *-R. The Madurese word is probably from the Kangean dialect (in which also the final R of *niuR, p. 98 and *tiDuR, p.150 became ~):

*wlaiR (PAN, To., Fu., Sm. vai 'water': wahiR123 ), OJv. wai, wway 'water', NJv. wé in the compound wé-daD 'warm, hot water', \ Mal. aer 'water', Mad. aeD 'water'.

2.14.4.3. PMJ [rR21. Wh en an Old and New Javanese cognate is lacking and Sundanese has rand there is a Malay and/or Madurese containing r, an ambiguous reconstruction with *[rR2] is employed:

*pa[rr, R2R2]un (PWI, NgD. pehon 'to prepare a field by burning': pa[r]un), Snd. pirun 'to light a fire'; Mal. paron 'to burn off weeds'.

*[rR2]am[Bb]ay, Snd. r4mbay, Mal rambay 'to flow (of tears),

*[rR2]amas,Snd. ram4s, Mal. ramas 'to knead'.

*ham[Bb]a[rR2], Snd. hambar, Mal. (h)ambar 'tasteless (of water)'.

2.14.4.4. A PMJ doublet, one with initia1 *Ru-, the ot her with initia1 *R13i-. In one instance a PMJ doublet, one exhibiting penu1timate *u, the other exhibiting penu1timate *i is perhaps to be reconstructed. Snd. qimah 'house' points to *R13 imah, when taken together with U1awa nima 'house'. The fo110wing deve10pments might have taken p1ace. If the initia1 consonant was *R1' the sequence *yiwas reduced to i since Sundanese y does not occur preceding i. But if it was *R3' that consonant wou1d simp1y have disappeared. On the other hand OJv. umah, NJv. ómah (L), umah (0) (0), Mal. rumah, and Mad. Kang. r6ma299a 'house' are from *Rumah (PAN, Sa. nume 'house': Rumaq).

2.15 The 1aryngea1s. The PMJ phoneme inventory has three 1aryngea1s: *h, *q, and PMJ hiatus.

Dyen 1953a reconstructed three PAN 1aryngea1s: *q, *h, and *nothing. 200 The essentia1 members of each correspondence are given below:

Dyen 1965 proposed the following additional reconstructions based on Formosan evidence: *Ql (e.g., his *paQliC[ J; cf. *pahit, p. 175); *Q2 (e.g., his *[d~JaRaQ2; cf. *DaRah, p. 147); *51 (e.g., his *[ Ji51aq; cf. *qiah, p. 75); *52 (e.g., his *ka52iw; cf. *kaiw, p. 87); *53 (e.g., his *daw53a; cf. *Duaq, p. 148); *54 (e.g., his *54iRup; cf. *hiRup, p. 117); *55 (e.g., his *55ayup); *x1 (e.g., his *x1apuy[ J; cf. *qapuy, p. 94); *x2 (e.g., his *x2apat; cf. *qappat, *qampat, p. 193); *X (e.g., his *?iXu); *H (e.g., his *baRaH; cf. *SaR2aq, p. 124); and *? (e.g., his *ki~a?; cf. *ki~aq, p. 121).

Dyen 1971 distinguishes *hl through *h4 (the original correspondence for Dyen 1953a *h now assigned to *h1), *ql and *q2: *hl (e.g., his *h1unus; cf. *hunus, p. 63); *h2 (e.g., his *h2anin; cf. *haDin, p. 173); *h3 (e.g., his *h3a[nJdap; cf. *haDap, p. 149); and *h4 (e.g., his *h4atur; cf. *hatur, p. 105); *ql (e.g., his *q1atay; cf. *hatay, p. 93); and *q2 (e.g., his *q2ijun; cf. *hilU!), p. 102).

Dyen>s initia1 *h1 through *h4' *q1 and *q2' and *54 merge in any case in PMJ h:

*h . 1 . e.g. , Dyen's *h1unus; PMJ hunus, p. 63

*h . 2· e.g. , Dyen's *h2a~in; PMJ ha~in, p. 173. *h . 3· e.g. , Dyen's *h3adap, h3andap; PMJ haDap, p. 149. *h . 4· e.g. , Dyen's *h4atur; PMJ hatur, p. 105. *q . 1 . e. g., Dyen's *q1atay; PMJ hatay, p. 93. *q • 2· e.g. , Dyen' s *q2 i ju!:» PMJ hilu~, p. 102. *5 • 4· e.g. , Dyen's *54iRup, PMJ hiRup, p. 117. Dyen's initia1 *x1, *x2, and *1 merge in PMJ q: *x1: e. g., Dyen's *x1apuy; PMJ qapuy, p. 94. *x . 2· e.g. , Dyen's *x2apat; PMJ qappat, qampat, p. 193.

*?: e.g., Dyen's *?anam; PMJ qannam, p. 72.

Dyen's intervocalic *h, *Sl' *S2' *S3 and *~othing merge in PMJ hiatus: *h: e.g., Dyen;s*qahWan, PMJ Daun, p. 148. *S • 1· e.g. , Dyen's *[ ]iS1aq; PMJ qiah, p. 75. *S . 2· e.g. , Dyen's *kaS2iw; PMJ kaiw, p. 87. *S . 3· e.g. , Dyen's *dawS3a; PMJ Duaq, p. 148. *nothing: e.g., Dyen's *buaq; PMJ Buah, p. 127. *?: Dyen's intervocalic *? in a word like *ka?an presumably disappeared in the pre-Proto-Malayo-Javanic period and the two vowels contracted to PMJ a: *kan, Snd. ha-kan CL) 'to eat', OJv. ma-kan-a 'in order to eat', NJv. pa-kan 'fodder', Mal. ma-kan 'to eat', Mad. pa-kan 'fodder'. Dyen's intervocalic *q and *Ql merge in PMJ h: *q: e.g., Dyen's *Caqi?; PMJ tahiq, p. 175. e.g., Dyen's *paQliC[ ]; PMJ pahit, p. 175. *Q • 1· Dyen's fina1 *h, *q, and *Q2 a1so merge in PMJ h: *h: e.g., Dyen's *rapuh; PMJ rapuh, p. 63. *q: e.g., Dyen's *s2u1uq; PMJ suluh, p. 177. *Q • 2· e.g., Dyen's *[dq]aRaQ2; PMJ DaRah, p. 147. Dyen's final *Sl' *S2' *H, and *? merge in PMJ q: *S . 1 . e.g. , Dyen's *tuqaS1; PMJ tuhaq, p. 176. *S . 2· e.g. , Dyen's *kuS2kuS2~ PMJ kukuq, p. 119. *H: e.g. , Dyen's *baRaH; PMJ BaR2aq, p. 124. *?: e. g. , Dyen's *kita?; PMJ ki~aq, p. 121 . 2.15.1. PMJ h. PMJ h is reconstructed in initia1, intervoca1ic, preconsonantal, postconsonantal, and fina1 position.

2.15.1.1. PMJ h in initial position. PMJ h appears as Sundanese h. Old Javanese orthography of ten exhibits competing forms, one with initial h, the other lacking it. Three possibilities of explaining them exist: (1) The spelling with h is an archàism and no initial h appeared in the pronunciation during the Old Javanese period, (2) the spelling with initial h represents the pronunciation of the words at an earl ier stage of Old Javanese and the spelling lacking the h the pronunciation at a later stage of Old Javanese, and (3) the two spellings reflect two different dialects and New Javanese is from the h-losing dialect, since New Javanese exhibits h only in expressive words (see p. 9). PMJ initial h also disappeared in Madurese. Wilkinson's Malay dictionary almost always exhibits doublets, one with initial h, the other lacking it. The existence of these doublets can be attributed to borrowings. A dialect (or language) which retained initial *h borrowed from a dialect which lost it, or conversely.

In only one comparison both the Old Javanese and the Malay cognate have an initial h without exhibiting a by-form:

*hias (PHN, Tag. hiyás 'gem': hiyas), Snd. hias201 , OJv. pa-hyas, NJv. pa-~s (0), Mal. hias, Mad. ~yas202 'embellishment'.

In a few comparisons both Old Javanese and Malay have competing forms:

~ *hayam (PHN, Tag. qa:yam 'dog': OD ayam, ID [1953a] qayam), Snd. hayam 'chicken', OJv. hayam-hayam-an 'moor-hen', ayam wana 'bush-hen', NJv. ayam (H), Mal. (h)ayam, Mad. ajám 'chicken'.

In most comparisons Old Javanese exhibits cognates with an initial h without by-forms and Malay exhibits competing forms: *halis. Snd. halis (L). OJv. halis. NJv. alés. Mal. (h)ales. Mad. alès 'eye-brow'. *hanat (PAN. Sa. m-ano 'to breathe': qa~at). Snd. han~t203. OJv. hanat. NJv. anat. Mal. {h)anat. Mad. a!laq 'luke-warm'. *hallaD' Snd. h~lan. OJv. halan. Mal. «h)a)laD' Mad. la!) 'eagle. hawk' . *hian204 (PWI. TBt. qian'majesty': hi[y]aD)' Snd. hia9205. OJv. hya9 'god'. a-hyan 'stately. beautiful. splendid'. NJv. a_~n206 'strange. beautiful. wonderful'. Mal. hia~. yan 'god'. *huta9' Snd. huta9 (L). OJv. huta!). NJv. uta,!} (L). Mal. (h)uta!J' Mad. ota!) 'debt'. See also *hampas. p. 117; *haBuq. p. 132; *hallay. p. 91 *haR2R2at. p.162; *hijaw. p. 54 ; and *huDa!). p. 150.

*hallat (PWI. TBt. qolat 'border': alat). Snd. h~l~t 'interval'. OJv. a-halat 'with an interval of ... '. lat. NJv. {a)l at 'interval '. Mal. {{h)a)lat 'periodicity. alternation'. Mad. {al)laq 'interval '.

See also *hatay. p. 93; *haDap. p. 149; *hiliR2• p. 163; and *huzan. p. 11l.

*huluq (PAN. Fi. qulu- 'head': qulu[?h]). Snd. huluq [(L) when used with human beingsJ. OJv. hulü. NJv. ulu (B). Mal. (h)ulu. Mad. olÖ(h) 'head'.

In one comparison Old Javanese exhibits a cognate with an initial h without a by-form and Malay exhibits a cognate with an initial vowel without a by-form:

*ha1uq (PHN, Tag. há:10q 'matter added to mixture': qaS1a1u),

Snd. ha1uq, OJv. ha1u, NJv. a1u, Mal. a1u 'rice-mortar'.

In five comparisons Sundanese exhibits competing forms, one with initia1 h, the other with initia1 q. The existence of these doub1ets is attributed to borrowings from a dialect (or 1anguage) which lost initia1 *h. Sundanese dia1ects in which initia1 *h disappeared are the dia1ects of Baros (north of Pamanukan) and the two enclaves, Le1ea and Parean. Initia1 h sometimes a1so disappeared in the Northern Banten and Cirebon dia1ects.

01d Javanese does not have a cognate in the first instanee, has competing forms in the second and third, and has a cognate with initia1 vowe1 without a by-form in the fourth and fifth. Ma1ay has in all five instanees cognates with initia1 vowel without a by-form:

*hazzan, p. 11 O.

*hassam207 , p. 18' .

*ha1an (PHN, Tag. há:raD 'obstacle'), Snd. (h)a1an, OJv. ma~-ha1aD-i 'to obstruct', a~-a1a~-i 'to oppose', NJv., Mal., Mad. a1aD 'to check, step in s.o.'s way'. *hasap208 (PHN, Tag. h6:sap 'smoke': 00 hasap, 10 [h0]asap), Snd. (h)as4p, OJv. asap, asöp, NJv. asap, Mal. asap 'smoke'. OJv. hasap also occurs; if it is a loan from earl ier Ma1ay, it indicates that an initial h was still present in Ma1ay at the time the item was borrowed. It is conceivab1e, however, that the second a is due to assimi1ation.

*hali!), Snd. (h)a1in' OJv .• 1i!), NJv. a1é'!), Ma1. Jak. a1eD' Mad. \ ale9 'to proteet one's face, screen'.

We also reconstruct *hwhen Sundanese exhibits an initial q, Old Javanese and Malay competing forms, one with initial h, the other lacking it, and New Javanese and Madurese an initial vowel. The Sundanese forms must be suspected of being borrowings from a dialect or language whjch lost *h-:

*hayun (PAN, Sa. esu-esu 'earthquake': ?ayun209 ), Snd. qayun 'to sway', OJv. ma-hayun-an 'to be shaken', ayun 'to swing, rock', Mal. (h)ayon 'to sway'.

*huju~ (PAN, Fi. qu~u 'cape': [qh]uzu~), Snd. qujuD 'point, end, spit', OJv. huju~ 'spit (of land)', ujuD 'leg, paw', NJv. ujÓ~, Mal. (h)ujoj 'point, extremity'.

*hulur (PHN, Tag. hu:log 'to fall': huluR), Snd. qulur 'to let go (a rope)', OJv. hulur-ulur-a 'may the fringes be of', / , , NJv. ulor, Mal. (h)ulor, Mad. olor 'to pay out (rope), let go' .

See also *hurup, p. 117.

PMJ initial h is also reconstructed wh en Malay exhibits hin correspondence with Sundanese initial q, and Old and New Javanese initial vowel:

*halus (PHN, Tag. há:los 'almost'), Snd. qalus (L), OJv. alus, NJv. a16s, Mal. (h)alos, Mad. alàs 'fine, smooth, delicate'. *hukur (PHN, Tag. qU:kol 'what a thing is intended for': OD ukur, ID [1953a] qukur), Snd. qukur, OJv. ukur, NJv. uk6r, Mal. (h)ukor, Mad. 6kOr 'to measure'.

See also *hilu~, p. 102.

In the following instance both Sundanese and Malay fail to exhibit a competitive form with initial.h, whereas Old Javanese has a

*hanin (PAN, To. fakaqani 'to check, stop the wind': h2a~in), Snd. qa~in, OJv. (h)a~in, NJv. a~én, Mal. a~en, Mad. a~~n 'wind'.

2.15.1.2. PMJ h in intervocalic position. PMJ intervocalic h became 01d Javanese h. lts reflexes in the other languages are discussed below.

Between like vowels PMJ -happears as h in Sundanese and in Malay and became Madurese -q-. Between *a's it also appears as h in New Javanese:

*[dO~]ahan (PAN, Sm. la 'twig': [dd]ahan), Snd., Mad. dahan . . 'tree-branch'.

*lahaq, Snd. lahaq, NJv. loho, Mad. laqa(h) 'fish-trap made of thin bamboo-sticks'.

*laha~, Snd., OJv. laha~, NJv. lahaD (0), Mal. Java, Mad. laqa~ 'palm-wine'.

*rahab, Snd. rahab 'to equip s.o. with clothes', OJv. aD-rahab-i 'to cover', Mal. rahap 'cloth laid over dead body'.

*saha~, OJv. saha3 'hot, peppery', Mad. saqa~ 'black pepper'.

Between *u's PMJ intervocalic h appears as New Javanese h or it disappeared and a homorganic glide was developed between the two li~e vowels:

*suhun, Snd. suhun 'to carry on one's head, ask', OJv. sakar suhun 'headdress', an suhun 'to ask', NJv. suhón (0) (0), suw6n (H) 'to ask (of one who is of higher rank)', Mal. Java sakar suhon 'headdress', Mad. soqon 'to carry on one's head, to ask'.

Between *a's PMJ intervocalic h appears as h in Sundanese. New Javanese exhibits competing forms, one with intervocalic h, the other with intervocalic w. The by-form with intervocalic h is marked as 'literary': 1 aha!), p. 75.

PMJ intervocalic ~ af ter *a and before *a appears as h in Sundanese and in Malay. lts New Javanese reflex is wand its Madurese reflex is -q-:

*sahaR, NJv. SUWÓ in kalam suw6 (0) 'pen of a palm-fiber', Mad. B.P. saqar 'stem, stalk, hard fibre in the outer leaf of the aren-palm'.

See also *dahaq, p. 74; *Dahak, p. 72.

PMJ intervocalic haf ter *a and before *a appears as h in Sundanese and in Malay. In Madurese PMJ -hbecame q when PMJ became Madurese a, i.e., before Madurese -q < *-p, *-t, *-k, *-1 (see the section on the Madurese reflexes of PMJ a). This h, however, was lost wh en the Madurese refl ex of PMJ a was a:

*pahat (PWI, NgD. pahat 'chisel': pahat), Snd. pah~t, Mal. pahat, Mad. paqaq 'chisel'.

*tahan (PWI, Mer. tana 'to hold fast': tahan), Snd. tahin 'what is used to catch, snare', OJv. tahan 'to resist, restrain' , Mal. tahan 'to resist, endure, restrain, set snares', Mad.

S. taan 'rope, to ambush'. Sundanese, New Javanese, and Madurese tahan 'to resist, endure' are all treated as borrowings from Malay.

PMJ intervocalic haf ter *a and before *i or *u disappeared in Sundanese except when the final vowel preceded final q. In a few instances competing forms occur. Thus. we can posit two dialects. one in which *h appears as hand another in which it disappeared. It is best to explain the competing forms as borrowings from the dialect in which h appears as h into the dialect in which it disappeared,or conversely. Intervocalic h appears as h in the dialect of Southern Banten. It disappeared in New Javanese. Malay. and Madurese. Wilkinson. however. of ten writes competing forms. one with intervocalic h. the other lacking it. In two instances only a cognate with orthographic intervocalic h occurs without a by-form lacking it. Following Oyen (1953a:7), we interpret the appearance of orthographic h between a and a different vowel as indicative of a real distinction in earl ier Malay which was preserved by the Arabic orthography:

*pahit (PHN. Tag. paft 'bitterness': 00 pahit, 10 [1971]

paQliC[ ]), Snd. pait, S.-B. pahit, OJv. pahit, NJv. paét. Mal. pa(h)et. Mad. pa~q 'bitter'.

I *pahul, Snd. paul, OJv. pahul, NJv .. paol (0) 'light-blue', Mad. paol 'dark-gray'.

*rahup (PAN, Sa. raqu 'to snatch up with the hands': raqup). OJv. / rahup. NJv. raop 'to wash one's face', Mal. ra(h)op 'to put the hands together cupwise', Mad. raap 'to wash one's face'.

*tahiq (PAN, Sa. ae 'excrement': Caqi?), Snd. nahiq 'to rust', taiq, Snd. S.-B. tahiq 'dirt', OJv. tahi, NJv. tai (L) 'excrement, dirt', tai-yeu 'rust', Mal. ta(h)i, Mad. taé(h) 'excrement, dirt'.

*tahil, Snd. tail. OJv. tahil. NJv. ta~l. sa-taél, Mal. ta(h)el,

Mad. taél 'a kind of weight'.

*tahun (PAN, Sm. tau 'season': taqwan), Snd, taun, Snd. S.-B. tahun, OJv. tahun, NJv. ta6n, Mal. ta(h)on, Mad. taon 'year,2l0.

Af ter *i and before *u or *a PMJ intervoealie h appears as Sundanese h in two instanees and disappeared in one instanee. It disappeared in Malay, even in the orthography. In New Javanese and in Madurese asemivowel homorganie to the preeeding vowel developed between the two vowels:

*plhak (PHN, Tag. piy~k 'split': 00 pihak, 10 [1953a] piqak), OJv. pihak 'pieee', Mal. sa-piaq 'one' (with sireh = 'a fresh quid of betel '), Mad. pèyak 'to (make a) split'. *sihu~, Snd., OJv. sihuD' NJv. siyó~, Mal. sio~, Mad. SèyOv'tusk'. *tihan (PAN, Fi. ndia 'handle': tiya~), Snd. OJv. tiha~, NJv. , tiya~, Mal. tia~, Mad. teyaD 'pole, mast'.

Af ter *u and before *a PMJ intervoealie *h disappeared in Sundanese exeept wh en the ultimate vowel preeeded *-q. In twoinstanees eompeting forms oeeur. Just like af ter i, it disappeared in Malay, generally also in theorthography. In New Javanese and in Madurese a semivowel homorganie to the preeeding vowel appears between the two vowels:

*Buhayaq211 (PAN, Sa. huasa 'eroeodile': buqaya [?h]), Snd. ouayaq, Western and Eastern Snd. bUhayaq212, OJv. wuhaya, NJv. buwoyo, boY0213, Mal. buaya 'eroeodile'.

*[rR2]uha~ (PAN, Sm. ua 'throat, neek': Ruqan), Snd. rO(h)aU 'space between two houses', Mal. ruaU 'hollow space'.

*tuhaq (PAN, Fi. tua 'grand-father': tuqaSl)' Snd. mi-tohaq

'parent-in-1aw', OJv. tuha, NJv. tuwa, Mal. tu(h)a, Mad. towa(h) 'old'. *tuhan (PAN, Fu., Sm. a-tua 'goddess'), OJv. tuhan, NJv. tuwan, Mal. tu(h)an 'master, lord', tohan 'God', Mad. t~wan 'master, lord'.

Madurese evidence makes it possib1e to reconstruct PMJ -hin a comparison for which the 01d Javanese cognate exhibits intervoca1ic w. The existence of this 01d Javanese form is probab1y a modernization of an 01 der spelling *luhar:

*luhar (PHN, Tag. 1uwá1 'outside': 00 luwar, 10 [1953a] luqar), Snd. luar 'outside', OJv. ka-luwar-an 'liberation', NJv. luwar (0) 'be let go, freed', Mal. luar, Mad. lowar 'outside'.

2.15.1.3. PMJ h in preconsonantal position. For putative examples, see the section on PMJ doubled monosyllables.

2.15.1.4. PMJ h in postconsonantal position. The only instance in which h appears in postconsonantal position is *ta~hiq. p. 49. It appears as h in Sundanese and Old Javanese, and disappeared in New Javanese. In Madurese it assimilated to the preceding consonant, resulting in a double consonant.

2.15.1.5. PMJ h in final position. PMJ -h appears as h in all languages except Madurese where it was lost:

*sisih (PAN, Sn. paqu-sisi 'side of the house': sisiq), Snd. sa-sisih 'partial', OJv. a-sisih 'be on the same side, party-partner', sisih-an 'husband', NJv. siséh, sisèh (0) 'at the side', Mal. siseh 'side', Mad. sèsè 'to give way'. *suluh (PAN, To. hulu 'torch': s2uluq), Snd. suluh 'firewood',

(D). Mal. su10h 'torch. f1are'. Mad. solo 'to light with a torch' •

2.15.2. ~. PMJ q is reconstructed in initia1 position. between identica1 vowe1s. and in fina1 position.

2.15.2.1. PMJ 9 in initia1 position. PMJ q in initia1 position became glotta1 stop in Sundanese. It became orthographic initia1 vowel in Ma1ay. but Dyen (1953a:40) states that of ten Malay orthographic initia1 vowe1 is preceded by a phonetic glottal stop. The usua1 orthography has been chosen here despite this facto because with Dyen we observe that "all Ma1ay words with initia1 glotta1 stop a1so occur with initia1 vowe1 (or smooth onset) and the regu1ar notation of both variants wou1d serve no usefu1 purpose. This practice leads to no confusion in notation for even those words written with initial vowe1 which have a by-form with initial h a1so occur with initia1 phonetic glotta1 stop. To accommodate this change in the formu1as wou1d require the substitution of the longer phrase 'Ma1ay initial vowe1 or q' for the phrase 'Ma1ay initia1 vowe1' wherever the latter occurs and a1so the regu1ar notation of the two variants."

Initia1 *q became 01d Javanese orthographic initia1 vowe1 and New Javanese and Madurese orthographic and phonetic initia1 vowe1:

*qakuq (PAN. To .• Fu. au 'I': ?aku? ku?). Snd. qakuq. OJv. aD-aku 'to recognize. acknowledge'. Mal. aku 'my own. i.e., taking responsibi1ity for', aku-an 'acknow1edgement', Mad. akó(h} 'to pretend'.

*qinum (PAN, To., Fu., Sm. inu 'drink': ?inum), Snd. qinum (L), OJv. inum, NJv. in6m (B) (D), Mal. inom, Mad. én'àm 'to drink' .

2.15.2.2. PMJ 9 between 1ike vowe1s. PMJ q occurs intervoca1ica11y on1y between 1ike vowe1s. It became glotta1 stop in Sundanese and Madurese. It disappeared in Javanese, and the two contiguous vowe1s contracted to an 01d Javanese long vowe1. New Javanese has a short vowe1. It a1so disappeared in Ma1ay and the two vowels contracted to a short vowe1:

*kuqu1 (PHN, Tag. kuhó1 'snail': kuul), Snd. kuqu1, OJv. kü1. NJv. kó1. Mad. koqo1 'snail'.

*laqas, NJv. las. Mad. 1aqas 'unhusked rice-grain'. *saqat (PAN. Sa. ma-ata 'dried out': [q?h]asat214 ), Snd. saqat, OJv. sat-ana 'in order to make dry'. NJv. a-sat 'dried out'. *tuqus. Snd. tuqus 'to drip', OJv. tus-tus 'to drip'. tûs 'sap'. NJv. tós 'to drip out'. Mal. Per. tos 'to drip unti1 dry'. *tiqis. Snd. tiqis. OJv. tis, NJv.' a-té's (D) (0) 'co1d'. *ruqum (PWI, NgD. ha rum 'fragrant': ha[r]um215 ), Snd. ruqum , ,216 - , sme11 • OJv. rum 'sme11', a-rum 'lovely, fragrant , NJv. a-róm, Mal. ha-rom, Mad. roqom 'fragrant'.

See also *duqum, p. 107; kaqa~, p. 73.

In one comparison the 01d Javanese form does not exhibit a long vowe1:

*taqal, OJv., NJv. tal, Mad. taqa1 'lontar-palm'.

2.15.2.3. PMJ 9 in fina1 position. PMJ q in fina1 position became glotta1 stop in Sundanese and disappeared in Ma1ay, 01d and New Javanese, and Madurese. In West Madurese, however, it became h before pause and disappeared e1sewhere:

*maluq, NJv. ma1u (D), Mal. ma1u, Mad. ma1Ó(h) 'feelings of shame, ashamed'.

*jamuq, Snd. ja-jamuq 'dried herbs', NJv. jamu (L) 'drink, medicine', Mal. jamu-jamu 'a plant', Mad. jh~mà(h) 'medicine'.

*taliq (PAN, Sa. ~li 'rope': taliSla), Snd. taliq, OJv. tali, NJv. tali (L), Mal. tali, Mad. talé(h) 'rope, cord'.

In one comparison the Javanese cognate exhibits an unexpected final hand the Madurese cognate an unexpected final -q:

*BaR2R2ay, Snd. bereq (L) 'to give', OJv. w-in-eh 'be given', NJv. wè-wèh (L), Mal. bari, Mad. barriq 'to give'.

PMJ BaR2R2ay would be expected to become earl ier Javanese *wi by the following development [see Dyen's discussion in Dyen (1953a:61)]. Af ter PMJ -ay became i (see the section on PMJ -ay), PMJ R2 disappeared in the two vowels contracted to i. As was described in 1.6.3.2., a base-final i alternates with y before the suffixes -akan, -ani, and -an. Thus, *wi+akan presumably became *wyakan, *wi+ani presumably became *wyani, and *wi+an presumably became *wyan. Generally, the semivowel of the base and the initial vowel of the suffix did not contract to a single vowel in Old Javanese in morphologically complex forms. It is, however, conceivable that there was an Old Javanese dialect in which contraction occurred, possibly only in derivations involving sequences of monosyllabic bases. Such a contraction would result in forms like OJv. *wekan, *weni, and *wen. Pigeaud (n.d.) lists both NJv. wèni 'to get' and NJv. wèn 'to get, given' as dialectal forms. OJv. *we might have replaced an older *wi because the latter was in the same relationship to a suffixed form such as *wekan 'was given' as OJv. limbe 'to spread out' was to l-um-imbekan 'tightened'.

OJv. weh,cou1d have acquired its final h as a result of the fo110wing ana10gica1 change. OJv. weh rep1aced an older *we because the latter was in the same re1ationship to a suffixed form such as *wean as NJv. kumbah 'to wash' is to NJv. kumbaan 'laundry' (see section 1.5.3.2.).

The additiona1 fina1 q in Madurese can a1so be attributed to an ana10gica1 change. Mad. barriq rep1aced an older form *barri because the latter was in the same re1ationship to a suffixed form such as

barriqi as Mad. 6diq . 'to live' is to Odiqi 'to kind1e (fire)'. . In a few comparisons Madurese exhibits competing forms, one with

a fina1 vowe1 (and fina1 h in West Madurese), the other with a fina1 glotta1 stop. Madurese has doub1ets for the numbers 'two, 'three', 'five', 'seven', 'eight', and 'nine' : *Duaq, Mad. 9UW'(h), ~uw'q 'two'; ~311uq, Mad. ta110(h), ta11~q 'three'; *limaq, Mad. 1èma(h), l~maq 'five'; *pituq, Mad. pèttÓ(h), pettoq 'seven'; *w2a1uq, Mad. bá11u(h), b~lluq 'eight'; *sa~aq, Mad. sa~a(h), saDaq 'nine'.

The numbers with fina1 q are used in counting and wh en emp10ying them in p1ace of nouns. They are a1so used preceding nouns in counting, fo110wing nouns which occur in sets, af ter intransitive verbals, before agghiq and -agghiq 'more', and before èb~9 'each'. For examp1es. see Stevens (1969:203).

The gain of the fina1 q can be attributed to the staccato speech sty1e which is used in counting. This by-form a1so came to be used in the environments just described.

2.15.3. PMJ hiatus. PMJ hiatus on1y occurs between nonidentica1 vowe1s. The vowe1s in hiatus remained in Sundanese and in Ma1ay. 01d Javanese exhibits contraction of the two vowe1s in hiatus when the first vowe1 was *a. When the first vowe1 was *i, *u, or *a, the semivowe1 corresponding to the vowe1 generally appears in its p1ace. In a few instances, 01d Javanese exhibits competing forms, one with asemivowel fo110wed bya vowe1, the other with contraction of this sequence. However, New Javanese on1y exhibits contraction. In Madurese the semivowe1 corresponding to the preceding vowe1 appears before the fo110wing different vowe1.

2.15.3.1. PMJ hiatus between *a and *i:

*kain, Snd. kaen 'woven fabric', OJv. ken 217 'underc10thes',

Mal. kaen, Mad. kaên 'woven fabric'.

*pais, Snd. pais 'what has been roasted in hot ashes', NJv. pès 'roasted', Mal. paes, Mad. paês 'spiced fish cooked in a banana-1eaf'.

2.15.3.2. PMJ hiatus between *a and *u:

*laun (PHN, Tag. 1aqón 'long duration'), Snd. 1aun218 , OJv. a-10n, NJv. a-1on, Mal. 1aon, Mad. laan 'slow'.

*sau~ (PWI, TBt. saon 'cover, umbre11a': saan), Snd. sau~ 'hut, cover', OJv. sOD 'what gives shadow, overshadowed by', NJv. ka-so~-an (B) 'overshadowed by', Mal. nao~219 'shadowing, shelter', Mad. naö~220 'shadow'.

*pauh (PAN, Fi. mbau 'kind of fruit-tree': pahuq), Snd. pauh, OJv. poh, NJv. pOh, Mal. paoh, Mad. pao 'kind of tree'.

2.15.3.3. PMJ hiatus between *i and *a:

*lian and *lain (PAN, Fi. 1ia 'to transform, change': 1iqan), Snd. 1ain, OJv. 1yan, 1en, NJv. 1en (B)221, Mal. 1aen, Mad. 1a~n, Mad. B. l~yan 'different, other'.

*lia~, Snd. 1ia~, OJv. 1yaD' 1e~. NJv. 1e~. Mal. 1ia~'a sma11 aperture' .

2.15.3.4. PMJ hiatus between *u and *a:

*tuak222 (PWI, TBt. tuak 'pa1m-wine: tuwak), Snd. tuak, OJv.

twak, tok, Mal. tuaq, Mad. tówaq 'palm-wine'.

*huan, OJv. a~-hwan, NJv. aD-on " ( L), Mad. " owan 'to guard sheep'.

2.15.3.5. PMJ hiatus between *i and *u:

*hiuq (PWI, TBt. qiu 'shark': hi[yJu), Snd. hiuq, OJv. hyü, Mal. (hi)yu 'shark'.

When the sequence *-iuwas preceded by *h it became OJv. hyu, as i11ustrated in the preceding comparison. lts New Javanese continuation is yuand not *u. Af ter OJv. h the anticipated contraction of -yuto -u- (cf. OJv. dyus, NJv. dós from *diRus, p. 165) did not take place. It did not occur probab1y because of the prior deve10pment of initial hyto ybefore OJv. u:

*hius, Snd. hius 'to blow, how1 (of wind)', OJv. (h)-um-yus, NJv. um-yós (B) 'to whizz, buzz'.

2.15.3.6. PMJ hiatus between *a and *a. There are on1y two comparisons which exhibit *hiatus in this environment. In the first comparison we on1y find a New Javanese cognate. There is no Old Javanese form in Juynboll. It presumably was *pwar (see section 2.6.4.3.). From other tomparisons (ss the section on *R) we know that earl ier Javanese hiatus whose first member was *a and whose second member was *a became Old Javanese -waand contracted to New Javanese o. PMJ hiatus in this environment became Malay h:

*paar, NJv. por 'tray', Mal. pahar 'pedestal tray'.

In the fo11owing compafison the Javanese and Ma1ay cognates are lading:

*haay, Snd. hiay, Mad. ~way 'to yawn'.

2.15.3.7. PMJ hiatus between *u and *a. Tne only comparison for *hiatus in this environment is the following. In Sundanese the ultimate vowel assimilated to the penultimate vowel and the two vowels contracted to OJv. u223 :

*luah (PHN, Tag. 16:haq 'tear' : only ID [1953a] luhaq), Snd.

luquh 'to arrive weeping' , OJv. luh, NJv. 16h 'tear'.

2.15.3.8. PMJ hiatus between *u and *i: see *Buit, p. 129.

2.15.3.9. PMJ hiatus between *i and *a: see *hian' p. 170.

2.15.4. Ambiguous reconstructions. An ambiguous reconstruction with intervocalic [h] is employed when a Javanese cognate is lacking and the forms in the other three languages are ambiguous as to whether PMJ had an *h or *hiatus: e.g. *li[h]at, Snd., Mal. liat, Mad. lèyaq 'tough, tenacious'.

2.16. PMJ double consonants af ter vowels ot her than *a. In section 2.5.1. we dealt with the reconstruction of double consonants af ter *a. In this section we shall deal with the reconstruction of PMJ double consonants af ter vowels other than *a. PflJ double consonants are reconstructed whenever Madurese exhibits double consonants and where there is no reason to suspect these Madurese words to be borrowings. The double consonants were reduced to single consonants in Sundanese, Malay, and Javanese.

2.16.1. PMJ -nn-:

*Bannak, OJv. banak, NJv. banaq 'entire', Mal. banaq, Mad. b'nnaq 'many, a great deal'.

*minnak (PWI, NgO. minak 'oi1 ': minak), Snd. minak (L), OJv. minak, NJv. minaq (0), Mal. minaq, Mad. mènnaq 'oil'. 2.16.2. PMJ -IJIJ-: *la1JlJit (PAN, Fi. la~i 'sky': la~it~), Snd., OJv. la!)it, NJv. la~é't, Mal. la~et, Mad. la~~èq 'sky'. *la~~uy (PHN, Tag. lanóy 'to swim'), NJv. laDi, Mad. lanDOY 'to swim' .

2.16.3. PMJ -11'-:

*qallih (PAN, Fi. yali 'be absent': ?aliq), Snd. qalih (H) 'to move, change', OJv. alih 'two, to change', NJv. alé'h (L), Mal. aleh, Mad. all~ 'to move, change'.

2.16.4. PMJ -ss-. The Madurese reflexes of *-ssare -ssor -cc-. Thus, we can posit two Madurese dialects, one in which *-ssappears as -ssand another in which *-ssbecame -cc-. In the following instances it is best to explain the existence of competing forms as borrowings from the -ssdialect into the -ccdialect, or conversely:

*qassin (PAN, To. m-ahi 'sour': ?asliLa), Snd. qasin, OJv. asin, NJv. asèn, Mal. asen, Mad. assèn, accèn 'salty'.

*Bassah (PAN, Fi. sava 'to wash': basaq), Snd. basih 'wet', OJv. wasah224 'washed', Mal. basah, Mad. bássa, b/cca 'wet'. *Bassuh225 (PAN, Fi. savu-i 'to wash the canoe': basuq), OJv. ~, / aO-wasuh, NJv. wasoh, wasoh (0), basoh (0), Mal. basoh, Mad. /, / '" basso, baccu 'to wash'.

In the following instance PMJ -ssprobably is from a PAN cluster -lswith assimilation of the first consonant to the second consonant:

*hassam (PAN, Sm. m-asa 'sour': qalsam), Snd. has~m226, OJv. hasam 'sour', asöm227 'tamarind', NJv. asam 'tamarind', 'sour' (0), Mal. asam 'sour, tamarind', Mad. assam, accam 'tamarind' .

In the following two comparisons the Madurese cognates occur without a by-form with -cc-:

*qassah (PAN, Fi. yaza 'to rub, whet': h24asaq), Snd. qasah, OJv. an-asah, NJv., Mal. asah, Mad. assah 'to grind down, whet' •

*qassak (PAN, Fu. moso 'done, well-cooked': masak), Snd. qasak, OJv. t-asak, Mal. m-asaq, Mad. m-assaq 'ripe, done'.

2.16.5. PMJ -dd-:

*qaddas (PW!, TBt. qadas 'fennel': adas), NJv., Mal. adas, Mad. addhás 'fennel'.

*qadduq (PHN, Tag. q~:ro 'to put forward in a prodding manner': adu), Snd. qaduq, OJv. a!}-adu, NJv. adu (L). Mal. adu, r~ad. addhu(h) 'to pit, match one against the other'.

*kaddut, Snd. kadut 'paunch, belly', also: 'a sack which is spread out', OJv. kadut, NJv. kad6t, Mal. kaen kadot, Mad. kaddhuq 'sack-cloth'.

*laddi!}, NJv. 1 adé!} , Mal. lade~, Mad. S. P. laddhi~ 'knife'.

2.16.6. PMJ -jj-: see *tujjaw, p. 65.

2.16.7. PMJ - pp-:

*luppaq (PWI, TBt. lupa 'to forget': lupa), OJv. k-in-a-lupa-n 'forgotten', NJv. lupÖ (0) (0), Mal. lupa, Mad. lOPPé!(h) 'to forget' .

2.16.8. PMJ -tt-: *qattas (PHN, Tag. taq's 'height': a(n)tas), NJv. atas (0) (0) 'clearly visible', Mal. atas 'position over or above', Mad. attas 'above, on top'.

*labbuh (PWI. TBt. labu 'to cast the anchor': labuh), Snd. labuh (L) 'to fall, cast anchor' , OJv. ma-labuh 'to throw o.s. into, plunge into', NJv. lab6h 'to cast anchor, throwaway', Mal. laboh 'to lower by means of a cord', Mad. labbhu 'to throw o.s. into the water in order to drown o.s.'. *rubbuh (PWI, TBt. robo 'to fall': Rubuh), Snd., OJv. rubuh, NJv. rUb6h, Mal. ruboh, roboh, Mad. robbhu 'to fall'. *gag~in (PHN, Tag. g~:riD 'ivory': gadiD)' Snd. gadin 'ivory. tusk', OJv. gaqin 'tusk', NJv. gag€D' Mal. gaden 'ivory, tusk', Mad. gh'9ghin 'ivory'. *gaqquD (PWI, TBt. gadoD 'earth-fruit': gaqun), Snd. gadu~, OJv. gagun' NJv. gagc5n' Mal. gadoD' Mad. ght99huD 'a wild earth-fruit' . *kagqal, Snd. kadal, OJv., NJv. kagal, Mal. Java kadal, Mad /. ka<;lcJha 1 'grass-l i zard' .

2.16.9. PMJ -kk-: see *[Bb]ukkaq, p. 193.

2.16.10. PMJ -bb-:

2.16.11. -- PMJ -dd-: .. -

*hag9an (PAN, Fi. yaran-i 'arrnament': h13aga~), OJv. um-a9aD 'opposition', haqa~-an 'waited for', NJv. aga~, Mal. (h)adaD' I Mad. aqghaD 'to block the way, lie in wait'.

I *pa99an (PHN, Tag. pa:ran 'meadow': pa(n)dan), OJv. a-macJa~ 'to clarify, clear', NJv. paqa~ 'light', 'laid open (of ground),

/ (0), Mal. pada~ 'treeless waste land', Mad. pa9ghaD 'clear, bright'.

2.17. The infixed nasal. In Malayo-Javanic languages cognates sometimes differ among themselves in that one reflects a single intervocalic consonant and another the same consonant preceded by a nasal. This phenomenon appears not only in cognate sets implying a PAN etymon, but also in those implying a PMJ etymon. The function of this difference in the presence or absence of a nasal is unclear. Oempwolff cites reconstructions whose cognates differ on1y in the pres en ce or absence of a nasal by enc10sing the nasal in parentheses, e.g. *ka(m)pit. However, when Merina exhibits for example -tsfor anticipated -sas reflex of *s (cf. *Bisik, p. 129) Oempwolff posits a 'reduced nasa1 cluster' (2.91 and 2.95). For a discussion of Dempwolff's 'reduced nasal cluster', see Oyen (1971:46). Following Oyen (first in Oyen 1953a), we cite two reconstructions. The reconstruction without the nasal is cited first followed by its reflexes. Then the reconstruction with the nasal is cited followed by its reflexes. We shall refer to this nasal as the 'infixed nasal '. The following are instances of reconstructions of this kind.

2.17.1. *-1-; and *-pl-:

*jal[iq, ayJ, NJv. jali, Mal. jali batu 'kind of tree'; and *ja~l[iq, ayJ, Mad. jh{91é(h) 'kind of tree'.

*sa1ir, Mad. B. P. sal~r 'to have apparently on1y one testicle or with testicles one of which is very small'; and *sa~lir, Snd. sa~lir 'to have only one testic1e', OJv. sa~lir 'to have a large scrotum at one side', NJv. sa~lér 'with a small testicle', Mad. S. san1ér 'to have on1y one testic1e, sma11 testic1e' . 2.17.2. *-s-; and *-os-: \ \ . *mus[iq, ay], Mal. musi, Mad. mose(h) 'caraway'; and *muOS[lq, ay] NJv. munsi 'caraway'. *musuh (PWI, TBt. musu 'enemy': mu(n).suh), Snd., OJv. musuh, NJv. mus~h (0), Mal. musoh, Mad. moso 'enemy'; and*munsuh, / OJv. muDsuh, NJv. munsoh 'foe, enemy'. 2.17.3. *-j-; and *-nj-: *qujur, Snd. k-ujur 'length of body', OJv. m-ujur 'lengthwise', NJv. ujór 'side, direction (in which s.t. is stretched ou~'; and *qunjur, Mal. unjor, Mad. Önjhur 'to stretch out, thrust out, proj eet' . 2.17.4. *-p-; and *-mp-: *hapaq, Snd. hapaq 'void, empty'; and *hampaq, Mal. (h)ampa , vo i d, empty'. *kapak (PHN, Tag. kapák 'sound of beatings': ka(m)pak),OJv. k-um-apak 'to haek', Mal., Mad. B. P. kapaq 'axe, hatehet'; and *kampak, Snd. kampak, NJv. kampaq 'axe, hatehet'. *kapur, NJv. kapór 'udder'; and *kampur, Mad. kampor 'udder'. *tapiq (PAN, Fi. tavi-a 'to sweep': tapi[?h]), Snd. tapiq 'to winnow', OJv. t-in-apy-an 'winnowed', NJv. tapèn (tapi+an) 'to.winnow'; and *tampiq (tampi[?h]), Mal. tampi, Mad. tampè(h) 'to winnow'. 2.17.5. *-t-; and *-nt-: *latun, NJv. 1anà 1ató~ (0) 'mineral oi1'; and *lantuD ' Snd. minak 1antun, NJv. 1an6 1antón' Mad. m~nnaq 1anto9 'mineral oil' . *qutah (PAN. Sa. m-oa 'to vomit': [q?h]utaq). Snd. qutah. OJv. m-utah. NJv. m-utah. Mad. öta 'to vomit'; and *quntah ([q?h]untaq). Mal. m-untah 'to vomit'.

2.17.6. *-k-; and *-nk-:

*Bukus (PWI. TBt. bukkus 'bundle': bunkus). Mal. bukos 'bundle. packet'; and *Bunkus. Snd. bunkus 'bundle'. OJv. w-in-unkus / / 'seales'. NJv. wunkos 'wrapped in aleaf'. bunkos 'package'. *jukun (PW!. NgO. jukon 'dug-out canoe': ju(n)kun). Snd .• OJv. jukuD' NJv. juktln. Mal. jukon. Mad. jhukÖn 'dug-out canoe'; and *junkun. NJv. junk6n (0). Mal. junkon 'dug-out canoe'. *takis (PWI. NgO. tankis 'to parry': ta(n)kis). Snd. takis. OJv. t-um-akis-akan 'to parry'; and *tankis. NJv. ta~kés 'defense' (B). 'shield' (0), 'dam. dyke' (0). Mal. tankes 'to parry by striking aside'. Mad. tankés 'to parry'. also Mad. S. 'dyke along the sides of a road'.

*tukup (PAN. Fi. tuku 'to cover the head with ashes'). OJv. tukup 'to cover'. NJv. tukóp 'to hold the hand on or in front of s.t.; and *tunkuP. Snd. tunkup 'to hold the hand on s. t.'.

2.17.7. *-B-; and *-mB-:

*BuBun (PAN. Fi. taqo-fufu 'ridge-pole': bubun). OJv. wuwun. NJv. WUWÓn 'ridge (of the roof)'. Mal. bubon 'swelling up in a dome-like mass from below, roof'. Mad. bubu~ 'ridge (of roof)'; and *BumBu~. Mal. bumbon 'roof'. *guBin. NJv. guwén 'with a hare-lip'. gówéa 'of earthenware and similar objects. where a piece is broken off at the edge';

and *gumBi9' Mad. ghumbi9 'with a piece broken off (of porce1ain, earthenware)', a1so: 'hare-lip'. *tuBuh (PAN, Fi. tumbu 'to grow': tubuq), OJv. tuwuh, NJv. tuwÓh 'to grow', Mal. tuboh 'body'; and *tumBuh, Snd. tumbuh, Mal. tumboh, Mao. tombu 'to grow, sprout'. *taBaR2 (PWI, lBt. tambar 'antidote': ta(m)baR), Snd. tawar 'to make onese1f immune to poison', pan-awar 'antidote' , OJv. tawa 'antidote', NJv. pan-owo 'antidote', Mal. pan-awar 'antidote', Mad. pan-abár 'antidote'; and *tamBar, OJv. tamba 'remedy, medicine', NJv. tamba (L), Mal. tambar 'to hea1, cure'. NJv. pan-awar 'antidote' is treated as a 10an from Ma1ay. Snd. tawaq 'to make onese1f immune to poison' and tambaq (H) 'remedy, medicine' and Mad. tambhá(h) 'remedy, medicine' are treated as 10ans from Javanese. *huBiq (PAN, Fi. quvi 'yam': qubi[?h]), Snd. huiq, OJv. (h)uwi, NJv. uwi, Mal. ubi, Mad. ~bi(h) 'yam, tuber'; and *qumbiq (qumbi[?h]), Mal. (h)umbi 'bu1b, bu1b-root'.

2.17.8. *-b-; and *-mb-:

*guba1, NJv. guba1 (0) 'to be caught on'; and *gumba1, Mad. ghumbha1 'to ho1d each other, stick'. *sabu~, Mad. sabhu9,'to prolong by taking another piece'; and *sambu~, Snd. sambu~ 'to prolong', OJv. sambu~ 'contributions', NJv. sambó~ (L), Mal. samb09' Mad. sambhu~ 'to prolong by taking another piece'.

2.17.9. *-9-; and *-n9~: *gida1, .. Mad. ghidhá'l 'loose'; and*gindal, .. OJv. gindal 'separate', NJv. gincJal (0) 'loose, separated', Mad. B. ghindhá'l 'loose'.

2.17.10. PMJ doublets, one exhibiting a double consonant, the other a nasal cluster. The following two examples are instances of Madurese competing forms, one with a double consonant, the other with a nasal cluster. The cognates in the other Malayo-Javanic languages all exhibit a nasal cluster. We reconstruct a PMJ doublet, one exhibiting a double consonant, the other a nasal cluster. The Madurese competitive form with the nasal cluster can be attributed to a borrowing from Javanese or Malay:

/ , *gattu~ (PWI, TBt. gattun 'to hang'), Mad. ghatto~ 'to hang'; and *gantu~, Snd. gantu~ 'to hang', OJv. ma-gantuu-an 'hanging', NJv. gantö~, Mal. ganto~ 'to hang'; cf. Mad. ghántà~ 'to hang'.

*likku~ (PWI, TBt. lehu~ 'be bent': li(~)ku~), Mad. lekk~~ 'to form a circle around'; and *li~ku~, Snd. li~ku~'to surround', NJv. lè~k~~-an 'bay', Mal. li~ko~, le~ko~ 'curve, curved, hollow'; cf. Mad. lè~kà~ 'to form a circle around'.

There are two other possible hypotheses with regard to forms like these: (1) one could propose the reconstruction of a PMJ doublet *gattu~, *ganttuu' for which all languages but Maduresecontinue the form with an infixed nasal (where the double consonant was reduced to a single consonant), while Madurese continues the form lacking it. Such a hypothesis can, however, not be made lightly, since it involves the violation of what has been said earl ier about the PMJ morpheme structure (and also applies to the PAN morpheme structure), i.e. clusters consist of no more than two consonants. (2) One could argue that the following alternation appeared in PMJ. When a nasal was infixed in a base with a double consonant, the double consonant was replaced by a single consonant: e.g •• *gatu~ alternates with *gattu~ in a form *gantuD.

Other reconstructions of doublets. one with af! intervocalic nasal cluster. the other with an intervocalic double consonant are the following:

*Battis (PWI. TBt. bitis 'lower leg': bitis. ba(n)tis). Snd. bitis. OJv. watis. Mal. bates. Mad. battés 'lower leg'; and *Bantis. NJv. want{s 'lower leg'.

*bakkah (PAN. Sa. hoka 'to burst': bakaq). Snd. bikah 'to expand. open up. blossom'; and *ba~kah (ba9kaq). NJv. baDkah 'to split'. Mal. ba~kah 'to become cracked. split'. Mad. B. P. and Mad. Kang. bha~ka 'to split'.

*qattah (PWI. TBt. qata 'to eat raw meat': a(n)tah). Snd. qatah. Mal. Ked. and Mal. N.S. m-atah. Mad. m-atta 'raw. not ripe'; and *qantah. Mal. antah 'husk. sheath (of grain)'.

*mattah. OJv. matah 'raw. not ripe'; and *mantah. NJv .• Mal. mantah 'raw. not ripe'.

*[Bb]ukkaq (PAN. Sa. huqe 'to uncover': buka?). Snd. bukaq 'to open. begin'. NJv. buko 'introduction'. bukaq 'open. begin'. Mal. buka. bukaq. Mad. bukkaq 'open'; and *[Bb]u~kaq. Mad. Kang. bu~kaq 'open,228.

*habbun (PHN. Tag. qambÓn 'drizzle': a(m)bun). Snd. qibun. OJv. (h)abun. NJv. (a)b6n. Mad. (ab)bhun 'dew'; and *hambun. Mal. ambon 'dew'.

*qappat (PAN. To .• Fu .• Sm. fa 'four': x2apata). Snd. qopat229 • OJv. pa-pat, NJv. pa-pat (L) 'four'; and *qampat. Mal. ampat, Mad. ampaq 'four'.

*qa~~ik (PWI, TBt. qotik 'few': a(n)~ik), Snd. sa-4tik 'few', 1-4tik (L) 'small'; and *qan~ik, NJv. m-an~éq 'extremely small' .

*kappit (PAN, Fu. kopi-kopi 'to be squeezed in': kapit), Mal. kapet 'pressure between two connected surfaces, e.g .. between two fingers or between the arm and the body', Mad. kappèq 'to carry under the arm'; and *kampit, NJv. kampét 'to carry under the arm'.

, / *kattus, Mad. kattos 'to brag'; and *kan~us, NJv. kan~os 'to brag, big talk'.

*qukkil (PWI, TBt. qukkil: u!Jkil), Mad. Ökk~l 'to lever up slightly'; and *qu!Jkil, Snd. qU!Jkil, NJv. uJké'l, Ó~k~l (0), Mal. u9kel 'to lever up slightly:

2.18. PMJ doubled monosyllables. In section 2.5.3. we shortly dealt with PMJ doubled monosyllables in connection with the discussion of the origin of Madurese double consonants. In this section we shall deal with the reflexes of Pt4J doubled monosyllables in all four languages.

2.18.1. The Sundanese reflexes. The Sundanese reflex of a PMJ doubled monosyllable is a word of the shape C1V1C1V1C2 if *C2 of the PMJ word of the shape *C1V1C2C1V1C2 is a consonant other than a nasal, *g or *k. If *C2 is one of the latter, the original shape is retained: *OagOag, Snd. d4gd4g 'to run in masses', NJv. ~a~ag (0), Mal. dadaq, Mad. ~akgak, Mad. P. ~a~~ak 'crowded together'. *siksik, Snd. siksik 'to cut up fine', NJv. siséq 'to scrape off', Mad. sèksèk 'to slice'.

*ta~ta~ (PHN, Tag. tantá9 'pulled'), Snd. ta~ta~, OJv. ma-na~ta~, NJv. tanta9' ta~ta~ (0), Mad. ta~ta9 'to cha11enge'. *tultu1, Snd. tutu1, NJv. tut61, Mad. t~ltb1 'spotted'.

Sundanese has a number of words of the shape C1V19C1V1n. The question is whether these words are continuations of PMJ doubled monosyllables with ~ as preconsonantal nasalor whether they are continuations of a doubled monosyllable with n as the preconsonantal nasal. If the preconsonantal nasal was *~, then one might posit that all languages but Sundanese exhibit partial assimilation of this consonant to the following stop. If the preconsonantal nasal was *n, then one might posit that Sundanese exhibits dissimilation.

One can argue that evidence from New Javanese supports the reconstruction of *9' In one instance New Javanese exhibits doublets, one with preconsonantal 9' the other with preconsonantal n. The competitive form with 9 is marked by Pigeaud (n.d.) as archaic (see under *tu9tun). Furthermore, there is an instance of a comparison between Tagalog and Malayo-Javanic forms (see *ciDcin) suggesting that *9 perhaps appeared in preconsonanta1 position a1so in proto-1anguages of higher order such as PHN.

The following comparisons are reconstructed with a preconsonantal

*baDban (PHN, Tag. bambán 'bast, fiber': banban), Snd. baDban, NJv. bamban 'kind of reed'.

*buDbun (PHN, Tag. bumbón 'dam of branches and twigs to attract fish in rivers and lakes': bunbun), Snd. bU9bun 'a deep hole in the water closed up with a curtain of bamboo-sticks', Mal. bumbun 'hut like a bee-hive or bell-tent used by fpw1ers, hunters, or fishermen to concea1 themse1ves'. *ci~cin (PHN, Tag. si~s{~ 'ring': cincin), Snd. ciDcin (H), OJv. cincin 'ring', NJv. cinc(n 'ring' (0), 'sewing ring', Mal. cincen 'ring'. Taga10g fina1 ~ cou1d be exp1ained as an assimi1ation of the fina1 consonant to the fina1 D of the initia1 sy11ab1e.

*da~dan (PWI, NgO. dandan 'ship-bui1ding': dandan), Snd. da~dan (L) 'to get dressed, ready, c1ear', OJv. ma-dandan 'got ready', NJv. dandan 'to get dressed, ready', Mal. dandan 'straightaway, there and then', Mad. dhándhin 'to get dressed' •

*ja~jan, Snd. jaDjan 'crocodi1e-co1ored', NJv. janjan (0) 'pa1e (yellow)', Mad. jh(njh{n 'white-ye11ow'.

*tu~tun (PHN, Tag. tunt6n 'conclusion': tuntun), Snd. tu~tun (L), OJv. tuntun, NJv. tuntón, tUDt~ (0), Mal. tunton, Mad. t~nt~n 'to lead'.

2.18.2. The Javanese and Ma1ay reflexes. In Javanese and Malay the reflex of the PMJ doubled monosyllable is C1V1C1V1C2 if *C2 is a consonant other than a nasal. If *C2 is a nasal, the original shape of the PMJ doubled monosyllable is retained, but the preconsonantal nasal assimilates to the following consonant if it is a stop in Malay and standard Javanese. In Old Javanese and in a dialect of New Javanese the nasal does not assimilate:

*BalBal, NJv. babal (0) 'difficult to extract, stupid', Mal. babal 'dull, stupid', Mad. bal bal 'tough (e.g., of wood which is difficult to be prepared for usage), obstinate, not p 1 easant ' .

*Bu9Bu~ (PAN, Sm. pupu 'quiver': bU9bun), OJv. bU9bun 'ho110w bamboo', NJv. bumbó'n 'quiver, boiler', Mal. bumbo~ 'a water-vesse1 made from a joint of bamboo', Mad. bunbu~ 'cooker in the shape of a cylinder (made of bamboo)'. *caDca~, Snd. canca9 'bind, tether', OJv. c~in-a~can 'tethered', NJv. canca~, ca~ca9 (D), Mad. canca9 'to tether (a horse)'. *DamDam (PHN, Tag. liml{m 'to sit over eggs': damdam), NJv. ~an~am, ~am~am 'to plan secret1y, in quiet', Mal. dandam 'yearning, feeling of love or spite', Mad. ~am9am 'friendly, nice (of s.o. who usual1y is not ta1kative)'.

If *C1 is *r and *C2 is *m, *C2 became New Javanese ~, but was retained as m in Malay:

/ " , *rumrum, NJv. runrom (B), Mal. rumrom, Mad. romrom 'to show love to, fond1e'.

If both *Cl and *C2 are nasals no assimilation occurs:

*namnam, Snd., NJv., Mal., Mad. namnam 'kind of tree'.

2.18.3. The Madurese reflexes. As was pointed out in section 2.5.3. preconsonantal *C2 in a doubled monosyllable assimilates to the following consonant in some instances. This assimilation, it was said, probably is the regular development in the B. and P. dialects. The evidence presented below shows that such an assimilation only occurs if *C2 is a stop, *r, or *s. Further research will have to be carried out to determine whether this assimilation also occurred when *C2 is a consonant other than those just mentioned.

In section 2.5.3. two comparisons were presented, one in which the B. dialect shows assimilation (*taptap), and one in which both the B. and P. dialects show assimilation (*tastas).

In the following comparisons Madurese exhibits doublets, one showing assimilation, the other the absence of it:

*cipcip, OJv. an-icip 'to kiss', NJv. cic{p-cic{p (0)(0), Mal. cicep, Mad. c~pcèp, c~cc~p 'to test with the tip of the tongue'; cf. Snd. qi-cip 'to try, taste (food), test'. / *cupcup, OJv. an-ucup 'to suck', NJv. cucop 'to put the nozzle to the mouth, drink', Mal. cucop 'ta sip, smell, kiss', Mad. cbpcop, cèccop 'to suck at the nozzle of s.t.'. *butbut (PAN, Sa. huhu 'to pull out'), Snd. bubut 'a person who operates a lathe', OJv. b-in-ubut 'ta provide with turner's work', NJv. bubót 'to pull out (hair), turn (wood on a lathe)', Mad. bhutbhut 'to pull out (hair, feathers)', bhubbhuq 'to turn (on a lathe)'. In the following instances Madurese exhibits a cognate which has assimilation without a by-form lacking assimilation: *~apQap (PAN, Fi. rara 'kind of tree'), Snd. dadap, OJv., NJv. ~a~ap, Mal. dadap, Mad. qh{qqháq 'kind of tree'. *gargar (PHN, Tag. gadgád 'shelled out': gaggag), Snd. ragrag (metathesis) 'to fall (off)', NJv. gagar 'without leaves, bare (of a tree)', Mad. ghágghár 'ta fall (off)'. *siksik230 (PAN, To. hihi 'muscle': sisik, siksik), OJv. sisik, NJv. siséq, Mal. siseq, Mad. sèssèq 'scale of a fish'. *s-ul-upsup, Snd. s-ul-usup, NJv. s-l-usop, / Mad. s-al-ossoq " " 'ta crawl in'.

In instances like the following Madurese exhibits cognates which do not show assimilation of preconsonantal *C2 to the following consonant and for which a by-form showing assimilation is lacking: *juljul (PHN, Tag. duldó'l 'to force into': ,zulzul), NJv. juj61, . Mal. jujol, jojol, Mad. jhuljhul 'sticking out'.

*kudkud, OJv. a-kukud-an 'to pack together and clear out', NJv. kuk6d 'to pack up', Mad. kotkot 'to pack up, gather, glean'. *kurkur (PAN, Fi. kukuva 'to scratch': kugkug), Snd. kukur 'to / rasp', OJv. a-kukur 'to scratch away, smooth', NJv. kukor 'to scratch', 'rasp, file' (0) (0), Mal. kukor 'to rasp', Mad. k6rkor 'to scratch off (out)'.

*sitsit, Snd. sisit 'scale of a fish', Mad. sètsèt 'covered with scales, scales'.

In the following instances the Madurese double consonant might be attributed to the assimilation of a final h of the first syllable of a PMJ doubled monosyllable to the initial consonant of the second syllable. It is difficult to determine whether this is a dialectal development.

Thus, we might consider the possibility of reconstructing a form such as *cahcah (PAN, Sa. tata 'spread all over, dispersed': cacaq) for the comparison of the following forms: OJv. a-caca-cacah-an 'hacked', NJv. cacah 'to hack into fine pieces', Mal. cacah 'to prick a pattern on a flat surface, e.g. on pastry', Mad. cacca 'to carve, cut into small pieces'. Similarly, we may reconstruct a form *tuhtuh for the comparison of the following forms: Snd. tutuh, OJv. ma-nutuh, NJv. tutÓh, Mal. tutoh, Mad. t~ttÖ 'to poll a tree'.

2.18.4. A list of PMJ doubled monosyllables. In the following list some more PMJ doubled monosyllables are cited to further illustrate their reflexes in the four languages:

*parpar, NJv. papar (D) (0) 'equa1, to flatten' , Mal. papar 'flat, smooth', Mad. B. parpar 'to flatten'. *paspas (PHN, Tag. pasp's 'to dust'), Snd. papas 'to chop off', OJv. ka-papas 'destroyed', NJv. papas 'broken off' (0), 'cut off', Mal. papas 'to strip off, remove', Mad. paspas 'to cut off'. *paspas-an, Snd. papas-an, Mad. paspas-an 'kind of c1imber'. *sapsap (PAN, Sa. toto 'to suck up'), Snd. s~s~p (L), NJv. sasap, Mal. sasap, Mad. sapsap 'to suck'. *OanOan, OJv. 9anga~, NJv. gangan (B), Mal. buron danda9' buroD dandaD' Mad. ~á~9án 'crow'. 231 *ma~maD' Snd. manman, NJv. manman (0) (0), Mad. manman 'uncertain, to doubt'.

*BarBar, Snd. babar in babar 1ayar 'to unro11 the sai1s', OJv. b-in-abar 'spread out', NJv., Mal. bab~r 'to expand, open out, boom out a sai1', Mad. b'rb~r 'to spread out'.

*[O~]in[O~]in (PAN, Fi. riri 'screen against wind': ~i~gi~), Snd. di~di9' NJv. ~ing'~ (0), (0), 9iD~én (0), Mal. dinde~ 'screen, inner wa11 '.

*ju~jun, Snd. ju~juD 'to lift up, pick up', OJv. pa-ju~jun'to pride o.s. on (?)', NJv. junjó~, jU9jÖ~ (0) 'to lift up', Mal. junjoD 'support on the head', Mad. jhuDjhun 'to lift up to over the head'.

3. CONCLUSION

This study has been an attempt to contribute to a better knowledge of the past history of four important Austronesian languages, toward the reconstruction of their last proto-language, and also toward the reconstruction of the proto-language of highest order.

We have been able to reveal the patterns of the reflexes of PAN phonemes in PMJ and in its four daughter languages. The following tables recapitulate our results. These tables are intended to give a general overview of the correspondences between PAN and PMJ phonemes and their reflexes in the four languages. The tables contain the PAN phonemes as reconstructed by Dyen. The second column contains the PMJ phonemes, taking account of a number of mergers prior to this stage and a number of new proto-phonemes. The next five columns contain the main reflexes of these proto-phonemes in Sundanese, Old Javanese, New Javanese, Malay, and Madurese. Dialectal differences are not indicated. Furthermore, only the reflexes in initial, intervocalic, and final position are listed.

~ In the final closed syllable.

Q/ See section 1.8.3.1.

Y Before pause.

rY When followed by u.

!V Before h.

fJ In the final syllable.

gj Before q.

W Before i .

jj q between identical vowels.

jj w after u, o.

11 y after i, e.

11 w after u.

!!!I y after i.

For many proto-phonemes the evidence seems undisputab1e. Some major prob1em areas, such as the origins of the Sundanese and Old Javanese pepets have been solved. We have been ab1e to suggest corrections and disambiguations of some of Dempwolfr's and/or Dyen's reconstructions.

Some major prob1em areas remain. It is interesting that the resu1ts of this study offer on1y slight support for the subgrouping of Ma1ay with Madurese which was arrived at by the 1exicostatistica1 method. The on1y phono1ogica1 evidence that has appeared in support of such a subgrouping is the merger of *R and *r ~ r shared by Ma1ay and Madurese. On the other hand, one might argue that the percentages of cognation between Ma1ay and Madurese are inf1ated because possib1y many of those Madurese words which are ambiguous as to whether they are inherited or borrowed are actua11y borrowings from Ma1ay. Evidence against the subgrouping of Ma1ay and Madurese is perhaps to be found in the morpho1ogy. The causative suffix is -kin in Sundanese and -kan in Ma1ay. It is -akan in Old Javanese, -aké (L) and -akan {H} in New Javanese, and -aghi in standard Madurese, -aghan in Kangean Madurese, and -akan in Bawean Madurese. Furthermore, both Javanese and Madurese share an imperative locative suffix: OJv. -ana, NJv. -{n}bnb, Mad. -áná~-ana. This suffix does not appear in Sundanese or Ma1ay.

Other prob1em areas which remain are the appearance of both u and 0 and i and e in Ma1ay non-fina1 sy11ab1es for PMJ u and PMJ respective1y, and u, Ó, 0, and i, é, è in Javanese non-fina1 sy11ab1es and 0 and è in Javanese fina1 sy11ab1es for PMJ u and PMJ i respective1y. We have treated them as dialecta1 ref1exes of single protophonemes. In the case of Malay we had to resort to a hypothetical dialect; in the case of Javanese we were able to identify a dialect in which 6 and é in non-final syllables apparently are the regular reflexes of PMJ u and PMJ i respectively.

Also problematic is the tentative reconstruction of *Rl' *R2' and *R3' The evidence presented here is not as solid as might be desirable, especia11y for the correspondence assigned to *R3' The weakness in this reconstruction has already received comment.

It cannot be determined at this point whether PMJ double consonants or the distinction between PMJ Band PMJ band the distinction between PMJ D and PMJ Q also appeared in earl ier proto-1anguages or in the ear1iest proto-1anguage. Many of the PMJ reconstructions which contain double consonants cannot be associated with PAN reconstructions made by Dempwo1ff and Dyen. The instances in which such an association can be made are the fo110wing: PMJ laU9it, p. 185; PMJ qassin, p. 185; PMJ Bassah, p. 185; PMJ Bassuh, p. 185; PMJ qassah, p. 186; PMJ qassak, p. 186. The PAN reconstruction *h13aqa~ with which we associated the PMJ reconstruction *haqqan can be regarded as weak because of the poor semantic fit of the eastern with the western form. This difference in meaning raises the possibility that the eastern form, despite its correspondences, may have been incorrectly associated with the western words. Furthermore, we have been ab1e to show that in one instance a PMJ double consonant is from a PAN cluster of non-identica1 consonants, name1y *-ls- (cf. *hassam, p. 186). It is therefore not inconceivab1e that perhaps a1so other PMJ double consonants might be the reflexes of clusters of nonidentica1 consonants that appeared in earl ier proto-1anguages or in the earliest proto-language.

In respect to the PMJ Bib and PMJ DI9 distinctions it appears that most PMJ reconstructions containing PMJ b or PMJ 9 cannot be associated with PAN reconstructions. Furthermore in most of those instances in which such an association can be made, the PAN reconstrlJctions are weak, because the meaning of the eastern words are different from those of the western words. In the following instances one might argue that the PMJ reconstructions containing *b and *9 can be associated with sound PAN correspondences: PMJ b: PMJ butbut, p. 198; PMJ rabbah, p. 140; PMJ lambut, p. 141; PMJ baffaw, p. 88; PMJ balay, p. 90; PMJ 9: PMJ 9apgap, p. 198. The following PMJ reconstructions can only be associated with rather weak PAN reconstructions: PMJ b: PMJ bakal, p. 136; PMJ bakkal, p. 137; PMJ bulad, p. 138; PMJ labur, p. 139; PMJ tabbuk, p. 140; PMJ kambaU' p. 140; pr~J tambaD' p. 141; PMJ kabbaw, p. 85; PMJ 9: PMJ 9ampar, p. 153; PMJ tan9aq, p. 156; PMJ [D~]agaq, p. 157.

Furthermore a PMJ reconstruction containing "U or *9 can only in a few instanees be associated with a sound PHN reconstruction: PMJ tabbas, p. 77; PMJ tambak, p. 141; PMJ baR1ah, p. 137.

As for PWI, it is not impossible that the PMJ Bib and PMJ D/g distinctions will have to be assigned to this level of reconstruction. A quick check through a Buginese dictionary revealed that the Bib distinction perhaps also appears in this West Indonesian language: e.g., Bug. were 'to give' (cf. PMJ BaR1R1ay, p. 126), Bug. wuta 'blind' (cf. PMJ Butaq, p. 130), but Bug. bakuD 'lily' (cf. P~lJ bakun), Bug. bata 'brick (cf. pr~J bataq, p. 137). However, this connection is not obvious, since in many instanees Buginese exhibits b for an anticipated w: e.g., Bug. batu 'stone' (cf. PMJ Batuq, p. 128). As for the D/~ distinction, evidence from, Makassarese seems to indicate that this distinction is not 1imited to Malayo-Javanic 1anguages: e.g., Mak. rau~ 'leaf' (cf. PMJ Daun, p. 148), Mak. ruwa 'two' (cf. PMJ Duaq, p. 148), Mak. tanruq 'horn' (cf. PMJ tanDuk, p. 156), but Mak. dasereq 'f100r' (cf. PMJ çasar, p. 153), Mak. daparaq 'anything that is used to stand or 1ie on such as wood or stone' (cf. PMJ ~ampar, p. 153).

We conc1ude that the distinction between *B and *b, and between *D and *~ must be reconstructed for PMJ. It is hard1y conceivab1e that in all instances in which we have reconstructed *b and *9, the Sundanese, Javanese, and Madurese forms cou1d be exp1ained as due to secondary deve10pments. The evidence for the *B/b distinction seems better attested, since it is supported by Sundanese morphophonemics (see section 2.12.4.). It might be argued that these two distinctions did not appear in earl ier proto-1anguages (except perhaps in PWI) or in the ear1iest proto-1anguage, given the few instances in which a PMJ reconstruction containing *b or *~ is associab1e with sound reconstructions made for earl ier proto-1anguages or the ear1iest proto-1anguage. If, however, we shou1d succeed in discovering that one of these distinctions can be assigned to an earl ier proto-1anguage, the chances of it being the *B/b distinction are better, since the evidence for *b is stronger than for *~.

A1though there is good reason to be cautious about the phonetics of proto-phonemes, one might attempt to put forth hypotheses as to the phonetics of those proto-phonemes whose ref1exes show no or on1y very 1itt1e phonetic variation. The proto-phonemes *p, *t, *!. *c. and *k probably were all voiceless stops, *p being bilabial, *t dental, *~ alveolar, *c palatal, and *k velar. PMJ 1 probably was.a lateral, and PMJ r probably was an apical trill. The proto-phonemes *m, *n, *n, and *0 probably were bilabial, dental, palatal, and velar nasals. respectively. The phonetic values of what has been reconstructed as *b, *d. *9, *j. and *g are more problematic. Their reflexes are voiced lenis stops in Sundanese and Malay; in Javanese their reflexes are voiceless stops not essentially different from the pronunciation of P. t. ~. c. and k. but the following vowel is murmured. The prenasalized phonemes mb. nd, nj. and ~g. however. phonetically resemble Sundanese or Malay mb. nd, nj, and og respectively; in Madurese their reflexes are "vo iceless stops with indifferent tension followed by strong aspiration".(Stevens 1968:38) One might argue in a most tentative way that *b, *d. *~, *j, and *g were the voiced aspirated counterparts to *p, *t, *~, *c, and *k. Possibly *B and *0 were voiced unaspirated sounds, since their reflexes in initial and postconsonantal position in Sundanese and Malay, in postconsonantal position in Javanese, and in initial, intervocalic and postconsonantal position in Madurese are voiced lenis stops.

We intend to continue the investigation of the historical relationships among the Malayo-Javanic languages. We also intend to extend our attention both to the ot her languages of the 'Javo-Sumatra Hesion' and to the languages of other West Indonesian subgroups. By doing so, we might be able to resolve some of tbe problems mentioned above.

FOOTNOTES

1 A work on Lampung has just appeared: Walker, Dale F. 1973. A sketch of the Lampung language: the Pesisir dialect of Way lima. Unpublished Cornel1 University dissertation. This book became avai1- able to me on1y af ter the completion of my research for this work.

2 These figures were obtained from Mr. I. Dyen in a persona1 communication. They are based on on1y one 200-word list per language and therefore differ fr om the figures which appear in Dyen 1965b, in which Dyen sometimes used more than one list per 1anguage in the ca1cu1ation of the percentages of re1ationship.

3 The 100-word list contains 92 items of Swadesh's 215-word list and 7 additiona1 items; see Bergsland and Vogt.

4 The 200-word 1ists for Sundanese, Javanese, Ma1ay, and Madurese can be found in the appendix.

5 I investigated the Sundanese 1anguage borders whi1e doing field work in Indonesia during the academic year 1971-72.

6 For a detai1ed account of Sundanese phono10gy, see Van Syoc.

7 The phonemes c, j, and nare fropta1 stops. They are pronounced with the b1ade of the tongue touching the a1veo1ar ridge. The phoneme q is the glotta1 stop.

8 For a detai1ed account of Sundanese morphophonemics, see Van Syoc and Coo1sma 1904. For an ana1ysis of verbal forms, see Robins.

9 This change has been made, since my own material shows that in this position wand y do not appear in the speech of my informants.

10 This is the number given in Pigeaud 1967.

11 For a detai1ed ana1ysis, see Horne.

12 The phonemes ~, 9, and n9 are apica1 stops. They are pronounced with the tip of the tongue touching the a1veo1ar ridge. The phonemes mb, nd, n9, nj, and ~g are 1ike the consonants b, d, 9, j, and 9 with nasa1 onsets.

13 Sumukti, p. 5.

14 For a detai1ed ana1ysis, see Horne.

15 This data is taken from Zoetmulder. For a more detai1ed ana1ysis of 01d Javanese morphophonemics, see the same work.

16 This information is taken from Dyen (1945:17).

17 For a more detailed analysis, see Dyen 1967.

18 The phonemes bh, dh, 9h, and jh are voiceless stops with indifferent tension followed by strong aspiration.

19 For a detailed analysis of Madurese morphophonemics, see Stevens 1968.

19a Pigeaud's entries which are followed by the abbreviation gr (=grof) are considered to fall in this category.

20 Kiliaan 1904 does not distinguish such a level.

21 See the list in the appendix.

22 With an inexp1icab1e -ghinstead of anticipated -k-.

23 See Ras (1970:429).

24 Snd. wa1iraU 'su1phur' probably is a 10an from Javanese. PMJ initia1 B regu1ar1y became Snd. b.

25 Snd. waniq (L) 'brave' is treated as a loan fr om Javanese.

26 Mad. oji 'to test si1ver or gold' probab1y is a borrowing from Javanese or Ma1ay. The regu1ar reflex of *-jis Mad. jh (see section 2.9.2.).

27 Mal. balira, balera 'weaver's sword' must be considered a loan, since the regu1ar reflex of *-1- is Ma1ay d (see section 2.8.6.).

28 This word, if it is a 10an from Javanes~ presumab1y acquired this figurative meaning af ter it had been borrowed.

29 For more examp1es, see the section on PMJ hiatus.

30 See section 1.4.2.

31 Taga10g exhibits a instead of an anticipated i.

32 Dempwo1ff (1.86) wrong1y assigned NJv. woq to his *buuk. The occurrence of b is treated by Dempwolff as an "Unexp1ained exception".

33 Tag. pu16q 'island' is considered not to be a reflex of uempwo1ff's *pu1aw.

34 With inexplicab1e i for anticipated u [Dempwo1ff (2.155)].

35 Ma1ayo-Javanic evidence points to areconstruction with *-ay rather than with *-ay as suggested by Dyen.

36 This co11ection contains notes on Javanese dia1ects and can be found at the Universitas Indonesia, Jakarta. I was ab1e to brief1y page through these notes.

37 The fact that NgD. exhibits u perhaps suggests that in the last proto-language which Ngaju-Dayak and PMJ continue the vowe1 was *u.

38 See fn. 37.

39 Mad. kOSOD 'empty, vacant' is marked as a borrowing from Ma1ay in Kiliaan 1904.

40 Mad. rombaq, rumbaq 'to tear down' are treated as 10ans. Mad. rombaq exhibits a instead of an anticipated a, and rumbaq exhibits u instead of an anticipated 0 and a instead of an anticipated ( (see the vowel alternation rule, section 1.8.3.1.). There is a1so a form ràmbtq 'to tear down' which cou1d be interpreted as a borrowing under the hypothesis that the vowe1s 0 and á were substituted for origina11y different vowe1s.

41 For more examp1es, see the section on PMJ hiatus.

42 OJv., NJv. tanah 'earth' are treated as 10ans from Ma1ay. See fn. 48.

43 The origin of the distinction between short and long pepet in 01d Javanese orthography is perhaps best exp1ained as having been formed on the ana10gy of the orthographic distinction between 01d Javanese short and long i, u, and a. It is interesting to note that the traditional Sundanese writing system does not distinguish the two Sundanese phonemes a and 4.

44 For NJv. -was reflex of *-h-, see the section on *h.

45 Mad. pbppd 'to beat' is treated as a 10an from Javanese.

46 Dyen (1953a:12) gives NJv. pèh as reflex. The word pèh, however, is not the continuation of OJv. p-ayöh, p-öyah, or p-uyuh, but of OJv. pyah 'groins, side'. NJv. pèh has the meaning 'water pipe through an opening on the side of a dam, the sides of the upper be11y'.

47 Snd. tawak 'to stab' is treated as a 10an from Javanese.

48 The reconstruction with fina1 *-ah instead of Dempwolff's *-ah is based on evidence from Sundanese and Javanese and makes a reconsideration of the correspondences which Dempwo1ff assigned to his *-ah and *-ah necessary. Dempwo1ff assigned the fo110wing ref1exes to his *-ah: Tag. -iq, TBt. -0, NgD. -ah, Mer. -i, Mal. -ah. Dempwolff was unab1e to suggest its Javanese, Fiji, and Samoan reflexes, since in the on1y reconstruction exhibiting fina1 *-ah for which there is a cognate in these 1anguages, the two vowe1s *a and *acontracted in all three languages af ter the 10ss of *-R-.

The fo110wing ref1exes were assigned to his *-ah: Tag. -aq, TBt. -a, NgD. -ah, Mer. -a, Mal., NJv. -ah, Fi., Sa., Fu., Sm. -a. Dempwo1ff reconstructed five etyma with *-ah:

(1) *ia[h], Tag. qi':hiq, TBt. eo 'urine'. , (2) *baRaq, Tag. bá:gaq 'u1cer', TBt. baro, NJv. woh, Mal. barah, NgD. baha, Mer. vai = bai, Fi. mbo, Sm. mata-fä 'u1cer, abscess'.

(3) *[t]aba[h], Tag. tabáq, TBt. tabo, Mer. tavi 'fat, we11- tasting'. /

(4) *[t]u(~)ka[h], Tag. tunk1q 'limb, joint', TBt. tuho 'landmark', tukko 'stump of a tree', Mal. toukah 'plank used as a support on mud flats', NgD. tunkih 'chopped-off piece', Mer. tuhi 'union, combination'.

(5) *tanah, TBt. tano, NgD. tanah, Mer. tani, Sa. ano 'earth, land' •

Besides *tanah Dempwo1ff reconstructs a by-form with fina1 *-ah: (6) *tanah, NJv., Mal. tanah 'earth, land', NgD. tana 'field'. Other examp1es of reconstructions with fina1 *-ah are the foll owing:

(7) *lamah, Mal. 1amah 'weak', NgD. l~-lamah 'soft', Mer. 1emi 1eminä 'weakness', Sm. 10ma 'to decrease (of wind)'.

(8) *basah, Tag. basáq 'moistened', TBt. baso 'watery', NJv. basah 'disso1ve, decay', Mal. basah, NgD. bisa, Fi. sava 'wet'.

Dempwo1ff exp1ains the occurrence of Tag. aq as reflex of his *-ah in (2) and (3) as distant assimi1ations. He considers Mer. -i as reflex of his *-ah in (7) an unexp1ained exception. The same exp1anation is resorted to in the case of the appearance of TBt. -0 as reflex of his *-ah in (8). Thus, Dempwo1ff has to refer to assimi1ations [in (2) and (3)] and unexp1ained exceptions [in (7) and (8)], and has to reconstruct a doublet [(5) and (6)] in order to account for the ref1exes of his *-ah and *-ah. No reference to unexp1ained exceptions and no reconstruction of a doublet has to made, if we assign the fo110wing ref1exes to *-ah: Tag. -aq, TBt. -0, NgD. -e, Snd. -4h, OJv. -ah, -öh, -uh, NJv. -oh, 11a1. -ah, Mad. -a, Mer. -i, Fi., Sa. -0, Sm. -a (1). Tag. -iq in (1) can be exp1ained as assimi1ation of the u1timate to the penu1timate vowe1. NgD. -a in (2) is exp1ained as assimi1ation of the u1timate to the penu1timate vowe1. Reconstruction (4) can be ignored, since its ref1exes are not or on1y margina11y re1ated in respect to their meanings. The two meanings that seem c10sest are the Taga10g and Merina ones. The Toba-Batak, Ma1ay, and Ngaju-Dayak meanings are neither re1ated among themse1ves nor to the Taga10g and Merina meanings. If we argue that the Taga10g and Merina meanings are in fact close enough to be considered cognate, then a reconstruction *[t]u(U)kiq can be made. In respect to (5) and (6) we on1y reconstruct *tanah. Under this hypothesis New Javanese and Ngaju-Dayak tanah are treated as 10ans, probab1y from Ma1ay. NgD. tana is treated as the inherited word, exhibiting assimi1ation of the u1timate to the penu1timate vowe1. Reconstructions (7) and (8) are now reconstructed with *-ah. All ref1exes are regu1ar. NJv. basah 'disso1ve, decay' is treated as a 10an. It is not c1ear whether it is from Ma1ay, since modern ~la1ay basah does not have the meaning of the Javanese form.

If we now consider our reconstruction *paR2R2dh which can be associated with Oempwolff's *paRah, we see that TBt. -0 in poro 'squeezed out' and NJv. 6 in pÓh (0) 'squeezed out' are regular reflexes under the new hypothesis. Oempwolff had to refer to unexplained exceptions in both instances, since *-ah regularly became TBt. -a and *a ... a regularly contracted to NJv. 0 af ter the loss of *R (cf. *taR2R2ab, p. 77 ).

Also in the case of a reconstruction made by Oyen (1953a:12) with *-aq areconstruction with *-ah will .resolve a hitherto unexplained exception. Oyen reconstructed *luhaq, Tag. lJ:haq, Bisayan as recorded in Encarnacion loha, Hiligaynon Bisayan luhaq, NJv. 16h 'tear'. In a footnote, Oyen observes that the Javanese cognate has Ó inexplicably for anticipated o. However, if we reconstruct PHN luhaq the Javanese form is regular, since the sequence *uha contracts to NJv. ó (the PMJ reconstruction is *luah, p. 184).

49 With an analogical d-.

50 Map 3 in the appendix shows the dialectal distribution of the Sundanese forms for 'fat'.

51 For an explanation of the citation of two forms, see the section on the 'Infixed nasal'. Mad. dhuk, addhuk 'black sugar-palm fiber' is treated as a loan from Javanese, since *z regularly became Mad. jh.

52 With ~ for an anticipated Ó.

53 Snd. kaboq 'buffalo' is treated as a loan, probably from Javanese. The standard Sundanese word for this meaning is mundiO.

54 Professor R. Hendon in a personal communication has suggested a different line of reconstruction. He posits the reconstruction *tamuay on the basis of the comparison of TBt. tamue and NgO. tamuei with the Old Javanese and Madurese forms, since *-ay regularly became OJv. i and Mad. i or ~ (distributed according to the vowel alternation rule) [see section 2.7.2.4.2.]. OJv. tamuy regularly developed into NJv. tami. The origin of NJv. tamu is explained in the following way. The derivative *katamuyan perhaps became katamuwan by assimilation of the y to the u. By back-formation a new stem tamu could have been formed on the analogy of forms such as kalakuwan or kasuciyan (whatever their origin). According to Hendon the Malay form tatamu 'guest' is perhaps a loan from Javanese and the inclusion of Mal. tamu 'to regale, feed up' is perhaps open to question.

55 Snd. Bad. wayuq 'fermented drink' is treated as a loan, probably from Javanese.

56 Map 4 in the appendix shows the dialectal distribution of the Sundanese forms for 'wood'.

57 Snd. samuq (L) 'appearance, exterior' is treated as a loan, probably from Javanese.

58 Snd. bankeq 'corpse, carcass' is treated as a loan, probably from Javanese.

59 Snd. buleq in mundin buleq 'white buffalo' is treated as a loan, probably from Javanese.

60 Snd. gadeq and Mad. ghá~hi(h) 'pawn, pledge' are treated as loans, probably from Javanese.

61 Snd. gaweq (L) 'work' is treated as a loan, probably from Javanese.

62 Snd. S.-B. laweq 'to yarn katuns' is .treated as a loan, probably from Javanese.

63 Snd. pakeq (L) 'to put on clothes' and Mad. pakè(h) are treated as loans, probably from Javanese.

64 Snd. ranteq 'chain' is treated as a loan, probably from Javanese.

65 Snd. saleq 'dried fruit, esp. dried banana' and Mad. salè 'dried banana' are treated as borrowings from Javanese.

66 Mad. gárami 'kind of fish' is treated as a loan, probably from Javanese. The regular reflex of *gis Mad. gh (see the section on PMJ g).

67 Mad. pa~i(h) 'rice' is treated as a loan, probably from Malay. The regu 1 ar refl ex of *-]- is Mad. 1 (see the sect i on on P~IJ 1).

68 Snd. kandiq (L) 'water-goglet' is treated as a loan from Javanese or Malay.

69 Snd. qapiq in taliq qapiq 'fuse, slow-match' probably is a borrowing from Javanese or Malay.

70 Mad. api in báto api 'flint' is treated as a borrowing from Malay.

71 Mad. báli(h) 'to return' is treated as a loan, probably from Javanese.

72 Snd. suriq 'weaving-comb' is treated as a loan, probably fr om Javanese or Malay.

73 Mad. B. tOlé(h) and Mad. B. P. S. dhuli (L) 'quickly, immediate; are treated as loans from Javanese.

74 Snd. turiq 'kind of tree' is treated as a loan from Javanese or Malay.

75 There is also a Sundanese form qamis 'sweet'. Possibly, this is a metathesized form with analogical loss of n-.

76 For the final n as reflex of *R, see section 2.14.4.2.

77 Mal. initial ~ in nilu 'nerve pain' is treated as a dialectal reflex.

78 J. Dyen in a personal communication has pointed out that Peninsula Malay has ani-ani 'harvesting knife'. Presumably, this word is from the dialect in which *n before *i became n.

79 Presumably, the ultimate vowel assimilated to the penultimate vowel before *nn before *i became Snd. n.

80 This form is considered to contain a prefix which changed the initial b to the corresponding nasal.

81 With analogical loss of the initial ~.

82 The Madurese evidence points to areconstruction with final *1 rat her than to areconstruction with final *d. OJv. h-in-affut 'passive of 'had let float", its New Javanese continuation anót 'to float' (this form only appears in Jansz), and NgD. hanut 'to float' under this hypothesis are treated as borrowings, probably from Malay.

83 Of Dempwolff's languages only Toba Batak, Javanese, and Ngaju-Dayak have different reflexes for PAN -d (our *d) and PAN -j (our *1). In this instance a Toba Batak and Javanese cognate are lacking. Ngaju-Dayak shows a form with final t which is the regular reflex of *d. Javanese evidence, however, points to areconstruction with PAN -j (= PMJ 1), since the latter regularly became Javanese r, while PAN -d (= PMJ·d) regularly became Javanese d (see the section on PMJ dl. Final PAN j, however, regularly became NgD. r. NgD. bukit under this hypothesis must be suspected of being a borrowing, probably from Halay (for a discussion on the inherited reflexes of PAN phonemes in Ngaju-Dayak, see Dyen 1956).

84 Mad. pasar in p~sar bhum~ 'center of earth' is treated as a loan from Javanese.

85 The Madurese form points to areconstruction with final *1, since final *d became Madurese t, and not q. Snd. laut 'sea, oceàn' is treated as a loan from Malay.

86 Malayo-Javanic evidence suggests areconstruction with final *d rather than with final *t.

87 Snd. dalua~ and Mad. dhälub~n 'native paper' are treated as loans from Javanese.

88 NJv. jlujór 'taek, bast' is treated as a loan, probably from Malay.

89 Mad. édhap (H) 'eye-lash' is treated as a loan from Javanese.

90 Snd. kidan and Mad. kèddhá9 'barking deer' are treated as loans from Javanese.

91 Snd. kiduD 'kind of prayer to keep evil spirits away' and Mal. kidoU 'chanting' perhaps are borrowings from Javanese.

92 Snd. qondan (L) 'to invite' is treated as a loan, probably from Javanese.

93 Snd. katag 'heart-beat' is taken to be a loan from Javanese.

94 Snd. galut 'to fight' is treated as a loan from Javanese or Malay.

95 Snd. pantil 'knob, fruit-bud' is treated as a loan, probably from Javanese.

96 Snd. waliwis 'teal' is treated as a borrowing, probably from Javanese.

97 Snd. warun 'small shop' is treated as a loan, probably from Javanese.

98 Snd. watas 'border' is treated as a loan, probably from Javanese.

99 Snd. warah 'to announce' is treated as a loan from Javanese.

100 Snd. walas (H) 'pity, mercy' is treated as a loan from Javanese.

101 Snd. balut 'eel' is treated as a loan, probably fr om Malay.

102 Snd. wanaU 'what can or may be, but does not have to be; can, able, to have the power and right to' is treated as a borrowing from Javanese.

103 Mad. bharrás 'rice' is treated as a loan, probably from Malay.

104 For a discussion of these forms, see p. 180f.

105 Snd. wilah 'bamboo-lath' is treated as a loan, probably from Javanese.

106 With an inexplicable -ho

107 Mad. bu~Ó(h) 'kind of tree' is treated as a loan from Javanese.

108 Mad. bakul 'basket' which exhibits irregular vowel reflexes is treated as a loan from Javanese or Malay.

109 Snd. walan 'grasshopper' is treated as a borrowing from Javanese.

110 Fi. mbali-mbali 'be crazy' is treated by Dempwolff as a reflex of his *balik. However, it is probably better treated as a reflex of *baliw, since the Tagalog reflex of the latter (b!:liw) has the same meaning as the Fiji form.

111 Snd. balas, walas 'requital' are treated as loans from Javanese.

112 Mad. bhántén 'to dash against, throw against' is treated as a loan, probably from Javanese or Malay.

113 Snd. wayan 'shadow' is a loan from Javanese.

114 Snd. banis and Mad. bhaUUès 'heartless' are treated as borrowings, probably from Javanese or Malay.

115 Snd. wisik 'to whisper' is a borrowing, probably fr om Javanese.

116 Dempwolff explains Mer. -tsas a 'reduced nasal cluster' (see the section on the 'Infixed nasal').

117 Snd. bibit 'seed, seedling', Mad. B. P. bibit 'to begin', and bhibhit 'seedling, to originate from' are treated as borrowings from Javanese.

118 Snd. wa-waUun-an 'building' is treated as a loan from Javanese.

119 Snd. balaU' walaU 'spotted' are treated as loans, probably from Javanese.

120 Mad. bháiak 'to pull (out)' is treated as a loan from Javanese.

121 OJv. wrat and its New Javanese continuation wrat (B) 'heavy' probably are borrowed from a dialect or language in which *B- became wand *R2 became r.

122 With penultimate a instead of anticipated i.

123 Snd. ribut 'storm' probably is a borrowing from Malay.

124 Snd. qobah 'to move' is treated as a loan, probably from Javanese.

125 Mad. obhá 'to change, alter' is treated as a loan from Javanese or Malay.

126 Mad. tabbhu(h) 'sugarcane' is treated as a borrowing from Javanese or Malay.

127 Mad. tobbh{(h) 'fish-poison' is treated as a loan from Javanese or Malay.

128 Mad. rabbhuQ 'bamboo-shoot' is treated as a loan from Malay.

129 With a weakening of the penu1timate vowe1.

130 Snd. kambun 'puffed up, swo11en' probably is a loan from Javanese or Malay.

131 Snd. bandaD 'kind of fish' probab1y is a borrowing from Javanese.

132 Snd. bantar 'quick' is treated as a loan from Javanese.

133 This PHN reconstruction does not appear in Dempwolff.

134 Snd. baka1 (L) 'provisions' is treated as a loan from Javanese.

135 With assimi1ation of the ultimate vowel to the penultimate vowel.

136 Snd. bandun-an 'dam, dyke' is treated as a borrowing from Javanese.

137 Evidence from Sundanese and Madurese supports a reconstruction with final *-ad rather than with final *-at. The Ngaju-Dayak form bulat 'round' probably is a borrowing from Malay, since PANa became NgD. e (see Dyen 1956).

138 Snd. kabas 'to flick, jerk away' probably is a loan from Javanese or Malay.

139 With an analogical gh-.

140 Mad. S. labur 'lime mixed with water' is treated as a loan from Javanese or Malay.

141 Snd. rabut 'to fight' is treated as a loan from Javanese.

142 Mad. obin, ubin 'flagstone' are treated as loans from Javanese or Malay.

143 Snd. rabah 'to fall' is treated as a loan from Javanese or Malay.

144 Snd. kambar 'twins' is treated as a borrowing from Javanese.

145 Snd. lambut 'soft, small' is treated as a borrowing from Javanese or Malay.

146 Mad. bhábá 'territory under one's authority' is treated as a loan from Javanese.

147 Snd. banan 'thread' is treated as a loan from Javanese or Malay.

148 Snd. sambar 'to pounce to seize and carry off' is treated as a borrowing from Javanese.

149 Coolsma (1913:61).

150 Coolsma (1913:54).

151 Snd. raiq, ra-ray (L), and ra-rahi 'face' and Mad. raè(h), ra-raè(h) 'face' are treated as loans from Javanese.

151a NJv. 9arah (0) 'blood' is treated as a loan from Malay.

152 With dissimilation of the initial consonant.

153 OJv. a-darat 'on foot' and NJv. darat-an 'land' are treated as loans, probabiy from Malay. Mad. dhár~t, dhártt 'to walk' are probably loans fr om Javanese or Malay:

154 Snd. rataq, Mal. rata, and Mad. rata(h) 'flat' are treated as loans from Javanese.

155 Snd. ratuq, Mal. ratu, and Mad. rato(h) 'prince, ruler, queen' are treated as borrowings from Javanese.

156 With an analogical -q.

157 Mad. ronrOn 'foliage, leaves' is treated as a borrowing from Javanese.

158 Snd. raDuq 'disturbed' is treated as a borrowing from Javanese.

159 Snd. paroq (L) and Mad. parO(h) 'one half' are considered borrowings from NJv. pa-ró (L) 'one half'. The Old Javanese form is sa-pa-rwa 'the one half'.

160 See fn. 152.

161 OJv. 9uri 'thorn' is treated as a borrowing from Malay.

162 With assimilation of the ultimate to the penultimate vowel.

163 Snd. (h)adap (M) 'to be in front of' is treated as a loan, probably from Javanese.

164 Mad. aghap 'facing towards' is treated as a loan, probably from Javanese or Malay.

165 Snd. badak 'rhinoceros' is treated as a loan from Malay or from Javanese. Mad. bárák 'rhinoceros' is treated as a loan from Javanese.

166 Snd. wuriq-wuriq 'late1y' is treated as a 10an from Javanese.

167 Mal. murah 'cheap' is treated as a 10an, probab1y from Javanese. For a simi1ar treatment, see Oyen (1953a:16, fn. 73).

168 For the fina1 n as reflex of *R, see section 2.14.4.2.

169 Snd. kandor 'slack, slow' is treated as a 10an from Malay.

170 This form cou1d have acquired its final ó as the resu1t of an 'ana10gica1 wrong division'. NJv. kan~ó under this hypothesis rep1aced an older form kan~u (0) because the latter is in the same relationship to a suffixed form such as NJv. kan~àni 'to be loosened' as NJv. tinjÓ (0) (0) 'to go visit' is to tinjoni (0) (0) 'to be visited' .

171 The preconsonantal nasal probably was *n and not *0 as Oyen reconstructs. Preconsonantal *0 appears as 0 in Sundanese, see section 2.8.4.3.

172 Snd. pandam 'to bury' is treated as a loan from Javanese.

173 With an analogical bh-.

174 NJv. sa-sindir-an (0) 'erotic song with allusions' is treated as a loan, probably from Malay.

175 Snd. tindas 'to crack a flea or louse on the finger-nail' is treated as a loan from Javanese.

176 Snd. dampul 'a composition used for caulking boats' is treated as a loan from Javanese or Malay.

177 . , Mad. B. P. guga~ and Mad. S. P. guga~ 'warehouse' are treated as borrowings from Javanese or nalay.

178 Mad. koguD 'to wear over one's head' is treated as a loan from Javanese or Malay.

179 Snd. pada~ 'sword' is treated as a loan from Javanese or Malay.

180 Snd. sa dan 'intermediate' is treated as a loan from Javanese.

181 Snd. sadap 'tasty' is treated as a loan fr om Javanese.

182 Snd. Bog. and Cian. sadih 'sad' are treated as loans from Javanese or Malay.

183 With an analogical gh-.

184 Mad. Qáqá(h) 'breast, chest' is treated as a loan from Javanese or Malay.

185 Mad. quyuu 'sea-cow' is treated as a loan from Javanese or Malay.

186 Mad. B. gón~ól, gh&n~Ól, S. P. ghunqhul 'bare, bald' are treated as borrowings from Javanese.

187 "/' / / Mad. B. gonqel, ghonqel, P. ghun~il, S. ghun~hil 'without or with only few feathers, bald' are treated as loans from Javanese.

188 With assimilation of the *1 to the final *Rl'

189 NJv. barah 'ulcery skin-disease' is treated as a loan from Malay.

190 With an analogical -q.

191 With a doubling of the initial semivowel. The second member of the Old Javanese double consonant contracts with the following vowel to New Javanese à.

192 Snd. qotot 'marrow' must be considered a borrowing, perhaps from Javanese.

193 This form probably is a Pampangan loan.

194 See fn. 193.

195 The level of reconstruction of this form is dubious, since the meaning of the Fiji form is quite different fr om that of the Western forms.

196 The spelling with a must be considered a scribal corruption for an etymologically correct u.

197 Mad. os~k 'rafter' is treated as a loan from Javanese.

198 Mad. adts 'to take a bath af ter intercourse' is treated as a loan from Javanese.

199 Snd. talur (VL) 'scrotum' possibly is a loan from Malay.

199a Mad. B. oma in saoma 'household' is treated as a loan from Javanese.

200 For a review of how Dempwolff's reconstructions compare to Dyen's, see Dyen 1953a and Dyen 1971.

201 5nd. qaes 'embellishment' is treated as a loan from Javanese. 50 is Snd. paesan 'grave-stone'.

202 Mad. 5. P. paèsan and Mad. B. maèsan 'grave-stone' are treated as loans from Javanese.

203 With an inexplicable n.

204 Mad. yau in ka-yau-an 'home of the gods' is treated as a loan from Malay ka-yan-an 'home of the gods'.

205 Snd. yan 'god' is treated as a loan from Malay.

206 NJv. yan (B) is treated as a loan from Malay.

207 For a dialect map of the Sundanese words for 'sour', see map 5 in the appendix.

208 For a dialect map of the 5undanese words for 'smoke', see map 6 in the appendix.

209 Oyen 1953a has *qayun.

210 For a dialect map of the Sundanese words for 'year', see map 7 in the appendix.

211 Mad. bájá(h) 'crocodile' is treated as a loan from Javanese.

212 For a dialect-map, see map 8 in the appendix.

213 For a hypothesis in respect to this Javanese form, see Oyen (1953a:ll, fn. 27).

214 PMJ exhibits metathesis of the first two segments.

215 Evidence from Malayo-Javanic languages shows that in this instance Oempwolff's criterion languages were not sufficient to distinguish between a simple vowel and a sequence of two vowels.

216 Snd. qarum 'fragrant' is treated as a loan from Javanese.

217 OJv. kahin and its New Javanese continuation kaén (0) 'cloth' are treated as loans, probably from Malay.

218 Snd. qalon 'slow' is treated as a loan from Javanese.

219 With an analogical n-.

220 See fn. 219.

221 NJv. laén (0) 'different' is treated as a loan, probably from Malay.

222 NJv. tuwaq 'palm-wine' is treated as a loan, probably from Malay.

223 See fn. 48.

224 The deviant meaning of this form may be due to asemantic merger with OJv. wasuh 'to wash' from *Bassuh (see the next entry). The merger may have been caused by the phonetic (wasah versus wasuh) and semantic ('wet' versus 'to wash') similarity of the two items. For NJv. basah, see fn. 48.

225 Snd. wasuh (H) 'to wash' is treated as a loan, probably from Javanese.

226 Snd. qasam 'tamarind' is treated as a loan from Javanese.

227 For the two Old Javanese spellings, see section 2.6.4.3.

228 It seems likely that the final q can be explained to be of secondary origin.

229 With an inexplicable o.

230 Snd. sisik 'scale of a fish' is treated as a loan from Javanese or Malay.

231 NJv. maman (0) 'uncertain' must be regarded as a reduplicated monosyllable.

APPENDIX I

THE 200-WORD LISTS

English Sundanese Javanese Malay Madurese

all kabeh kabèh semua kabbhi

and j4n lan dan bán, kalabán

animal binata!) ké'wan binata!) hèwan, bhurun alas

ashes labuq awu abu l~lOD

at diq , no~ di è

back tukau, gagar balaka~ abáli tOlJgolJ

bad \ , gore!) eleq jahat, jhub(q buroq

bark kulit kaiq / kulet kulet kolèq (na kaju)

because kusabab jalaran, sabap, sabá'b sabab karana

belly bH~~ wata~ parot tabuq

big gadeq gadé' basar . " raJa

bird manuk /. manoq buro!) manoq

bite Degel cOkOt giget ~èkk'èq

black hid~!) i ra!) hitam céll aD blow niup sab6l bartiop ~ , , narepo

blood gatih /. gateh darah dá'rá' •

bone tula!) /. baloj tula!) tol a!)

breathe !)anapas ambakan barnapas anabá'

burn m4Hm kObO~ manala, ~~lat, barrlala nomat

child budak anaq budaq anaq, naq-anaq

English Sundanese Javanese Malay Madurese

cloud megaq mego " '- awan mêgghá'

cold tiris, a9 am sajoq cèll ap tiqis

come 9ata~ data~ tak~ data~ /

count !)itu!) Ditó1J bila9 mèt~~

cut 04rH Di ris poto~ Oarraq, fièks~k

day b4ra!) awan sia9 ar~

die maot mati mati matè

dig naliq na9~q ma!)gali ~alè

dirty kotor ragad kotor k~t~r

dog qanji!J asu anje!J patèq

drink ninum " I !jombe minom ~ènÖm

dry gari!J garé!J karen karrè!j

dull mintul kattl tumpol tom pO 1

dust kakabul bal adóg aboq abu

ear c4liq kupé'!j tali!Ja kopèD

earth tan4h 1 amah tanah tana

eat dahar ma!)an ma kan !)akan

qandog , egg an~og talor tal1~r

eye mataq, mÖt?> ma ta mataq panon

fa 11 ragrag tibb jatoh ghagghar " /

far jauh adbh jaoh 'h/ J au

fat gajih ./h gaJe lamaq gh{jhi

father bapaq bapaq bapaq ammaq, appaq

fear sHn wadi takot tak6q

English Sundanese Javanese Malay Madurese

feather buluq wulu bulu bulu

few saHik si~éq, sadiket sak~nèq satit~q

fight galut / galot barkalahi, atokar bartumboq

fire s=in=iq gani api apoy

fish lauk iwaq ikan jhukoq

five limaq limÓ 1 ima l~maq

float lJambalJ kamamba~ barapo~ / namban

flow Da 1 ir mili ma!)aler aghili

flower kamba~ kamba~ bU!Ja kambhá'n

fly hibar, mabó'r tarba~ ~abbhar !Japu1J

fog kabut pacJó't kabot ambun

foot sukuq si ké'l kaki sOko

four qopat papat ampat ampaq

fruit buah woh, buah / buwa, / / wohwohan wáqbuwaqan

give mereq wei kasi abarriq

good qalus / apeq bagos, bá'ghus baeq

grass jukut sukat rumpot rabbhá'

hejoq .. / green lJO hijaw bhiru

guts jar6an parot panjalJ , parroq

hair buquk ramb6t rambot bbuq

hand l=i!)=in ta~an ta!Jan ta na!)

he manehna cJéwèqé' dia abá'qna

head huluq, sirah kapala cè~ak sirah

Eng1ish Sundanese Javanese Ma1ay Madurese

hear Uadegeq karunu daUar Dè~i~

heart hateq . / Janto!) janto~ até

heavy b-irat ab~t barat barrá'q

here didHq kené' sini è<}innaq

hit n-i-ag-i1 !Ö!Öq kana, napadhi puko1

ho1d naka1 gagam pagau nagghuq

how kumahaq kapariyé' bagaimana, ~áqr'rrma macam mana

hunt /. moroq mbage1 mamburu abhuru

husband sa1akiq b6jd 1aki 1aké

dewek aku aku ' \ seDkoq

if 1amun , ka1aw \ yen mon, ka1amOn

in diq , ~0!l da1am èdá'l am '/ Jaro

ki 11 maehan mat~ni bun oh matéqé

know nahoq -aarti tau tao

1ake situq , , ta10go taseq / ta1agha

1augh s-iriq ~guyu katawa agha11á'q

1eaf daun 9~9öD daon <}áun

1eft kencaq kiw~ kiri kacér

leg bitis sikél kaki sOk6

1ie na9gigir tur6n bare!)

1 ive hirup / urep hidop ocjiq

1iver bayah ati hati até

long panja9 dowd panja9 jháu

10use kutuq 1 i "Os?> kut u kóto

English Sundanese Javanese Malay Madurese

man lalakiq lana!) ora~ 1 akil aki Örè!) lakèq

many lobaq akêh banaq bánnaq

meat dagi!) / dagen dage!) dhághi~

mother qindu!) ibu maq ambuq, èbhu

mountain / gun u!) gunon guno!) ghun3n

mouth SU!)ut caVkam mulot cbloq

janaV '" name !)aran nama nama

narrow h4rH ciY6t sampet c~pèq

near d4kH ca~aq dakat sammaq

neck punduk gulu leher lèqèr

new qanar anar baru anar

night pHi!) balJi, malam malam wan i

qirun / nose iro~ hido!) èlo!)

not hanHq I ora tidaq anjá'q

old kolot tuwo tua tÖwa

one hij iq sij i satu sètto!)

other lain 1 iyané laen laèn

jalmaq , , person WO!) ora!J ore!)

play qulin d6lanan maen amaèn

pull narik taréq tareq narèq

push , suru!) suro!) tolaq n~tOk

rain hujan udan hujan ojh6n

red Mr4m aba~ merah m'èra

right ban ar banar batol bhan<jar (correct)

Eng1ish Sundanese Javanese Malay Madurese

right katuhu ta!)an kanan ka!)an (hand)

river wa1u!)an kali , su!)ay so!)ay

road jalan dalan jalan lorO!)

root qakar ~y~d akar ram~q

rope taliq tali tali tam~ar, tal

rotten bobrok bosoq lapoq, buccbq buroq

rub !)agosok !)gos~q gosoq k~sok

salt quyah uyah garam bujá

sand pasir wa~i paser ba<j~hi

naritaq könqö kata , say !)ocaq

scratch .- !)agaroq ngaroq garu, agháru garoqgaroq

sea 1aut sagoro laot tas~q

nenjoq , see ndaloq te!)oq, ~an~la nampaq

seed binih biné'h baneh bighi

sew 9aput nd~nd~m jaet ajháiq

sharp sikH 1an~ap tajam tajham

short pondok canqaq pendeq , / pen~aq

sing tamba!) namba!) manai'li !J~j hu!)

sit diuk 1ungóh dudoq tajuq

skin kulit kul e't kulet k~l~q

sky la!)it 1ané't la!)et la~3~q

sleep heqes, turu tidor tê<ju!J sareq

smal 1 Htik cilé'q kacel kènèq

English Sundanese Javanese Malay Madurese

smell s4~it mambu mamba u abá'u

smoke has4p kaból asap kokos

smooth 1 amas alós licen, alos halos

snake qoray ulo ular dlar

some qayaq qogeq salah sijiné'9 sadiket pan-bárámpan

spit niduh idu maludah , acopa

spl it ~b4lah bal ah bal ah N\b/ ne aq

squeeze marHt paras picet 9~péq

stab nojos jÓjdh tikam rlÓco

stand na~tu~ ~adag bardiri manjha!)

star benta~ linta!J binta!) binta!)

stick qiUk kayu kayu kaju

stone batuq watu batu báto

straight lampa~ jajag luros lorOs

suck 9anot sarot maDisap ~~, 9annot

I I matahari , sun panon poeq sra~e~e are, mata are

swell nambahan, mal ambó~ kambaD ' bárá '3alobaqan ba~kaq (of skin)

swim !)ojay ~ala9i barna!) ala~~oy

tail buntut buntót ekor buntoq

that qetaq, kaé itu arOwa, qituq kaqissa

there didituq iku, situ, èdissa k6nó, . sana konD

they maranehananaq gèwèqé kabéh dia, mareka

English Sundanese Javanese Malay Madurese

thick kandal kanda 1 tabal tabbal

thin qipis ti pe's nipes t~p~s

think mikir pik{r piker m~kk~r

this qHq i ki ini r~~a, areya

thou qanj4n kówe' a!Jkaw b{qna

three tiluq talu tiga talloq

throw ~alu.9 mbalan bua!J' !}6t'èppaghi ma 1 empa rkan

tie nalik4n ikat ikat nalèq~

tongue letah i lat lidah jhilá'

tooth huntuq untu gigi ghighi

tree ta!)kal uwe't pokoq kaju

turn robah mé~goq beloq mdtar

two duaq lÖrÖ dua / ~uwaq

vomit qutah mutah muntah !)Öta

walk jalan malaku barjalan ajhálán

warm panas panas panas panas

wash s4s4h !)umbah cuci abacco

water caiq banu aer a~~

we qura!) aku kab~h kita, sè!)kÖq kabbhi sarereaq kami

wet bas4h talas basah btcca

what? naon ~p~ apa apa

when? qirahaq kapan bila bilá

where? dimanaq andi mana ~c}imma

white bodas / puteh puteh potè

NOTE: Words for the meanings 'freeze', 'ice', and 'snow' were not elicited.

APPENDIX II

MAPS

Cl- .. \ ' .. ",,,. ~-f1 ~~ {) "'" r ' I~

-17 JAVA

1:8000 000 OST -JAVA 1:4000000 3 b 1 = Javanese 3 b 2 -- Sundanese 11 A 3 b 3 = r.ladurese 11 A _ Djakarta 11 A 3 a 5 b - Malay Rl)

(!) ~ @ @) ~ ~ 0 • j~ \- 0 ~ @ e e u:: (?) G) (!) @ €> e @ (!) @ G 0 0 ~ @ @ • e • • • • • e 0 • @ 0 0 00~(~)· ~"';o<)~.ri

UI El [3 I!J ;1 @ [iJ l!/ l!1 lil 1!1 ~ El ~ lil El @ @ @ @ (!j@ @ @~ @ 239

@ @ cel @ G e @ -~\ @ Gl e ~ e @ e @ (?) @ e e (!) • e e e @) 0 e (!) e (!) ® @ In (!) @ ~ @ @ (!) 240

ec. :(.

@ 0 (!) <è) 0 0 0 0 (!) -l\ (!) ~ G> 0 0 - 0 G> ~ 0 ij) • e @ e C!l e 0 e @ @ <!l G \9 (!) e 0 ® @ (!) @) ~~ @ iè@ @'! @ e ~ 241 e '\!) <!l el e @ <!J @ @ Gl ® @ <ê> ~ @) e 0 @) ® 0 @ (!) <ê> 0 (!) @ (!l e (!) (!) @ e <ê) @ ® (!) ~ @ (!) @) (!) 0 t'- @ Gl 0 (!) (!) 242

= 11 --.......

@ ~ ~ .. @ <i> <i> @ 0 @/0\ 11 ~ .-<!.l ~ ~ ~ ~ ~. ~ <é) ~ ~ ~ ~ ~ ~ ~ /@.~ ~ ~ . @ ~$I ~ ~ \~ c?) ~ @) ei ~ <!> @@ @ @<t @ @ 243

INDEX OF PROTO-~~LAYO-JAVANIC RECONSTRUCTIONS (Numbers following the reconstructions refer to pages in the text.)

haBi s. 132 hanzuaD' 111 haBuq. 132 qaiiam. 99 haBuk. 132 haniq. 99 qaddas. 186 qanul. 103 qadduq. 186 ha!)at. 170 haDap. 149 qa~gay. 90 qakuq, 178 hapaq. hampaq. 189 hadda~. 187 ha~in • 173 hala!). 171 qapaq, 51 qa11 i h, 185 hapak. 120 hal i!), 171 hapit, 118 halis, 170 qapuy, 94 haluq, 171 qassah. 186 qalur. 105 qassak, 186 halus. 172 hass3m, 186 ham[Bb]a[rR2], 166 hasap, 171 qamBat, 135 qassin, 185 hampas, 117 qattah, qantah, 193 qampah, 74 qattas, 187 qanak, 120 hatap, 73 qanay, 90 hatay, 93 qan[Dç]a~, 157 hatur, 105 hanay, 93 qawak, 82 qantiq, 118 hayam, 169 qanuq, 97 hayun, 172

cukur, 113 DagDag , 194 cu!)ur, 67 [D~]akkat, 157 cupcup, 198 DamDam, 197 [dD4]ahan, 173 DaO!laR1, 76 [dD~]agaw, ga[d~]aw, 88 Da~9uq, 148 damar, 107 Dappaq, 149 da~dan, 196 Diq, 148 dahaq, 74 [D~]i~[D~]i~, 200 diR2iq, 162 Duaq, 148 diRus, 165 [D~]ugal, 147 duqum, 107 Du1aD' 149 [D~]a~aq, 157 DuR1 iq, 148 Dahay, 147 [D~]uyuD' 157 Dalam, 149 ga~u~, 153 DaODaO' 200 gai~, 61 Da~DaR2' 100 gampar, 153 Dapur, 149 gapqap, 198 Daraq, 147 gasar, 153 DaRah, 147 gayu~, 153 DaR2at, 147 gampul, 153 DataR2, 148 gassak, 153 Data~, 149 gugaq, 153 Datuq, 148 gukuh, 153 Dahuq, 148 gukun, 153 Daun, 148 gusun, 153 Daya~, 149 haay, 73, 184 Dahak, 72 habbun, hambun, 193 ha11a~. 170 gapit. 53 ha11ay. 91 garaw. 88 hall at. 170 gargar. 198 qamBuk. 135 gattu!J. gantu!). 192 -an. 45 gaway. 91 qannam. 72 gaqam. 77 qappat. qampat. 193 ga11 ah. 75 haR2R2at. 162 ga11ut. 119 qaHik. qan~ik. 194 gi~a 1 • gin~al. 191 hazzan. 110 gilaq. 54 hazzuk. hanzuk. 80 gil i~. 102 gabus I. 138 gilir. 112 gabus 11. 139 gimbal. 140 ga[D~]ay. 91 giwa!l' 58 gaga~. 154 guBa 1. 132 gaggi!). 187 gUBi!J. gumBi!J. 190 gaguh. 154 gubal. gumbal. 191 gag~u~. 187 guga~. 154 galaw. 85 gul impa~. 71 ga 1 i h. 111 gul i!J. 53 galugaq. 52 gun [Dg] il • 158 gambar. 140 gun[D~]ul. 158 gambir. 140 guramay. 93 gandar. 107 gusuk. 70 ganDul. 151 hia!l' 170 ganj il • 109 hias. 169 gantiq. 53 qiBaR. 76. 132

laha!l' 173 1 abbat, 139 laqas, 179 1 ahan, 75 1 abbuh, 187 lamBah, 135 lahaq, 173 laBBih, 76 labur, 139 lambi~, 141 lakiq, 101 lanzaR laddi!l' 186 1 ambut, 141 lamBat, 135 lassah, 75 lambu!l' 51 1 assu!l' 48 l a1a1, 73, 103 1 a'J~an, 72 lamun, 97 lian, lain, 182 lanear, 113 1ian' 182 lanDak, 151 1 i[h]at, 184 lan~un' 155 1 i hat, 119 lantay, 91 lieik, 56 laU!lit, 185 li ei n, 113 1aOkah, 101 ligar, 57 1 a 'G!luy, 185 1 ikku!j' 1 iDku~, 192 1, 162 1 imaw, 55 laris, 101 1 imBay, 92 laruq, 101 lingih, 155 2iw, 87 lintah, 55 1atu'J' lantu!J' 189 li~sir, 58, 106 laPk, 121 liruq, 56 1 au] , 103 liwat, 57 laun, 182 luhar, 177 1 away , 91 luBa!), 133

2, 164

lapaR

laR

pahul, 175 pisaw, 85 paya!}, 90 pitik, 56 payu!l' 89 pizak, 111 paar, 72, 183 pucuk, 66 pa[DD, 11Jih, 159 pu[D!Ji!l' 159 paddaU' 154 pudak, 154 pa11uh, 63 pukaw, 64 panas, 102 puki q, 119 panDam, 152 pulaw, 65 pan!il , 122 pulay, 92 pan!un, 122 pusa! ' 103 pannuh, 79 pusar, 106 pappah, 75 putus, 118 pa[rr, R2R2]ay, 92 rahab, 173 pa[rr, R2R2]un, 166 raBun, 133 2R2ah, 77 raksuk, 119 paR1 Rl am, 81, 161 [rR pattay, 93 [rR 2]amas, 166 pihak, 176 rampas, 51 pi c ak, 58 rampuD' 104 pilih, 56 rantay, 92 pin[D~]ah, 158 ra!)kul, 104 pinda~, 156 rapih, 53 pintaq, 55 rapuh, 63 pipiq, 117 rahup, 175 pipis, 106 rabbah, 140 pisah, 57 rabbut, 139

paR

2]amBay, 92

[rR2]am[Bb]ay, 92, 166 salay, 92 ran [Dc}]ah, 158 salin, 97 riBut, 134 salir, sa!Jl ir, 188 rim[Bb]as, 57 sam[Bb]ar, 143 rimBit, 135 sampay, 92 ririh, 104 sanDa!}, 152 [rR2]uha~, 176 sanDar, 152 [rR2]uay, 65 sancJi!J, 156 rubbuh, 187 saUl iD' 102 rubun, 140 saput, 105 ruksak, 120 saR1a!J' 161 rumpu!J' 68 sau~, 182 rumrum, 197 sawah, 82 runtuh, 118 sahaR, 174 rurub, 105 sabbah, 139 ruqum, 179 sa[dd, c}c}]uq, 159 Rl aBBu!J' 135 s ac}c} aD' 154 R13imah, 166 sac}c}ap, 154 Rumah, 166 sac}c}ih, 154 R3usuk, 165 saggu!l' 79 sahaD' 173 sambur, 141 saqat, 179 samm[uq, aw], 89 saBuD' 133 sanDuk, 152 saBut, 133 sappah, 106 sabu!l' sambu~, 191 sapsap, 200 sa[DlJay, 157 sassah, 72 saguq, 112 siBur, 133 siksik, 198 sunsan' 106 siDaq, 150 SU!)guy, 94, siksik, 194 SU!)ut, 100 sidik, 107 su~gi!), 68 simpay, 92 suru!), 66 simpa!), 57 surambiq, 72 2, 152 suhun, 173 2, 164 suruy, 95 singah , 51 taqal, 179 si!jkab, 142 taBaR siram, 57 taBur, 133 sisih, 177 tahéln, 174 sitsit, 199 tagih, 112 sihu~, 176 tahiq, 175 suBél!), 133 tahil, 175 suDuq, 150 tajiq, 108 sugih, 67 tajin, 54 sulam, 96 takis, ta!Jkis, 190 suluh, 177 taman, 96 suligiq, 71 taliq, 180 s-ul-upsup, 198 tam[Bb]a~, 143 sumBiq, 135 tamBir, 135 sumba!.l' 141 tambak, 141 sumbuq, 141 tamba!), 141 sumbul, 141 tami a!:j , 52 sun[D<J]u~, 67 tamuq, 86 sun9a~, 156 tanDak, 152

sinaR

sinDiR

2, tamBaR 2,191 tanDuk. 152 taBBas, 77 tandaq, 156 taBBak, 77 tandi9' 156 taBBuq, 134 tanduq. 156 tabbah, 139 tanah, 73 tabbas. 140 taniq, 97 tabbuk, 140 tanju~. 109 talluq, 79 tanaq, 99 ta11uR. 165 ta!Jgap, 112 tanmuq, 79 ta!Jgu~, 100 tannun. 79 ta!Jhiq, 49 tappu!J' 79 2R2as, 164 [t~]a!Jkay, 92 taR ta!Jkub, 142 tastas, 49 ta!Jta~, 195 tiha!), 176 tapak, 118 tiDuR, 150 tapay, 93 tiqis, 179 tapiq, tampiq, 189 timah, 118 taptap, 49 timBaq, 136 tarimaq, 52 timba!). 141 2uh, 163 timun, 63 2um, 163 tinDas, 152 tasik, 120 tin[Dq]ih, 158 tatal, 102 tinjaw, 85 tahun, 176 ti~ga1u!J' 62 tawaR2 I, 84 ti!Jgil i!J' 62 ta!)i s, 100 taR taR taR tawaR2 rr, 84 ti!Jkah, 120

2R2ab, 77

[tt]i[rR2]ay, 55 tuR2ut, 163 titis, 57 tuhtuh, 199 tiwas, 58 [t~]uhur, 68 tuhaq, 176 tuqus, 179 tuak, 183 ~uk~uk, 66 tuhan, 177 ~ukul, 121 tuBaq, 134 huan, 183 tuBuh, tumBuh, 191 quBah, 134 tujj aw, 65 huBan, 134 tukaR2, 164 huBiq, humBiq, 191 tukup, tu!)kup, 190 qubin, 140 tultul, 195 qucap, 117 tul U!), 67 huDa~, 150 tuluy, 95 huDip, 151 tumaq, 96 huay, 65 tumBu!), 68 qujiq, 53 tumpa~, 64 huju!), 172 tumpuk, 68 qujur, qunjur, 189 tun[D~]aq, 158 qukkil, qU!Jkil, 194 tunDuh, 152 hukur, 172 tunDuk, 153 qu1 aR1 , 162 tunqu~, 156 hulal, 103 tu!}aw, 85 hu1uq, 170 tU9ga 1 , 102 hu1ur, 172 tu!)tun, 196 qumbal, 141 turuy, 95 qundur, 97 turus, 106 qun~aq, 156

2um, 110 qun~a'(), 157 zauh, 110 qun~uh, 157 z.a11as, 110 huntuq, 38 hunus, 63 za11i!), 110 zilat, 110 hunza!J' 111 qupah, 67 zual.ll0 zukut. 110 qurab, 124 3uh. 110 qurug, 11Z qurup, 117 2uq. 163 2at, 163 huR1 am, 161 190 qutah, quntah, huta!l' 170 qutak, lZ0 huzan, 111 wzaDaq, 84 w1aiR, 165 w2aluq, 84 w1aR3iq, 83 wzayaq, 84 w2uyah, 84 yuyuq, 89 zalan, 109 zaluha!), 109 zalujur, 109

zaR

zuR

qURzaD' 163

quR

zuR

YU!JYun, 89

WORD INDEX

(In this index h appears in its proper alphabetic position except when it occurs in parentheses in which case it is ignored. )

attas Mad.187

a-tua Fu.,Sm.177

qatur Snd.105

a-bas OJv.164

au To. ,Fu.l78

awak OJv.82

qawak Snd.82

awaq NJv.,Ma1.82

OJv.125

a -wa Qa k-wa ~a k

a -wa nis OJ v . 1 29

a-wil at OJv. 125

(h)awis OJv.132

awu NJv.132

ayam NJv.169 (h)ayon Mal.172 báccd Mad.185 Snd.221

ayam wana OJv.169

(h)ayam Mal.169

qá:yam Tag.169

qayaq Snd.84

ayun OJv.172

qayun Snd.172

a-yuyu OJv.89

bábá Mad. 142

ba-bakal-an Snd.136

ba-bantal Snd.136

babar,NJv.,Mal.200

babar layar Snd.200

ba-batan Snd.121 ba-ba~aD OJv.121 b~-buro!) NgD.69 bácca Mad. 185

badak TBt.150,

ba da q Ma 1. 150

báQá(h) Mad.84

baQaq NJv.150

bá:ga Tag.124

bagá!) Tag.132 bá:gaq Tag.137,214

baha NgD.214 bájá(h) Mad.224

bájár Mad.143

bakal Snd. ,0Jv.,

NJv. ,Mal •• Mal.Java,

Pal.136

bakal-an NJv.136

bakaw Mal.85

bakol Ma1.128

bakól NJv.128

bá:kol Tag.128 bako!) Mal. 136

bakón NJv.136

bá: kon Tag. 136

Mad.218

bakoq Snd.85 bakul Snd.128. baku!l Snd. ,0Jv. 136. Bug.208 bálá Mad.124 balanak Snd. ,0Jv. 132 bä:lanak NgD.132 bá1ánaq Mad.132 ba la!) OJv. 128 bá:la!) Tag.128 bá1áO Mad.128 bá:laq Tag.124 balasMal.128 balay Snd.44,90, OJv.,Mal.90 bale OJv. 90 baH NJv.44,90 baleh NgD.128 baleq Snd.44, Mal. 128 balêq NJv.128 balas Snd.219 bálas Mad.128 balas-an NJv.128 bali NJv.94 bálibis Mad.124 báli(h) Mad.216 balik Snd.128,144 Tag. ,OJv.128 balikat Snd.129 bal':kat Tag.129 bálikat Mad.130 balilá Tag.55 bálilá(h) Mad.55 báliq Mad.128 bálirá!) Mad.52 bál1u(h) Mad.49,84, 181 bál1uq Mad.49,181 balón NJv.136 bá:lo!}-bá:100 Tag. 124 balu!) Snd.,OJv.136 bamban NJv.195 bambán Tag.195

banál Tag.74

bandan Mal. 136

bandao 2.!!.

bande!) Mal.136

taliq banda!) Snd. 155 bandan OJv.,NJv.136, Snd.220 bandin Snd.136,144 ban~an-an OJv.,NJv. 155 banoq Snd.41,88 banta 1 Snd. ,0Jv. , NJv. ,Mal.136 bantar Mal.137 banten Snd.60, Mal. 60,128 bantél) NJv.128 ban~en OJv.60 ban~èl) NJv.60 bantar NJv.137, Snd.220 bantin NgD.60,Snd. 128,144 banual) Tag.62 banak OJv. 184 barlaq NJv.,Mal.184 báiiî'laq Mad. 184 bai'iu OJv. ,Mal.88, NJv.41,88 baPiuq Snd.41 banao NgD.85 ba!)aw Mal.85 ba!)ban Snd.195 bankay 2.!!. sawan baOkay Mal., Snd.91 bankáy Tag.91 baOké NJv.91 bankeq Snd.216 bano OJv.85 banó NJv.85 ba!)on Ma 1.130 banón NJv.130 bándn Mad.B.130 baOoq Snd.85 baijsal NJv. ,Mal. 137 ba!}un Snd.130,144 OJv.130 baok Snd.64 baoq Mal.64

11i Sa.180 è1o!) Mad.102 e1u Sa.163 ènom Mad.178 èn~è1 Mad. 122 èna(h) Mad.98 eo TBt.214 e-ohu Sa.134 ~pi Sa.118 qepok Snd.61 epoq Ma1.61 époq NJv.61 èpoq Mad.B.61 èrèd NJv.58 èrè(h) Mad.54 èrèn Mad.100 eret Ma1.58 èrèt Mad.58 èroq Mad.117 esu-esu Sa.172 èyas Mad.169

(a)bón NJv.193 (h)abun OJv.193 (ab)bhun Mad.193 (a)dan NJv.110 addhuk Mad.215 ajan Ma1.110 (aj)jhan Mad.110 «h)a)lan Ma1.170 (a1)laq Mad.170 «h)a)lat Mal. 170 (a)lat NJv.170 ambat Ma1.135 ambon Mal. 193 amboq Ma 1 .135 amboq NJv.135 ambuq Mad.135 ampaq Mad.193 ampat Ma1.193 -an OJv. ,NJv. ,Mad. 95 anam Ma1.72 qandog Snd.43 an40g NJv.43 (an)nam Mad.72 annam Mad.50 qangh Snd.81 «h)a)rat Ma1.162 ayah OJv.75 ayöh OJv.75

fa To. ,Fu. ,Sm.193 fa-fanu To.,Fu., Sm.130 fakaqa"!)i To.173 faka-1 ifu-1 ifu To.134 fantsunä Mer.113 fe-kite Fu.121 feri Mer.159 fifi Mer.117 fi ndra ~ler. 158 fine TO.131 fo-aqi Sm.126 fu1u To. ,Fu., Sm.129

gaoos Ma1.138 gabos in kayu giDos Ma1.138 Han gabos Ma1.139 gabós NJv.138,139 gabus Snd.139 gabus in kayuq gabus Snd.139 gada!) Snd.154 gaday Ma1.91 gade TBt.91 gade!) Ma1.187 gadeq Snd.216 gadgád Tag. 198 gadin Snd. 187 gado!) TBt. ,Ma1.187 gadoq Snd.88 gaduh Snd.154 gadun Snd.187 gaQan-an NJv.154 gac;je OJv.91 ga4e NJv.91 gac;jen NJv. 187 gaQi!) OJv.187 ga46h NJv.154 ga4ón NJv. 187 gastu"!) OJv. 187 gagar NJv.198 ga1eh MaLJava 112 ga1eh NJv.111 ga1ih Snd.,OJv.111 ga1ó NJv.85 ga10q Snd.85 ga1uga OJv.52 gambar NgD. ,Snd., NJv.,Ma1.140

gamber Mal. 140

gamber NJv.140 gambi1 Tag. 140 gambir Snd.140 gandar Snd.,NJv., Ma1.107/l08 gando!) Snd.69 gandu1 in gadan gandu1 Snd.151 ganQó1 NJv.151 gando1 NJv. 151 gan~àn NJv.69 ganje1 Ma1.109 ganjel NJv.109 ganjil Snd.109 ganti Mal.,NJv.53 gant' Tag. 53 gantiq Snd.53 ganton Mal. 192 gantó!) NJv.192 gantu!) Snd.192 gá:pas Tag.51 gapet NJv.53 gapèt NJv.53 gapit Snd.Bad.53 gappit TBt. 53 gari1 Tag.109 gá:ri!) Tag.187 garoq Snd.43,88 garu OJv. ,NJv., Ma1.88 ga tt U!) TBt. 192 gawai OJv.91 gaway Mal.91 gá:way Tag.91 ga we OJv.91 ga we NJv.91 gaweq Snd.216 gewa!) Ma1.58 gaste NJv.43 gadeq Snd.43 ga 1 empa!) Ma 1. 7l galah OJv.75 galimpa!) Ma1.7l g(a)limpa!) NJv.71 galot Ma1.119 galót NJv.43,119 galöh OJv.75 galuga Mal.52 g(a)lugo NJv.52 ga1ut Snd.43,218 ganto!) Ma1.69 gateh NJv.43

jamu NJv.180

jamu-jamu Ma1.180

janjan NJv.196

janji TBt.NJv.,

Ma 1 .100

jáo-báján-an Mad. junjon Ma1.200

janjan Snd.196 jaoji NJv.100 janjiq Snd.100 j aoh Ma 1 .110 jarah Snd.,Ma1.108 jarah-rayah NJv.108 jaró NJv.85 ja rom ~la 1 . 11 0 jaroq Snd.85 jarum Snd.110 jauh Snd.110 jejer Snd.,Ma1.61 jèjèr NJv.61 jenke1 Snd. ,Ma1.61 jè!Jkè1 NJv.61 je!)ko1 Snd. ,Ma1.61 jenkö1 NJv.61 j a las Ma 1. 110 j al en Mal. 11 0 ja1uao ~la1.109 ja1ujor Ma1.109 j i 1 a t Ma 1 . 11 0 j-in-ahit OJv.53 j1ujór NJv.217 joged Snd.61 jógèd NJv.61 joget Ha1.61 jojo1 Ma1.199 joneq Snd.93 jua1 TBt. ,Snd., Ma 1 • 11 0 jujo1 Ma1.199 jujó1 NJv.199 jujór NJv.64 juju NJv.63 jujuq Snd.63 jujur Snd.64 juko!) NgD. ,Ma1.190 jukón NJv.190 juku!) Snd.,OJv.190 jukut Snd.110 jun i Ma 1 .93

junjón NJv.200

jU!jjó!j NJv.200

jU!)jU!) Snd.200

ju!)ko!) Ma1.190

ju!) kón NJv. 190 ju ro h Ma 1 . 110 juru OJv. ,NJv., Ma1.108 juruq Snd.108,163 juruq tu1is Snd.108 juquh Snd.ll0 jháiq Mad.53 jhá1án ~lad.109 jhá1ujhur Mad.109 jhámö(h) Mad.180 jhánjhán Mad.196 jhánjhi Mad.100 jhá!)lè(h) Mad.188 jhárá Mad.108 jhárum Mad.110 jháruy Mad.85 jháu Mad.110 jha11i!) Mad.B.P.110 jhijhir Mad.B.61 jhilát ~'ad.110 jhujhu(h) Mad.63 jhujhur Mad.64 jhukön Mad.190 jhu1jhu1 Mad.199 jhu!)jhul) t~ad.200 jhuru Mad.B.110 jhuru(h) Mad.108 jhuru köncè Mad.108 jhuwá1 Mad.110

kabar NJv., Mal. M1c.139 kaca!) Snd. ,OJv. , NJv. ,Mal. ,Mad. kada1 Snd. ,Mal. kada!)-kadan Ma1.154 kadot in kaen käëfot Ma 1.186

kabo!) Ma1.133

kabhár Mad.139

ka-bhuru Mad.138

B.P.1l3

Java 187

kada!) Snd.107,154,

OJv.,NJv.107

kadöt NJv.186

ka-döha OJv.74

kadut Snd.,OJv.186

kadhábu!) Mad.154

kadhá!) Mad. 107

kadhá!j bárghá(h)

Mad.107

kaddhuq Mad.186

kada10Jv.,NJv.187

ka~a!) NJv. 154

ka-Qatw-an OJv.148

kaQawó!l NJv.154 kaQhá~ Mad.154

kaQQha 1 Mad. 187

kaen NJv.224

kaen Snd.,Ma1.182

kaèn Mad. 182

ka-göm NJv.77

kahin IJv.224

kaiq Snd.87

ka-göm-an OJv. 77 kaja!j Snd. ,OJv., NJv. ,Ma1.109 kajhá!) Mad.109 k-a1aq Mad.115 ka-1uwar-an OJv.177 kambá1 Tag.140 kamba!) Snd., NJv. , Ma1.143 kamba!)-kamba!) NJv. 143 kambá!) Tag. ,Mad.143 kamban-an NJv.143 kam-ba1as OJv.128 kambhá!) in bháláy kambhá!) Mad. 143 and in dhámar kaiiiD1ïá~ ~la d. 143 kamönè!) Mad.53 kampak Snd.189 kampaq NJv.189 kampel ~la1.117 kampel NJv. 117 kampèl Mad.117 kampil NgD. ,Snd., OJv.117

kaju(h) Mad.87

ka-junjun Mad.89

kampör Mad.189 kamuni !) Snd., OJv.53

-kan Mal.45,206

kance!) Mal.97

kance!) NJv.97

k-in-urn OJv.77 kion Ma1.56 ki:pil Tag.79 k i ra Ma 1 . 119 kiraq Snd.119 kirern Ma1.96 kirérn NJv.96 kirirn Snd.96 kirirn-an OJv.96 ki rà NJ v. 119 kiruh Snd.79 kita To. ,Fu. ,0Jv., Ma 1 .51 kitaq Snd.51 kità NJv.51 k+qirn Snd.77 kiq;!! Snd.73 -kin Snd.45,206 k;p;l Snd.79 kirH Snd.41,48,73 k-1ayu NJv.87 k-1èru NJv.56 k-1iru NJv.56 kàdu!) Mad.222 kàkà(h) Mad.119 kó1 NJv.179 kà1èq Mad.114,116 ko10n Snd., Ma1.70 kà1àn NJv.70 kórn NJv. 77 kàrnban NJv.135 kàrnbán Mad.135 kàrnpà1 Mad.96 kàncar Mad.B.P.165 konduh OJv.157 kàn~ur Mad.151 kànqhán Mad.155 konen Snd.54 kànè n Mad.54 kontol Snd.,Ma1.70 kóntà1 NJv. ,Mad. B.P:70 kó!! NJv.73 kóqo1 Mad. 179 koqo!) Mad.73 kopa Fu. 140 kopi - kopi Fu. 194 kora(h) Mad.161 kora!) Mad.105 korkor Mad.199 korup OJv.117 koso!) Snd.,Ma1.70

kósóO NJv.70,

Mad.213 kota~ Mad.38 kótkot Mad.199 kótó(h) Mad.118 krat OJv. 73 k-runu NJv.76 kuban Snd., NJv. , Ma1.139 kû:bo Tag.133 kubón Tag.135 kubu Ma1.133 kudu Mal. 154 kudun Snd.154 kuduq Snd.154 kUQón NJv.154 kUQu OJv.154 kuhó1 Tag.179 k-ujur Snd.189 kukód NJv.199 kukor Ma1.199 kukór NJv.199 kuku OJv.,NJv., Ma 1.119 kuku- Fi.119 kukuq Snd.119 kukur Snd.199 kukuva Fi. 104,1 Q9 kÜ10Jv.179 kü:1a!) Tag.105 ku1i Fi.114 ku1ót Tag.150 k-urn-arnba!) OJv.143 k-um-apak OJv.189 kurnbaan NJv.181 kurnbah NJv.181 kurnba!) OJv., NJv. , Ma1.135 kumpo1 Ma1.96 kurnpó1 Tag. ,NJv.96 kumpu1 Snd. 96 k-um-rut OJv.150 kuncó NJv.165 kunda!) Snd.155 kundan-an raja Ma1.155 kundó1 Tag.151 kundor Ma1.151 kundur Snd.151 kunQa!) NJv.155 kunQur OJv.151 kuné!) NJv.54

kunin OJv.,Ma1.54

kün OJv.73

kura-kura Ma1.161

kuranjiq Snd.71

kura!) Snd.,OJv.,

NJv .• Ma1. 105

kusuka Mer.70

kuta!) Snd.,NJv.,

Ma 1 .38

kutu Fi. ,Mal.,

NJv.118

kutuq Snd.118

kuqu1 Snd.179

kuwa1 NJv.132

kuwu NJv.133

kuwuq Snd.Cir .• Ind.,

kuyaq Snd.161

Surn.133

la Srn.173 laqa(h) Mad.173 1aqan Mad.173 laqas Mad.179 1abáy Mad.91 1á:bay Tag.91 lab; Tag.76 laboh Ma1.187 1abóh NJv.187 1abor Ma1.139 1abór NJv.139 1abu TBt.187 labuh Snd.187 1abur Snd.139, Mad.S.220 labbhu Mad.187 1abhur Mad.B.P.139 lade!) Ma 1.186 1adé!) NJv.186 1adre Sa.91 1addhi!) Mad.S.P.186 1aqe To.147 laen Ma 1. 182 1aén NJv.224 1aèn Mad.182 lafa To.153 lafu To.149 1agyóq Tag.87 1 aha!) Snd. ,0Jv. , NJv.173 1ahaq Snd.173 1ain Snd.182

wa!)és NJv.129

wasi OJv.,NJv.127

wata!J OJv. ,NJv.126

watis OJv.193

watu NJv.40,126,

OJv.126

(wa)walas NJv.128 wa-wanan NJv.125 wilah OJv.,NJv.126, Snd.218 wila!) Snd.,NJv.130 wilat NJv.125 w-in-eh OJv.126,180 winéh NJv.126 winèh NJv.126 win i OJ v. 131 winih OJv.126 w-in-ila!) OJv.130 winta.!) OJv.131 w-in-uni OJv.99 w-in-u!Jkus OJv.190 wi ré!) kuné!} NJv.126 wi rè!) NJv. 126 wirè!) kuné!J NJv.126 wiséq NJv.129 wisik Snd.219 wlikat NJv.130 wöh NJv.127,162,214 wok OJv.64 wölu NJv.84 WÖ!} NJv. 163 wóq NJv.161 wöq NJv.64,212 wör NJv.124 wös NJv.126,162 wötNJv.1l9 wöwö NJv.125 wrat OJv.219 wudal NJv.110 wuqug OJv.154 wuhaya OJv.176

watuq Snd.40

wino!) OJv.62

winö!J NJv.62

wiré!J NJv.126

wit-an OJv.129

wlas OJv. 125

wli hapu OJv.125

wö-wö NJv.124

wük OJv.161

wukér NJv.l03,127

wukir OJv.l03,127

wuku OJv. ,NJv. 127

wulan OJv.,NJv.129

wu 1 i in

yana-yana Fi.99 ya~ in ka-y~-an Mad.224

sa-wuIT NJv.127

wulih OJv.127

wulóh NJv.127

wulöh NJv.127

wulu OJv.,NJv.129

wuluh OJv.127

wu!)kal OJv.,NJv.132

wu!)kóq NJv.65

wU!Jkós NJv.190

wU!Jkuk OJv.65

wU!Jsu OJv.68

wU!Ju NJv.127

wU!Jü OJv.127

wuri OJv.,NJv.150

wuri -wuri NJv. 150

wuriq-wuriq

Snd.222

wusu NJv.127

wusü OJv.127

wuta OJv.130,

Bug.208

wutaq Snd.130

wutö NJv.130 WUWÓ!) NJv.190

WUWU!) OJv.190

wuyah OJv.84

wwah OJv.127

wwalu OJv.84

wwa!) OJv.163

wwara OJv.84

wwas OJv.126,162

wwat OJv.1l9

wway OJv.165

wwaya OJv.84

wwit OJv.129

wyah OJv.126

yali Fi.185

~ava Fi.82 ~ina Fi .164

ya!) Mal. 170,

Snd. ,NJv. 224

yara!)-i Fi.187

yuyu NJv.89

yatu Fi.l05

yaza Fi.186

Shortly af ter the completion of the preceding pages received a copy of D. J. Prentice's paper "Vet another PAN phoneme?" in which he also suggests the reconstruction of two labial stops. namely *S and ·b. instead of Dempwolff's single *b. Prentice compares Javanese (Jav.). the Kadazan dialect of Coastal Dusun (Kdz.). and the Timugon dialect of Lowland Murut (Tmg.). Kadazan and Timugon "are both members of the ldahan group of languages spoken in Sabah and have close affinity to languages spoken in the Philippines and northern Celebes. They are representative of the two main subgroups of the ldahan languages. respectively Dusunic and Murutic." (p. 1)

Prentice reconstructs ·S and ·b to account for the correspondences shown in the following table:

In Timugon initial *B disappears af ter a CV-prefix: e.g., * BaRqat, Tmg. bagat 'weight', ma-agat 'heavy'. Intervoeal ie * B disappears in Timugon preeeding or following u and in the environment a •.. i: e.g., *luBan, Tmg. luaD 'hole' and * RaBii(h), Tmg. m'aiq 'evening'. Examples illustrating the *B: *b distinetion and eited by Prentiee are the following 3 : *B-: * Bulan, Tag. buwán, OJv. wulan, NJv. wulan, ram-bulan, Mal. bulan, Kdz. vuhan, Tmg. bulan 'moon'. *-B-:*laBan, Tag. lá:ban 'to fight', OJv., NJv., Mal. lawan, Kdz. havan, Tmg. lawan 'rival, opponent'. *-(m)B-: *tu(m)Buq, Tag. tubóq 'grow', OJv. tuwuh 'body, growth, life', NJv. tuwóh 'grow', Mal. tuboh 'body', tumboh 'grow', Tmg. -tuuq 'grow'. *b-: *baDal, Tag. biDi 'deaf', NJv. ba!)al 'headaehe', Mal. banal, Kdz. bODo, Tmg. bo!)ol 'deaf' . *-b-: *tabus, Tag. tubós, OJv. tabus, NJv. tabós, Mal. tabos, Kdz. tobus, Tmg. -tabus (perhaps loan from Malay) 'to redeem'. *-(m)b-: *la(m)ban. Tag. libi!) 'grave', NJv. laba!) 'aqueduet, irrigation' (perhaps not eognate), Mal. lamba!) 'wheel-rut'. Kdz. hobo!), Tmg. lobo!) 'grave'. Prentiee proposes ambiguous reeonstruetions with -(Bb) whenever the posited eognates have the reflexes shown in the following table:

Examp1es are the fo11owing:

(1) • (Bb)a!}kay, Tag. ba!)káy, OJv. wankay, NJv. wa!)ké, ba!Jké, MaL, Kdz., Tmg. ba!)kay 'corpse';· 13(Bb)iq, Tag. lab; 'excess', OJv. 13wih 'excess, superior', NJv. luwéh, 13béh (the latter is probably a loan from Ma1ay), Mal. 13beh, Kdz. hobiq 'more', Tmg. -labiq 'overflow'.

(2) ·(Bb)uRuk, Tag. bugók 'rotten', OJv. wük, NJv. wóq 'smelling, fou1 odor', Mal. buroq 'decayed', Kdz. b-in-uuk, Tmg. buuk 'add1ed egg'; ·ta(Bb)H, OJv. tawilk, twH, NJv. tuwH 'to stab, pierce through', Mal. tebaq (-eirregu1ar; perhaps not cognate)4 'to chop, hack', Kdz., Tmg. -tobok 'stab'.

(3) • (Bb)a!)un, Tag. bá:!)on 'erected', OJv. wa!Jun 'build', ba!)un 'stand up', NJv. wa!lón 'shape, form', ba!)ón 'build', Mal. ba!)on 'rise, raise', Kdz. a-va!Jun, Tmg. ma-a!)un 'become, create'; ·i(Bb)aq, Tag. qibá 'different', NJv. ébah 'move', éwah 'change', Kdz.

t-iva-n, Tmg. iwa-n 'parent-1n-law'.

(4)* (Bb)alay, Tag. bá:hay 'house', OJv. bal~y 'hall', NJv. balé 'ber 'I', Mal. balay 'hall', Kdz. v-in-ahay 'beesnest', sam-bahay 'live in one's fiancé's house', Tmg. walay 'house'; * Ra(Bb)ut, Tag. gá:bot 'tear away', NJv. rabót 'lift up, out', Mal. rabot 'tear away', Tmg. -a ut 'carry, convey' (loss of R- irregular; perhaps not cogna te) .

(5) *tu(Bb)a, Tag. tu:ba, NJv. tubö, Mal. tuba, Kdz. tubo, Tmg. tuo 'fish-poison'.

(6) *(Bb)a(Bb)a, Tag. babá 'carry on the back', OJv. wawa, NJv. wöwö, Mal. bawa 'carry', Kdz. -babo 'carry on back or shoulders', Tmg. lim-bawo 'shoulder' (possibly not cognate).

Prentice feels that his evidence "strongly suggests the presence of a voiced fricative series" (p. 35) in PAN of which *B 1s posited to be the bilabial member (the velar one being *R). He thus puts forth a hypothesis as to the phonetics of this proto-phoneme, just as I do on p. 210 where I suggest that PMJ B might be taken to be a voiced bilabial unaspirated stop (*b being the aspirated counterpart). There seems to be little doubt about *B being labial, but the d1fference in the interpretation as to it being a fricative (Prentice) or a stop (Nothofer) indicates that Dyen is correct when he says that one should be cautious about the phonetics of proto-phonemes at this time: " ... the determination of the phonetic nature of a proto-phoneme depends increasingly on the subgrouping as the phonetic variation of the correspondences increases. Since the subgrouping of the Austronesian family still is in large part indeterminate, particularly at the highest levels, it is perhaps incautious to attempt to deal with the phonetics now whenever the phonetics of the reflexes show great variation, except in the most tentative way." (Oyen 1971:23)

Whatever the exact phonetic nature of *B, one might tend towards Prentice's hypothesis of interpreting it as PAN, since a large number of languages shows that a distinction between *B and *b is necessary. However, one should always keep in mind that ihe reconstruction of *B in addition to *b has to be posited to account for correspondences which - so far - have only been found in WestIndonesian languages. This *B can perhaps be considered to be a member of an additional consonant series of which *0 is another member. The reflexes of • 0 are closely patterned with those of *B.

Prentice and I agree in the reconstruction of the *B : *b distinction in 30 of 54 comparable instances. Complete disagreement appears in only two instances for which Prentice reconstructs *b and I reconstruct *B. In two instances Prentice reconstructs *b and I propose ambiguous reconstructions with ~(Bb), and in twenty instances

Prentice proposes an ambiguous reconstruction with *(Bb) and I that of *B. Comparing the criteria which lead to ambiguous reconstructions, one notes that one of the main reasons leading to the last type of divergent reconstructions is to be found in the difference of interpretation of the Javanese evidence. Prentice treats the Javanese evidence as inconclusive and reconstructs *(Bb) whenever Old and/or New Javanese has (have) a doublet, one member exhibiting w, the other exhibiting b, whether Idahan (=Kadazan and Timugon) exhibits the reflex of *B or *b. I reconstruct *B whenever Old and/or New Javanese has (have) a doublet, whatever the reflex in the other languages.

Prentice and I agree in the reconstruction of * B and *b in the following instances:

*B: *BahuR 'mix' *Bajaq 'i nform' *Balas 'repay' *Baliw 'change' *BaRa(h) 'ember' *BaRqan 'molar,5 *Batu 'stone' *Bauk 'chin whiskers' *Bana!) 'win' *BaRqat 'heavy' *Buaq 'fruit'

*BuBun 'roofri dge' *Buku 'knot' *Bulan 'moon' *Bulu 'hair' *aBuh 'ash' *iB3R 'appetite. saliva' *luBan 'hole' *quBan 'grey hair' *Ra(m)Bun 'cloud' *saBu!) 'cockfighting' *ta(m)BaR 'antidote' *tu(m)Buq 'grow' *b: *bawan 'onion' *bi bi 'duck' ·binkuk 'crooked' *butbut 'pl uck' *t3(m)buk 'pierce' *t3bas 'cut down'

Let us now consider ·w. The Kadazan reflex of this proto-phoneme is w in initial and intervocalic position and the Timugon reflex is b or ~ (the latter af ter a CV-prefix) in initial position and w in intervocalic position.

The Sundanese reflex of what has hitherto been reconstructed as intervoca1ic *w is not on1y w as stated on p. 84. but we a1so find nc or competing forms. one exhibiting w. the other exhibiting nco

Sundanese has w as reflex of PAN intervoca1ic w in the instances presented below:

(1) *awak. TBt. ak 'backbone'. OJv. awak. NJv. awaq (l). Mal. awaq. Snd. qawak (l). Mad. abäq (l) 'body', Fi. 1-ewe 'f1esh. content'.

(2) *awan. TBt. aoa9' OJv .• NJv .• Mal. awa9-awa9' Snd. qawa!)-awan. Mad. bän-abä9 'atmosphere'. Mer. avana 'rainbow'. Fi. yawa 'distance' .

(3) *bawan. Tag. bä:wan 'gar1ic'. TBt. baoan. NJv .• MaL. Snd. bawan. Mad. bhäbä!) 'onion'.

(4) *lawas. TBt. laos 'still'. OJv. 1awas. NJv. 1awas (l). Mal. 1awas. Snd. 1awas (l) 'to last long', NgD. lawas 'a long time ago'. Mer. lava 'without interruption' .

(5) *i'iawa. Tag. qu-nä:waq 'consideration'. NJv. nöwö. r~al. nawa. Snd. nawaq, Mad. nabä 'soul'. To .• Fu .• Sm. ma-nava 'to breathe'.

(6) *tiwas. Tag. tiwas 'be fau1ty'. OJv .• NJv. tiwas 'to perish'. Mal. tewas 'being worsted'. Snd. tiwas 'to perish'. Mad. tèbäs 'accident. disaster' • NgD. tiwas 'guilty'. Fi. ndewa 'contagious'. To .• Fu. ma-tsiva. Sm. ma-tiva 'unhappy. poor'.

Sundanese has -ncas reflex of PAN intervocalic w in the following instanees:

(7) ·lawaq6, Tag. lá:wa, lawa-lá:wa, la-lá:waq, Mal. lawa-lawa, Snd. lancah, Mad. labá-labá 'spider, web'.

(8) ·sawa, Tag. sawá, TBt. sa, OJv. sawa, NJv. söwö, Ma 1. sawa, Snd. sancaq, Mad. sabá 'python'.

In one instanee Sundanese exhibits competing forms, one with -nc-, the other with -w-:

(9) ·kiwa, Tag. k-al-iwáq, OJv. kiwä, NJv. kiwö, Snd. kencaq (L), kiwaq (H), Mer. havi-a 'left'.

In two instanees Sundanese has competing forms, one exhibiting -nc-, the other -w-, as reflex of Oempwolff's *b or *(m)b. In the first comparison Prentice reconstructs --B-. He does not consider the second:

(10) *laban (Prentice: -laBan), Tag. lá:ban 'to fight', OJv., NJv., Mal. lawan, Snd. lawan, lancan, Mad. labán 'opponent, rival'.

(11) ·ra(m)ba, TBt. ramba 'jungle', OJv. (-) rawa, NJv. röwö, Mal. rawa, Snd. rawaq, rancaq, Mad. rabá 'morass'.

Oempwolff proposes no reconstructions for the following comparisons. In the first instanee Snd. -nccorresponds to Malay and Javanese -wand to Sumenep and Bangkalan Madurese b. Toba-Batak has -0-:

(12) TBt. ta-raoa!l, NJv. k-rawa!J-an (0), Mal. ka-rawan' ta-rawan' Snd. ka-rancaD' Mad. S. ka-rabá n,

B. ta-rabán 'openwork'.

Snd. -nccorresponds to Ma 1. -mbin:

(13) Mal. lamba!) 'depressed of soil; low-lying, trodden underfoot; = (West-Sumatra) lamban. Also of roots swelling and becoming soft by long immersion in water', Snd. Hnca!j-;n 'red and soft (e.g., the feet of s.o. who has been standing in water for a long time), rotten (of rice which has been standing too long in water which is too deep), loose (through water)'.

The following comparison probably is to be associated with Sanskrit kutumburi, kustumbart 'coriander-seed'. Snd. -nccorresponds to TBt., NJv., and Mal. -mb-:

(14) TBt. katumbar, NJv., r1al. katumbar, Snd. katuncar 'coriander-seed'.

The data presented above can be accounted for by treating Snd. w as the reflex of intervocalic *w [ ( 1) - (6)] a n d Sn d. n cas th ere f 1 ex 0 f * wor * B aft er a nasal and - perhaps - before *a [(7) - (14)] 7. It is difficult to determine from this data alone whether the latter part of this hypothesis, namely that *-Nwor *-NBbecame Snd. nc only before *a, can indeed be maintained, but other evidence which will be given below tends to support it. An unambiguous reconstruction with ·w or *B can only be posited in case Tagalog and/or Toba-Batak have a cognate and this cognate has a single intervocalic consonant, since these languages do not merge *-wand

*-B-. We thus do not change reconstructions (1) - (6), but change reconstructions (7) - (11) to: (7) *la(N)waq, (8) ·sa(N)wa, (9) * ki(N)wa, (10) *la(N)Ban, and (11) *ra(N)(wB)a. For comparison (12) we reconstruct *ka-ra(N)(wB)an and *ta-ra(N)(wB)an, for (13) * laN(wB)an' and for (14) *katuN(wB)ar.

We have seen above that Sundanese exhibits competing forms in three instances: lancan/lawan, rancaq/rawaq, and kencaq (L)/kiwaq (H). With regard to the first two instances two hypotheses are possible concerning the origin of the occurrence of these competing forms: either the member of the doublet which has -wis treated as a borrowing from Malay or Javanese and the other member is treated as inherited or both are treated as inherited, the member which has -wbeing treated as continuing the member of an original doublet which had *-B- (in the case of (10» and *-(wB)- (in the case of (11», the other member being treated as continuing the other member of the original doublet which had *-NB- and *-N(wB)-, respectively. With regard to the third instance it appears that the high word kiwaq should be considered a borrowing from Javanese, given the fact that Javanese has a word of identical shape and meaning and that about three-fourth of the Sundanese high words can be interpreted as borrowed from this language (see p. 44).

This new hypothesis which argues that ·w and *B af ter *N and before *a became Snd. -ncwould make a reconsideration of some of my PMJ reconstructions with *-mBnecessary, if we regarded * N as representing the nasal * m. In the PMJ reconstructions *lamBat, *kumBa~, and *timBaq the Sundanese or the Madurese forms would have to be considered borrowings. If we maintained the reconstruction of *-mB-, the Sundanese form exhibiting -mbwould have to be considered a borrowing, since we would expect Snd. -ncas its reflex before * a. If, on the contrary. we reconstructed *-mbin these instances, the Madurese form exhibiting -mbwould have to be considered a borrowing, since the regular reflex of *-mbis Mad. -mbh-.

The Sundanese reflex of initial *w is not only ~, but we also find w or c. In my treatment of the PMJ semivowels on p. 83 I reconstruct *wl - and *w2- and take them to disappear in Sundanese. I now tend towards the hypothesis that *w l - became wand that ·w2- disappeared. However. the evidence leading to this new hypothesis is far from solid:

*w 1 -: (1) * w 1 a 0 a, Tag. wa la q. TB t. s 0 - a d a 't her eis not', OJv. wwara 'there is'. NJv. bra 'there is not'. Ma1. ada 'there is'. Snd. waraq 'to take place, realize', e.g., moal waraq nikah qay;naq 'there will be no wedding today', Mad. b&cJ&(h) 'there is'.

There are only two more instances in which the Sundanese form with wcannot be explained as a borrowing from Javanese. However, in both cases the imperfect meaning-fit raises doubts as to whether we have a cognate relationship. Dempwolff proposes areconstruction with ·wfor the first comparison:

(3) ·wliat, NJv. wèt-aké 'keep, preserve, conserve', Snd. wiat 'deposit, (en)trust'. ·w2-: (4) ·w2aya, OJv. wwaya, Snd. qayaq 'there is'. (5) *w2 uyaq, NgD. uyah, OJv. wuyah, NJv. uyah (L), Snd. quyah, Mad. buH 'salt'. Dempwolff's initial *w became Snd. c in: (6) ·waiR, TBt. aek, OJv. wai, wway 'water', NJv. wé in the compound wéda!) 'warm, hot water', Mal. aer, Snd. caiq Mad. aè!J 'water', NgD. aer-mati 'mud', Fi. wai, Sa. wei, To., Fu., SIl'. vai 'water'. Not only Dempwolff's initial • w, but also Dempwolff's initial *b (generally Prentice's ·B) became Snd. c in some instances:

(2)*wl aRi, TBt. ari 'day', nattu-ari 'yesterday', NJv. udan wé-wé 'rain while the sun is shining', Mal. (h)ari 'day', mata-hari 'sun', Snd. wa-wari-an8 in pestaq wawarian 'after-celebration, extra-fète, feast which finishes off a big feast', Mad. arè(h) 'day'.

(7) ·bahaq (Prentice: ·Bahaq), Tag. baháq, OJv. wah, NJv. wa-wah-an, Mal. bah, Snd. caqah, Mad. báqá 'flood', U. haa 'spring tide'.

(8) *ba!)kudu, Tag. ba!)kü:ro, TBt. bakkudu, Mal. bankudu, ma!)kudu, Snd. ca!)kuduq, Bal. wu!)kudu, NgO. mlnkudo 'kind of Morinda'.

(9) *binay, ba-binay (Prentice: *(Bb)inay, (Bb)a(Bb)inay), OJv. wini, bini, NJv. bini (B), Mal. bini 'woman', Snd. ca-weneq (l) 'virgin', a1so: 'maid', be-beneq 'fiancée', Mad.Kang. bá-bineq-an, Bug. wa-wine, Mak. ba-ine, Sm. ma-fine, To., Fu., Sm. fa-fine 'wife'.

(10) ·bani(r). Mal. baner 'buttress-1ike projection from a tree-trunk', Snd. cani r 'root-projection, projection on or at the root of a tree', NgO. baner' buttress-1 He projection from a tree-trunk'.

(11) ·bari!)in, TBt. bari!)in, OJv. wari!)in, NJv. wri!)én, Mal. bari!)en, Snd. cari!)in, wari!)in 'Ficus benjami na' .

(12) ·bata'] (Prentice: • Batan)' TBt. batalJ 'trunk', OJv. wata!) 'stick, trunk', NJv. wata!) 'long stick, po1e' (0), 'lying tree-trunk' (0), Mal. bata!) 'trunk', Snd. catalJ in cata!) kaiq 'tree-trunk stripped of its branches and 1ying on the ground', NgO. bata!), Mer. vatana 'treetrunk' •

Oempwolff proposes no reconstruction for the fo110wing comparisons in which we find identica1 initia1 correspondences:

(13) Mal. bayor, Snd. cayur 'Pterospernum javanicum'.

(14) Mal. ba1u10q, balo1oq, Snd. caruluk, Mad.B.

bá1u1uq 'fruit of the sugar-pa1m'.

In (8) on p.302 and in (3) on p.308 Ma1ay has competing forms, one with an initia1 b, the other with an initia1 m, matching a Sundanese form with initia1 c. In the fo11owing comparison Ma1ay and Javanese m matches Snd. c-:

(15) NJv. tamigi (metathesis), Mal. matigi, mantigi, Snd. cantigiq wUnuq 'Vaccinium varingifo1ium'.

We on1y find a Javanese and a Sundanese cognate in the fo11owing two comparisons:

(16) NJv. wa~bq 'be11y, stomaeh', Snd. caduk' (Vl) 'excrement, be11y, intestines', wa-waduk 'excrement, rectum' .

(17) NJv. wa~as, pa~as, Snd. cadas 'sandstone, rock'.

The comparisons 1isted be10w are to be treated as doubtfu1 cases because of the imperfect meaning-fit:

(18) OJv. a-bangun 'to compete', am-ban~un 'to equa l' , NJv. ban~óD-bancJbn (0) (0) 'one after the other, at the same time', Mal. bandon 'twin, dup1icate', Snd. candu!) 'to take a second wife', ban dUn 'one af ter the other', banduu-an 'do together'.

(19) Mal. baon 'catfish with venomous fins', Snd. eau!), baun 'name of a ri ver-fi sh' .

(20) OJv. baham, NJv. baham, bam 'molar' , Mal. baham 'to chew', Snd. cah;m (l) 'mouth', bah;m (Vl) 'put in mouth' .

From the data presented above we conc1ude that one can maintain the reconstruction with ·w1- in (1) - (3) and ·wZ- in (4) and (5). We a1so conc1ude - considering our hypothesis with respect to the origin of Sundanese intervoca1ic nc - that Sundanese initia1 c in (6) - (15) should probab1y be treated as a reduced nasa1 cluster reflecting a prenasalized ·wor ·B_ 9. In all instances the following vowe1 is ·a. However. we do not reconstruct a nasal cluster in this position. since we can put forth the hypothesis that Snd. carose from a nasa1 cluster which is not original. So. for example. one cou1d argue that. when postnasa1 medial ·w and·B became Snd. c. the combination of a preceding morpheme with a final nasal. e.g. an. and a base with initial ·w or· B wou1d be common. Since ·w or ·B is postnasa1 in the sequence • ... !)+base. it is superficia11y indistinguishable from a base with inherited c (from ·c). The following analogy could take place: ••.• !)+cbase : • cbase as •... nwbase : • x or * ... !)+Bbase :. x. At the time this change occurred there probably existed a large number of competing forms. one exhibiting C-. the other exhibiting w (from *w-) or b- (from *B-). In some instances the member with chas replaced the other member (e.g .• caqah). in other instances both members have been retained. only their meanings diverging (e.g .• cah;m. bah;m). and in stillother - unrecoverable - instances the member with bhas replaced the other member.

We thus maintain Dempwolff's reconstruction with *win (6), but change Dempwolff's *bin (7) - (12) to *B-: (7) *Bahaq, (8) *Bankudu, (9) *Bin3y, Ba-Bin3y, (10) *Bani(r), (11) *Barinin, and (12) *Batan. For (13) - (15) we reconstruct: (13) *Bayu(r), (14) *Baluluk, and (15) *Ba(n)tigi.

For (16) and (17) we reconstruct *(wB)-, because - so far - we have not been able to find a cognate in a language which does not merge *wand *B-. If we were to find a cognate which pointed towards areconstruction with *B, the Sundanese form wa-waduk in (16) would have to be considered a borrowing from Javanese, since *B- became Snd. b. The reduplication must then be posited to have occurred at a later stage of Sundanese. However, if we found a cogna te which pointed towards areconstruction with *w-, the Sundanese form would have to be considered ambiguous, since it could either be interpreted as a loan from Javanese or as an inherited form, given the fact that we have posited that Snd. wis the reflex of non-prenasalized *w-. We reconstruct: * (wB)aQuk and *(wB)aqas.

No reconstruction is proposed for comparisons (18) - (20) because they are doubtful instances.

Snd. be-beneq in (9) can be explained as a reduplicated form. The reduplication must have occurred at a stage of Sundanese at which *-B- had already become Snd. w. Otherwise, we would expect Snd. * be-weneq.

The Madurese form comparable to those cited in (10) is katumbhár 'coriander-seed' which, however, is treated as a loan from Malay or Javanese, since ·w and ·B regularly became Mad. b.

The Sundanese form with win (11), namely wari~in, is considered a loan from Javanese, since initial nonprenasalized ·B became Snd. b.

The initial p in the New Javanese form pa~as in (17) can be explained as being the result of an analogical change.

In the following comparison Sundanese has competing forms, one exhibiting initial and postnasal b, the other initial and postnasal c. We reconstruct initial and postnasal ·B: ·BalinBin, TBt. balinbin 'Averrhoa', NJv. blimbén 'the belimbing-fruit', blimbi~-an 'edged, ridged', Mal. balimbe~ 'ridged longitudinally', buah balimben 'name of the fruit of the Averrhoa bilimbi. As a plantname balimben covers among others: Averrhoa bilimbi, A. carambola', Snd. calincin 'Averrhoa', buah calincin 'unridged fruit of a kind of Averrhoa, rarely eaten', buah balinbin 'ridged fruit of a kind of Averrhoa, generally eaten'. In the Kadipaten-area buah caliDcin is the ridged fruit and buah balinbin the unridged fruit. Mad. bhálinbhi9' bhálimbhin 'Averrhoa' is treated as a loan from Javanese or Malay.

The postnasal c in Snd. calinci n will have to be explained as the result of an assimilation to the initial c, if we want to maintain the hypothesis that postnas~l ·B became Snd. c only before *a.

In a number of instances Malay has doublets, one member of which has an initial vowel or b, while the other member has an initial c, matching a Sundanese form with initial c. This evidence, although it is sparse, could be taken as pointing towards a hypothesis similar to that which is proposed for Sundanese. While initial ·w became Malay h/~ and initial *B became b, prenasalized initial ·w and *S before *a may have beéome Mal. c. The evidence is the following:

(1) Malay has a form caer 'diluted, watery (of viscous things)' besides aer 'water' as continuation of ·waiR. This form with ccannot be treated as a loan from Sundanese. since the Sundanese form exhibits the loss of final *R10 . Furthermore, also Jakarta-Malay has a form which exhibits c: èn-cèr 'thin and watery'. Note that in this instance c occurs af ter a nasal.

(2) Mal. banar 'a climber, Smilax species', canar 'a plant, gen. for Smilax species', Snd. canar 'a tree lying on the ground', Mad. bánar, ~áun bánar 'Smilax species (Liliaceae)'. In this case the Malay form with ccould be explained as a loan from Sundanese.

(3) In the following comparison Dempwolff and Prentice reconstruct *b-:

* 11 banku(q)an ,Tag. bankuan-bundok 'Pandanus dubius', Mal. ba!)kua.!), mankua!) 'Pandanus atrocarpus'; we also find Mal. ca!)kuan 'Chisocheton glomeratus or Dracontomelum mangiferum', Snd. ca.!)kuan 'name of a kind of reed from which things like baskets are made. lts leaves are used for the packing of aren-sugar', NgD. ba!)kuan 'kind of reed', Mer. vakuana 'Pandanus'. The Sundanese evidence leads to areconstruction with *B-. It is not clear whether the Malay form with ccan be associated with these forms. It seems, however, that all these forms are not directly associable with Mal. sankuan' Snd. bankuan' NJv. baDkbwa!), and Mad. b!!)köwa9 'yam-bean'.

There is one instance in which both Malay and Sundanese have competing forms, one with -w-, the other with -nc-. Dempwolff reconstructs *-w-:

(4) *kawaq, Tag. k!:wa 'saucepan', OJv. kawah 'pan', NJv. kawah 'crater', 'pot' (D), big cooking-pot' (0), Mal. kawah 'vat, cauldron, crater', kancah 'a narrow-mouthed cooking-pot for boi1ing rice', Snd. kawah 'a big and deep hole', also: 'crater', 'big cooking-pan', kancah 'a big iron pan', NgD. kawah 'pan'. The Malay form with -nccould be considered a loan from Sundanese, the Sundanese form with -wa loan from Malay. The Sundanese evidence suggests the reconstruction *ka(N)waq.

In the light of this data one might suggest in a most tentative way that in the followillg instance Mal. -crepresents a reduced nasal cluster, given the fact that it matches NJv. -w-: NJv. bawa!), Mal. (a)mbacan 'Mangifera foetida' . We tentatively reconstruct ·ba (N) (wB )a!).

Not only Malay, but also Javanese has doublets, one member of which has -w-, while the other has -nc-. In the following instance Javanese has a low word exhibiting -wand a high word exhibiting -nc-. Sundanese also has competing forms, one with -W-, the other with -nc-. The one with -wis marked by Coolsma as a loan from Javanese. It is interesting that the form with -ncis a high word. As pointed out earlier, Sundanese low words generally contain inherited reflexes as opposed to high words which generally contain reflexes associated with loan words. We thus find:

NJv. réwaD (L) 'companion, after-birth, help s.o.', rénca!) (H) of ba tór 'servant', (H) of réwa!) 'compani on' , Snd. rewa!); renca!) (H) of batur 'servant, companion, adherent'. In this case the Javanese form with -nccould be taken to be a loan from Sundanese, whereas the Sundanese form with -wprobably is a loan from Javanese as indicated by Coolsma. We tentatively reconstruct ·ri(N)(wB)aD.

Two comparisons seem to argue against our hypothesis that prenasalized ·w and ·8 became Snd. cand -nconly before *a:

OJv. w-in-uruk 'to be instructed', NJv. wuruk-an 'education, teaching, instruction', Snd. wuruk 'instruction'. Sundanese also has curuk 'index-finger' which might be considered cognate with the other forms.

New Javanese and Sundanese have a doublet, one member of which exhibits -b-, while the other exhibits -ncin:

NJv. obor, oncor. Snd. qobor, qoncor 'torch'.

It is apparent that the evidence presented here with regard to prenasalized *w and *B is not entirely solid. Questions arise with regard to what kind of nasal precedes these two proto-phonemes. We leave the question open by reconstructing *N. Comparison (14) seems to suggest that in this instance at least the nasal probably was *m, because it can be associated with Sanskrit kutumburu, kustumbart. Furthermore, it is not entirely clear, whether prenasalized *w and ·B became Snd. c only before * a. We also will have to further investigate, whether this particular change can be posited to have occurred in Malay and perhaps also in Javanese or at least in dialects of these languages as we 11.

FOOTNOTES

lThe significance of the asterisk will vary from the equivalent of PAN to PMJ.

2prentice does not cite any reconstruction with *-mB-. --mb-. or *-m(Bb~. but only reconstructions with an optional nasal. i.e. with *-(m)B- or *-(m)b-. This optional nasal is never reflected in the Idahan languages. We know of the postnasal reflex of *B and *b in Kadazan and Timugon only through forms which have a prefix ending in a nasal: e.g .• *Buku. Kdz. vuku 'knuckle, knot in wood'. tim-buku 'hairbun'. Tmg. buku 'knuckle'. tim-buku 'knot'; *bawan, Tmg. am-bawan 'onion'.

3All comments included in parentheses are those of Prentice.

4The co~nate word probably is Mal. tabaq 'bar used for stone-breaking', see p. 77.

51 suggest the reconstruction of penul timate * a instead of -a, see p. 132.

60yen 1953a reconstructs a final vowel: *lawa. We reconstruct a final q and explain the absence of -q in Tag. lawa-lá:wa and the absence of -h in Mal. lawa-lawa as follows. The final *q of a root which became preconsonantal in a doubling is lost in Tagalog and Malay. By analogy the final q of the second member of the doubling also disappears. Tag. lá:wa is areformation from Tag. lawa-lá:wa.

7Snd . -ncalso is the reflex of PAN postnasal c, see p. 113.

80n p. 83 Snd. poeq 'day' is treated as the continuation of this etymon preceded by *pa-.

9Snd . calso is the reflex of PAN initial c. see p. 113.

lOThese new hypotheses regarding the origin of cin Snd. caiq and Mal. caer make the combination of my PMJ reconstructions *caiR13 and *wlaiR possible. We simply reconstruct *waiR.

110yen 1953a reconstructs intervocalic q, since he considers NJv. ba!lkówa!l 'yam-bean' a cognate. Since we exclude this form from tne comparison because of the poor meaning-fit. the reconstruction becomes ambiguous. i.e. the proto-form may have exhibited *hiatus or * -q-.

Besides the works cited in the main bibliography used the following works for this section:

Dahl, O.C. 1973. Proto-Austronesian. Lund. Encyclopaedie van Nederlandsch-Indië. 1917-1921. 4 vols. 's-Gravenhage, Leiden.

Matthes, B.F. 1859. Makassaarsch-Hollandsch woordenboek. Amsterdam.

Matthes, B.F. 1874. Boegineesch-Hollandsch woordenboek. Amsterdam.

Prentice, D.J. 1974. "Vet another PAN phoneme?" Paper presented at the First International Conference on Comparative Austronesian Linguistics, Honolulu.

p. x i , 1. 5 up; Javaans ( for: Javans) p. 16, 1. 7 down; wat ja (for: waca) p. 49, 1. 13 down; an-atas (for: a-natas) p. 49, 1. up; ta 11 öq (for: tellöq) p. 52, 1. 6 down; new note af ter Snd. tarimaq: According to Professor Noorduyn 01d Sundanese has tarema. p. 52, 1. 7 down; an-arima (for: a-narima) p. 57, 1. 6 up; an-impalJ (for: a-nimpa~) p. 66, 1. 2 down; man-uru!l (for: ma-nuru!)) p. 83, 1. 6 down; ... disappeared in Sundanese and Malay.(for: .•• disappeared in Sundanese.) p. 83, af ter 1ine 10 down; add: It became Madurese b. p. 84, 1. 9 down; Mad. bujá (for: buyä) p. 85, 1. 5 down; mangrove (for: mangrow) p. 86, 1. 3 down; PWI (for: PQI) p. 86, 1. 14 down; an-an-ami (for: a-na-nami) p. 91, 1. 7 up; hang rice (for: hand rice) p. 94, 1. 7 up; new note af ter 01d Snd. apuy: Prof. Noorduyn never found apuy in the manuscripts. It is 1isted by Coo1sma on the authority of P1eyte, who -according to Prof. Noorduyn - is a dubious source. Modern Sundanese has qapuy in samp;q qapuy 'roasted cassava'. The expression occurs in the Cianjur area. p. 95, 1. 5 down; OJv. an-u1uy-i 'to continue', an-u1i-n-uli (for: OJv. a-nu1uy-i 'to continue', a-nu1i-nu1i) p. 102. 1. 10 down; PMJ 1 in ••• (for: PMJ in ••• ) p. 102. 1. 6 up; ••• to PAN j. (for: ••• to Dyen's

p. 128. 1. 5 down; ••• in whieh 01d and/or New Javanese •.• (for: ••• in whieh New Javanese ••• )

symbo 1 j.) p. 108. 1. 13 down; •.• to PAN z. (for: ••• to Dyen's

symbo 1 z.) p. 109.1.1 down; nipah (for: nipa) p. 109. 1. 13 up; •.• to Dyen's PAN Z. (for: to

Dyen's symbo1 Z.) p. 112. 1. 11 down; an-agih (for: a-nagih) p. 124. 1. 10 down af ter *Ba1iBis 96 ; insert: (PHN.

Tag. ba1iw1s 'tea l' : baliwis) p. 125. 1. 1 down; am-awa ( for: a-mawa) p. 125. 1. 11 down; mam-a1i (for: ma-mali) p. 125. 1. 11 up; am-a1it (for: a-ma1it)

p. 128. 1. 12 down; sa-ba1ik-naq (for: sa-ba1ik-na) p. 130. 1. 7 up; 'to wake up' ( for: 'to make up' ) p. 132. 1. 13 up af ter baR2qa U).; insert: Snd. b;h;!J

( L ) 'neek'. p. 132. 1. 8 up; Mal. (h)abes ( for: Mal. (h)abis) p. 132. 1. 2 up; Mad. abu(h) 'ash' • ( for: Mad.

abu(h).) p. 132.1. 1 up af ter NJv. kuwa1; new note: With an ana1ogiea1 k-.

p. 142, 1. 3 down; an-in kab ( for: a-niukab) p. 145, 1. 6 up; ..• for final Q which accordinQ ... (for: ... for final which ac c 0 rd i n 9 ... ) p. 152, 1. 4 down; am-anc;la m (for: a-man~am) p. 164, 1. 8 down; man-a kw-a (for: ma-nakw-a) p. 173, 1. 11 down; Ma 1. dahan (for: Mad. dahan) p. 176, 1. 9 up; • •. in the orthography. (for: in theorthography.) p. 177,1. 5 up; Sm. paqu-sisi (for: Sn. paqu-sisi) p. 180, 1. 13 down; and the two vowels ..• (for: in the two vowels ..• ) p. 195,1. down; man-antan (for: ma-nantan) p. 195, 1. up; Mal. bumbon (for: Mal. bumbun) p. 199, 1. 5 up; man-utuh (for: ma-nutuh) p. 211, 1. 3 down; Way Lima (for: Way lima) p. 211, 1. 17 down; The phonemes c and jare frontal stops, ij is a frontal nasal (for: The phonemes c, j, and nare frontal stops.) p. 213, 1. 18 up; the script of the 01d Sundanese manuscripts ... (for: the traditional Sundanese writing system ... ) . p. 213, 1. 1 up; and *a contracted .•• (for: and *acontracted ... ) p. 214, 1. 5 down; *baRah (for: *baRaq) p. 228, 1. 6 up; insert the Sundanese word for 'guts ': p;jit.

p. 246, between BuDiq, 150 and Buit, 129; insert: BU9u9, 154 p. 246, 1. 7 up, left column; erase: Bukit, 103 p. 246, 1. 6 up, left column; Buki!, 103, 127 (for: BUkil, 127) p. 247, 1. 1 up, left column; erase: bu~ug, 154