ArticlePDF Available

Is there a Bima-Sumba Subgroup?

June 2008
Oceanic Linguistics 47(1):45-113

DOI:10.1353/ol.0.0006

Authors:

University of Hawaiʻi at Mānoa

For some seven decades a number of Austronesian languages in the Lesser Sunda islands of eastern Indonesia have been assigned to a "Bima-Sumba" subgroup. No evidence has ever been presented for this group, yet through sheer repetition it has come to be accepted by many scholars. A comparative analysis of "Bima-Sumba" languages shows clear support for a Sumba-Hawu group, and limited evidence for a larger genetic unit that includes many or all of the languages of western and central Flores. However, there is no support for a more inclusive subgroup that incorporates Bimanese, unless it also includes languages that were not assigned to the original Bima-Sumba group.

Content uploaded by Robert Blust

Content may be subject to copyright.

IS THERE A BIMA-SUMBA SUBGROUP?

Robert Blust

University of Hawai’i

For some seven decades a number of Austronesian languages in the Lesser Sunda islands of eastern

Indonesia have been assigned to a ‘Bima-Sumba’ subgroup. No evidence has ever been presented for this

group, yet through sheer repetition it has come to be accepted by many scholars. A comparative analysis of

‘Bima-Sumba’ languages shows clear support for a Sumba-Hawu group, and limited evidence for a larger

genetic unit that includes many or all of the languages of western and central Flores. However, there is no

support for a more inclusive subgroup that incorporates Bimanese, unless it also includes languages that

were not assigned to the original Bima-Sumba group.

1. Introduction. In 1938, at a time when it might have been imagined that colonialism

would become a permanent institution, the Dutch government commissioned the

publication of the Atlas van tropisch Nederland, a compilation of economic information,

vital statistics, etc. on Dutch colonial possessions in the tropics. As part of this larger

enterprise S.J. Esser, a language officer (taalambtenaar) of the colonial government in

what was then the Netherlands East Indies was asked to prepare a map of the languages

of this region, together with an accompanying key. In his key to the map Esser went

beyond a simple listing of language names, organizing the languages into 19 implicitly

genetic groupings: 1. Sumatra group, 2. Java group, 3. Dayak group (including Bajaw), 4.

Bali-Sasak group, 5. Philippine group, 6. Gorontalo group, 7. Tomini group, 8. Toraja

group, 9. Loinang-Banggai group, 10. Bungku-Laki group, 11. South Celebes languages,

12. Muna-Butung group, 13. Bima-Sumba group, 14. Ambon-Timor group, 15. Sula-

Bacan group, 16. South Halmahera-West New Guinea group, 17. Melanesian languages,

18. North Halmahera language family, 19. Papuan languages. He marked groups 1-17 as

‘Malayo-Polynesian’ (Austronesian), and groups 18 and 19 as independent families

unrelated to Austronesian or to one another. Given its form (a brief contribution to an

atlas), Esser presented no supporting evidence for his classification. In hindsight it can

be said that he made some important observations, as in recognizing the Tomini and

South Halmahera-West New Guinea groups. However, he also committed serious errors,

some of them quite surprising, as in separating Mongondow and Ponosakan, which he

assigned to a Philippine group, from Gorontalo, Buol, or Kaidipang, which he assigned to

a distinct Gorontalo group, when all of these belong to the fairly closeknit Gorontalic

group of Philippine languages (Sneddon and Usup 1986, Sneddon 1989, Blust 1991).

Esser’s Group 13, ‘Bima-Sumba’ contains the following languages:1

TABLE 1: Languages in the ‘Bima-Sumba’ Group of Esser (1938)

1 Although the Atlas did not appear until 1938, it had been in planning and production for some years prior

to this. Esser himself died in 1930, and it is likely that his classification was prepared during the 1920s.

Language Location

Bimanese eastern Sumbawa

Manggarai west Flores

Ngad’a-Lio central Flores

West Sumba western Sumba

East Sumba eastern Sumba

Hawunese Savu/Sawu

As with other units in his classification, no supporting evidence was given for a Bima-

Sumba group, and none has been given since. Labels for putative linguistic subgroups

nicely illustrate the power of names: once they have been committed to print it is difficult

not to take them seriously. One factor abetting this tendency is repetition. After a claim

has appeared in a credible source, even without supporting evidence, it is easily cited by

other writers as though it were on an equal footing with the best-supported hypotheses,

and repetition alone is capable of conferring a spurious air of legitimacy to assertions that

were never accompanied by arguments or data of any kind. The so-called ‘Bima-Sumba

group’ is a case in point. To illustrate with sample citations from the published literature,

Verheijen (1967:xviii) notes that he found some interesting points of comparison between

Manggarai and languages in the central Moluccas, but that he “paid closer attention to the

languages of the Lesser Sunda islands (Nusa Tenggara), especially to those of the Bima-

Sumba group.” Lebar (1972:69 ) noted that Bimanese “is related to Savunese and shows

some affinity with Manggarai” (where “related to” = “subgroups with”). Fox (1977:5)

notes that “The peoples of Sumba, Raijua, Savu, and Ndao speak separate but related

languages that belong to the Sumba-Bima grouping of eastern Indonesian languages.”

Ruhlen (1987:346) lists sixteen ‘Bima-Sumba’ languages, including Bimanese, Dhao,

Hawu, all languages of Sumba (Anakalangu, Kambera, Kodi, Lamboya, Mamboru,

Wanukaka, Weyewa), and six languages of western and central Flores (Ende, Li’o,

Manggarai, Ngad’a, Palu’e, Riung). More recently Klamer (1994:1) stated that Kambera

“is part of the Sumba-Bima group of Austronesian languages,” and Grimes (2000:1:507)

holds that the ‘Bima-Sumba group’ consists of 26 languages, including Bimanese, all

languages of Sumba (the seven cited by Ruhlen plus Laura of west Sumba), Hawu, Dhao,

Komodo and 14 languages of western and central Flores (Ende, Ke’o, Kepo’, Li’o,

Manggarai, Nage, Ngad’a, Palu’e, Rajong, Rembong, Riung, Rongga, So’a, Wae Rana).

The first indication that something is wrong with the Bima-Sumba hypothesis is found in

a comparison of basic vocabulary. As noted in Blust (1996a:275), it would be surprising

if languages that subgroup together show no exclusively shared innovations in basic

vocabulary. Appendix 1 uses a variant of the Swadesh 200-word list to examine this

question for Bimanese, Manggarai, Kambera and Hawu (Savu). PMP reconstructions are

given first, followed by the equivalent of the test-list meaning in each language, with

letter coding of cognate decisions. The results are summarized in Table 2:

TABLE 2: Innovations in basic vocabulary that are exclusively shared among languages

of the ‘Bima-Sumba’ group (B = Bimanese, M = Manggarai, K = Kambera, H = Hawu)2

2 In addition, Manggarai tilu, Hawu wo-dilu ‘ear’, and Manggarai lema, Kambera làma ‘tongue’ are post-

PMP innovations, but are shared with languages in the ‘Ambon-Timor’ group, as Tetun tilu-N, Sika,

2

B M K H

branch x x

breathe x x

flower x x x

left side x x?

long x x

sew x x

elder sibling x x?

sky x x

star x x

thick x x?

thou x x

wind x x

In short, Kambera and Hawu share four items (branch, breathe, long, wind) on a variant

of the Swadesh 200-word list that are unattested elsewhere except in other languages of

Sumba, and in Dhao. In addition they share an innovation for ‘flower’ that has been

found elsewhere only in the languages of western and central Flores. There is thus some

prima facie evidence for a Sumba-Hawu subgroup that may link more distantly with the

languages of western and central Flores. Most other pieces of evidence plotted in Table 2

are weak: Manggarai léo, Hawu keriu ‘left side’ may reflect an innovation that replaced

PMP *ka-wiRi, but the sound correspondences are irregular, and it is unclear whether the

Hawu form contains a prefix; Bimanese nda

?

u ‘sew; needle’, Hawu



yau ‘sew’,



yi



yau

‘needle’ reflect *zaRum ‘needle’, and the semantic extension to ‘sew’ carries little weight

in subgrouping; Bimanese sa

?

e, Manggarai ka

?

é ‘elder sibling’appear to show an

innovative final vowel, but the initial consonant correspondence is irregular, and the

history of these forms remains unclear; Manggarai awaN, Kambera awaNu ‘sky’ reflect

PMP *awaN ‘cloud’ with a semantic innovation that, again, is too commonplace to carry

much weight for subgrouping; Bimanese ntara, Manggarai ntala ‘star’ has replaced PMP

*bituqen, and must be taken seriously as subgrouping evidence, but Tagalog tála

?

‘bright

star, planet’, Nias madala ‘morning/evening star’, Sa’a madala, Arosi madara ‘morning

star’ point to PMP *mantalaq ‘Venus’, with apparent syllable loss in Bimanese and

Manggarai; it is unclear whether Bimanese te

≡

e, Kambera tímbi ‘thick’ are cognate (<

*t↔(m)bay) or convergent; finally, Bimanese ndai mu, Kambera nyu-mu ‘thou’ use the

PMP 2s genitive pronoun *-mu in innovative ways, but it is not clear that this reflects a

single innovation. The evidence for including Bimanese in Esser’s ‘Bima-Sumba’ group

is thus particularly weak, consisting of two semantic innovations (‘needle’ to ‘sew’ and

‘cloud’ to ‘sky’), one formal and semantic innovation (*mantalaq ‘Venus’ to ntara

‘star’), one possible functional innovation (use of *-mu in combination with some other

morpheme as a non-genitive pronoun), and two possible lexical innovations (elder

sibling, thick), both of which are problematic. Perhaps the most striking of these

innovations is the word for ‘star’, but this also appears in Sika dala ‘star’, hence in a

member of Esser’s ‘Ambon-Timor’ group (cp. *taneq > tana ‘earth’, *timuR > timu

‘east, east monsoon’, *qateluR > telo ‘egg’, *tuktuk > tutu ‘knock, pound, beat’, *tasik >

Lamaholot tilu-N, Kédang til ‘ear’, and Kemak lama-r, Mambai lama-lau ‘tongue’.

3

tahi ‘sea, saltwater’, *telu > telu ‘three’ for regular reflexes of *t- in Sika, and *ma-tasak

> daha ‘ripe’, *ma-tuquR > du

?

ur ‘dry’ for other reflexes of *nt-).

The primary purpose of this paper is to show that there is strong evidence for a Sumba-

Hawu group, and more restricted evidence for a larger subgroup that includes many or all

languages of western and central Flores, but that Bimanese can be included in this group

only if many of the languages of Esser’s ‘Ambon-Timor’ group are included in it as well.

In pursuing this question I will look at four languages representing all major areas within

the proposed ‘Bima-Sumba’ group: Kambera of eastern Sumba, Hawu, spoken on Sawu

island, Bimanese of eastern Sumbawa, and Manggarai of western Flores.

In describing the historical phonology of these languages it will be more practical to use

Proto-Malayo-Polynesian (PMP) as the primary reference point rather than the more

distant Proto-Austronesian, since all Austronesian (AN) languages outside Taiwan share

certain phonological mergers. PMP had vowels /i, u, e, a/, where *e represents schwa,

and consonants /p, t, c, k, q, b, d, z, j, g, m, n, ñ, N, s, h, l, r, R, w, y/ (where *c was a

voiceless palatal affricate, *q a uvular stop, *z a voiced palatal affricate, *j probably a

palatalized voiced velar stop ([gy]), *r an alveolar tap, and *R an alveolar trill, which

became uvular in many languages). The dominant canonical shapes were CVCVC and

CVCCVC, where the medial cluster was a homorganically prenasalized stop (*tumbuq

‘grow’, *punti ‘banana’), or the abutting consonants in a reduplicated monosyllable

(*butbut ‘pluck, uproot’, *selsel ‘regret’). A few forms appear to have contained other

types of clusters, as with *beRsay ‘canoe paddle’, or *paNdan ‘pandanus’. Prominent

distributional constraints include: 1. the palatals *c, *z and *ñ did not occur in final

position, 2. *j did not occur word-initially, and 3. the schwa did not occur word-finally.

In addition, *c, *g and *r are rare or absent in the basic vocabularies of many languages.

All of these languages show a number of loanwords. Where these can be identified with

confidence they are primarily from Malay, or from non-AN languages through the

medium of Malay. Although some of these loans may postdate the foundation of the

Republic of Indonesia, the great majority appear to have entered the local languages

centuries earlier, and so are attributed here not to Bahasa Indonesia, but rather to Malay.

2. Kambera. Lebar (1972:74) notes that the island of Sumba “may be regarded as a

single cultural unit, although it is customary to distinguish between East and West

Sumba, with Kodi in the extreme west considered separately.” Of the seven or eight

languages that are generally recognized for Sumba ---Anakalangu, Kambera, Kodi,

Lamboya, Mamboru, Wanukaka, Weyewa, Laura ---, adequate descriptive materials exist

only for Kambera, or East Sumbanese (Onvlee 1984, Klamer 1994).3 Although not

3 Onvlee (1984:ix) states that the Kambera spoken around Waingapu, the capital of the regency of East

Sumba, is ‘clearly related’ (duidelijk verwant) to dialects spoken over most of east and central Sumba, and

the eastern part of west Sumba, and more distantly related to six other ‘dialects’ on the island, namely

Mangili in eastern Sumba, Lewa in central Sumba, Anakalangu and Mamboru in the eastern part of western

Sumba, and Weyewa and Kodi in the westernmost part of the island. Klamer (1994:3) questions Onvlee’s

use of ‘dialect’ and states of these language communities that “there is no consensus on whether these

should be considered ‘languages’ or ‘dialects’.” Lansing et al (2007:16026), on the other hand, state

confidently that there are at least 29 languages on Sumba, a figure that far exceeds all earlier estimates.

4

always stated explicitly, it is generally assumed that the languages of Sumba constitute a

subgroup (Lebar 1972:74, Onvlee 1984:ix, Lansing et al 2007). Some evidence for this is

found in unpublished Swadesh 200-word lists for Kodi (Kod), Lamboya (Lmb),

Weyewa/Wewewa (Wey), Anakalangu (Ank) and Kambera (Kmb) collected by the

anthropologist James J. Fox (Fox n.d.), which show such exclusively shared cognate sets

in basic vocabulary as Kod bokol, Ank bakul, Kmb bakulu ‘big’, Kod, Lmb, Wey katilu,

Ank, Kmb kahilu ‘ear’ (other languages of the Lesser Sundas have a reflex of *tilu, but

without the prefixed element ka-), Kod, Kmb rumba ‘grass’, Kod loNge, Wey loge, Ank

logi, Kmb luNgi ‘hair’, Kod katakku, Lmb kataku, Ank, Kmb katiku ‘head’ < *kat↔ku

(Hawu, Dhao katu ‘head’ may be related, but if so show irregular syllable loss), Kod

bahen-jaka, Kmb jeaka ‘if’, Lmb kali, Kmb kalai ‘left side’, Kod daNa, Ank, Kmb daNu

‘much/many’, Wey buddi/wuddi, Kmb bidi ‘new’, Kod hudoNo, Ank riduN, Kmb

ruduNu ‘night’, Lmb, Wey, Ank kaweda, Kmb kawiada ‘old’, Lmb tuku, Kmb tukku

‘throw’, Kod Nandu, Kmb Nàndu ‘tooth’, Kod paNa-daNo, Ank paNa-daN, Kmb pata-

ndaNu ‘tree’, Kod jamma, Ank njiama, Kmb njuma ‘we (excl.)’, Kod haddu, Kmb hidu

‘sick, painful’, Kod jemmi, Wey yemmi, Ank njiami, Kmb njumu ‘you (pl.)’. In addition

the languages of Sumba share irregular developments in PMP *ikuR > Kod, Lmb, Kmb

kiku, Wey kikku, Ank kaiku, with unexplained k-, and in *naNuy (> naNi) > Kod, Ank

Nani, Kmb Neni ‘swim’, with a metathesis of the consonants that is surprisingly absent in

Lmb, Wey naNi. It is argued here that Hawu-Dhao subgroups with the languages of

Sumba, but in view of the limited information available for all of these except Kambera it

is not clear whether Hawu-Dhao is equidistant from all languages of Sumba, or is more

closely related to some of these languages than to others. For purposes of this paper it

will be assumed without further comment that Hawu-Dhao forms one genetic unit, and

the languages of Sumba another. The latter will be represented solely by Kambera, as

documented in Onvlee (1984).

Kambera is spoken in the eastern two-thirds of Sumba (Needham 1980:24, Klamer

1994:2-3). As just noted, the languages of Sumba appear to form a closed group, but data

in Fox (n.d.), which includes 200-word Swadesh lists for Kodi, Lamboya, Weyewa

(Waijewa), Anakalang (Anakalangu) and Kambera, suggests that linguistic diversity is

somewhat greater in western Sumba. Sect. 2 will touch on those aspects of synchronic

phonology that are most relevant to the aims of this paper, and provide an account of

Kambera historical phonology.

2.1. Synchronic phonology. The Kambera phoneme inventory, following Klamer

(1994:12, 16), appears in Table 3:

TABLE 3: The phoneme inventory of Kambera

Consonants Vowels

p t k i/i: u/u:

≡  j

m n ny N

mb nd nj Ng e/ai o/au

h

5

l

r a/a:

w y

Kambera p, t, k are voiceless unaspirated bilabial, alveolar and velar stops;

≡

and



are

bilabial and alveolar implosives, and j is a voiced non-implosive palatal affricate.

Because the implosives do not contrast with non-implosive counterparts they are written

as b, d in Onvlee (1984).4 The nasals have their expected phonetic values except that ny

is said to be a “prenasalized semivowel” rather than a palatal nasal. The prenasalized

stops mb, nd, nj and Ng are treated as unit phonemes for reasons that need not concern us,

and r is an alveolar trill. All lexical data are drawn from Onvlee (1984), who writes the

contrast of short and long vowels as i (short) vs. í (long), u (short) vs. ú (long), and a

(long) vs. à (short). In addition he writes a rare vowel á, as in á 1. ‘yes’, 2. short form of

aya ‘older sibling’, 3. ‘shriek, howl’, 4. ‘exclamation of astonishment’, yawáNu ‘cheer,

especially in religious rituals’, or ulá ‘stroke over’, but without further explanation.

Because of its rarity and apparent irrelevance to the topics treated here this vowel is

ignored. In addition Onvlee writes aü for what appears to be [awu]. Klamer (1994:16-

18) suggests that the diphthongs ai and au are long counterparts of e and o, and she

points out that there are three additional vowels that are confined to ideophones. Neither

of these points need concern us further. Fox (n.d.) writes a few Kambera words with

geminate consonants (as tukku ‘throw’), but geminates do not occur in any other source.

For convenience in locating forms in the original source I follow Onvlee’s orthography

except that ng is written N.

Plain and prenasalized voiced stops contrast, as in bai ([≡ai]) ‘female, of animals’ vs.

mbai ([mbai]) ‘musty, rotten’, or dàda ([↔a]) ‘sound of stuttering’ vs. ndànda ([nd↔nda])

‘fine, penalty; trembling, shaking’. In addition, Kambera has a process of prenasalization

by which ‘non-controlled intransitive verbs’ are derived from their transitive

counterparts, as with pata ‘break X’ : mbata ‘be broken (e.g. chair)’, or baha ‘wash X’ :

mbaha ‘be wet’ (Klamer 1994:255-58). In these alternations voiceless stops alternate

with their plain voiced counterparts (p- : mb-, t- : nd-, k- : Ng-), and h- (from *s)

alternates with nj-. This process appears especially common with bases that begin with b

or p, as these make up around 24% of the entries in Onvlee (1984) that begin with a stop,

yet constitute 18 of the 24 examples that Klamer gives to illustrate valency-reducing

prenasalization. In other cases the plain : prenasalized contrast reveals evidence of

morphology that has become at least partly fossilized, as with bulaku ‘open the eyes

wide’ vs. mbulaku ‘uncovered, without covering’, dàki ‘follow, stay nearby’ vs. ndàki

‘follow’, hauku ‘scoop up water’ vs. njauku ‘motion of scooping up water’, or ketu

‘hook’ vs. Ngetu ‘stay hooked’.

Prenasalized initial obstruents have not been reconstructed for PMP, but are an areal

feature in much of Sulawesi and in the languages of the western Lesser Sundas. In native

4 Klamer (1994:12) states that Kambera has “three implosive consonants”, but she lists only

≡

and



. In

addition, her phoneme chart lists j, nj, ny and y in the ‘alveolar’ column, while leaving her ‘palatal’ column

completely empty. I assume this is a printing error.

6

Kambera words many of these segments appear to originate from syncope of the prefixal

vowel in p- or b-initial verbs prefixed with *ma- ‘stative’ (*ma-panas > mbanahu ‘warm,

hot’, *ma-baRani > mbeni ‘brave’, etc.). Prenasalized initials may have been introduced

to Kambera, then, through a change of the form PMP *ma-p, *ma-b > PSH *mp-, *mb-.

This change could be motivated by the well-known avoidance of dissimilar labial onsets

in successive syllables, an avoidance that has given rise to such diverse phonological

phenomena as ablaut in western Borneo, and pseudo nasal substitution in Taiwan, the

Philippines, and various parts of western and central Indonesia (Chrétien 1965, Blust

1997:23-27, 2004:76-80). It appears likely that a similar process is responsible for many

of the prenasalized initials in other languages of the Lesser Sundas, and perhaps in

Sulawesi. Non-labial prenasalized initials would necessarily have another origin. Few of

these can be traced etymologically, but some of those that can, such as PMP *taqun >

Kambera ndauNu ‘year’, may derive from proto-forms that were preceded by the nominal

article *na (*na taqun > ndauNu). Such a development may have been favored by the

prior existence of mp- and mb-, but this cannot explain forms such as PMP *diaq >

Kambera ndia ‘no, not’. Finally, the plain voiced stops in Malay loanwords are often

borrowed as prenasalized stops both initially and medially, as in mbayaru (< Malay

bayar) ‘to pay’, karambua (< Malay kerbau) ‘carabao’, ndandu (< Malay dadu) ‘dice’,

bàndilu (< Malay bedil) ‘weapon’, njamu (< Malay jam) ‘hour’, banju (< Buginese

bajaw) ‘Bajaw’, or buNgihu (< Malay bugis) ‘Buginese’. Klamer (1994:13) makes this

adaptation absolute: “in loan substitutions, /b, d, g, dZ/ are always prenasalized.”

However, Onvlee (1984) gives ajaru (< Malay ajar) ‘instruction, learning’, bawa (<

Malay bawaN) ‘onion’, habu (< Malay sabun, from Portuguese sabão) ‘soap’, jala (<

Malay/Sanskrit jala) ‘casting net’, jàriku (< Malay jeruk) ‘citrus fruit’, and jawa (< Jawa)

‘foreign’ (from either Malay or Javanese), which contradict this statement.

Kambera is unusual among languages of insular Southeast Asia in having a length

contrast in the vowels which cannot be interpreted as contrastive stress. Examples

include li ‘go, walk’ vs. lí ‘sound, word, story’, liri ‘list, tilt sideways’ vs. líri ‘carry on

the hip’, himbaNu ‘dipnet’ vs. pímbuNu ‘winnow rice’, tu ‘so, be such’ vs. tú ‘lay

something down’, buba ‘sound of bubbling up’ vs. búbu ‘eye of a needle or fishhook’,

huluru ‘moving rapidly’ vs. húluru ‘sloping, as a path’, and àha ‘sound of clearing the

throat’ vs. aha ‘clean’. Long vowels generally occur in the first syllable regardless of

word length (lí, líri, húluru as noted above, húburuNu ‘pour out’, ndítikuNu ‘penetrate’,

wúnjuluNu ‘wind or twist around’), although exceptions are discussed below.

A close comparison of Onvlee (1984) with Klamer (1994) reveals several discrepancies

which require a brief discussion. Some of these are minor, but others are not, and must

be addressed before it is possible to consider the historical phonology:

(1) Klamer (1994:19) states that “(C)VV syllables can occur only under main stress,

where they contrast with (C)V syllables. In other than main stress positions only (C)V

occurs.” By “(C)VV syllables” she means syllables in which the nucleus is either a long

vowel (rú|bu ‘deep’) or a diphthong (pai|ta ‘bitter’), where the bolded syllable is

stressed. In other words, it is claimed that a stressed syllable can be (C)V or (C)VV, but

a heavy syllable cannot appear in unstressed position. This statement seems to rule out

7

forms like larí ‘young (of female animals)’, lawú ‘fontanel’, maNú ‘grope, feel for s.t.

with the hand’, paní ‘flying fox, fruit bat’, rupí < Malay rupiah ‘basic unit of currency’,

and ulá ‘stroke over’ as possible roots, since elsewhere (1994:25) Klamer states

unequivocally that “Main stress is without exception on the initial syllable of the root”

(italics original).

(2) Klamer (1994:29) states provisionally that “a vowel may not be followed by a

vowel/consonant of equal height,” hence ruling out forms such as **[ri:ja], **[ki:wa],

**[tu:wa], **[tu:ja], **[ri:i], **[ki:u], **[tu:i], **[tu:u] or **[ra:a]. She then notes two

counterexamples, ha.ri.u ‘a thousand’, and ha.ru.i ‘have trouble’, and so qualifies this

statement as “a long vowel cannot be followed by a vowel/consonant of equal height.”

However, in view of forms such as túya ‘mother’s brother’ (phonemically identical to the

hypothetical **[tu:ja] which she gives!), wúya ‘crocodile’, yíyipu ‘by bits, a little at a

time’, or yúyuNu ‘jerk, shudder’ this statement must be abandoned.

(3) Klamer (1994:20) states that “Trisyllabic roots do not occur.” Although forms such

as buNgahu ‘open s.t.’ are trisyllabic on the surface, the last vowel is a ‘paragogic vowel’

that is not part of the underlying form; its synchronic status is reportedly confirmed by

the fact that it is introduced into many loanwords, as ajaru ‘study’ (< Malay ajar), huratu

‘letter’ (< Malay surat), or timbaNu ‘scale’ (< Malay timbaN). Klamer adds that forms

such as paita ‘bitter’ are also not trisyllabic, since –ai- or –au- following an initial

consonant function like a syllable nucleus (and are considered the long equivalents of the

mid vowels e and o). In effect, she states that CVCVCV roots that do not end in –Cu and

do not contain an –ai- or –au- in the first nucleus position --- hypothetical forms such as

**rimuna, **obali or *puita --- do not exist. This is a decidedly odd claim, since many

forms in Onvlee (1984) contradict it, and any attempt to describe the historical phonology

of Kambera must contest it. Among other forms that exemplify canonical types that

Klamer rules out for Kambera are 1. halapa ‘sandal woven of lontar leaves’, 2. kalaü

‘mouse’, 3. kapita ‘k.o. large marine fish: Balistes mitis’, 4. kapíta ‘sling’, 5. kapuka

‘extremity, top’, 6. kareni ‘near’, 7. katana ‘to ask’, 8. kawana ‘right side’, 9. laNira ‘k.o.

tree along riverbanks’, 10. laNoda ‘exhausted’, 11. laNoda ‘k.o. salmon’, 12. laNora

‘wipe off’, 13. lawina ‘k.o. bean: Cajanus cajan’, 14. layia ‘ginger’, 15. mahàna ‘still,

quiet’, 16. mahapi ‘depressed, downcast’, 17. mahawa ‘at intervals’, 18. mandauta

‘afraid’, 19. maneni ‘k.o. vegetable’, 20. manera ‘tree with strong wood’, 21. mania ‘too

large, misshapen’, 22. manila ‘k.o. peanut’, 23. manua ‘wounded’, 24. palili ‘taboo’, 25.

pamula ‘to plant’, 26. parua ‘molar tooth’, 27. patola ‘climber with edible fruit: Luffa

cylindrica’, 28. tabibi ‘on the side’, 29. taleli ‘flute’, 30. taliNa ‘ear; hear’, 31. tanai

‘intestines’, and 32. taNara ‘look upward’. Some of these, as kawana, mandauta and

pamula, reflect affixed forms (*ka-wanan, *ma-takut, *pa-mula), but there is no evidence

in Onvlee (1984) that they still contain a morpheme boundary, and others, as katana (<

*kutana), layia (< *laqia), taleli (< *tulali), tanai (< *tinaqi), and taNara (< *tiNadaq)

were PMP trisyllables that retain their canonical shape despite sound changes that have

altered individual consonants or vowels.

(4) One last point seems worth bringing up, at least briefly. After stating that trisyllabic

roots do not occur, Klamer (1994:22) holds that “The only unambiguously trisyllabic (i.e.

8

(C)V CV CV) roots are roots with a final paragogic vowel /u/.” She places this vowel

within parentheses, and states that “the consonants /l, r, h, t, k, N/ are the only ones to

occur in the root-final position, preceding the /u/ in brackets.”

Since Onvlee (1984) lists words such as alahu ‘forest’ (< *halas), búburu ‘porridge’ (<

*buRbuR) or wulaNu ‘moon’ (< *bulan) with the same invariant –u that is found in e.g.

hamaNu ‘life-force, soul’ (< *sumaNed), kataru ‘caterpillar, worm’ (< *katadu ‘large

green hairless stinging caterpillar’, or manahu ‘cook’ (< *ma-nasu), it is difficult to see

what synchronic evidence there is for treating –u in the first three forms as a paragogic

vowel without also treating the last three forms in the same way. To do this, however,

would create unnecessary complications in the historical derivation of these forms, since

surface trisyllables like alahu, búburu or wulaNu added a paragogic vowel historically,

but surface trisyllables like hamaNu, kataru, or manahu did not. No evidence is given to

support the claim that the ‘paragogic u’ is non-phonemic, apart from its insertion in

loanwords. However, since Kambera lacks final consonants and the most common final

vowel is –u it can be expected that the –C of loanwords will be altered to –Cu to conform

to existing phonotactic restrictions. Onvlee (1984) sometimes gives variant forms with or

without the final syllable, but here it is the entire syllable that varies, not just the vowel:

bula ~ bulaku ‘open the eyes’ (cross-referenced), kubu(ku) ‘sound of plopping into water’

(single entry), pànda ~ pàndaNu ‘pandanus’ (cross-referenced), etc.

2.2. Historical phonology. Kambera historical phonology is sufficiently complex to

initially mask the cognation of such Malay loanwords as halamatu (< selamat) ‘well-

being’ and buNgihu (< bugis) ‘Buginese’, or native terms like líru < *layaR ‘sail’ and

riNu < *duyuN ‘dugong’, but in general this does not present serious problems. As will

be seen below, most final consonants are preserved by the addition of a supporting vowel

–u, but only after some loss of contrast had begun to occur in this position. Syllable-final

consonants disappeared word-medially unless the first was a nasal and the second a stop:

*buRbuR > búburu ‘rice porridge’, *bejbej > búburu ‘bundle’, *beRsay > búhi ‘paddle a

canoe’, *suksuk > huhuku ‘penetrate’, *putput > puputu ‘puff, blow on’, *tuktuk > tútuku

‘knock, pound, beat’, but *beNkuq > bàNgu, wàNgu ‘bent, curved’, *diNdiN > dindiNu

‘wall’, *kempuN > kambu ‘belly’, *naNka > naNga ‘jackfruit’, *pandak > pandaku

‘short’, *paNdan > pàndaNu ‘pandanus’, *sendiR > hàndi ‘lean against’, *tumbuq >

tumbu ‘grow’. While words such as dindiNu or tumbu reflect PMP forms with a syllable

boundary between the nasal and the stop, then, in Kambera it has shifted leftward

(di.ndi.Nu, tu.mbu), since this language permits only open syllables (Klamer 1994:19).

A number of voiced obstruents show non-etymological prenasalization. In some cases

this can be explained as a product of prefixation with *ma- ‘stative’ followed by syncope

of the prefixal vowel, as in *ma-baseq > mbaha ‘wet’, *ma-baRiw > mbai ‘tainted,

spoiled’, *ma-belaN > mbàlaNu ‘spotted, striped’, *ma-baqeRu > mbaru ‘new’, *ma-

panas > mbanahu ‘warm, hot’, *ma-baRani > mbeni ‘brave’, or *ma-penuq > mbínu ‘full,

of a container’. The stative prefix also appears to be fossilized in *ma-takut > mandauta

‘afraid’, but without syncope. As noted earlier, syncope leading to prenasalization is

generally confined to bases with an initial labial stop, and may have been motivated by an

inherited constraint against dissimilar labials separated by a single vowel (whereas

9

homorganic prenasalization is widespread). The history of prenasalized obstruents in a

few other examples is far less clear, as with *salaq > njala ‘wrong, in error’, *saNa >

njaNa ‘branch’, or *taqun > ndauNu ‘year’. It is possible that njala and njaNa reflect

forms with verb-forming nasal accretion (*n-salaq, *n-saNa), and that ndauNu reflects a

form that was preceded by an article *na (*na taqun), but this is speculative. In still other

examples there is evidence for historical prefixation with *maN-, as with *kali > maNeli

‘dig up’ ( < *maN- + *kali), but in some forms the occurrence of such an apparent prefix

is completely unexpected, as with *anay > maNaní ‘termite’ (< *maN- + *anay?).

*p. When not prenasalized *p became p: *panaw > panu ‘leucoderma’, *paNdan (>

pandan > p↔ndan) > pàndaNu ‘pandanus’, *qapeju (> p↔du) > pídu ‘gall’, *pija > pira

‘how much/how many?’, *pusej > puhu ‘navel’; *hapuy > epi ‘fire’, *epat > patu ‘four’,

*sepaq > hàpa ‘chew betel’, *ma-nipis > manipihu ‘thin, tenuous’, *puput > puputu

‘blow on (a fire, etc.)’; *ma-qudip (> mudip) > miripu ‘living, alive’. When it followed a

PMP nasal, or came to follow a nasal as a result of syncope, *p merged with b: *kempuN

> kambu ‘belly’, *empu > umbu ‘grandparent’ (but àpu ‘grandmother’), *ma-panas (>

mpanas) > mbanahu ‘hot’, *ma-penuq (> mpenuq) > mbínu ‘full, of a container’, *sa-Na-

puluq (> sa-mpuluq) > mbulu ‘unit of ten’.

*t. If not prenasalized reflexes of *t are almost always t: *taliNa > taliNa ‘ear’, *telu >

tílu ‘three’, *timuR > timiru ‘east, east wind’, *tuna > tuna ‘eel’; *datu > ratu ‘clan

priest’, *qatay > eti ‘liver’, *betas > botahu ‘break off’, *kuRita > (ka)wita ‘octopus’,

*qutan > utaNu ‘forest’; *epat > patu ‘four’, *kawit > kaitu/ketu ‘hook’, *kulit > kalitu

‘skin’, *paqit > paita ‘bitter’. When it originally followed a nasal or came to follow a

nasal as a result of syncope, *t merged with *d: *muntay > mundi ‘citrus fruit’, *na taqun

> ndauNu ‘year’, *-nta > -nda ‘1px’, *pintu > pindu ‘door’. Some reflexes show

sporadic medial prenasalization, as found in the history of many AN languages (Blust

1996b:139-145): *bitik ‘snare trap’ > bindiku ‘spring up’, *ma-takut > mandauta ‘afraid’.

Three forms unaccountably reflect *t as h: *bities (> bitis) > wihi ‘foot’, *ma-tuduR >

mahuru (cf. Kodi manduru) ‘sleep’, *pitu > pihu ‘seven’. A consideration of more

locally restricted cognate sets reveals at least one further example in ka-hilu ‘ear’ (cp.

Manggarai, Palu’e tilu, Kodi ka-tilu, Tetun tilu-n ‘ear’). The change *t > h presumably

went through s, but no clear condition for this development has been found. In addition

*t has disappeared without stateable conditions in one form: *Ramut > amu ‘root’.

*c. Reflexes of PMP *c are difficult to find in many languages. The only candidate

located to date for Kambera is kapihaku ‘mud, muddy’ (with monosyllabic root *-cak

‘muddy; sound of walking in mud’). It is perhaps also worth noting in addition that the

voiceless palatal affricate of Malay loanwords is replaced by h, as in hapuru ‘to mix’

(Malay campur), hina ‘Chinese’ (Malay cina), or mahaNu ‘tiger’ (Malay macan).

*k. PMP *k split into k and zero without stateable conditions: *kali-maNu > kalimaNu

‘mangrove crab’, *kepit > kàpitu ‘fasten’, *kima > kíma ‘giant clam’, *kumis > kumihu

‘moustache’; *lakaw ‘walk, go’ > laku ‘bring, escort’, *dekit > ka-dàkitu ‘adhere to’,

*ikuR > k-iku ‘tail’, *bukas > bukahu ‘to open’; *pandak > pandaku ‘short in height’,

10

*tasik > tehiku ‘sea, saltwater’, *ma-buhek > maüku ‘drunk’, but *kahiw (> kaiw > kayu)

> ai ‘wood’, *kita > ita ‘see’, *kutu (> utu) > wutu ‘louse’; laki > lei ‘man, male’, *paku

> paü ‘fern’, *ma-takut > mandauta ‘afraid’, *siku > hiu ‘elbow’, *hikan > iaNu ‘fish’,

*paniki > paní ‘flying fox’, *manuka > manua ‘wounded’, *kuhkuh (> kuku) > wú

‘fingernail, toenail, claw’; *anak > ana ‘child, *balik > weli ‘return, go back’, *manuk >

manu ‘chicken’. Several forms with *-k- > k, as kiki ‘file’ (Malay kikir) and paku ‘nail’

(Malay paku) may be loans from Malay. Following a nasal *k became g: *beNkul >

bàNgu ‘lower the head’, *buNkas > buNgahu ‘to open’, *naNka > naNga ‘jackfruit’.

In two known cases *k is reflected as N, presumably through the fossilization of an

earlier process of verbalizing nasal substitution: *kali > maNeli ‘dig, dig up’ (< *maN- +

*kali), *kaen > NaNu (*maN- + *kaen > maNaen > maNan > maNaN > maNaNu >

NaNu).

*q. PMP *q disappeared in all 41 known etymologies in which it is found: *qaRta ‘alien

people’ > ata ‘slave’, *qauR > au ‘bamboo’, *qatay > eti ‘liver’, *qiris > irihu ‘cut into

strips’, *qihu > iu ‘shark’, *quban > uwaNu ‘gray hair’; *ma-qitem > mitiNu ‘black’,

*taqun > ndauNu ‘year, season’, *paqit > paita ‘bitter’, *taqi > tai ‘feces’; *Rumaq >

uma ‘house’, *baseq > baha ‘wash’, *piliq > pili ‘choose’, *pahuq > pau ‘mango’.

*b. PMP *b split into b or w without stateable conditions: *bayu > bai ‘pound rice’,

*bales > balahu ‘return in kind’, *beNkuq > bàNgu ‘lower the head’, *bikaq > bika

‘split’, *buNkuk > buNgu ‘hunchbacked’; *babaq > baba ‘short’, *sebu > hàbu ‘douse a

fire’, *tebuh > tíbu ‘sugarcane’, but *batu > watu ‘stone’, *beli > wíli ‘buy; cost, price’,

*b-in-ahi (> binay) > wini ‘female’, *bulan > wulaNu ‘moon’; *abu > aü ([awu]) ‘hearth;

ash’, *kulabaw > kalaü, ‘mouse, rat’, *ma-buhek (> mabuk) > maüku ‘drunk’, and

*quban > uwaNu ‘gray hair’, *babuy > wei (*-b- > 0) ‘pig’. In a few cases b/w doublets

are found, as with *bekaq > bàka ‘to split’, but wàka ‘split open (state)’, *baRu > ka-

baru ‘Hibiscus tiliaceous’, but waru ‘k.o. hibiscus’, or *babaq > ka-baba ‘short, low’, but

wawa ‘below, west’. Word-finally the only known example shows loss: *ma-huab (>

mawab) > mawa ‘yawn’. When prenasalized *b is reflected only as a stop: *tumbuq >

tumbu ‘grow’, *tumbuk > tumbuku ‘hit, pound’, *ma-belaN > mbàlaNu ‘striped, spotted’,

*ma-baseq > mbaha ‘wet’, *ma-baRiw > mbai ‘rotten’, *ma-baRani > mbeni ‘brave’.

*d. Word-initially *d is reflected as d or r without stateable conditions: *dalem > dalu

‘in, inside’, *daya > dia ‘upstream, toward the interior’, *dindiN > dindiNu ‘wall’, *di

atas ‘above, on top’ > dítahu ‘go up’, but *dahun > rau/ru ‘leaf’, *daRaq (> daya) > ria

‘blood’, *deNeR > reNu, roNu ‘hear’, *ditaq > rita ‘a tree: Alstonia scholaris’, *duRi

‘thorn’ > rí ‘bone’. In intervocalic and final positions it is reflected as r: *katadu ‘large

green stinging caterpillar’ > kataru ‘caterpillar, worm, snake’, *tiNadaq > taNara ‘look

upward’, *Nuda > Nura ‘young’, *ka-mudehi (> kamudi) > kamuri ‘rear part, rudder’;

*kidkid ‘rasp’ > kikiru ‘shave, scrape off’ (also kiki ‘rasp; to rasp’), *lahud > lauru, luru

‘downriver, toward the sea’, *natad > nataru ‘cleared area around house’, *batad ‘millet

sp.’ > wataru ‘maize’. When prenasalized *d did not change: *pandak > pandaku ‘short

in height’, *paNdan (> pandan > p↔ndan) > pàndaNu ‘pandanus’, *sendiR > hàndi ‘prop

up, support’. In Malay loanwords d is generally borrowed as nd: bàndilu ‘weapon’ (<

11

Malay/Tamil bedil), hurundandu ‘soldier’ (< Malay/Portuguese serdadu), ndahi ‘scale,

balance’ (< Malay/Minnan daciN), ndewa ‘godlike power’ (< Malay/Sanskrit déwa).

One known etymology shows *d > d ~ r: *duha > dua ~ rua ‘two’.

*z. Only four likely reflexes of *z have been found; in three of these *z became r and in

the other it remained unchanged: *qazani > k-areni ‘near’, *zauq > rau ‘far’, *quzan >

uraNu ‘rain’, but *zaraN > jaraNu ‘widely spaced’. In Malay loanwords j is borrowed

both as j and as nj: ajaru ‘study’ (< Malay ajar), iju ‘yellow’ (< Malay hijaw ‘green’),

jala ‘casting net’ (< Malay/Sanskrit jala), but banju ‘Bajaw’ (< Malay/Buginese bajaw),

njamu ‘hour’ (< Malay/Persian jam). This suggests that jaraNu may be a Malay loan.

*j. Following penultimate schwa *j became d: *qalejaw (> lejaw) > lodu ‘day, sun’,

*qapeju > k-apídu, k-apadu ‘gall’. Following other vowels *j became r: *huaji (> uaji >

waji) > eri ‘younger parallel sibling’, *Najan > Nara ‘name’, *pajay > pari ‘rice’, *pija >

pira ‘how much/how many?’, *p-ijan > piraNu ‘when?’. Only one etymology with final

*j has been found, and here it disappeared: *pusej > puhu ‘navel’.

*g. There are no known reflexes of PMP *g in Kambera. However, the absence of a

non-prenasalized g in the phoneme inventory of the language can be taken as evidence

that *g changed to some other sound. In Malay loanwords g is borrowed as Ng: buNgihu

‘Buginese’ (< Malay bugis), NgataNu ‘unit of measure for volumes’ (< Malay gantaN),

Nguru ‘teacher’ (< Malay/Sanskrit guru), NgutiNu ‘scissors’ (< Malay guntiN).

*m. PMP *m remained unchanged initially and medially: *manuk > manu ‘chicken’,

*maja ‘evaporate’ > mara ‘dry’, *ma-qitem > mítiNu ‘black’, *um-utaq > muta ‘vomit’;

*tama > tama ‘enter’, *qala-metak > lamakatu (< met.) ‘leech’, *kima > kima ‘giant

clam, Tridacna sp.’, *lumut > lumutu ‘moss, algae’. Final *m became –N: *ma-qitem >

mítiNu ‘black’, *demdem ‘dark’ > ruduNu ‘dark, night’ (with irregular reduction of the

medial cluster), *tanem > taniNu ‘bury’, *inum (> unum) > unuNu ‘drink’. In one case *-

m is reflected as m, and in three others it disappeared: *enem > nomu ‘six’, *dalem >

dalu ‘in, inside, in the sea’, *padem (> p↔dam) > pàda ‘extinguish a fire’, mbàda

‘extinguished, of a fire’, *tiRem ‘oyster’ > tiu ‘k.o. mussel’.

*n. PMP *n generally did not change in initial and medial positions: *nabuq > ka-nabu

‘fall’, *nepuq > nípu ‘stonefish’, *nusa > nuha ‘island’; *panaq > pana ‘shoot (an

arrow)’, *tenun > tíniNu ‘weave’, *ina > ina ‘mother’, *tuna > tuna ‘freshwater eel’. In

final position *n merged with the other nasals as –N: *puqun ‘base of a tree’ (> pun) >

puNu ‘stem, trunk’, *quzan > uraNu ‘rain’, *kuden > wuruNu ‘cooking pot’, *qulin >

uliNu ‘rudder’. In at least three cases *-n disappeared, and in two others there are

doublets with final vowel or -N: *ka-wanan > kawana ‘right side’, *ma-qasin ‘salty’ (>

masin) > mehi ‘salt’, *Najan > Nara ‘name’, *dahun ‘leaf’ > rau ‘leaf, hair’ (also rauNu

‘perfumes made of redolent leaves and grass’?), *paNdan > pànda, pàndaNu ‘pandanus’.

12

*ñ. Only two reflexes of the PMP palatal nasal have been identified in Kambera. One

became y and the other n: *ñiluq ‘pain in teeth from eating something very sour’ > yilu

‘sour’; *miñak > mina ‘fat, grease, oil’.

*N. PMP *N almost always remained a velar nasal: *NaNa > ha-NaNa ‘gape, open the

mouth’, *naNuy > Neni (< met.) ‘swim’; *kali-maNu > kalimáNu ‘mangrove crab’,

*leNa > làNa ‘sesame’, *taliNa > taliNa ‘ear’, *tiNadaq > taNara ‘look upward’,

*deNeR > reNu

‘hear’; *liaN > liaNu ‘cave’, *qajeN > aruNu ‘charcoal’, *teriN > tàriNu ‘bamboo sp.’,

*duyuN > ríNu ‘dugong’. In three known cases final *N disappeared: *kempuN >

kàmbu ‘belly; womb’, *taRutuN (> tautuN) > tetu ‘porcupine fish’, *ujuN > uru ‘nose’.

*s. PMP *s is usually reflected as h: *salaq > hala ‘wrong, in error’, *sepaq > hàpa

‘chew betel’, *sakay > hei ‘climb’, *silak > hilaku ‘glow, flame’, *sisuq > híhu ‘k.o.

freshwater snail’, *susu > huhu ‘breast’; *asu > ahu ‘dog’, *seksek (> s↔s↔k) > hàhaku

‘stuff or cram in’, *isa > iha ‘one’, *nusa > nuha ‘island’; *halas > alahu ‘forest, bush’,

*bales > balahu ‘retaliation, compensation’, *kumis > kumihu ‘moustache’. Following a

nasal *s became j: *salaq > hala ‘wrong, in error’, but also njala ‘fault, mistake, in error’,

*saNa ‘bifurcation’ > haNa ‘space between spread legs, groin’, but also njaNa (in some

dialects) ‘branch’, *siku > hiu ‘elbow’, but also njiu ‘hit with the elbow’, *sebu ‘contact

of fire and water’ > hàbu ‘plunge in, immerse’, but also njàbu ‘splash, sound of

something coming into contact with water’.

In two known examples *-s disappeared: *badas ‘gravel’ > wara ‘sand, gravel’, *bities

‘calf of the leg’ (> bitis) > wihi ‘leg, foot’ (with h from *t).

*h. PMP *h invariably disappeared in Kambera: *halas > alahu ‘forest, bush’, *hapuy >

epi ‘fire’, *hipun > ipiNu ‘small fish that swims in seasonal swarms’, *hituq > itu ‘catfish

with poisonous spines’; *kahiw (> kayu) > ai ‘wood, tree’, *pahuq > pau ‘mango’, *qihu

> iu ‘shark’, *duha > dua ‘two’; *tumah > ka-tima ‘clothes louse’, *paRih > pai ~ pari

‘stingray’, *kuhkuh > wú ‘nail, claw’.

*l. PMP *l regularly became Kambera l: *labuq > labu ‘anchor’, *lasuq ‘scrotum’ >

lahu ‘penis’, *layaR > líru ‘sail’, *leNa > làNa ‘sesame’, *liuq ‘circumvent, go around’ >

liu ‘go outside’, *lumut > lumutu ‘moss, algae’; *walu > walu ‘eight’, *balik > weli

‘return, go back’, *telu > tílu ‘three’, *piliq > pili ‘choose’, *hulaR > ularu ‘snake’.

Final *l is attested in only one word, where it disappeared: *dumpel > ka-dumba ‘dull,

blunt’.

*r. PMP *r is found in four known forms, where it became r: *zaraN > jaraNu ‘widely

spaced’, *rusrus ‘slip or slide off’ > rúruhu ‘pull, drag’, *teriN > tàriNu ‘k.o. large

bamboo’, *terus > turuhu ‘directly through’.

*R. PMP *R generally is lost, but is reflected as r in ten forms and r ~ 0 in one other:

*Ramut > amu ‘root’, *Rumaq > uma ‘house’; *kaRat-i > kati ‘bite’, *baRiw > mbai

‘rotten, spoiled’, *baRani > mbeni ‘brave’, *daRaq > ria ‘blood’, *biRaq > wí ‘itching

13

taro: Alocasia macrorhiza’, *kuRita > wita ‘octopus’; *deNeR > reNu/roNu ‘hear’,

*bibiR > wiwi ‘lip’, *wahiR > wai ‘water’, *qauR > au ‘thick bamboo: Dendrocalamus

sp.’, *kuluR > kulu ‘breadfruit’, *muRmuR > mumu ‘rinse the mouth’, but *Rakit >

rakitu ‘tie together; raft’, *baRu > ka-baru ~ waru ‘hibiscus’, *baqeRu > mbaru ‘new’,

*habaRat > waratu ‘west, west wind’, *layaR > líru ‘sail’, *hulaR > ularu ‘snake’, *uRat

> ura/uratu ‘lines in the palm; tell fortunes’, *buRbuR > búburu ‘porridge’, *saliR >

haliru ‘flow’, *timuR > timiru ‘east, east wind’, *paRih > pai/pari ‘stingray’. The

change *daRaq > ria ‘blood’, parallel to *daya > dia ‘upstream, toward the interior’,

suggests that *R became y in at least some forms before disappearing.

*w. PMP *w is reflected as w in seven of the eight etymologies that have been found:

*wahiR > wai ‘water’, *walu > walu ‘eight’, *qawa > awa ‘milkfish: Chanos chanos’,

*awaN > awaNu ‘atmosphere, space between earth and sky’, *lawaq ‘spider’ > lawa-Nu

‘web’, lawa NgíNgi ‘spider web; spin a web’, *ma-huab (> mawab) > mawa ‘yawn’,

*siwa > hiwa ‘nine’. In the eighth etymology it has disappeared: *kawit > kaitu/ketu

‘hook’. The sequence *-aw is treated separately below.

*y. PMP *y occurred only medially and finally. Medial *y has no consistent outcome in

Kambera. In two cases it remained a palatal glide: *aya ‘father’s sister’ > aya ‘elder

sibling’, *buqaya > wuya ‘crocodile’. In two other forms *-ay- or *-uy- contracted to i:

*daya > dia ‘upstream, toward the interior’, *duyuN > riNu ‘dugong’. In one other form

the sequence *-ay- contracted to a long vowel i: *layaR > líru ‘sail’. Finally, in *bayu >

bai ‘to pound, as rice’, *-yu contracted to i. A similar change probably should be

recognized in *kahiw (> kaiw > kayu) > ai ‘wood, tree’.

*-ay. In many AN languages *-ay, *-aw, *-uy and *-iw have been monophthongized.

This usually affects only tautosyllabic –VC sequences, but monophthongization may also

affect heterosyllabic sequences of vowel + glide. As already seen, some examples of the

latter change are found in Kambera. The diphthong *-ay invariably became i: *qatay (>

ati) > eti ‘liver’, *sakay (> hai) > hei ‘climb, ascend’, *quay > iwi ‘rattan’, *m-atay (>

mati) > meti ‘die, dead’, *pajay > pari ‘rice’, *qumay > umi ‘unicorn fish’, *b-in-ahi (>

binay > bini) > wini ‘female, feminine’.

*-aw. PMP *-aw became u in all known etymologies: *kaRaw > kau ‘scratch an itch’,

*lakaw > laku ‘go, walk’, *qalejaw > lodu ‘day, sun’, *panaw > panu ‘leucoderma’.

*-uy. PMP *-uy became i in the three known relevant cases: *hapuy (> api) > epi ‘fire’,

*naNuy (naNi > Nani) > Neni ‘swim’, *babuy (> babi > wawi) > wei ‘pig’.

*-iw. PMP *-iw became i in *laRiw > lai ‘run’ and *baRiw > mbai ‘rotten’. In *kahiw >

ai ‘wood, tree’ it is most likely that this form passed through an intermediate stage in

which it had the shape *kayu, and that the –i is thus a reflex of –yu- rather than –iw.

Vowel reflexes will be treated somewhat differently from consonant reflexes. First, some

discussion is needed regarding the origin of vocalic length contrasts. Second, all vocalic

oppositions were neutralized in prepenultimate syllables, and it therefore makes sense to

14

treat these together rather than under the separate headings of ‘reflexes of *a, reflexes of

*e’, etc. Third, since vowels have far higher frequencies than consonants only a small

fraction of relevant cases will be given to illustrate regular developments.

Origins of vowel length. Where etymologies are available, nearly all long vowels in

Kambera appear to have one of three sources: 1) a low vowel + adjacent high vowel or

glide contracted to a single long high vowel, 2) the vowel preceding a medial cluster was

lengthened if the cluster was reduced, 3) PMP *e (schwa) shows unconditioned phonemic

split, one reflex being í. The first type of change is seen in *di-atas > dítahu ‘upward’,

*layaR > líru ‘sail’, *ma-iRaq > mí ‘red’, *ma-qitem (> ma-item) > mítiNu ‘black’,

*biRaq (> biaq) > wí ‘itching taro’, *buqaya > wúya ‘crocodile’, and in Malay loanwords

such as rupí < rupiah ‘basic unit of currency’. However, other forms did not develop í or

ú from earlier vowel sequences, as seen in *kahiw (> kayu) > ai ‘wood, tree’, *maRi >

mai ‘come’, *daya > dia ‘upriver’, *laqia > layia ‘ginger’, *qauR > au ‘large bamboo

sp.’, *bahuq > wau ‘stench’, *duha > dua ‘two’, or *buaq > wua ‘fruit’. The second type

of change is seen in *bejbej ‘wrap by tying around’ > búburu ‘bundle’, *buRbuR >

búburu ‘porridge’, *tuktuk > tútuku ‘knock, pound, beat’, and possibly *rusrus ‘slip or

slide off’ > rúruhu ‘pull, drag’. The third source is best discussed in connection with

reflexes of PMP *e. In one known case a long vowel has arisen in the final syllable as a

result of the loss of a consonant between like vowels: *paniki > paní ‘flying fox, fruit

bat’. In another it has arisen in a monosyllable for the same reason: *kuku > wú ‘nail,

claw’. Other examples are unexplained, as with *sisuq > híhu ‘freshwater snail’.

V3. Many AN languages in western Indonesia have merged at least PMP *a and *e ([↔]),

and more often all vowels as schwa in prepenultimate syllables. Kambera shows a

similar merger of antepenultimate vowels, but as a rather than schwa, as in *sumaNed >

hamaNu ‘life-force, soul’, *ka-mudehi > kamuri ‘last part; rudder’, *hazani > k-areni

‘near’, *kutana > katana ‘ask’, *katadu > kataru ‘k.o. caterpillar’, *ka-wanan > kawana

‘right side’, *qali-matek (> limatek) > lamakatu ‘leech’ (with *t/k metathesis), *laqia >

layia ‘ginger’, *ma-tuduR > mahuru ‘sleep’, *ma-takut > mandauta ‘afraid’, *ma-tuqah

> matua ‘mature, old’, *ma-buhek > maüku ‘drunk, intoxicated’, *tulali > taleli ‘flute’,

*taliNa ‘ear’ > taliNa ‘ear; hear’, *t-in-ahi > tanai ‘intestines’, or *tiNadaq > taNara

‘look upward’. In *kali-maNaw > kalimaNu ‘mangrove crab’ the second vowel of the

prefix unaccountably remained a high front vowel, and in *kulit > kalitu ‘skin’ a merger

of *u with *a evidently occurred after the addition of the paragogic –u.

Prepenultimate vowels that were originally initial or that came to be initial through loss

of *q or *h dropped: *qali-matek (> limatek) > lamakatu ‘leech’, *qalejaw > lodu ‘day’,

*qunap-i > nepi ‘fish scale’, *qateluR > tílu ‘egg’, *habaRat > waratu ‘west, west wind’.

*a. PMP *a generally remained unchanged. However, in the environment __Ci it often

shows partial assimilation to the following syllable peak: *hapuy (> api) > epi ‘fire’,

*huaji (> *wari) > eri ‘younger parallel sibling’, *qatay (> ati) > eti ‘liver’, *sakay (>

haki) > hei ‘climb’, *hazani > k-areni ‘near’, *labiq > lebi ‘excess’, *ma-baRani > mbeni

‘brave’, *ma-qasin ‘salty’ (> masin) > mehi ‘salt’, *m-atay (> mati) > meti ‘die; dead’,

*qunap-i (> napi) > nepi ‘fish scale’, *naNuy (> naNi) > Neni ‘swim’, *tulali > taleli

15

‘flute’, *tasik > tehiku ‘sea’, *tamiN > temiNu ‘shield’, *babuy (> wawi) > wei ‘pig’,

*balik > weli ‘return, go back’ . Although this is a recurrent change there are a number of

exceptions to it in forms that appear to be directly inherited: *kahiw > ai ‘wood, tree’,

*bahi > bai ‘female (animals)’, *laRiw ‘flee, run away’ > lai ‘run’, *laqia > layia

‘ginger’, *maRi > mai ‘come’, *paRih > pai/pari ‘stingray’, *paqit > paita ‘bitter’,

*palihi > palili ‘taboo’ (?; expected **palí), *paniki > paní ‘flying fox’, *pajay > pari

‘rice’, *taqi > tai ‘feces’, *taliNa > taliNa ‘ear’, *t-in-ahi (> tinay) > tanai ‘intestines’,

*tahep-i > tàpi ‘winnow’, *wahiR > wai ‘water’. Low vowel assimilation in Kambera is

both irregular and asymmetric, since rounding assimilation did not take place when the

next syllable contained *u: *asu > ahu ‘dog’, *qalehu > alu ‘pestle’, *Ramut > amu

‘root’, *qauR > au ‘bamboo sp.’, etc.

*e > zero. Word-initially and in vowel sequences PMP *e disappeared: *enem > nomu

‘six’, *epat > patu ‘four’; *bities ‘calf of leg’ > wihi ‘leg, foot’, *binehiq (> bineiq) >

wini ‘seed’. In non-initial position the reflexes of PMP schwa are puzzling, as they show

what appear to be multiple phonemic splits.

*e > à. In a number of cases *e in a stressed (penultimate) syllable has become a short

low vowel: *empu > àpu ‘grandmother’, *bekaq > bàka ‘split’, *bekul > bàNgu ‘bend

the head low’, *sendiR > hàndi ‘support, prop up’, *sepaq > hàpa ‘chew betel’, *dekit

‘joined along the length’ > ka-dàkitu ‘attach to, stick to’, *kepit ‘pinch together’ > kàpitu

‘pinch between, insert’, *leNa > làNa ‘sesame’, *belaN > mbàlaNu ‘spotted, striped’,

*paNdan (> pandan > pendan) > pàndaNu ‘pandanus’ *tekik > tàki ‘k.o. gecko’, *teriN

> tàriNu ‘k.o. slender bamboo’, *bekas > wàkahu ‘loosen, untie’, *beluk > wàluku ‘to

bend, as rattan’. A similar development occurs in Malay loanwords, as with àmahu

‘gold’ (< Malay emas), bàhi ‘iron’ (< Malay besi), bànaNu ‘thread’ (< Malay *benaN),

or jàriku ‘citrus fruit’ (< Malay jeruk). Since PMP schwa was extra-short and is often

implicated in consonant gemination in daughter languages this reflex is in keeping with

expectation.

*e > í. In a number of other cases the schwa has become a long high front vowel, directly

contrary to what might be expected based on the clear evidence of its original shortness:

*beNkuq > bíNgu ‘bend, bent’, *qapeju > kapídu ‘gall’, *ma-penuq > mbínu/pínu ‘full,

of a container’, *nepuq ‘stonefish’ > nípu ‘k.o. sea fish’, *telu > tílu ‘three’, *qateluR >

tílu ‘egg’, *tenun > tínuNu ‘weave’, *beli > wíli ‘cost, price’. Most cases of *e > í occur

before alveolar consonants, and it is possible that phonological conditioning lies behind

this split, but some examples of *e > à also occur before alveolars, and a statement of

predictable conditions remains elusive.

Other outcomes. In addition to à and í, PMP *e has several other outcomes in Kambera.

In original final syllables it appears as a short i in *tanem > taniNu ‘bury’, and in stressed

syllables à varies with short i in bàka/bika ‘split’, and with long í in bàNgu/bíNgu ‘bend’,

although in these cases it is possible that the words contain a common monosyllabic root

(*-kaq ‘split’, *-kul ‘bend’), but are otherwise unrelated.

16

In still other cases *e is reflected as a, o or u: *baseq (> *basaq) > aha ‘wash’, *bales >

balahu ‘return in kind (revenge, compensation)’, *kempuN > kambu ‘belly’, *qali-matek

(> limatek) > lamakatu ‘leech’, *taneq > tana ‘earth’, *bates > watahu ‘boundary’;

*betas > botahu ‘break off’, *enem > nomu ‘six’, *qalejaw > lodu ‘day, sun’; *qajeN >

aruNu ‘charcoal’, *dalem > dalu ‘in, inside’, *sumaNed > hamaNu ‘life-force, soul’,

*pusej > puhu ‘navel’, *tiRem ‘oyster’ > tiu ‘k.o. mussel’, *terus > turuhu ‘directly

through’, *añem > unaNu ‘plait, as mats’ (*a/e metathesis), *qunej > unu ‘pith, core’,

*qetah > uta ‘rice husk, chaff’, *kuden > wuruNu ‘cooking pot’. Since *-e > a/_q is a

conditioned change in many AN languages, one generalization emerges: PMP *e became

Kambera u whenever it preceded a lost final consonant.

*i. PMP *i is almost invariably reflected as i: *iluR > ilu ‘saliva’, *kawit > kaitu ‘hook’,

*pipi > pipi ‘cheek’. The one exception that has been noted is *ma-nipis > manipa ‘thin,

of materials’ (but also manipihu ‘fine, of woven material’).

*u. PMP *u usually is reflected as u. However, in a small number of forms it appears

instead as i: *quay (> uwi) > iwi ‘rattan’, *tumah > ka-tima ‘body louse’, *ma-qudip (>

mudip) > miripu ‘living, alive’, *timuR > timiru ‘east, east wind’, *buliR > wili ‘ear of

grain’, *buni > wini ‘ringworm’. In each of these forms *u > i has occurred adjacent to a

labial consonant (non-phonemic in *uwi). However, *u has remained back and rounded

next to labials in other forms, as with *Ramut > amu ‘root’, *empu > àpu ‘grandmother’,

*apuN > apuNu ‘to float’, *mula > pa-mula ‘to plant’, *buaq > wua ‘fruit’, or *bulan >

wulaNu ‘moon’. The change *u > i must, therefore, be considered unconditioned, but

apparently was favored by an adjacent labial consonant (perhaps together with *i in an

adjacent syllable), and hence involved a type of labial dissimilation.

Glide accretion. Three words show addition of a labiovelar glide to a following *u after

loss of initial *k: *kuhkuh (> kuku > uu) > wú ‘nail, claw’, *kuden (> uden) > wuruNu

‘cooking pot’, *kutu (> utu) > wutu ‘hair louse’. A similar change is seen in the apparent

loanword wutaNu ‘debt’ (< Malay utaN). Surprisingly, words that originally began with

*u-, or that began with *qu- or *hu- (where the consonant presumably was lost very

early), do not show this change: *qulam > ulaNu ‘side-dish eaten with rice’, *hulaR >

ularu ‘snake’, *qulin > uliNu ‘rudder of a boat’, *umpu > umbu ‘grandfather’, *qumay >

umi ‘unicorn fish’, *añem > unaNu ‘plait, as mats’ (with vowel metathesis), *qunej >

unu ‘pith, core of something’, *upas > upahu ‘poison, snake venom’, *uRat > ura/uratu

‘lines in the palm; tell fortunes’, *quzan > uraNu ‘rain’, *qutan > utaNu ‘woods, forest’,

*quban > uwaNu ‘gray, of hair’, and ulaNu ‘repeat’ (< Malay ulaN),

utuNu ‘profit, gain’ (apparently < Malay untuN). This is puzzling, since it suggests that

*ku- became wu-, an unexpected change, and one that is contradicted by *kuluR > kulu

‘breadfruit tree’, and *kumis > kumihu ‘moustache’. No evidence has been found for

palatal glide accretion before i- with the possible exception of yíla ‘pull, haul’, a form

that may be a borrowing of Malay héla ‘drag, pull’.

The paragogic vowel. Although little evidence exists for treating non-etymological –u

as rule governed in the synchronic grammar of Kambera, it was clearly added after final

consonants in the history of the language. There are three known cases in which –a was

17

added instead: *lahud ‘downstream, toward the sea’ > laura ‘sea’ (but also lauru

‘upstream, toward the interior’) *ma-takut > mandauta ‘afraid’, *paqit > paita ‘bitter’.

As shown below, this is a change in some of the languages of Sumba spoken to the west

of Kambera, and the forms in questions may be loans.

Ordering. There is not much of interest that can be said about the ordering of sound

changes in Kambera. Some inferences are extremely transparent, as in concluding that

the change *a > e/_Ci could only have taken place after monophthongization had created

a last-syllable i in forms such as *hapuy > epi ‘fire’ or *qatay > eti ‘liver’. Perhaps the

most significant inference about ordering in Kambera concerns the paragogic vowel.

Although a similar development appears throughout Sumba this evidently is an areal

feature, since Kambera paragogic –u corresponds to paragogic –o or –a, and Kambera

nasal + paragogic –u corresponds to zero in most other languages of central and west

Sumba: *epat > Kod pato, Lmb, Wey pata, Ank, Kmb patu ‘four’, *ma-nipis > Kod

manipiho, Lmb manipha, Kmb manipihi ‘thin, of materials’, *ma-panas > Kod mbanaho,

Kmb mbana(hu) ‘warm, hot’, *inum > Kod, Ank inu, Lmb, Wey enu, Kmb unuNu

‘drink’, *hikan > Kod igha, Wey ia, Ank yiaNu, Kmb iaNu ‘fish’, *quzan > Kod, Lmb,

Wey ura, Ank uraN, Kmb uraNu ‘rain’. In addition, some lenition of final consonants

had already begun in Kambera before the paragogic vowel was added, as seen in the

changes *-m > -Nu and *-n > -Nu documented above. Surprisingly, word-final nasal

lenition is not matched by a corresponding lenition of word-final stops through merger as

glottal stop. Rather, final stops remain unchanged, or have disappeared entirely.

Finally, as noted by Klamer (1994:15) published sources suggest that the change *s > h in

Kambera took place between 1876 and 1909. In Anakalangu, which is closely related to

Kambera, PMP *s is preserved as a sibilant (*asu > Ank asu, Kmb ahu ‘dog’, *bities >

Ank wisi, Kmb wihi ‘leg, foot’, etc.).

3. Hawu. Hawu is spoken on the islands of Sawu and Rainjua between Sumba and

Timor. According to Walker (1982:1) Hawunese speakers recognize five dialects

“approximating the former kingdoms of Seba, Mesara, Timu, Liae and Rainjua.” Useful

published information on the language can be found in Kern (1892), a historically-

oriented overview of the language with a vocabulary of about 1,200 items, Wijngaarden

(1896), a small dictionary of approximately 1,800 base forms, Walker (1982), the only

published grammar, based primarily on the Seba and Mesara dialects, and C. Grimes (to

appear). Most of the material cited here is from Wijngaarden, but with orthographic

modifications based in part on Walker (1982), and in part on my own analysis.

It is universally agreed that Hawu is closely related to Dhao, spoken on the island of

Ndao off the western tip of Roti, where it has come under fairly strong contact influence

from Rotinese, a language that Esser assigned to his ‘Ambon-Timor’ group. Writers

from Jonker (1903) to Fox (1977:268) have described Dhao as a dialect of Hawu.

Walker (1982:57-63), who provides a brief comparison of the two, concludes that Hawu

and Dhao are closely related languages, a view also expressed by B. Grimes (2000). This

issue has been most thoroughly explored by Charles E. Grimes (to appear), who

concludes that although Hawu and Dhao share much vocabulary, differences in word

18

order, high-frequency grammatical functors, and the typology of case marking (Hawu is

ergative, Dhao accusative) render the two forms of speech mutually unintelligible. In

addition, as Grimes (to appear), and others have pointed out, much of the earlier literature

refers to ‘Savu’ and ‘Ndao’, but these spellings are inappropriate. Savu had an s

phoneme when first recorded that no longer exists, and the native pronunciation of the

language name has consequently become Hawu. Similarly, Dhao has no nd- sound or

sequence, and Grimes consequently writes the language name as ‘Dhao’. Following

Grimes I adopt the more modern spellings of these language names here. This introduces

an inconsistency, since Kambera, the principal language of Sumba, has also altered its

nineteenth-century s to h, leaving the Kambera pronunciation of the island name as

‘Humba’. However, since some other languages of Sumba retain the earlier s phoneme,

and since place names tend to become fixed through their use on maps, I refer to the

islands as ‘Sumba’ and ‘Sawu’, but to the language spoken on the latter island as ‘Hawu’.

3.1. Synchronic phonology. Walker gives the phoneme inventory of Hawu as follows:

TABLE 4: The phoneme inventory of Hawu

Consonants Vowels

pt k?i u

b d j g

≡  y♥e↔o

m n ñ N

h a

l

r

w

As in many other AN languages t is dental, but d, n and l are alveolar. The rhotic is

described as an alveolar trill or flap, and ≡, , y and ♥ as implosives made at bilabial,

alveolar, alveopalatal and velar places of articulation. Although he gives no examples of

stress in words that contain a penultimate schwa, Walker (1982:7-8) notes that Hawu has

a “clear preference for penultimate stress,” and it is assumed that this pattern was also

true at earlier stages of the language.

Unlike Kambera, Hawu has no prenasalized stops, or any phonetic consonant clusters.

For purposes of this paper little else need be said about the synchronic phonology, except

to note that Hawu, like Kambera, has only open syllables. However, although this result

has been achieved in Kambera through the addition of a supporting vowel –u, in Hawu it

has been achieved through loss of syllable-final consonants. One morphophonemic

feature is also important. Hawu has a fairly large class of what Walker (1982:65) calls

‘agreement verbs’. These verbs, which are overwhelmingly transitive, differ in singular

and plural forms, generally in agreement with what Walker anayzes as the absolutive NP.

Examples, with the plural preceding the colon and the singular following, include: b

↔

la :

b

↔

le ‘extend, stretch out’, duri : dure ‘drink (soup)’, hiu : hio ‘to tear’, nono : none ‘dry in

the sun’, t

↔≡

u : t

↔≡

o ‘pierce, stab’, and u



yu : u



ye ‘tie up (bundle)’. According to Walker

19

(1982:23), if the plural ends in VCu, where V = any vowel except u, the singular form

will end in –o. All other singular forms end in –e, but in addition if the plural form ends

in a HCa, where H = high vowel i, u, the singular will end in MCe, where M = mid vowel

e, o, as in hi

≡

a : he

≡

e ‘splash’, or peluja : peloje ‘take care of’. Synchronic predictability

and comparative evidence show that plurals are typologically unmarked and historically

conservative. Where data on sound change is cited for agreement verbs it is therefore

taken from these forms wherever they are available in the primary sources.

Walker notes that the Dhao (Ndao) phoneme inventory differs from that of Hawu in three

particulars: it has c and s, and lacks w. In cognate forms Hawu h corresponds to Dhao c

or s, and Hawu w corresponds to Dhao h, as shown in Table 5:

TABLE 5: The phonological correspondences Hawu h : Dhao c, s, and Hawu w : Dhao h

PMP Hawu Dhao English

sakay ha?e ca?e climb

esi ↔hi ↔ci one

siwa heo ceo nine

tasik d↔hi d↔si sea, saltwater

susu huhu susu breast

babuy wawi hahi pig

beli w↔li h↔li buy

bulan w↔ru h↔ru moon

3.2. Historical phonology. Hawu historical phonology shows moderate complexity,

with the theoretically most interesting changes occurring in the vowels. Since Hawu has

lost syllable-final consonants, it is impossible to use final consonant correspondences in

constructing a subgrouping argument.

*p. PMP *p is reflected as Hawu p or zero. Except as noted below, this split cannot be

predicted from the phonological environment, and both reflexes occur in forms that

appear to be directly inherited: *pani > ani ‘bait’, *pajay > are ‘rice (generic)’, *peñu >

↔

ñu ‘turtle’, *pia > ie ‘good’, *paniki > ni

?

i ‘fruit bat’, *pusuq > uhu ‘heart’; *hapuy > ai

‘fire’, *nipi > ni ‘dream’, *qapuR > ao ‘lime (for betel)’, but panas > pana ‘warm, hot’,

*pahuq > pau ‘mango’, *pija > p

↔

ri ‘how much, how many?’, *piliq > pili ‘choose’,

*putiq > pudi ‘white’; *papan > papa ‘plank, board’, *epat >

↔

pat ‘four’, *depa > r

↔

pa

‘fathom’. From the limited available evidence it appears that *mp is reflected as p:

*empu >

↔

pu ‘grandparent, grandchild’. PMP *qapeju ‘gall’ has two reflexes, one with

*p and one without (p

↔

u ‘bitter’,

↔

u ‘gall’), and next to pili ‘choose’, the unexplained

form

≡

ili also occurs. Following PMP penultimate *e (schwa), and initially in words that

took *ma- ‘stative’, *p did not lenite. The phonetic basis for resistance to lenition seems

clear. In Hawu penultimate schwa geminates a following consonant (Walker 1982:6). If

the same was true of pre-Hawu, as seems likely, the lenition of *p was blocked after a

stressed schwa as a result of phonetic gemination. With *ma-p- sequences syncope

20

occurred, giving rise to prenasalized stops that would also have been more resistant to

lenition than their simplex counterparts, even though ultimately they were simplified by

loss of the nasal.

Grimes (to appear) suggests that *p was retained in Hawu-Dhao prefixes and numbers,

but was lost initially and medially in other forms. While it is true that *p is retained as a

stop in *pa- > pe- ‘causative’, and in the numerals *epat >

↔

pa ‘four’, and *pitu > pidu

‘seven’, this is a questionable set of conditions for a sound change, and is contradicted by

the loss of *p in *sa-Na-puluq (> s↔-N↔-pulu > s↔-N↔-ulu > s↔-N-ulu) > heNuru ‘ten’, and

the retention of *p in examples such as *papan > papa ‘plank’, *pauq > pau ‘mango’,

*pija > p

↔

ri ‘how much/how many?’, or *putiq > pudi ‘white’.

*t. The split of PMP *t into Hawu d or t is also unpredictable. Both reflexes occur

word-initially and word-medially in similar vocalic environments: *tasik ‘sea, saltwater’

> dahi ‘beach’, *taqi > dei ‘feces’, *teka > d

↔

ka ‘come, arrive’, *qatimun ‘cucurbit’ >

dimu ‘watermelon’, wo dimu poro ‘cucumber’, *tuna > d

↔

no ‘freshwater eel’; *mata >

mada ‘eye’, *qatay > ade ‘liver’, *ma-qitem > m

↔

di ‘black’, *um-utaq > m

↔

du ‘vomit’,

but *taNis > taNi ‘weep’, *teduN > t

↔

u ‘head covering’, *tinaqi > tenei ‘guts’, *ta-hiup >

tiu ‘to blow’, *tunu > tunu ‘burn’; *datu > ratu ‘prince’, *kali-wati > kelati ‘earthworm’,

*betak > w

↔

ta ‘split’, *tuktuk ‘knock, pound, beat’ > tutu ‘peck’. Earlier *nt became

Hawu t: PMP *taqun > PSH *ntauN > Hawu tou ‘year’.

In a single case *t has become



, and in another n: *tebuh >

↔

bu ‘sugarcane’, *takaw >

(me)-na

?

o ‘to steal’. The latter irregularity evidently is a residue of a once active process

of verbalizing nasal substitution (Blust 2004).

*c. Only one reflex of PMP *c is known: *picik ‘splash, spray, sprinkle’ > pihi ‘shake

water off the hands’. However, since Malay loanwords with c (< *c) are borrowed with

h (

≡

aha < Malay baca ‘read’, guhi < Malay guci ‘earthen vessel’, maha < Malay macan

‘tiger’), and since Hawu has neither a voiceless palatal affricate nor an s, it can be

concluded despite this slender evidence that PMP *c became Hawu h.

*k. As for the preceding voiceless stops, PMP *k shows multiple reflexes in Hawu

without statable conditions. Word-initially *k is reflected either as k- or as zero: *kaRaw

> kao ‘scratch’, *kali (> *keli) > kei ‘dig, dig up’, *kulit > kuri ‘skin’, but *kahiw (>

*kayu) > aju ‘wood’, *kahu > au ‘2s’, *kutu > udu ‘louse’. Medially *k is reflected

either as glottal stop or as –k-: *kaka > a

?

a ‘elder parallel sibling’, *zukut >



yu

?

u ‘grass’

*paniki > ni

?

i ‘fruit bat’, *sakay > ha

?

e ‘climb’, *takaw > (me)-na

?

o ‘steal’, *daki > ra

?

i

‘dirt on body, dandruff’, *takut > da

?

u ‘afraid’, but *buka > boka ‘open’, *bekaq > b

↔

ka

‘split’, *teka > d

↔

ka ‘come, arrive’, *lekuq > l

↔

ku ‘fold’. The only generalization to

emerge from this is that *k never lenites following a stressed schwa. As with reflexes of

*p, this appears to be a consequence of earlier phonetic gemination.

*q. The PMP uvular stop *q disappeared without a trace in Hawu: *qapuR > ao ‘lime

(for betel)’, *qabu > awu ‘ash’, *qelad >

↔

la ‘wing’, *qanitu > nidu ‘ghost’, *qulun-an (>

luna) > n

↔

lu (with regular vowel metathesis and sporadic metathesis of the consonants);

21

*taqi > dei ‘feces’, *ma-Ruqanay > mone ‘male’, *taqun > tou ‘year’, *tuqud > tu ‘knee’,

*baqeRu > wiu ‘new’. The etymologies *qudaN > k

↔

ru ‘shrimp’ and *quay > gui ‘rattan’

can be maintained only by assuming that the initial consonant of the Hawu forms is a

separate morpheme for which no other evidence is currently available.

*b. With one exception PMP *b is reflected as w,

≡

, or b, without apparent conditions,

although the first reflex is by far the most frequent: *baRa > wa ‘lungs’, *beli > w

↔

li

‘buy’, *bisul > wihu ‘boil, abscess’, *bulan > w

↔

ru ‘moon’, *busuR > wuhu ‘bow’; *qabu

> awu ‘ash’, *babaq > wawa ‘under, beneath’, *buaq > wo, wue ‘fruit’, *bubu > wuwu

‘conical bamboo basket trap for fish’, but *babaq >

≡

a

≡

a ‘short’, *besuR >

≡↔

hu ‘satiated’,

*buaq >

≡

ue ‘fruit’, *buta >

≡↔

tu ‘blind’; *sebu ‘meeting of heat and water’ > h

↔≡

u

‘steam’, *tebek > t

↔≡

u ‘stab’, and *abelaj > b

↔

la ‘wide, spread out’, *b-in-ahi (> binay >

bine) > b

↔

ni ‘woman’, *buka > boka ‘to open’; *tebuh >

↔

bu ‘sugarcane’. Again, the one

generalization that emerges from the data is that the lenis reflex w never occurs following

a stressed schwa, where gemination of the stop would be expected.

PMP *buaq gave rise to doublets wo ‘fruit’, used in names of fruits (wo pau ‘mango’, wo

mu

?

u ‘banana’, etc.), but

≡

ue/wue ‘numeral classifier for various objects, including fruits,

eggs, houses and humans’ (wo pau t

↔

lu

≡

ue ‘three mangos’, lit. ‘fruit mango three fruit’ ).

Both forms appear to be directly inherited, given etymologies such as *duha (> dua) >



ue ‘two’, since such low numerals are rarely borrowed. Finally, *buni is reflected as

huni and wuni ‘to hide’. The first of these variants may be a borrowing from Dhao.

*d. PMP *d generally became r, but occasionally became



: *daRaq > ra ‘blood’, *daki

> ra

?

i ‘dirt on the skin’, *depa > r

↔

pa ‘fathom’, *dahun > rou ‘leaf’; *tada > tara ‘natural

cockspur’, *ma-qudip > muri ‘living, growing’, Nuda > N

↔

ru ‘young’, but *dalem >



ara

‘in, inside’, *duha >



ue ‘two’, *kedeN > k

↔

i ‘stand up’, *teduN > t

↔

u ‘head covering’.

In one known form *d became d, and in another it disappeared: *diNdiN > didi ‘wall’,

*dikit > iki ‘small’. Although both r and



reflect *d word-initially, medial *d became



after PMP *e and r after other vowels. Subsequent vowel metathesis then produced some

schwas before an r reflex of *d, as in *Nuda (> Nura) > N

↔

ru ‘young’.

In one known form *d is reflected as n, perhaps as a result of assimilation to the onset of

the preceding syllable: *anaduq ‘long (of objects)’ > nano ‘long, of time’.

*z. PMP *z (probably a voiced palatal affricate) became



y, which Walker describes as a

voiced palatal implosive affricate, although phonemic palatal implosive affricates in other

AN languages are palatalized alveolars: *zalan >



yara ‘road’, *zaRum ‘needle’ >



yau

‘sew’,



yi-



yau ‘needle’, *ziRuq >



yiu ‘bathe’, *zukut >



yu

?

u ‘grass’.

*j. PMP *j did not occur initially, and hence has a non-zero reflex only word-medially in

Hawu. In this position it is reflected as



following the reflex of a PMP stressed schwa,

and as r elsewhere: *qapeju > p

↔

u ‘gall’, *qalejaw (> l

↔

o) > lo



o ‘day, sun’, *hapejes >

p

↔

a ‘pain, sickness’, *pajay > are ‘rice (in general)’, *huaji (> waji) > ari ‘younger

parallel sibling’, *Najan > Nara ‘name’, *pija > p

↔

ri ‘how much/how many?’. As in the

22

other cases already cited above, the resistance to lenition following a stressed schwa

probably resulted from phonetic gemination of most consonants in this position.

*g. No reflexes of *g have been noted in Hawu, but Malay loans with g are borrowed

without change to this consonant (

≡

agi < Malay bagi ‘divide up’, guhi < Malay gusi

‘earthen pot (not made on Sawu)’, piga < Malay piNgan ‘plate’).

The origin of implosion in Hawu (and allophonically in Kambera) is unclear. Implosion

is an areal feature in southeast Sulawesi and nearby portions of the Lesser Sundas, so

contact appears to be implicated in at least many cases. It is possible that at one time all

voiced obstruents in Hawu were automatically imploded, and that contrast only arose

through language-internal changes such as *t > d, and borrowing.

*m. PMP *m is invariably reflected as Hawu m: *m-ataq > mada ‘raw’, *mamaq >

mama ‘chew’, *ma-iRaq > mea ‘red, ripe’, *ma-qitem > m

↔

ti, ‘black’, *miñak > m

↔

ñi

‘fat, oil’, *ma-qudip (> mudip) > muri ‘alive, growing’; *ama > ama ‘father’, *Ramut >

amo ‘root’, *qatimun > dimu ‘cucumber, watermelon’, *lima > l

↔

mi ‘five’.

*n. With one exception, PMP *n remained unchanged: *nabuq > (me)-nawu ‘fall’,

*nanaq > nana ‘pus’, *nipi > ni ‘dream’, *niknik > nini ‘sandfly, midge’; *anak > ana

‘child’, *enem >

↔

na ‘six’, *buni > wuni ‘to hide’, *ma-Ruqanay > mone ‘male’, *inum >

N-inu ‘drink’. The exception is *niuR > ñiu ‘coconut’, with sporadic palatalization in an

apparently native form, a development that is independent of the similar change in Malay

ñiur and related forms in languages of central and western Borneo, as Kayan ñuh.

*ñ. Although it has merged with *n in most daughter languages, PMP ñ is retained as a

distinct phoneme in Hawu. Only three or four cases are known: * ñilu ‘set the teeth on

edge, of something very sour’ > me-ñilu ‘sour’, *ñamñam-i ‘taste’ > ñami ‘chew’;5 *peñu

‘green turtle’ > p

↔

ñu ‘turtle’, and possibly *añem >

↔

ñu ‘to plait’, with irregular *e > u.

*N. *N generally remained unchanged: *NaNa > me-NaNa ‘open mouth wide’, *Najan

> Nara ‘name’, *Nuda > N

↔

ru ‘young’; *saNa > haNa ‘fork (of road, etc.)’, *sumaNed >

hemaNa ‘soul, spirit of a living being’, *deNeR > r

↔

Ni ‘hear’, *taNis > taNi ‘weep, cry’,

*teNaq > t

↔

Na ‘middle’. In one form *N is reflected unaccountably as n: *ma-laNu

‘dizzy, intoxicated, poisoned’ > me-lanu ‘dizzy’.

*s. PMP *s is invariably reflected as Hawu h: *salaq > hala ‘fault’, *sebu > h

↔≡

u

‘steam’, *siwa > heo ‘nine’, *siku > wo-hi

?

u ‘elbow’, *susu > huhu ‘breast’, *suduk >

huru ‘ladle, spoon’; *baseq >

≡

aha ‘wash’, *isa >

↔

hi ‘one’, *pusej >

↔

hu ‘navel’, *isi >

ihi ‘contents’, *bisul > wihu ‘boil, abscess’, *musuq > muhu ‘war’. Referring to the

Dutch colonial literature on Hawu, Walker (1982:3) notes that “Donselaar’s (1872)

account is important because, in addition to 50 lexical items, it documents a period in

which s and h were interchangeable.” As in Kambera, then, historical records show that

this is a late change.

5 Similar suffixed forms of this word are found in Oceanic languages (Blust 1976:27).

23

*h. PMP *h disappeared: *hapuy > ai ‘fire’, *huaji (> wadi) > ari ‘younger parallel

sibling’, *habaRat > wa ‘west’; *kahiw (> kayu) > aju ‘wood’, *qahelu > alu ‘pestle’,

*duha >



ue ‘two’, *lahud ‘downriver’ > lou ‘sea’, *ma-buhek (> mabuk) > mawo

‘drunk’, *pahuq > pau ‘mango’, *dahun > rou/ru ‘leaf’, *bahuq > wou ‘smell, odor’.

*l. PMP *l generally became l, but about one third of the known reflexes show *l > r.

No conditioning environment for this split is evident, and both reflexes appear to be

found in native forms (in one case in the same form): *layaR > lai ‘sail’, *lahud > lou

‘sea’, *leku(q) > l

↔

ku ‘fold’, *lima > l

↔

mi ‘five’, *liaN > lie ‘cave’, *lulun > lule ‘roll up,

as a mat’; *telen > d

↔

la ‘to swallow’, *beli > w

↔

li ‘buy’, *iluR > ilu ‘saliva’, but *walu >

aru ‘eight’, *talih > dari ‘rope’, *suluq > huru ‘torch’, *kulit > kuri ‘skin’, *lalej > lara

‘housefly’, *buliR > wuri ‘ear of grain’, *bulu ‘body hair, downy feathers, floss on plant

stems’ > wuru ‘fibers, as the hairs on roots’. All examples of *l > r are intervocalic, but

as noted above, some examples of *l > l also occur in this position. In a single known

case *l has disappeared: *kali (> keli) > kei ‘dig’.

*r. A reflex of *r has been found only in *burit ‘hind part, rear; base or bottom’ > wui

‘base, bottom’, where it has disappeared, probably after merging with *R.

*R. PMP *R usually dropped: *Ramut > amo ‘root’, *Rumaq >

↔

mu ‘house’; *baRani >

≡

ani ‘brave, bold’, *taRum > dao ‘indigo’, *kaRaw > kao ‘scratch’, *maRi > mai ‘come’,

*duRi > rui ‘bone’, *baRa > wa ‘lungs’. Although its outcome is zero, there is reason to

believe that this change went through an intermediate stage with *y, since there is palatal

coloration of the chromatically neutral mid-central vowel preceding the reflex of *R, as

in *deNeR (> r↔N↔y) > r

↔

Ni ‘hear’, *beRay (> w

↔

yay > w

↔

ye) > wie ‘give’, or *baqeRu (>

w↔yu) > wiu ‘new’. Finally, in two known cases *R is reflected as r in words that appear

to be native: *Rusuk > ruhu ‘ribs’, *timeRaq ‘tin, lead’ > tem

↔

ra ‘lead’.

*w. Although few examples are available, it appears that PMP *w generally disappeared

word-initially: *walu > aru ‘eight’, *wahiR > ei ‘fresh water’, *wada > era ‘there is,

there are’, *huaji (> wari) > ari ‘younger parallel sibling’. In two known cases the

sequence *wa contracted to o: *siwa (sio) > heo ‘nine’, *wani > oni ‘honeybee’. In one

other form the derived sequence *wa became wo: *buqaya (> waya) > woe ‘crocodile’.

*y. Little data is available to determine the development of syllable-initial *y. In one

form is appears to be reflected as i: *layaR > lai ‘sail’. In another the sequence *ay

contracted to e: *buqaya (> waya) > woe ‘crocodile’.

*-ay. PMP *-ay was monophthongized as –e: *qatay > ade ‘liver’, *pajay > are ‘rice (in

general)’, *b-in-ahi (> binay > bine) > b

↔

ni ‘woman’, *sakay > ha

?

e ‘climb’, *ma-

Ruqanay > mone ‘male’, *beRay > wie ‘give’.

*-aw. PMP *-aw was monophthongized as –o: *panaw > ano ‘leucoderma’, *kaRaw >

kao ‘scratch’, *qalejaw > lo



o ‘day’, *takaw > (me)-na

?

o ‘steal’.

*-uy. PMP *-uy was monophthongized as –i: *hapuy > ai ‘fire’, *babuy > wawi ‘pig’.

24

*-iw. PMP *-iw was monophthongized as -i: *laRiw ‘run away, flee’ > (pe)-lai ‘run’.

The derivation *kahiw > kaiw > kayu > aju ‘wood’ shows *-iw > u, but this is clearly a

result of prior resyllabification.

V3. In many AN languages, especially those of Borneo and adjacent parts of western

Indonesia and the southern Philippines, either *a alone, or all vowels merged as schwa in

prepenultimate syllables. In Hawu, by contrast, the merger was to the mid-front vowel e.

Walker (1982:7) notes that V1 (a prepenultimate vowel, here called V3 to distinguish it

from the first-syllable vowel in disyllables) “can be any vowel except schwa. It is usually

e (80%), but sometimes o (10%).” Known reflexes of V3 include *sumaNed > hemaNa

‘soul, life-force’, *kamali ‘men’s house’ > kemali ‘house, household’, *kali-wati (> k↔l↔-

wati > k↔lati) > kelate ‘earthworm’, *minaNa > menaNa ‘mouth of a river, estuary’,

*qati-mela > tem

↔

la ‘dog flea’, *timeRaq > tem

↔

ra ‘lead, tin’, and *tinaqi > tenei

‘intestines’. Syllable reductions in forms such as *ma-qitem (> m↔-item > mitem) > m

↔

di

‘black’, or *ma-qudip (> m↔-udip > mudip) > muri ‘living, alive, growing’ suggest that

this merger took place through earlier schwa, since schwa is rarely permitted adjacent to

a vowel or semivowel in any AN language. This inference is further supported by *ma-

nipis (> m↔-nipis > m↔niis) > m

↔

ni ‘thin, tenuous’, where the historically secondary schwa

evidently came to be penultimate before the merger of all prepenultimate vowels as e.

Finally, in forms with a medial consonant that was lost before this change took place

contraction of a trisyllable to a disyllable prevented merger of the first-syllable vowel as

e: *b-in-ahi (> b-in-ai > binay > bine) > b

↔

ni ‘woman’, *binehiq (> biniq) > wini ‘seed

(for planting)’.

A prepenultimate vowel that was initial or that became initial through loss of a preceding

consonant dropped, almost certainly because of a prior change of *a to schwa, since

prepenultimate initial schwa is excluded in most AN languages: *qateluR/qiteluR > d

↔

lu

‘egg’, *qapeju ‘gall’ > p

↔

u ‘bitter’,

↔

u ‘gall’, *qalejaw > lo



o ‘day, sun’, *um-utaq (>

↔m-utaq > muta) > m

↔

du ‘vomit’, *qulun-an (> lunan > nulan) > n

↔

lu ‘wooden headrest’,

*qanitu > nidu ‘ghost’, *hapejes > p

↔

a ‘pain, painful’, *habaRat (> baRat) > wa ‘west’.

*a. PMP *a changed in Hawu under three conditions. First, it partly assimilated to an

adjacent high vowel: *taqi > dei ‘feces’, *tau > dou ‘person’, *wahiR (> wai) > ei ‘fresh

water’, *pia > ie ‘good’, *liaN > lie ‘cave’, *lahud ‘downriver, toward the sea’ > lou

‘sea’, *dahun > rou/ru ‘leaf’, *taqun > tou ‘year’, *bahuq > wou ‘smell, odor’. With one

known exception (*balik >

≡↔

li, w

↔

li ‘turn back’), this change occurred only when there

was no intervening consonant, and it failed to take place in some forms even when the

vowels were in contact: *kahiw (> kayu) > aju ‘wood’, *qahelu (> qalu) > alu ‘rice

pestle’, *pani > ani ‘bait’, *qabu > awu ‘ash, gray’, *kahu > au ‘2s’, *layaR > lai ‘sail’.

Second, the sequence *-ua became –ue in at least three forms: *buaq (> bua) >

≡

ue, wue

‘classifier for fruits’, *duha >



ue ‘two’, *tuak > due ‘palm wine’. In one known form

this sequence has contracted to o: *buaq > wo ‘fruit’. There may be another example of

this change, which is clearly similar to the change *wa > o already noted above, since

*ma-Ruqanay (> m↔Ruanay > m↔uanay > muanay) probably passed through a stage in

25

which it was *muane before becoming mone ‘male’. The third condition, which led to

centralization of *a, is seen in vowel metathesis, a topic that is treated separately below.

In addition to these apparently conditioned changes, PMP *a is irregularly reflected as e

in *wada > era ‘there is’ and *kali > kei ‘dig’. The second of these is an irregularity that

is shared with Eastern Malayo-Polynesian languages as a group (Blust 1983/84).

*e. PMP *e deleted if it came to abut another vowel, a change that evidently antedated

Proto-Sumba-Hawu: *qahelu (> qaelu) > alu ‘pestle’, *ma-buhek (> mabuek > mabuk) >

mawo ‘drunk’, *binehiq (> bineiq > biniq) > wini ‘seed for planting’. Apart from this, *e

was ordinarily retained as a mid-central vowel in penultimate position: *bekaq >

≡↔

ka

‘split’, *ma-belaj > b

↔

la ‘broad, flat’, *qelad >

↔

la ‘wing’, *enem >

↔

na ‘six’, *epat >

↔

pa

‘four’, *leku > l

↔

ku ‘fold’, *hapejes > p

↔

da ‘sting; pain’, etc. A schwa that came to be

word-final through consonant loss changed to –a (in verbs a/e in accordance with the

innovative system of plural/singular agreement) thereby conforming to an inherited

constraint against word-final schwa: *tanem > dana/dane ‘bury’, *dalem >



ara ‘in,

inside’, *telen > d

↔

la/d

↔

le ‘to swallow’, *enem >

↔

na ‘six’, *sumaNed > hemaNa ‘soul,

life-force’, *lalej > lara ‘housefly’, *nabek > nawa ‘wave, breakers’, *hapejes > p

↔

a

‘pain’. Irregular changes of *e are seen in *kedeN > k

↔

i ‘stand’, *deNeR > r

↔

Ni (pl.),

r

↔

Ne (sg.) ‘hear’, and *tebek > t

↔≡

u ‘pierce, stab’. In the second form *e > i/e may have

been conditioned by *R > y, then –

↔

y > i. In a single known case penultimate *e became

o, possibly through regressive assimilation following the monophthongization of *-aw:

*qalejaw (> l↔daw > l↔do) lo



o ‘sun, day’.

*i. PMP *i is almost invariably reflected as a high front vowel. The only exceptions

noted to date are *balik ‘return, go back’ >

≡

ale/balu ‘turned around, gone back’ (also

≡

ari/

≡

are ‘turn over in one’s sleep; change; repent’), *siwa > heo ‘nine’, *kawil > kawa

‘fishhook’, and *kali-wati > kelate ‘earthworm’.

*u. PMP *u normally remains unchanged, but has been unconditionally lowered to o in

the following forms: *Ramut > amo ‘root’, *qapuR > ao ‘lime (for betel)’, *buka > boka

‘to open’, *taRum > dao ‘indigo’, *tuna > d

↔

no (met.) ‘eel’, and *anaduq ‘long (of

objects)’ > nano ‘long, of time’.

Vowel metathesis. From the standpoint of linguistic theory the most noteworthy feature

of Hawu historical phonology is a set of changes that I will call ‘vowel metathesis’. This

development is theoretically important because 1) it is regular, 2) it affects only a subset

of vowel combinations, and 3) it appears to involve phonetically unrelated processes that

formed a single change. Table 6 presents the relevant data that support these claims:

TABLE 6: Vowel metathesis in Hawu6

PMP Hawu English

6 In addition to these examples, Kern (1892) suggested that à



yi ‘rain’ reflects *quzan through an irregular

intermediate form with penultimate *i (i



ya).

26

01. *b-in-ahi (> binay > bine) b↔ni woman

02. *buta ≡↔du blind

03. *tuna d↔no freshwater eel

04. *isa ↔hi one

05. *pusej (> uha) ↔hu navel

06. *hiket (> ika) ↔ki tie, bind

07. *kuden (> ura) ↔ru cooking pot

08. *Rumaq ↔mu house

09. *sukat h↔ku measure

10. *lima l↔mi five

11. *ma-qitem (> mitem > mita) m↔di black

12. *um-utaq (> mutaq) m↔du vomit

13. *miñak m↔ñi fat, oil

14. *qulun-an (> lunan) n↔lu (< met.) wooden headrest

15. *kita N↔di to see

16. *Nuda N↔ru young

17. *pija p↔ri how much/many?

18. *p-ijan p↔ri when?

19. *bulan w↔ru moon/month

There are two known exceptions: *buka > boka/boke ‘to open’, and *ma-isa (> misa) >

miha ‘alone’. This last item is surprising, since it contains the base *isa, which shows

vowel metathesis in

↔

hi ‘one’. The data in Table 6 reveal a pattern of the following form:

Given V1[+hi]CV2[-hi]

Then V1CV2 => ↔CV1

In other words, two vowels separated by a consonant regularly metathesized if the first

was high and the second was not. In nearly all cases this meant *iCa or *uCa, but b

↔

ni

‘woman’ (earlier *bine) shows that the input for metathesis was [+hi] C [-hi]. The non-

high vowel that was moved to penultimate position then centralized to schwa. This

change did not occur if no consonant intervened, and metathesis did not affect any other

V1CV2 combination: *buaq >

≡

ue/wue ‘fruit; numeral classifier’, *duha (> dua) >



ue

‘two’, *pia > ie ‘good’, *liaN > lie ‘cave’, *tuak > due ‘palm wine’; HH: *niniq > nini

‘midge’, *qanitu > nidu ‘ghost’, *pusu > uhu ‘heart’, *puki > u

?

i ‘vulva’, LH: *kahiw (>

kayu) > aju ‘wood’, *babuy > wawi ‘pig’, *tasik ‘sea/saltwater’ > dahi ‘beach’, *pani >

ani ‘bait’, *daki > ra

?

i ‘dirt on body’, *batu > wadu ‘stone’, LM: *qatay > ade ‘liver’,

*m-atay > made ‘die/dead’, *sakay > ha

?

e ‘climb’, *takaw > na

?

o ‘steal’, and LL: *salaq

> hala ‘fault’, *mata > mada ‘eye’, *nanaq > nana ‘pus’ (the last three forms, if

metathesized, should show schwa as the penultimate vowel).

The reality of Hawu vowel metathesis as a complex sound change is confirmed by its

occurrence in g↔ru, a borrowing of Malay gula ‘sugar’, although other Malay loans do not

show it (ruha < Malay rusa ‘deer’, piga < Malay piNgan ‘plate’), ke-hiwa < Malay séwa

‘hire, rent’). The same pattern of vowel correspondence also appears in a few cognates

without a known higher-level etymology, as in Kambera mbula, Hawu b

↔

lu ‘forget’,

27

Kambera unaNu, Hawu

↔

ñu ‘to plait, as mats’ (possibly from PMP *añem ‘plait’ through

intermediate *uñaN), Kambera uta, Hawu

↔

to (also Selaru ut) ‘chaff’, Kambera kuraNu,

Hawu k

↔

ru ‘shrimp’ (possibly from PMP *qudaN with a fused prefix reflecting *ka-), and

Kambera wira, Hawu w

↔

ri ‘snot’. Kern (1892) noted the vocalic interchange in several of

these forms, as with

↔

hu ‘navel’, which he compared with Malay pusat, and described as

a metathesis of earlier uha, or m

↔

di ‘black’, which he described as a metathesis of earlier

*mitem. In other cases he identified cognates without mentioning metathesis in the

Hawu form, as with

≡↔

du, which he compared with Malay buta ‘blind’. In still other

cases his proposals of cognation were erroneous, as with Hawu

↔

hi ‘one’, Makasarese

assi ‘contents’ (< PMP *hesi), and Malay asiN ‘foreign’ (origin unknown). All in all,

then, it is clear that although Kern correctly noted vowel metathesis in a few forms he did

not recognize either its specific form or its regularity in Hawu.

Although little published data is available for Dhao, it is clear from Walker (1982:57-63),

Grimes (to appear), and Fox (n.d.) that vowel metathesis also took place in this language:

*b-in-ahi (> binay > bine) > Hawu, Dhao b

↔

ni ‘woman’, *isa > Hawu

↔

hi, Dhao

↔

ci ‘one’,

*iket (> ika) > Hawu, Dhao

↔

ki ‘tie, bind’, *Rumaq > Hawu, Dhao

↔

mu ‘house’, *lima >

Hawu, Dhao l

↔

mi ‘five’, *ma-qitem (> mitem > mita) > m

↔

di ‘black’, *p-ijan > Hawu,

Dhao p

↔

ri ‘when?’, *bulan > Hawu w

↔

ru, Dhao h

↔

ru ‘moon’. This change alone thus

shows the close relationship of these two languages. The theoretical implications of

vowel metathesis in these languages will be treated in a separate publication. For now it

is sufficient to note that this appears to be a ‘complex sound change’, a rare type of

change in which phonologically unrelated processes combine in a single innovation.

It is worth noting that some earlier sources on Hawu show a few significant differences

from the material in Wijngaarden (1896) or later writers. It has already been noted that

an 1872 publication on Hawu by the Dutch missionary W.M. Donselaar described the

language at “a period in which s and h were interchangeable.” Kern (1892), who never

visited Sawu, but obtained his material from Dutch residents on the island, has forms

with ai and au that in Wijngaarden (1896) and Walker (1982) show partial assimilation of

the low vowel, as with ai (K), ei (W/W) ‘water’, lau (K), lou (W/W) ‘sea’, or tau (K), tou

(W/W) ‘year’. It is unclear whether these discrepancies are due to dialect differences or

to change by lexical diffusion (the short time span between these two publications makes

the first of these possible explanations more plausible).7 More strikingly, Kern lists

several forms with final glottal stop. These occur in 13 known items in a vocabulary of

about 1,200 words, or about 1% of the vocabulary. Where etymologies are available

these glottal stops reflect *-p, *-k, or *-s (never *-t): *ma-qudip (> mudip) > muri

?

(K),

muri (W) ‘living, alive’, *anak > ana

?

(K), ana (W) ‘child’, *ma-panas > pana

?

(K), pana

(W) ‘warm, hot’, *hapejes > p

↔

a

?

‘pain, sickness’. However, in some other forms *-k

7 Allthough Walker (1982) writes e.g. dou ‘person’, Grimes (to appear) writes t

↔

u ‘year’, with a mid-central

vowel or diphthongal nucleus. Walker (1982:3) indicates that his data was collected from speakers of the

following dialects: Mesara (360 minutes), Seba (210 minutes), Rainjua (90 minutes), Timu (60 minutes),

Liae (60 minutes). We can assume, then, that it is most strongly influenced by Mesara and Seba speech.

Grimes (to appear), also states that he worked with speakers of several dialects (Seba, Dimu, Raijua), but

had little contact with Liae and Mehara (Mesara) speakers. It is perhaps this differential contact with the

Mesara/Mehara dialect that is primarily responsible for these differences in the representation of vowel

sequences reflecting PMP *a(C)u.

28

and *-s were lost : *manuk > manu ‘chicken’, *ma-nipis > m

↔

ni ‘thin, fine’, etc. The

report of a rare glottal stop as the only final consonant in Hawu some 120 years ago is

puzzling but probably accurate, since Kern cites minimal pairs that are distinguished only

by this feature (ko

?

‘thorn’ vs. ko ‘once’, muri

?

‘living, growing’ vs. muri/muru ‘lord,

master’). Moreover, in two cases a final glottal stop matches the final –t or –k of a Dutch

loanword (koa

?

< Dutch kwart ‘quarter’, woro

?

< Dutch vork ‘fork’).

4. Bimanese. Bimanese is spoken over the eastern portion of the island of Sumbawa,

between Lombok and Flores. This area was subjugated by the Makasarese kingdom of

Goa in 1616, through which Islam was introduced, along with a small number of

loanwords. By contrast, borrowing from Malay/Indonesian has been much more intense.

Bimanese is closely related to Donggo (< Bimanese doNgo ‘mountain’, dou doNgo

‘mountain people’), spoken by a small non-Islamic population in the mountains of

eastern Sumbawa.

4.1. Synchronic phonology. The first substantial account of Bimanese was by the

Dutch Indonesianist J.C.G. Jonker, who published a grammar, dictionary and collection

of texts in a single integrated description (1893-1896). Little work was subsequently

done on the language for over three-quarters of a century. In January, 1972, I worked for

6-7 contact hours with Abdul Karim Sahidu (AKS), originally from the mountain village

of Maria, some 17 km. from the capital of Bima, who was then studying in Hawai’i.

During this time I collected about 600 words and 15 sentences, and was able to make

some fairly confident observations about phonetics, phonology and morpheme structure.

Subsequently Sahidu (1978), and Ismail, Azis, Yakub, Taufik and Usman (1985)

produced Bimanese-Indonesian dictionaries, and Owens (2000) devoted a monograph-

length treatment to several topics in Bimanese grammar. I will use my own fieldnotes as

my primary data source, and both dictionaries, along with the older work of Jonker as

supplementary material. The phoneme inventory of Bimanese is shown in Table 7:

TABLE 7: The phoneme inventory of Bimanese

Consonants Vowels

p t c k ?i u

(b) d j (g)

≡  e o

m n N a

f s h

w

l

r

To these we might add a prenasalized series, as in Owens (2000:5), but this is debatable.

The plain stops have their expected values except that t is postdental and d alveolar.

While

≡

is bilabial,



is apico-domal, thus supporting Greenberg’s (1970) observation that

coronal implosives tend to be strongly retracted. Although d is common, and [mb] and

[Ng] are found in many native words, b and g are rare, and are confined almost entirely to

29

loanwords. It is difficult to discuss the phonemic status of these segments without

considering both borrowing and complementation.

In my own data b was noted in three forms, all of which are loans from Malay/Bahasa

Indonesia (BI): balumba ‘waves at sea’ (BI belumbaN ‘billows’), bara ‘strong wind’ (BI

aNin barat ‘strong wind from the west; west monsoon’, bedi ‘shoot, fire a weapon’ (BI

bedil ‘firearm’). A plain voiced velar stop was noted in eight forms, of which at least six

are also apparent loans: gala ‘pole used to knock down fruits, etc.’ (BI galah ‘long pole,

quant’), galara ‘traditional village headman’ (BI gelar ‘title’, gelar-an ‘titular

designation’), gamba ‘picture’ (BI gambar), ganja ‘piece of wood used to keep a car, etc.

from rolling downhill’ (BI ganjaN ‘tighten or tauten by means of a wedge; fix a kris-

blade firmly in its hilt by means of cloth wrappings’), garu ‘old-fashioned way of ironing

clothes, using the juice of a fruit and pressing with a shell’ (BI garus/gerus ‘a shell,

Cypraea tigris; giving a gloss to cloth by use of this shell; calendering), gendi ‘eyebrow’,

geo ‘carry a child on the back’ (source unknown), golo ‘k.o. knife’ (BI golok ‘machete or

sword with a convex cutting edge’). A similar picture emerges from the more extensive

materials in the two available dictionaries. The evidence to hand thus suggests that non-

prenasalized [b] and [g] are confined largely to loanwords.

Although it has no word-final consonants Bimanese permits homorganically prenasalized

obstruents initially and intervocalically. Surface clusters include mb, mp, nd, nt, nc, nj,

Ng and Nk, although the last of these is unattested initially and in my data it was recorded

medially only in roNko ‘smoke a cigarette’ (BI me-rokok), and ruNka ‘to change’. Most

words that contain the sequences mb and Ng have no known loan source, and so appear to

be native, whether these occur in initial position, as with mbai ‘rotten’, mba

?

a ‘year’,

mbei ‘to give’, mbinti ‘cockfight’, mbocu ‘satiated’, mbuda ‘blind’, Ngalo ‘hunt with

dogs’, Ngomi ‘you (addressing younger interlocutor)’, Nguda ‘to plant’, or medially, as

with komba ‘spotted, as a young animal’, rombo ‘straight’, tagambe ‘useful tree resin,

dammar’, umbu ‘to bury’, duNga ‘k.o. tangerine’, kaNge ‘ring finger’, ruNgi ‘to push’,

hoNgo ‘head hair’, or saNgolu ‘rainbow’.

Phonological alternations include k ~ g (haju ka

?

a ‘firewood’ : N-ga

?

a ‘to burn’), and

≡

~

b (sa-

≡

ua uma = one-CL house ‘a house’ :



ua m-bua uma = two-CL house ‘two houses’).

Given these alternations and the tenuous evidence for [b] and [g] in native forms, [mb]

and [Ng] can be analyzed as /m≡/ and /Nk/ either underlyingly, or at a stage prior to the

acquisition of loanwords that disrupted the complementation of [≡] with [mb] and of [k]

with [Ng]. Other alternations include f with p (fu

?

u ‘base of a tree’ : m-pu

?

u ‘classifier

for trees’, s with c (saNa ‘fork of a branch’ : n-caNa ‘to branch, bifurcate’), h with g

(heko : N-geko ‘surround, encircle’), v with b (vali ‘again’ : m-bali ‘repeat, alternate’, and

r with d (ra

?

a ‘blood’ : n-da

?

a ‘bleed’). As is often the case, these alternations suggest

some of the historical changes that have occurred in the language.

Like b and g, the voiced palatal affricate j is rare, and is usually found in loanwords, as

joge ‘traditional dance’ (BI joget ‘pas de deux, generally danced by two girls’, < Javanese

joged ‘classical Javanese dance; woman street dancer’). However, unlike the case with b

and g, there is strong evidence that j also occurs in native forms as haju < PMP *kahiw

30

(through *kayu) ‘wood; tree’ or loja < *layaR ‘sail’. The only other phonological issue

that merits comment is that I generally recorded [v] where both dictionaries and Owens

write w. In the speech of AKS these phones varied freely, with the fricative being the

more common sound. I follow the published sources and use w here.

Stress is penultimate, with no apparent variation or conditioning: [véi] ‘female, woman’,

[máde] ‘die, dead’, [≡í?a] ‘to split’, [vóu] ‘stench’, [tambári] ‘turn the head’. It was

impossible to determine whether stress shift occurs under suffixation, since only prefixes

were recorded. Although Bimanese has a symmetrical five vowel system, the low vowel

a varies freely with schwa in prepenultimate position, as in sa-mpuru ~ s

↔

–mpuru ‘ten’,

sambore ~ s

↔

mbore ‘hammer’, or ka

?

udu ~ k

↔?

udu ‘to heap, pile up’. Phonetic vowel

length was not noted in any position, but sequences of unlike vowels differing in one

degree of height are fairly common, as in wei and wou.

4.2. Historical phonology. Bimanese has lost all word-final consonants, even in

loanwords from Malay: gadi ‘ivory’ (Malay gadiN), hami ‘Thursday’ (Malay kamis),

maria ‘cannon’ (Malay meriam), sura ‘letter; writing’ (Malay surat), etc. Reflexes of

PMP consonants will therefore be given only for non-final position.

As in Kambera, a number of Bimanese words contain obstruents with non-etymological

prenasalization of initial obstruents. Examples include PMP *pispis > mpifi ‘temple

(anat.)’, *baRiw > mbai ‘spoiled, rotten’, *baseq > mbeca ‘wet’, *beRay > mbei ‘give’,

*besuR > mbocu ‘satiated’, *buta > mbuda ‘blind’, *danaw > ndano ‘lake’, and *depa >

ndupa ‘fathom’. The prenasalization of bases that begin with a labial obstruent appears

to derive from syncope in PMP *ma- ‘stative’, but the explanation for prenasalization in

other type of bases is unclear. Like Kambera, Bimanese also shows alternations of word-

initial *s with a palatal affricate following a nasal: saNa ‘branch’ : ncaNa ‘fork,

bifurcate; forked’ (cp. Kambera haNa ‘branch’ : njaNa ‘to fork, bifurcate’).

*p. PMP *p is reflected as f or p, without stateable conditions: *paRih > fai ‘stingray’,

*paku > fahu ‘fern’, *paNdan (> pandan) > fanda ‘pandanus’, *peñu > fonu ‘turtle’,

*puqun > fu

?

u ‘base of a tree’; *hapuy > afi ‘fire’, *nipi > nifi ‘dream’, *qalu-hipan (>

alipan > lipan) > rifa ‘centipede’, *upi > ufi8 ‘blow’, but *pa-kaen > paha ‘feed’,

*hapejes (> pejes) > pili ‘pain’, *pitu > pidu ‘seven’; *qapit > api ‘pinch, squeeze

between’, *lepet > lipi ‘fold, wrap’, *nipis > nipi ‘thin, of materials’, *epat > upa ‘four’.

PMP *-mp- remained unchanged: *lampaq > lampa ‘walk, go’, *empu > ompu

‘grandfather, grandson’, *tampak > dampa ‘blunt (tip)’, *sumpit > sumpi ‘blowpipe’.

There are two historically derived *mp sequences, and the reflexes of these diverge: *sa-

Na-puluq (> Na-puluq > mpuluq) > mpuru ‘ten’, but *panas > pana ‘hot’, *ma-panas >

mbana ‘humid, hot. In addition, if valid, *ma-penuq >

≡

ini ‘full, as a container’ shows an

irregular development *p >

≡

.

*t. PMP *t became Bimanese d or t, without known conditions: *taneq > dana ‘earth’,

*tepiR > dipi ‘mat’, *tau > dou ‘person’, *tebuh > do

≡

u ‘sugarcane’, *qatimun > dimu

‘cucumber’, *tuna > duna ‘freshwater eel’; *qatay > ade ‘liver’, *pitu > pidu ‘seven’,

8 *upi is a Proto-Central-Eastern Malayo-Polynesian innovation that replaced PMP *hiup (Blust 1993:283).

31

*kutu > hudu ‘louse’, *ma-buta > mbuda ‘blind’, but *taqi > ta

?

i ‘feces’, *teken ‘punting

pole’ > tiki ‘staff, cane’, *telu > tolu ‘three’, *tuqah > tua ‘old, of people’; *bataN >

≡

ata

‘tree trunk, log’, *qutek > iti ‘brain’, *tuktuk > tutu ‘knock, pound, beat’. PMP *-nt-

remained unchanged: *qali-metaq (> limeta > limta) > linta ‘paddy leech’, *mantalaq

‘Venus’ > ntara ‘star’, *ma-tasak > ntasa ‘ripe’, *pantaR > panta ‘platform bed’.

One form shows d/t variation: *talih > dari ‘rope from which things are suspended’, tari

‘bridle rope’. Another shows *t > s: *bities (> bitis) > wisi ‘calf of the leg’. Two others

show *t > n, almost certainly by fossilization of an earlier process of verbalizing nasal

substitution: *taNis > naNi ‘cry, weep’, *tuRun > ntu

?

u ~ nu

?

u ‘descend’.

*c. No reflexes of PMP *c have been identified in Bimanese.

*k. PMP *k is reflected as h or k, without stateable conditions: *kaRat > ha

?

a ‘bite’,

*kena > hina ‘hit, struck by s.t.’, *kikik ‘giggle’ > hihi ‘neigh, whinny’, *kenduR >

hondo ‘slack, of a rope’, *kunij > huni ‘turmeric’; *takut > dahu ‘afraid’, *aku > n-ahu

‘1s’, *laki ‘man, male’ > rahi ‘husband’, *paniki > panihi ‘flying fox’, *siku > cihu

‘elbow’, but *kaRaw > kao ‘scratch an itch’, *kedeN > ki



i ‘stand up’, *kiaw > ka-kio

‘peep, of chicks’, *kilat > ka-kila ai ‘lightning’; *bakaw > wako ‘mangrove’ *ma-bukuq

> mbuku ‘bent’. PMP *-Nk- generally became Ng: *biNkuN > biNgu ‘adze’, *kiNkiN >

kiNgi ‘little finger, pinky’, *laNkaN > laNga ‘step, stride’, *naNka > naNga ‘jackfruit’.

Two forms show unexplained loss of *k: *lakaw > lao ‘go’, *kuhkuh (> kuku) > uhu

‘nail, claw’. One other shows *k > N: *kali > Nari ‘dig up, excavate’. As with *taNis >

naNi ‘cry, weep’, the nasal of Nari is almost certainly the residue of a formerly active

process of verbalizing nasal substitution. Finally, one form shows no change to the

cluster *-Nk-: *Rakit (> raNkit) > raNki ‘raft’. This may be a borrowing of Malay rakit,

but if so prenasalization was innovated during or after the transfer between languages.

*q. PMP *q disappeared in initial position, but is regularly reflected as medial glottal

stop: *qaRta ‘outsider, alien person’ > ada ‘slave’, *qabu > awu ‘ash; hearth’, *qijuN >

ilu ‘nose’, *quzan > ura ‘rain’, *qubi > uwi ‘yam’; *paqet > pa

?

a ‘chisel’, *paqit > pa

?

i

‘bitter’, *taqi > ta

?

i ‘feces’, *tinaqi ‘intestines’ > na

?

e ‘stomach’, *biqak >

≡

i

?

a ‘split s.t.’,

mbi

?

a ‘crack open, as a hatching egg’, *tuquR ‘dry up’ > du

?

u ‘dry’, *puqun > fu

?

u,

mpu

?

u ‘base of a tree’, *tuqud > tu

?

u ‘stand’, ta-tu

?

u ‘knee’.

In one known form medial glottal stop disappeared: *tuqah ‘old, of people’ > dua

‘father’s elder brother, mother’s elder brother’. It is possible that *q disappeared between

a high vowel and a vowel of different quality, but if so some other explanation must be

offered for *biqak >

≡

i

?

a ‘split’.

*b. PMP *b became w, or occasionally ≡, without stateable conditions: *babuy > wawi

‘pig’, *beli > weli ‘buy’, *biRaq > wia ‘itching taro’, *buaq > wua ‘fruit’; *qabu > awu

‘ash; hearth’, *Rabiqi ‘evening’ > awi-na ‘yesterday’, *ka-labaw > karawo ‘rat’, *tabuni

> tawuni ‘placenta’, *sebu > suwu ‘douse a fire’, *ibah-an (> iban) > iwa ‘friend’, *qubi

> uwi ‘yam’, but *bataN >

≡

ata ‘tree trunk, log’, *bejbej >

≡

a

≡

a ‘tie by encircling’, *ma-

32

baseq ‘wet’ >

≡

eca, mbeca ‘wet s.t.’, *biqak >

≡

i

?

a ‘split’, *ma-baqeRu >

≡

ou ‘new’;

*labuq > la

≡

u ‘anchor’, *tebuh > do

≡

u ‘sugarcane’. PMP *mb or the similar cluster

derived by syncope did not change: *ma-baRani > mbani ‘brave’, *ma-besuR > mbocu

‘satiated’, *ma-buta > mbuda ‘blind’.

In one known case *b when prenasalized is reflected as p: *ma-beNel > mpiNa ‘deaf’.

*d. PMP *d generally became r, but occasionally became



in apparently native words,

without stateable conditions: *daRaq > ra

?

a ‘blood’, *demdem > rindi ‘dark’, *deNeR >

riNa ‘hear’, *dahun > ro

?

o ‘leaf’; *wada > wara ‘be, exist’, *anaduq > naru ‘long’,

*siduk > ciru ‘spoon, ladle’, *hadiRi > ri

?

i ‘housepost’, *ma-qudip > mori ‘living, alive’,

but *diNdiN >



indi ‘wall’, *duha >



ua ‘two’, *kedeN > ki



i ‘stand’. In one known case

*d is reflected as r in one affixed form of a morpheme, and as



in another: *m-uda >

moro ‘young’, *N-uda > No



a ‘young’. PMP *nd or the similar sequence that resulted

from secondary change became nd: *danaw > ndano ‘lake’, *diRuq > ndeu ‘bathe’,

*depa > ndupa ‘fathom’; *paNdan (> pandan) > fanda ‘pandanus’, *linduN > lindu ‘eel

sp.’, *demdem (> dendem) > rindi ‘dark’.

*z. PMP *z merged with *d, and so is reflected both as r and as



: *tuzuq > turu ‘point

at, point out’, *quzan > ura ‘rain’, but *zauq >



o

?

o ‘far’, *hazani >



eni ‘near’. In some

cases *z is reflected with prenasalization from an unknown source: *zaRum ‘needle’ >

ndau ‘needle’, nda

?

u ‘sew’.

*j. PMP *j occurred only in medial and final positions, and hence has a non-zero reflex

only intervocalically in Bimanese. In this position it is reflected four times as l, and five

times as r: *qapeju (> peju) > folu ‘gall (bladder)’, *qijuN > ilu ‘nose’, *pija > pila ‘how

much/how many?’, *hapejes (> pejes) > pili ‘painful’, but *huaji (> waji) > ari ‘younger

sibling’, *pajay > fare ‘riceplant, rice in the field’, *balija > lira ‘batten of the loom’,

*qalejaw (> lejo) > liro ‘sun’, *Najan > Nara ‘name’.

Although it may not be immediately obvious, the Bimanese reflexes of PMP *j are of

considerable interest. PMP *d and *j have merged in nearly all languages of eastern

Indonesia, but this apparently is not true of Bimanese. Neither *d nor *z is reflected as l

in any recognized etymology. This suggests that although *d and *z merged as r or

occasionally



, PMP *j merged with *l and then began to undergo the change of *l > r

which is incomplete for reflexes of PMP *j and *l.

*g. PMP *g is attested in a single form, where it became k: *gatel > kadi ‘itch, itchy’.

*m. PMP *m remained unchanged: *maRi > mai ‘come’, *m-atay > made ‘die, dead’,

*ma-qasiq > meci ‘mercy, pity’, *minak > mina ‘oil, grease’, *ma-qudip > mori ‘living,

alive’, *mula > mura ‘plant rice’; *ama > ama ‘father’, *Ramut > amu ‘root’, *qatimun >

dimu ‘cucumber’, *qumah > oma ‘cultivated field’, *Rumaq > uma ‘house’.

*n. PMP *n remained unchanged: *anaduq (> nadu) > naru ‘long’, *nipaq > ‘a palm:

Nipa fruticans’, *nipi > nifi ‘dream’, *nusa > nisa ‘island’; *anak > ana ‘child’, *anay >

33

ane ‘termite’, *hazani (*zani) >



eni ‘near’, *kena > hina ‘struck by’, *ina > ina

‘mother’, *b-in-ahi (> binay) > wine ‘female’, *tuna > duna ‘freshwater eel’.

*ñ. PMP *ñ merged with *n: *ñawa > nawa ‘breath’; *añam > N-ana ‘to plait’, *miñak

> mina ‘oil, grease’, *peñu > fonu ‘turtle’.

*N. PMP *N remained unchanged: *Najan > Nara ‘name’, *NaRaq > Nara ‘wild duck’,

*Nuda > Noda ‘young’; *saNa ‘bifurcation’ > saNa ‘branch’, *haNin > aNi ‘wind’, *ma-

RaNaw (> maaNaw) > maNo ‘dry’, *ma-beNel > mpiNa ‘deaf’.

*s. PMP *s became Bimanese c or s without stateable conditions: *sipsip > cici ‘insert,

slip between’, siku > cihu ‘elbow’, *siduk > ciru ‘spoon, ladle’, *siwa > ciwi ‘nine’;

*ma-qasin ‘salty’ > maci ‘sweet’, *baseq >

≡

eca/mbeca ‘to wet s.t.’, *isa > ica ‘one’,

*ma-besuR > mbocu ‘satiated’, but *salaq > sala ‘error’, *saleR > sari ‘bamboo floor’,

*sawa ‘python’ > sawa ‘snake’, *selsel > sinci ‘regret’, *sebu > suwu ‘douse a fire’, *si-

ia > sia ‘3s’, *qasiRa (> siRa) > sia ‘salt’; *ta-esak (> tasak) > ntasa ‘ripe’, *isi > isi

‘contents’, *nusa > nisa ‘island’, *susu > susu ‘breast’. Following a nasal *s became c:

*hasaN > N-anca ‘gills’, *qalesem (> lesem) > N-onco ‘sour’, *selsel (> sesel > sensel) >

sinci ‘regret’.

The s/c distinction in Bimanese apparently already existed at the time Malay loanwords

began to enter the language, since Malay c is invariably borrowed as c: Malay campur,

Bimanese campo ‘to mix’, Malay ceNkeh, Bimanese ceNke ‘clove’, Malay, Bimanese

cuka ‘vinegar’; Malay (< Chinese) daciN, Bimanese daci ‘scale, balance’, Malay (<

Sanskrit) kaca ‘glass’, Bimanese kaca ‘mirror; glass’, Malay kacau, Bimanese kaco

‘chaotic’, Malay (< Dutch) laci, Bimanese laci ‘drawer in chest of drawers’.

*h. PMP *h disappeared: *hapuy > afi ‘fire’, *haNin > aNi ‘wind’, *halas > ara ‘forest’,

*hipaR > ipa ‘other side’; *bahi ‘woman’ > wei ‘wife’, *duha >



ua ‘two’, *kahiw (>

kayu) > haju ‘wood/tree’, *wahiR > oi ‘water’, *luheq (> lueq > luq > ↔lu) > olu ‘tears’.

*l. PMP *l is reflected as Bimanese l or r without stateable conditions: *laNit > laNi

‘sky’, *lawaN > lawa ‘passageway, gate of a fence’, *lepet > lipi ‘fold’, *qalejaw (> lejo)

> liro ‘sun’, *lima > lima ‘five’, *linaw > lino ‘calm, of the surface of water’, *luab ‘boil

over’ > lua ‘pour’; *qateluR (> teluR) > dolu ‘egg’, *telu > tolu ‘three’, *abelaj > wela

‘broad, wide’, *beli > weli ‘buy’, *kilat > ka-kila ai ‘lightning’, but *laki > rahi

‘husband’, *qalima > rima ‘hand’, *qalu-hipan (> qalipan > lipan) > rifa ‘centipede’;

*halas > ara ‘forest’, *mantalaq ‘Venus’ > ntara ‘star’, *qahelu (> alu) > aru ‘pestle’,

*talih > dari ‘rope’, *walu > waru ‘eight’, *biliN > mbiri ‘revolve, turn’, *kulit > huri

‘skin’, *mula ‘to plant’ > mura ‘plant rice’, *sa-Na-puluq > mpuru ‘ten’, *qulu ‘head;

headwaters’ > uru ‘headwaters’, *bulan > wura ‘moon’, *buliR > wuri ‘head of grain’.

Two forms show variation between l and r: *balik > mbali ~ mbari ‘return, go back’,

*salaq > sala ‘error’, sara ‘miss the mark’. The form sala ‘error’ may derive from Malay

salah ‘wrong, in error’, since Bimanese sara ‘miss the mark’ appears to reflect *salaq,

and this form cannot so easily be explained as a product of borrowing. If there is any

34

possible condition for this split it can be noted that intervocalically following *e (schwa)

PMP *l invariably became l (four supporting cases, no exceptions).

In one known form *l is reflected as n: *lesuN > nocu ‘mortar’. If Nonco reflects PMP

*qalesem ‘sour’ then this form shows an irregular development *l > N (with sporadic

prenasalization of *s), or loss of *l followed by the addition of an unidentified affix.

*r. Reflexes of PMP *r are rare. The only examples identified in Bimanese are *rusuk

‘pierce, stab’ > rusu ‘skewer’, and *nara > nara ‘a tree: Pterocarpus indicus’.

*R. PMP *R generally disappeared, but is occasionally reflected as r in forms that do not

appear to be Malay loanwords: *Ramut > amu ‘root’, *Rumaq > uma ‘house’; *qaRta

‘outsider, alien person’ > ada ‘slave; I’, *waRi > ai ‘day’, *saRu > cau ‘comb’, *taRum

> dau ‘indigo’, *taRutuN ‘porcupine fish’ (> tautuN > tutuN) > dudu ‘porcupine’, *paRih

> fai ‘stingray’, *paRaw > ka-fao ‘hoarse’, *kaRat > ha

?

a ‘bite’ (with secondary glottal

stop), *maRi > mai ‘come’, *beRay > mbei ‘give’ (?), *ma-Ruqanay (> mauqanay >

muanay) > mone ‘male’, *tuRun ‘descend’ > ntu

?

u ‘to perch, of birds’, ntu

?

u ~ nu

?

u

‘descendants’, *laRiw ‘flee’ > rai ‘run, flow’, *hadiRi > ri

?

i ‘housepost’, *qasiRa (>

siRa) > sia ‘salt’, *biRaq > wia ‘itching taro’, but *Rakit > raNki ‘raft’; *qaRus > aru

‘current’ (< Malay (h)arus ‘current’?), *daRa ‘young girl, virgin’ > ndara ‘young female

animal’, mandara ‘young girl’, *NaRaq > Nara ‘wild duck’.

*w. PMP *w is normally reflected as w: *wanan > wana ‘right side’, *wada > wara ‘to

be’, *walu > waru ‘eight’; *siwa > ciwi ‘nine’, *kawil > hawi ‘fishhook’, *lawaN > lawa

‘passageway, gate of a fence’, *ma-huab (> mauab > mawab) > mawa ‘yawn’, *ñawa >

nawa ‘breath’, *sawa ‘python’ > sawa ‘snake’. In two known forms it appears to have

disappeared outright, and in another it effectively ‘disappeared’ through fusion: *waRi >

ai ‘day’, *wani > ani ‘honeybee’, *wahiR > oi ‘water’.

*y. PMP *y did not occur in initial position. Intervocalically it invariably became the

voiced palatal affricate j: *kahiw (> kaiw > kayu) > haju ‘wood/tree’, *layaR > loja

‘sail’, *ma-hiaq (> maiaq > mayaq) > maja ‘shy, embarrassed’, *bayu > mbaju ‘pound

grains’, *duyuN > ruju ‘dugong’. A similar change is seen in the substitution of j for y in

loanwords from Malay, as with paju ‘umbrella’ (Malay payuN)

*-ay. PMP *-ay was monophthongized to Bimanese –e: *qatay > ade ‘liver’, *anay >

ane ‘termite’, *pajay > fare ‘rice plant’, *m-atay > made ‘die, dead’, *ma-Ruqanay (>

mauanay > muanay) > mone ‘male’, *b-in-ahi (> binay) > wine ‘female’.

There is one possible exception: if Bimanese mbei derives from PMP *beRay ‘give’, it

appears that *-ay is reflected as –i in this form.

*-aw. PMP *-aw was monophthongized to Bimanese –o: *panaw > fano ‘leucoderma’,

*ilaw > ilo ‘lamp, torch’, *paRaw > ka-fao ‘hoarse’, *kiaw > ka-kio ‘to peep, of a chick’,

*kaRaw > kao ‘scratch an itch’, *lalaw > lalo ‘too much, exceed’, *lakaw > lao ‘go’,

35

*linaw > lino ‘calm, of water’, *qalejaw (> lejaw) > liro ‘sun’, *ma-RaNaw > maNo

‘dry’, *danaw > ndano ‘lake’, *babaw ‘top, uppermost part’ > wawo ‘top, peak’.

*-uy. PMP *-uy was monophthongized to Bimanese –i: *hapuy > afi ‘fire’, *babuy >

wawi ‘pig’.

*-iw. PMP *-iw merged with *-i and *-uy as Bimanese –i: *ma-baRiw ‘tainted’ > mbai

‘rotten’, *laRiw ‘flee’ > rai ‘run’.

*V3. As in nearly all other languages of eastern Indonesia, prepenultimate syllables that

began with a vowel or a laryngeal (*q or *h) disappeared in Bimanese: *qatimun > dimu

‘cucumber’, *qateluR > dolu ‘egg’, *hazani >



eni ‘near’, *qapeju > folu ‘gall’, *qalejaw

> liro ‘sun’, *anaduq > naru ‘long’, *qaninu > ninu ‘shadow’, *hapejes > pili ‘painful’,

*qalu-hipan (> qaluipan > qalipan) > rifa ‘centipede’, *hadiRi > ri

?

i ‘housepost’, *qasiRa

> sia ‘salt’. Although Bimanese does not have a schwa, and permits prepenultimate a,

this change suggests that prepenultimate vowel contrasts were once neutralized as schwa,

and that prepenultimate schwa was not allowed in initial syllables. In the few available

examples prepenultimate vowels that were not word-initial are reflected as a: *ka-labaw

> karawo ‘rat’, *tabuni > tawuni ‘placenta’.

*a > a. PMP *a generally remained unchanged in all positions: *qaRta ‘outsiders, alien

people’ > ada ‘slave’, *qapuR > afu ‘lime’, *wani > ani ‘honeybee’, *qahelu (> alu) >

aru ‘pestle’, *saRu > cau ‘comb’, *tuqah > dua ‘father’s elder brother’, *paRih > fai

‘stingray’, *wanan > wana ‘right side’.

*a > o. PMP *a(C)u is reflected as o, o

?

o or ou if the consonant disappeared. Cases with

fusion include *ma-Ruqanay (> mauqanay > muane) > mone ‘male’, and *ma-qudip (>

maudip > maurip) > mori ‘living, alive’. Cases without fusion include *zauq >



o

?

o ‘far’,

*pauq > fo

?

o ‘mango’, *qauR > o

?

o ‘bamboo’, *dahun > ro

?

o ‘leaf’, *baRut ‘crop, throat

of a bird’ > wo

?

o ‘neck’, and *baqeRu (> baRu > bau) >

≡

ou ‘new’, *tau > dou ‘person’,

*bahuq > wou ‘stench’. In at least one form *w and *a also fused to o: *wahiR > oi

‘water’. Finally, in one form that shows fusion of *a and *u, a low vowel in the next

syllable is raised and rounded to o, and in another form *a is raised to o without any clear

conditioning: *ma-uda > moro ‘young’, *layaR > loja ‘sail’.

It is clear that most of these changes (all but *layaR > loja) are conditioned. However,

despite the evidence of conditioning other lexical items that contain *wa-, or *-aCu with

loss of the consonant show no change of the low vowel (cp. *wahiR > oi ‘water’, but

*wanan > wana ‘right side’, *baqeRu (> baRu > bau) >

≡

ou ‘new’, but *saRu > cau

‘comb’ or *taRum > dau ‘indigo’). The most common pattern appears to be *-a(C)u >

-o

?

o, seen in the etymologies for ‘far’, ‘mango’, ‘bamboo’, ‘leaf’ and ‘neck’. Somewhat

less common is the pattern *-a(C)u > -ou, seen in the etymologies for ‘new’, ‘person’,

and ‘stench’. Least common is the pattern *-a(C)u > -au, as seen in ‘comb’, and ‘indigo’.

It is noteworthy that *-aCu never shows rounding assimilation if the consonant is

retained: *qapuR > afu ‘lime’, *Ramut > amu ‘root’, *qabu > awu ‘ash; hearth’, *takut >

dahu ‘afraid’, etc. Likewise, there is no rounding assimilation of *-u(C)a, even if the

36

consonant is lost: *tuqah > dua ‘father’s elder brother’, *duha >



ua ‘two’, *uRat > ka-

u

?

a ‘vein, vessel’, *buaq > wua ‘fruit’, etc.

*a > e. In a small number of forms *a has been raised and fronted preceding a syllable

that contained *i: *hazani >



eni ‘near’, *ma-qasiq (> masiq) > meci ‘mercy, pity’, *bahi

> wei ‘wife’. As with *a > o/_Cu, this change appears to be conditioned, but is irregular:

*waRi > ai ‘day’, *paRih > fai ‘stingray’, *maRi > mai ‘come’, *paqit > pa

?

i ‘bitter’,

*laRiw ‘flee’ (> laRi > lai) > rai ‘run’. Also, there is no fronting of *-i(C)a, even if the

consonant is lost: *si ia > sia ‘3s’, *qasiRa > sia ‘salt’, *biRaq > wia ‘itching taro’, etc.

In a single example *a became e without apparent conditioning: *baseq ‘wet’ >

≡

eca ~

mbeca ‘to wet something’.

*e > zero. PMP *e (schwa) has multiple reflexes in Bimanese, with little evidence of

conditioning. The one reflex that is conditioned is loss of *e adjacent to another vowel,

but this change is shared with almost all languages outside Taiwan and the Philippines,

and almost certainly happened long before the existence of Bimanese as a separate

language: *qahelu > aru ‘pestle’, *baqeRu (> baeRu > baRu) >

≡

ou ‘new’, *ma-buhek >

mawu ‘drunk’, *luheq (> lue > lu > ↔lu) > olu ‘tears’, *pa-kaen (> pakan) > paha ‘feed’.

*e > a. In a small number of forms *e is reflected as a: *baseq ‘wet’ > mbeca ‘to wet

s.t.’, *taneq > dana ‘earth’, *bejbej >

≡

a

≡

a ‘wind around’, *beNel > mpiNa ‘deaf’, *peRes

> pua ‘squeeze out’. In at least the first two examples it is likely that the sequence *-eq

first became *–aq, a conditioned change that has affected many languages. It is possible

that the last example is of the same type, since its etymology is ambiguous for *peRes or

*peReq ‘squeeze out’.

*e > i. This is one of the two most common Bimanese reflexes of the PMP schwa,

affecting the mid-central vowel in all positions: *enem > ini ‘six’, *kena > hina ‘struck

by’, *qutek > iti ‘brain’, *gatel > kadi ‘itch’, *kedeN > ki



i ‘stand’, *lepet > lipi ‘fold’,

*qalejaw (> lejaw) > liro ‘sun’, *hapejes (> pejes) > pili ‘painful’, *demdem > rindi

‘dark’, *saleR > sari ‘bamboo floor’, *selsel > sinci ‘regret’, *teken > tiki ‘cane, staff’.

*e > e. Bimanese e (a mid-front vowel) is a rare reflex of PMP *e (schwa): *beRay >

mbei ‘give’, *abelaj (> belaj) > wela ‘wide’, *beli > weli ‘buy’. Bimanese wela ‘wide’

and wira ‘spread out’ appear to be reflexes of the same form, suggesting that at least one

of them is a loan. However, it is possible that the latter form reflects PMP *belaj ‘spread

out, dry in the sun’ or its doublet *bilaj.

*e > o. This is the second most common reflex of PMP schwa, appearing in nearly as

many forms as *e > i: *tebuh > do

≡

u ‘sugarcane’, *qateluR (> teluR) > dolu ‘egg’,

*qapeju (> peju) > folu ‘gall’, *peñu > fonu ‘turtle’, *kenduR > hondo ‘slack, as a rope’,

*ma-besuR > mbocu ‘satiated’, *luheq (> lue > lu > ↔lu) > olu ‘tears’, *qalesem > Nonco

‘sour’ (?), *empu > ompu ‘grandfather’, *telu > tolu ‘three’, *betu ‘callus, blister’ >

wodu ‘swell up slowly’.

37

There is a tendency for *e > o to occur adjacent to labial consonants in contrast with *e >

i (6 cases out of 11, as opposed to 3 cases out of 11), but no regularity can be established.

*e > u. This is a rare reflex, but one that is attested in some very secure etymologies:

*depa > ndupa ‘fathom’, *peRes > pua ‘squeeze out’, *sebu > suwu ‘douse a fire’, *epat

> upat ‘four’. All four examples are found in the penult, but this does not ensure

predictability, since PMP *e is often reflected as other vowels in this position as well.

*i. PMP *i is invariably reflected as i: *haNin > aNi ‘wind’, *qapit > api ‘pinch’, *siku >

cihu ‘elbow’, *talih > dari ‘rope’, *kulit > huri ‘skin’, *ijuN > ilu ‘nose’, *linaw > lino

‘calm, of water’, *nipis > nipi ‘thin, of materials’, *niuR > ni

?

u ‘coconut’, etc.

*u. PMP *u is reflected as u in nearly all cases: *siduk > ciru ‘spoon, ladle’, *taRutuN

‘porcupine fish’ (> tautuN > tutuN) > dudu ‘porcupine’, *tuna > duna ‘eel’, *kutu > hudu

‘louse’, *luqar > lua ‘outside’, *bayu > mbaju ‘pound rice’, *puluq > mpuru ‘ten’, *susu

> susu ‘breast’, *butuq > wudu ‘penis’, etc.

In one known case *u is inexplicably reflected as i: *nusa > nisa ‘island’. In another it is

reflected as o, apparently as the result of assimilation: *kenduR > hondo ‘slack, as a

rope’. The latter example agrees with the recurrent change *-a(C)u > -o

?

o in suggesting a

tendency to avoid high and mid back vowels within a word/morpheme. Like many

changes in Bimanese, however, this one is irregular, since high and mid back vowels do

occur in other forms (*tebuh > do

≡

u ‘sugarcane’, *qateluR > dolu ‘egg’, etc.).

Secondary glottal stop. As noted above, a secondary glottal stop developed between

derived sequences of mid-back vowels, as in *zauq > do

?

o ‘far’ or *pauq > fo

?

o ‘mango’.

A similar change affected other vowel sequences, as with *niuR > ni

?

u ‘coconut’, and

*buaq > u

?

a ‘betel nut’, as well as forms that have lost *R: *kaRat > ha

?

a ‘bite’, *daRaq

> ra

?

a ‘blood’, *hadiRi > ri

?

i ‘housepost’ and *uRat > ka-u

?

a ‘vein, vessel’.

A failed canonical conspiracy? Bimanese shows two peculiarities that suggest a failed

canonical conspiracy. Several trisyllabic loanwords have added a-: Malay (Arabic)

dunia, Bimanese adunia ‘world’, Malay durian, Bimanese aduria ‘durian’, Malay

(Arabic) juma

?

at, Bimanese ajuma

?

a ‘Friday, day of formal prayer’, Malay mentua

‘parent-in-law’, Bimanese amantua ‘grandfather’, Malay (Persian) nak(h)oda, Bimanese

anakoda ‘captain of a boat crew’, Malay rebana, Bimanese arubana ‘small drum, tabor’,

Malay (Arabic) rezeki, Bimanese arujiki ‘sustenance, means of livelihood’, Malay

(Hindi) rupiah, Bimanese arupia ‘rupiah’. Epenthesis does not affect loanwords which

are disyllabic: Malay badak, Bimanese bada ‘rhinoceros’, Malay (Dutch) dasi, Bimanese

dasi ‘necktie’, Malay jaguN, Bimanese jago ‘corn, maize’, Malay (Arabic) mawar,

Bimanese mawa ‘rose water’, Malay, Bimanese raga ‘rattan ball used in sport’, Malay

rokok, Bimanese roNko ‘cigarette’.

Vowel epenthesis is not unusual in AN historical phonology, but is almost always used to

restore a lost disyllabism in content morphemes that have been reduced to monosyllables

by sound change (Blust 2007). Trisyllabic bases are rare, and quadrisyllables almost

38

unknown in most AN languages. What, then, would motivate vowel epenthesis in

trisyllables, but not in disyllables, or in quadrisyllables such as Malay, Bimanese

manusia ‘mankind’? It is noteworthy that some native trisyllables show irregular loss of

the first syllable: *balimbiN > limbi ‘starfruit: Averrhoa carambola’, *balija > lira

‘batten of the loom’, *minaNa > naNa ‘estuary, river mouth’, *dalikan ‘trivet’ > riha

‘hearth’. What seems to unite epenthesis and syllable deletion is a preference for bases

with an even number of syllables. However, forms such as *paniki ‘flying fox’ > panihi

‘cave bat’, *qaninu ‘shadow, reflection’ > s-aninu ‘mirror’, or Malay (Portuguese),

Bimanese sapatu ‘shoe(s)’ show that the effort to satisfy this target (if that is the correct

account of the facts) was not sufficiently thoroughgoing to achieve the desired effect.

5. Manggarai. Manggarai is spoken in a number of dialects over the western quarter of

the island of Flores (see supplementary map at end of Verheijen 1967). Verheijen and

Grimes (1995) note that there are “Approximately forty-three subdialects of Manggarai”

that “cluster into five clear dialect groupings: West Manggarai, West-Central Manggarai,

Central Manggarai, East Manggarai, and Far-east Manggarai. The latter is located in

north-central Flores, separated from the other Manggarai dialects by the Rembong

language.” Manggarai appears to subgroup with other languages of western and central

Flores, as Keo, Ngadha, Riung, Palu’e, Ende, and Lio. Like other areas of the western

Lesser Sundas, Manggarai was subject to the domination of the Makasarese-speaking

kingdom of Goa for some centuries, as well as to the sultanate of Bima after about 1727.

5.1. Synchronic phonology. Allowing for minor orthographic changes, Verheijen and

Grimes (1995:586) give the phoneme inventory of Central Manggarai as follows:

TABLE 8: The phoneme inventory of Manggarai

Consonants Vowels

p t c k ?i u

b d j g

mp nt nc Nk é e o

mb nd nj Ng

m n N

s h

l a

r

w y

Prenasalized stops occur initially and medially, and n does not occur word-finally. This

source gives v rather than w, which is said to be “a lightly fricativised viced labio-dental

fricative”, but since Verheijen (1967), which is the primary source of lexical data, uses w

I follow the orthography of the dictionary here. Little information is given on phonetic

detail, but the consonants appear to have their expected values, and, following an

orthographic convention adopted from Indonesian, e represents the schwa.

39

Stress is penultimate unless this vowel is schwa followed by a simplex (non-prenasalized)

consonant, and there is no rule of rightward stress shift when postclitic elements are

incorporated into the phonological word (Manggarai reportedly has no affixes of any

kind). Little else needs to be mentioned here except perhaps that the vowels of

prepenultimate proclitics often become schwa, as in [ča:] ‘one’, but [č↔-mpulu] ‘ten’.

5.2. Historical phonology. Manggarai historical phonology is relatively transparent.

Because of the size of the primary source, a Manggarai-Indonesian dictionary of 772

pages in double columns and ten point type (Verheijen 1967), as well as the lexically

conservative nature of this language, a very large number of etymologies is available, and

only a small sampling of the relevant material is presented here.

Unlike many other languages of eastern Indonesia, Manggarai preserves most final

consonants. The principle qualification to this statement is that stops undergo final

devoicing, and nasals show a strong tendency to merge as –N (exceptionless for *-n,

sporadic for *-m). Like several other languages of the region, medial clusters were

preserved if they consisted of a prenasalized obstruent, but otherwise were simplified by

elimination of the first member (*punti > punti ‘banana’, *qeNkem ‘enclose’ > eNkem

‘clasp, embrace’, but *suksuk > cucuk ‘insert’, *testes > tetas ‘rip, tear, break open’).

*p. Non-finally PMP *p is reflected as p: *paNdan ‘pandanus’ > pandaN ‘pineapple’,

*padem (> p↔dam) > pesa ‘extinguish a fire’, *pecel > pecel ‘squeeze in the hand’, *pitu

> pitu ‘seven’, *pudun > puruN ‘ball of string’; *apa > apa ‘what?’, *epat > pat ‘four’,

lepet > lepet ‘wrap, fold over’, *hipaR > ipar ‘spouse’s sibling, sibling’s spouse’, *upas

> upas ‘poisonous’. Few examples of final *p are available, and these show –p or zero:

*kilap > hilap ‘flash, sparkle’, *silap > cilap ‘dazzled, of the eyes’, *ukup > ukup ‘brood,

sit on eggs’, but *qatep > até ‘thatch, roof’, *ma-qudip (> mudip) > mosé ‘living, alive’.

Although mosé is said to be general to Manggarai dialects, até is reported only for the

Patjar dialect. PMP *-mp- remained unchanged: *hampil ‘go together’ > ampil ‘stay

together’, *empeN ‘block, obstruct’ > empeN ‘restrain a person who wants to fight’.

*t. PMP *t is invariably reflected as Manggarai t: *taqi > ta

?

i ‘feces; defecate’, *tahep-i

> tepi ‘winnow’, *telu > telu ‘three’, *qatimun > timuN ‘cucumber’, *titis > titis ‘drip’,

*tunu > tunu ‘burn’; *qaRta ‘outsider, alien person’ > ata ‘person’, *terter > tetir

‘tremble’, *betas ‘tear, rip’ > wetas ‘cut through (rope, etc.)’, *pitu > pitu ‘seven’, *kutu

> hutu ‘hair louse’; *epat > pat ‘four’, *lepet > lepet ‘fold’, *laNit > laNit ‘sky’, *bunut >

wunut ‘coconut husk’. PMP *-nt- remained unchanged: *bintaNuR > ntaNor ‘a tree:

Calophyllum sp.’, *punti > punti ‘banana’.

In three known forms *-t is reflected as –k: *zaqit > ja

?

ik ‘sew’, *paqet > pa

?

ak ‘chisel’,

*uRat > urak ‘vein, vessel’. In another, *-t is reflected as –k in some dialects, and as –t

in others: *laNit > laNik ~ laNit ‘sky’. Manggarai pa

?

ak ‘chisel’ shows a for expected

schwa in the ultima, an irregularity which may indicate borrowing from Malay pahat.

*c. Reflexes of PMP *c are not found in basic vocabulary, but occur in a number of

Manggarai words. Where they have been identified they show merger with *s as c:

40

*caket > caket ‘stick to; viscous’, *cepuk ‘dull sound’ > cepuk ‘break (teeth, tubers),

crunch (corn cob)’, *cerik > cerik ‘scream shrilly’; *pecak (doublet *peceq) > pecak

‘hatch from an egg’.

*k. Prevocalically PMP *k became k or h without stateable conditions (but *k > -h- is

rare): *kaNeqa ‘fissured, slightly cracked’ > kaNa ‘open, of the mouth’, *kamuniN >

kemuniN ‘a tree: Murraya exotica’, *kenas ‘preserved meat or fish’ > kenas ‘roasted meat

preserved in a bamboo container’, *kilep ‘glimpse’ > kilep ‘a moment; pass by rapidly’,

*kukun > kukuN ‘a tree: Schoutenia ovata’; *aku > aku ‘1s’, *laki > laki ‘male, man’,

*ekeb ‘cover’ > ekep ‘brood, sit on eggs’, *siku > ciku ‘elbow’, *bituka > tuka ‘stomach,

intestines’, *buka > wuka ‘to open’, but *kami > hami ‘we (excl.)’, *kena > hena ‘struck

by’, *kilat > hilat ‘lightning’, *kima ‘giant clam’ > hima ‘oyster’; *ekap > hap ‘gasp for

air’. Word-finally *k is reflected as k or zero (< -h): *anak > anak ‘child’, *qutek > utek

‘brain; marrow’, *sisik > cicik ‘to shell, to scale (as fish)’, but *minak > mina ‘oil,

grease’. PMP *-Nk- is usually retained without change, but sometimes shows postnasal

voicing: *baNkal ‘a tree: Nauclea sp.’ > baNkal ‘tree sp.’, *waNka > waNka ‘canoe,

boat’, *buNkaR > wuNkar ‘swell up and split open, as the soil when something pushes up

from below’, but *waNkaN ‘spread legs apart’ > waNgaN ‘walk with legs wide apart’.

In two forms *k- disappeared before u, and w- developed as an onglide: *kuku > huku ~

wuku ‘nail, claw’, *kunij > wunis ‘turmeric’. In one other it voiced intervocalically:

*wakaq > waga ‘split’.

*q. PMP *q is lost in initial and final positions, but is retained intervocalically: *qaRa >

ara ‘a tree: Ficus sp.’, qatay > ati ‘liver’, *qeRut > erut ‘tighten, bind tightly’, *qilaw

‘torch; illuminate’ > ilo ‘clear, bright’, *qulu > ulu ‘head’; *paqit > pa

?

it ‘bitter’, *taqan

> ta

?

aN ‘set a trap’, *biqak > wi

?

ak ‘to split, as bamboo’, *tuqah > tu

?

a ‘old, elder’, *tuqu

> tu

?

u ‘true’; *daRaq > dara ‘blood’, *baseq > baca ‘wet’, *piliq > pilé ‘to choose’,

*puluq > pulu ‘ten’.

*b. In non-final position PMP *b is is reflected as Manggarai b or w without stateable

conditions. Although statistical data are not available, *b > w appears to be far more

frequent than *b > b: *baseq > baca ‘wet’, *beNel ‘deaf’ > beNel ‘speechless, unable to

make a sound’, *bilaN > bilaN ‘to count’, *buta > buta ‘blind’; *Rabut > rabut ‘pluck,

pull out’, *lebet > lebet ‘dense, luxuriant (of vegetation)’, but *babaq > wa ‘under’,

*bahaq > wa

?

a ‘flood’, *balu > walu ‘widow(er)’, *beli > weli ‘buy’, *beRas ‘husked

rice’ > weras ‘seed, grain’, *betas ‘tear, rip’ > wetas ‘cut (rope, etc.)’, *bisul > wicul

‘boil, abscess’, *biNis > wiNis ‘show the teeth’, *bibiR > wiwir ‘lip’, *buaq > wua

‘fruit’, *buhek > wuk ‘head hair’, *bulan > wulaN ‘moon’; *qabu ‘ashes’ > awu ‘sparks;

ashes’, *labaw > lawo ‘mouse, rat’, *Ribu ‘thousand’ > riwu ’10,000’, *qubi > uwi

‘yam’. A few morphemes with initial and intervocalic *b show b-, but -w-: *ba-labaw >

b

↔

lawo ‘rat, mouse’, *bubun ‘spring, source’ > buwuN ‘well’, *buNbuN ‘ridge of the

roof’ > buwuN ‘peak’. Occasionally, b and w are in free variation: ba ~ wa ‘carry, bring’.

Word-finally PMP *b either devoiced or disappeared: *ekeb ‘cover’ > ekep ‘brood, sit on

eggs’, *ma-huab > N-oap ‘yawn’, *luab > lua ‘boil over, as rice in the pot’. The cluster

41

*mb remained unchanged: *hu(m)bun > umbuN ‘heap or pile up’, *imbaN > imbaN

‘weigh’.

*d. Word-initially PMP *d split unconditionally into d or s: *daRaq > dara ‘blood’,

*deNeR > deNé ‘hear’, *diqa > di

?

a ‘good’, but *daki ‘dirt on the body’ > saki ‘dirty’,

*danaw > sano ‘lake’, *dapuR > sapo ‘hearth, cooking place’,9 *dahun > sauN ‘leaf’,

*duha > sua ‘two’. Intervocalically and finally *d generally became s: *kedeN > hesé

‘stand’, *teduN > tesoN ‘take shelter’, *ma-qudip (> maudip > modip) > mosé ‘living,

alive’, *ma-udehi (> maudi > mudi) > musi ‘after, behind’, *budaq > wusa ‘foam,

bubbles’; *natad > natas ‘yard, cleared area near house’, *tuqud > tu

?

us ‘knee’. The

cluster *-nd- remained unchanged: *anduN > anduN ‘ancestor’, *paNdan ‘pandanus’ (>

pandan) > pandaN ‘pineapple’.

In one known form *d became –r-: *qudaN > k-uraN (but also kusé) ‘shrimp, lobster’.

*z. PMP *z usually became s, but sometimes j or d: *zalan > salaN ‘path, road’; *pezem

> peseN ‘close the eyes’, *tuzuq > toso ‘point out’, *quzan > usaN ‘rain’, but *zaqit >

ja

?

ik ‘sew’, *zemak > demak ‘feel, grope’, *haRezan (> Rezan) > redaN ‘ladder’.

*j. PMP *j became –s-: *huaji (> waji) > asé ‘younger parallel sibling’, *qalejaw (>

lejaw) > leso ‘day, sun’, *Najan > NasaN ‘name’, *qapeju > pesu ‘gall’, *pija > pisa

‘how much/how many?’. Word-finally *j is reflected as –r in two cases, but as –s and –t

in one each: *hapij > d-apér ‘twin’, *qunej ‘pith of plants’ > uner ‘palm pith; marrow’,

but *kunij > wunis ‘turmeric’, *pusej > putes (met.) ‘navel’.

*g. PMP *g became g, h or occasionally k in Manggarai, with no apparent conditions:

*gagar > gagar ‘brave, plucky’, *gatuk > gatuk ‘chatter, of the teeth’, *gerger > geger

‘shiver with chills; tremble’, *gerit > ‘sound of scratching or squeaking’ > gerit ‘scratch,

claw; scream’; *agu > agu ‘and, also’, but *gikgik > hihik ‘sound of laughter’, *bagal

‘overgrown, clumsy’ > bagal ~ bahal ‘obese, fat’, *tegas > tehas ‘clear, straightforward’,

*gatel > katel ‘itch, itchy’. Word-finally *g appears as –k: *Reneg > renek ‘press down’.

*m. PMP *m did not change in non-final position: *mata > mata ‘eye’, *ma-qeti > meti

‘evaporate, dry up’, *minak > mina ‘oil, grease’, *ma-udehi (> maudi > mudi) > musi

‘after, behind’, *muntay > munta ‘citrus fruit’; *ama ‘father; father’s brother’ > ama-N

‘mother’s brother, wife’s father, husband’s father’, *mamaq > mama ‘chew’, *lima >

lima ‘five’, *tumah > tuma ‘clothe’s louse’, *quma > uma ‘swidden’. In final position it

is reflected as –m or –N: *dalem > délem ‘deep’, *enem > enem ‘six’, *elem ‘shade,

darkness’ > lem ‘new moon; dark’, *tanem > tanem ‘bury a corpse’, but *inum > inuN

‘drink’, *zaRum > jaruN ‘needle’. One known reflex shows doublets: *anam > n-anam ~

n-anaN ‘to plait, as mats’.

*n. In non-final position PMP *n remained unchanged: *natad > natas ‘yard, cleared

area around a house’, *nipi > nipi ‘dream’, *niuR > nio ‘coconut’, *nusa > nuca/nunca

‘island’; *taneq > tana ‘earth, land’, *tenun > tenuN ‘weave’, *ina-y > iné ‘mother’,

9 Manggarai dapur ‘kitchen’ appears to be a Malay loan.

42

*nunuk > nunuk ‘banyan’. Word-finally *n normally merged with *N: *wanan > wanaN

‘right side’, *kaen > haN ‘eat’, *halin > aliN ‘change, transfer’, *hikan > ikaN ‘fish’,

*puqun > pu

?

uN ‘origin’.

In one known form it disappeared: *paNdan (> pendan) > pedé ‘pandanus’.

*ñ. PMP *ñ merged with *n: * ñamuk > namok ‘mosquito’; *peñu > penu ‘turtle’.

*N. PMP *N usually remained unchanged: *NaRaq > Nara ‘wild duck’, *Najan >

NasaN ‘name’, *NeN > NeN ‘buzz, hum’, *NisNis > NiNis ‘grin, show the teeth’; *laNit

> laNik ~ laNit ‘sky’, *ceNap > ceNap ‘catch in the mouth’, *buNas ‘first, first fruits’ >

wuNas ‘first, just now’; *awaN > awaN ‘atmosphere, space between earth and sky’,

*liaN > liaN ‘cave’, *keNkeN ‘shrink, as clothing’ > keNkeN ‘embrace tightly’, *qijuN >

isuN ‘nose’.

In one known case final *N disappeared: *kedeN > hesé ‘stand’. In another it

disappeared in one of two apparent reflexes of a proto-form: *qudaN > kusé (next to

kuraN) ‘shrimp’.

*s. Non-finally PMP *s became c: *salaq > cala ‘wrong’, *sepaq > cepa ‘chew betel’,

*siku > ciku ‘elbow’, *susu > cucu ‘breast’; *asu > acu ‘dog’, *baseq > waca ‘wash’,

*esa > ca ‘one’, *isi > ici ‘contents’, *nusa > nuca ~ nunca ‘island’. In final position *s

always became –s: *qapelas > pelas ‘tree with leaves like sandpaper: Ficus sp.’, *deRes

> nderes ‘flow swiftly’, *bities > witis ‘calf of the leg’, *anus > nus ‘smoke’.

*h. PMP *h disappeared in all positions: *hapuy > api ‘fire’, *hasek ‘dibble’ > acek

‘drive in a stake or post’, *heRet > eret ‘firm, tight, of binding’, *hipaR > ipar ‘spouse’s

sibling, sibling’s spouse’, *huaR > uar ‘a vine: Flagellaria indica’, *huaji > asé ‘younger

parallel sibling’; *wahiR > aé ‘water’, *kahiw > haju ‘wood’, *duha > sua ‘two’, *bahuq

> wau ‘stench’; *talih > tali ‘rope’, *kuhkuh > wuku ‘nail, claw’.

*l. PMP *l became Manggarai l: *lawi > lawi ‘tail feathers’, *qalejaw > leso ‘sun’,

*laqia > lia ‘ginger’, *liaN > liaN ‘cave’, *lulun > luluN ‘roll up, as a mat’; *alaq > ala

‘fetch, get’, *eliN ‘high-pitched sound’ > liN ‘to sound, as a flute’, *telu > telu ‘three’,

*kilat > hilat ‘lightning’, *qulu > ulu ‘head’; *balbal ‘hit with a stick’ > babal ‘whip, to

whip’, *pecel > pecel ‘squeeze, crush with fingers’, *etel ‘constipation’ > tel ‘packed

hard and firm’, *dempul > dempul ‘blunt, dull’.

In one known case *l- is unaccountably reflected as N-: *lesuN > NencuN ‘rice mortar’.

*r. PMP *r is reflected as r: *rawan ‘strong emotion’ > rawaN ‘sad, anxious’, *reNus

‘wrinkled, frown’ > reNus ‘knit the brows’, *riqas > ri

?

as ‘crack, split’, *rumpak >

rumpak ‘collide with s.t.; *barik > barik ‘striped, streaked’, *periN > periN ‘a bamboo:

Bambusa vulgaris’; *balar ‘scratch or wale on the skin’ > balar ‘long, of scratch mark’,

*gerger > geger ‘shake, shiver with chills’.

43

*R. In non-final position PMP *R is reflected as r: *Ratus > ratus ‘hundred’, *Remek

‘crush, pulverize’ > remek ‘soft, of earth around a termite mound’, *eRiqi (> Riqi) > ri

?

i

‘sword grass, Imperata cylindrica’, *RuNkeb > ruNkep ‘cover over, cover up’; *baRah >

wara ‘glowing ember’, *habaRat ‘west monsoon’ > warat ‘rainy season’, *deRes >

nderes ‘flow swiftly’, *baqeRu > weru ‘new’, *hiRup > irup ‘sip’. Word-finally *R is

reflected as r or zero without known conditions: *banaR > wanar ‘a thorny vine: Smilax

sp.’, *bibiR > wiwir ‘lip’, *bintaNuR > ntaNor ‘a tree: Calophyllum sp.’, *bauR ‘trigger

on spring trap’ > baur ‘spring-set trap’, *huluR > ulur ‘lower by rope’, but *ikuR > iko

‘tail’, *kuluR > kolo ‘breadfruit’, *wahiR > waé ‘water’, *wakaR > waké ‘root’.

As in all or nearly all CEMP languages, *R has irregularly disappeared in *maRi > mai

‘come’. In one other form it is inexplicably reflected as s: *baRu > waso ‘hibiscus’.

*w. PMP *w generally was retained in non-final position: *wahiR > waé ‘water’,

*wakaR > waké ‘root’, *walu > walu ‘eight’, *wani > wani ‘bee sp.’, *wiNis > wiNis

‘open the lips so the teeth show through’; *hawaN > awaN ‘atmosphere, space between

earth and sky’, *tawa > tawa ‘laugh’, *lawi > lawi ‘tail-feathers of a rooster’.

In one known form a historically secondary *w disappeared in initial position: *huaji (>

waji) > asé ‘younger parallel sibling’.

*y. PMP *y did not occur initially. Intervocalically, *y became Manggarai j: *kahiw (>

kayu) > haju ‘tree, wood’, *layaR > lajar ‘sail’, *ma-hiaq (> mayaq) > maja ‘shy,

embarrassed’, *bayuR > wajur ‘a tree: Pterospermum sp.’.

*-ay. As noted in Blust (1993:265) PMP *-ay generally became –a: *m-atay > mata ‘die,

dead’, *Ruqanay (> ruanay > ronay) > rona ‘male, man’, *b-in-ahi (> binay) > wina

‘female, woman’. In one known case it is reflected instead as –i: *qatay > ati ‘liver’.

*-aw. PMP *-aw became –o: *ba-labaw > belawo ‘rat’, *lakaw > lako ‘go, walk’,

*qalejaw > leso ‘day; sun’, *panaw > pano ‘leucoderma’, *takaw > tako ‘steal’.

*-uy. In the only known reflex PMP *-uy became Manggarai –i: *hapuy > api ‘fire’.

*-iw. PMP *-iw became Manggarai –i: *baliw ‘dual division; moiety’ > bali ‘friend;

enemy; divide in halves’, wali ‘return in kind (gifts, etc.)’, *baRiw ‘tainted, of food left

uneaten for too long’ > wari ‘poisonous, dangerous’.

V3. Antepenultimate vowels merged as schwa. Where a schwa came to precede another

vowel it dropped: *buqaya (> w↔qaya > waya) > waja ‘crocodile’, *baqeRu (> b↔q↔Ru >

b↔↔ru) > weru ‘new’, *binehiq (> win↔iq > win↔i) > wini ‘seed’,

As in all other languages considered here, prepenultimate syllables that began with a

vowel, *q or *h disappeared in Manggarai: *qalejaw > leso ‘day; sun’, *qalu-hipan >

lipaN ‘centipede’, *anuliN > nuliN ‘a fruit-bearing plant: Pisonia umbellifera’, *qapelas

> pelas ‘tree with leaves like sandpaper: Ficus sp.’, qapeju > pesu ‘gall’, *haRezan >

44

redaN ‘ladder’, *qani-Ruan > ruaN ‘small bee sp.: Apis indica’, *adamay > sama ‘a tree:

Pipturus argenteus’, *qateluR > telo ‘egg’. This suggests that prepenultimate vowel

contrasts were neutralized as schwa at an earlier stage in the history of the language, and

that initial schwa dropped three or more syllables from the end of the word.

*-a. PMP *a is normally preserved unchanged: *aku > aku ‘1s’, *baRani > rani ‘brave,

bold’, *laNu ‘dizzy, intoxicated’ > laNu ‘drunk’, *tapa > tapa ‘smoke or dry food to

preserve it’, *bahaq > wa

?

a ‘flood’.

In six known forms it unaccountably became é: *dalem > délem ‘deep’, *zauq > déu

‘far’, *qudaN > kusé (next to kuraN) ‘shrimp’, *lima > limé ‘hand’ (next to lima ‘five’),

*haRezan > redé (next to redaN) ‘ladder’, *wakaR > waké ‘root’.

*-e. PMP initial *e usually disappeared, but is retained in some forms. It is often lost

before an obstruent and retained before a sonorant, but counter-examples can be found in

both types of environment: *esa > ca ‘one’, *e(N)guk > guk ‘sound of water going down

the throat’, Nguk ‘sound of water being swallowed’, *ekap > hap ‘gasping, panting’,

*elem ‘shade, darkness’ > lem ‘new moon; dark’, *eliN ‘high-pitched sound’ > liN ‘to

sound, as a flute, or a clap of the hands’, *epaN > paN ‘notch’, *epat > pat ‘four’, but

*ekep > ekep ‘brood, sit on eggs’, *elak ‘separate two things’ > elak ‘alternating, spaced

widely; seldom’, *enem > enem ‘six’, *qentek ‘stamp on’ > entek ‘go into a rage while

stamping the feet’, *heRem > eret ‘tight, firm, of a binding’. In other positions *e is

consistently reflected as a mid-central vowel: *beli > weli ‘buy’, *qajeN > aseN

‘charcoal’, *teleb ‘sink, vanish from sight’ > telep ‘be hidden, disappear for a moment’.

*-i. PMP *i is normally reflected as i, but occasionally appears instead as a mid-front

vowel: *inum > inuN ‘drink’, *lima > lima ‘five’, *binehiq > wini ‘seed’, *bibiR > wiwir

‘lip’, but *huaji (> waji) > asé ‘younger parallel sibling’, *ma-qudip > mosé ‘living,

alive’, *qaninu > nénu ‘shadow, reflection’, *uliq > olé ‘restore, repair’, *piliq > pilé

‘choose’, *wahiR > waé ‘water’.

*-u. PMP *u usually became u, but occasionally became o: *u(N)kaq ‘open’ > uka ‘pry

up, open’, *ukup > ukup ‘brood, sit on eggs’, *puni > puni ‘tree fern: Cyathea sp.’,

*qampus > ampus ‘destroy, wipe out’, *telu > telu ‘three’, but *niuR > nio ‘coconut’,

*uak ‘sound of retching’ > oak ‘nausea, retching’, *ukuk ‘cough’ > okok ~ okuk ‘sound

of coughing’, *penuq > peno ‘full’, *dapuR > sapo ‘hearth, cooking place’, *tuzuq >

toso ‘point out’.

6. The Bima-Sumba hypothesis revisited. In this section the Bima-Sumba hypothesis is

assessed by considering 1) lexicostatistical percentages, 2) shared irregularities in

phonology, and fossilized morphology, 3) exclusively shared lexical innovations, and 4)

phonological innovations. The discussion of phonological innovations is deferred until

the end, as parallel sound changes evidently have taken place repeatedly throughout the

Lesser Sundas, greatly complicating problems of interpretation.

45

6.1. Lexicostatistical percentages. Table 9 summarizes cognate percentages linking

Bimanese, Manggarai, Kambera and Hawu with one another, and with Tetun, a language

assigned to Esser’s ‘Ambon-Timor’ group. These percentages are based on the following

cognate counts: Bima-Tetun = 60/200, Bima Hawu = 51/196, Bima-Kambera =56/200,

Bima-Manggarai = 55/200; Manggarai-Tetun = 66/200, Manggarai-Hawu = 57/196,

Manggarai-Kambera = 63/200, Kambera-Tetun = 63/200, Kambera-Hawu = 70/196,

Hawu-Tetun = 64/196). Cognate decisions are coded by letter in Appendix 1:

TABLE 9: Cognate percentages linking five languages of the Lesser Sunda islands

Tetun Hawu Kmb Mgg

Bima .30 .26 .28 .275

Mgg .33 .29 .315

Kmb .315 .358

Hawu .327

As seen here, the Bima-Sumba hypothesis is not supported by lexicostatistics since, apart

from Kambera-Hawu, Tetun scores higher on average with ‘Bima-Sumba’ languages

(.318) than these do with one another. It is known that lexicostatistics is a crude tool for

subgrouping because it confounds innovations and retentions, and accurate results can be

expected only when the retention rate for basic vocabulary is approximately equal for all

languages compared (Blust 2000). What is more relevant to developing a subgrouping

argument are exclusively shared replacement innovations in basic vocabulary. The data

summarized Table 2 suggest, at least tentatively, that there is a Sumba-Hawu group, and

perhaps a larger grouping of S-H with several languages of western and central Flores,

but that Bimanese is excluded from this and any other group that does not also include

‘Ambon-Timor’ languages such as Tetun. More extensive comparison supports this first

impression, but before reviewing the lexical evidence we must examine the distribution

of irregularities in phonology and fossilized morphology in light of their possible

subgrouping significance.

6.2. Shared irregularities in phonology, and fossilized morphology. Shared sporadic

sound changes carry great weight as subgrouping evidence, given the low probability that

they could be products of convergence. The following irregular developments are shared

by Kambera and Hawu, but are unknown elsewhere (except in other languages of Sumba

and in Dhao). They are attributed to a previously unrecognized proto-language, called

‘Proto-Sumba-Hawu (PSH). All available sources of lexical material for languages of the

area have been searched for relevant comparisons.

6.2.1. Shared irregularities in phonology.

1) PMP *qalejaw > PSH *loo (expected *l↔o) ‘day, sun’. This form shows

sporadic *e > o in both Kambera lodu and Hawu lo



o, possibly as a result of regressive

assimilation. Cp. Bimanese liro, Komodo ro, Manggarai leso, Ngadha leza, Lio leja,

Sika lero-N, Roti ledo ‘day, sun’, Helong lero, Atoni neno, Tetun loro ‘sun’, loro-n ‘day’

(Tetun o is the regular reflex of PMP *e), Kemak, Galoli lelo, Mambai lela, Lamaholot of

46

Larantuka lera, Lamaholot of Adonara rera, but Kodi, Weyewa lodo, Dhao lo



o ‘day’

(where o is not the regular reflex of *e).

2) PMP *dapuR > PSH *ra?u (expected *rapu) ‘hearth’. The assumption that

PSH *ra?u reflects PMP *dapuR with irregular *p >

?

may be incorrect. If so Kambera

raü, Hawu ra

?

u may be a shared lexical innovation. Cp. Manggarai sapo, Ngadha dapo,

Palu’e lapo ‘hearth’ (cognates not available for most other languages of the area).

3) PMP *qulun-an > PSH *nulaN (expected *lunaN) ‘wooden headrest, pillow’.

After regular loss of *qu- the expected PSH reflex of PMP *qulun-an would be *lunaN.

Kambera nulaNu, Hawu n

↔

lu thus share a sporadic metathesis of the first two consonants.

Reflexes in other languages of the Lesser Sundas are difficult to find. Manggarai luné

may reflect *lunaN with the same type of irregularity seen in *qudaN > kusé ‘shrimp’, or

*haRezan > redé ‘ladder’. On the other hand, the Manggarai form may be cognate with

Tetun luni, Kedang luni-n ‘rest the head; headrest’, with sporadic lowering of *i.

4) PMP *añem > PSH *uñaN (expected *añ↔N) ‘plait, weave’. Kambera unaNu,

Hawu

↔

ñu reflect *uñaN, a form that seems to show non-chance similarity to *añ↔N, the

expected PSH reflex of PMP *añem. The vowel correspondences are irregular, and may

be a product of both irregular *e > u, as in PMP *qajeN > Kambera aruNu ‘charcoal’,

and metathesis (cp. Manggarai n-anaN ‘plait, weave’, Roti ane ‘braid, plait, twine’, with

cognates unavailable for other languages of the area).

6.2.2. Shared fossilization of morphology.

1) PMP *kita > PSH *Nita ‘see’. This PSH word appears to show fossilized verb-

forming nasal substitution in both Kambera Nita and Hawu N

↔

di, although Onvlee (1984)

gives both ita and Nita ‘see’ for the first language. Cp. Bimanese eda, Komodo, Sika ita

‘see’, with cognates generally unavailable in other languages of the area.

2) PMP *qunap > PSH *napi ‘fish scale’. Kambera nepi, Hawu nai reflect *napi,

possibly from *qunap-i ‘to scale fish’ a verb with Oceanic reflexes such as Sa’a unehi

‘scale a fish’, Nanumea unafi ‘fish scales’, Rennellese

?

unahi ‘fish scales; scale a fish’.

No other reflexes of *qunap-i, or of *qunap have been found in languages of the Lesser

Sundas, although Paulohi (Seram) unai ‘refuse of, scrapings of’ may be related.

3) PMP *di atas > PSH *diatas ‘above, on top’. Both Kambera dítahu ‘upward

movement’, and Hawu



ida ‘high’ reflect this PMP compound preposition (*di ‘generic

marker of location’ + atas ‘top part’) as a single unit, with loss of morpheme boundary.

4) PMP *qeti > PSH *m↔nti ‘used up, finished’. Kambera mindi, Hawu m

↔

ti ‘used

up, finished’ appear to reflect *qeti with a prefix that has become part of a reanalyzed

base.

5) PMP *laRiw > PSH *pa-lai ‘flee, run away’. Onvlee (1984) gives Kambera

palai ‘run fast, flee’, but notes that lai occurs in the fixed expression njara lai ‘racing

47

horse’. Wijngaarden (1896) gives Hawu rai and perai ‘run fast, flee’ as one entry, noting

that each form occurs in certain expressions. PSH *palai probably contains the causative

prefix *pa-, which has become partly fossilized with this base in both languages.

6) PMP *maN-kaen > PSH *NaN ‘eat’. Kambera NaNu, Hawu Naa/Nae ‘eat’

might be treated as a lexical innovation, but both can be derived from PMP, with loss of

morpheme boundary (*maN-kaen > maNan > *maNaN > NaN). Roti na

?

a ‘eat’ is

assumed to have a different source.

6.3. Exclusively shared lexical innovations. In addition to the irregularities noted in

6.2, Kambera and Hawu share a substantial amount of vocabulary that has not yet been

found in languages outside the Sumba group and Hawu-Dhao. All exclusively shared

cognate sets found to date are listed in Appendix 3. The vowel correspondences in some

of these are problematic. However, many other forms are completely straightforward,

and together they provide strong evidence that at least Kambera and Hawu-Dhao if not

Hawu-Dhao and all languages of Sumba, form an exclusive subgroup.10

Where vowel correspondence are irregular or weakly attested the PSH reconstruction is

followed by a question mark. In most environments PSH mid vowels were raised to high

vowels in Kambera. However, this apparently did not happen in all cases, and exceptions

are especially common adjacent to *w. Forms with irregular consonant correspondences,

as Kambera tola, Hawu mola ‘upright, erect’, were generally discarded. Some forms that

initially appear to be exclusively shared lexical innovations turn out on closer inspection

to be loanwords, as with Kambera hedi, Hawu hedui ‘sad’ (< Malay sedih?), Kambera,

Hawu hela ‘saddle’ (< Malay/Portuguese sela), Kambera, Hawu dupa ‘incense’ (<

Malay/Sanskrit dupa), or Kambera karuNu, Hawu ka



u ‘sack’ (< Malay karuN ‘sack’).

A few other words that are not known to be loans were dismissed due to their meanings,

which are associated with introduced concepts, as with Kambera pareta, Hawu pareda,

or Kambera, Hawu rapa, both sets meaning ‘reins, bridle’.

6.4. Phonological innovations. The historical phonology of the languages of the

Lesser

Sunda islands seems almost calculated to perplex. There has been a strong tendency

throughout the region for stops to lenite and final nasals to merge as -N, but many shared

changes appear to be products of drift. To illustrate, although PMP *b is retained as a

voiced bilabial stop in Roti and Helong, it has lenited in nearly all other languages from

Bima in the west to at least the Leti archipelago in the east. The most common outcome

of this lenition is w (Bimanese, languages of west, central and east Flores, the Solor

archipelago, Kisar and the Leti-Moa and Babar archipelagos), although a more common

10 Kern (1892:157-58) had some inkling of the Sumba-Hawu relationship, noting that in phonological

system, grammar and vocabulary Hawu is most closely related to ‘Sumbasch’ (= Kambera). However, he

stated this only in passing, and offered no more than three or four lexical comparisons as supporting

evidence. Moreover, he added (vacuously) that Hawu has many points of similarity with Makasarese and

Buginese, but fewer with Javanese. Dyen (1965:28) included Hawu (Sawu), but not Bimanese, Manggarai

or Kambera in his lexicostatistical classification of the AN languages, finding the closest connection of

Sawu to be with his ‘Moluccan Linkage’.

48

outcome in the languages of Timor is f or h (or both, as in Tetun f-, but –h-). Several

other changes are common to languages that Esser assigned to his ‘Bima-Sumba’ and

‘Ambon-Timor’ groups. Given this tendency for sound changes to recur, shared

phonological innovations cannot be safely used to form subgroups in this region.

Table 10 summarizes the main patterns of sound change from sections 2-5. Changes that

are found in all of the languages and are widespread elsewhere in AN are not mentioned.

These include the loss of PMP *e which came to be adjacent to another vowel, and loss

of prepenultimate vowels that were initial or came to be initial through loss of *h or *q.

TABLE 10: Patterns of sound change in Kambera, Hawu, Bimanese and Manggarai

PMP Kmb Hawu Bima Mgg

*p p p/0 p/f p

-p/0

*mp mb p mp mp

*t t d/t d/t t

*nt nd nt nt

*c h? h ? c

*k k/0 k/0- k/h k/h

-k/?-

*Nk Ng Ng Nk

*q 0 0 0 0-, -0

?/V_V

0?/V_V

*b b/w w/≡/b w/≡w/b

-p

*mb mb mb mb

*d d/r r/r/s/d

*nd nd nd nd

*z r yr/s

*nz nd

*j d/r r/l/r s

*Nj

*g ? g? k? g/h

*Ng ?

*m m m m m

-N -m/N

*n n n n n

-N -n/N

*ñ n ñ n n

*N N N N N

*s h h c/s c

-s

*ns nj nc

*h 0 0 0 0

49

*l l l/r l/r l

*r r 0? r r

*R r/0 0 0/r r

*w w 0 w w

0 w/#_u

*y y? i? j j

*-ay i e e a

*-aw u o o o

*-uy i i i i

*-iw i i i i

-C -Cu 0 0 -C

*CV3 Ca- Ce- Ca-?

*a a a a a

e/_Ci e/_i e/_Ci

o/_u o?o

*e 0- ↔ie/0-

à -a e e

í o

i u

o

-u

*i i i i i/é

*u u u/o u u/o

6.4.1. Shared phonological innovations. Both Hawu and Bimanese show a split of

PMP *p in initial and medial positions. Although the outcome of this split is different in

the two languages (p/0 in Hawu, p/f in Bimanese), in theory this could be a single change

in an immediate common ancestor, with further changes *f > h > 0 in Hawu. The

temptation to follow this line of argumentation is increased by the observation that this

split initially appears to be unconditioned in both languages, and the fortis/lenis pattern of

reflexes agrees fairly closely between them, as shown in Table 11. Type = fortis (F) or

lenis (L); plusses and minuses mark agreement/disagreement:

TABLE 11: Parallel irregular split of PMP *p in Hawu and Bima

PMP Hawu Type Bima Type Gloss

*hapuy ai L afi L + fire

*epat ↔pa F upa F + four

*empu ↔pu F ompu F + grandparent

*lepet l↔pa F lipi F + wrap, fold

*ma-nipis m↔ni L nipi F -- thin, tenuous

*ma-panas pana F pana/mbana F + warm, hot

*nipi ni L nifi L + dream

*paniN ani L pani F -- bait

50

*paniki ni?i L panihi F -- flying

fox

*panaw ano L fano L + leucoderma

*pajay are L fare L + rice

*pahuq pau F fo?o L -- mango

*hapejes p↔a F pili F + sting; pain

*qapeju ↔u, p↔u L/F folu L +/-- gall; bitter

*peñu ↔ñu L fonu L + turtle

*pija p↔ri F pila F + how many?

*pitu pidu F pidu F + seven

*sa-Na-puluq he-Nuru L mpuru F -- ten

*depa d↔pa F ndupa F + fathom

*tepiR d↔pi F dipi F + mat

An adequate statistical test would require that the relative frequencies of fortis and lenis

reflexes first be determined in each language to frame the null hypothesis. It initially

appears that the fortis/lenis agreements in Table 11 deviate from chance (14 agreements,

5 disagreements, one pair with both plus and minus). However, closer attention to

environments suggests that much of this agreement is due to common conditioning. In

particular, stops that otherwise undergo lenition in medial position tend to resist it when

following a stressed schwa (PMP *e), since this gave rise to phonetic gemination.

Moreover, it is likely that the vocalic syncope in *ma- ‘stative’ before PMP bases that

began with a labial stop either happened early in the history of the Central Malayo-

Polynesian languages, or was recurrent, giving rise to prenasalized stops that also resisted

lenition, but were subsequently reduced to simplex stops in some daughter languages. If

we eliminate all examples that appear to show agreement due to common conditioning

(reflexes of *epat, *empu, *lepet, *ma-panas, *depa, *tepiR), and count reflexes of

*qapeju as both ‘plus’ and ‘minus’, the picture changes dramatically to one of nine

agreements (eight if PMP *hapejes was intermediate *ma-pedes), and six disagreements.

Of course, the conclusion that *p lenition was independent in Hawu and Bimanese also

follows from the simple fact of subgrouping: since Sumba-Hawu is a fairly clearcut

group and Kambera shows no evidence of *p lenition the simplest hypothesis is that *p

lenition in Hawu must have occurred independently of the similar change in Bima.

It would be tedious to provide a detailed demonstration of the same point for the parallel

change *t > d in Hawu and Bima, the split of *k, *b and *d into fortis and lenis reflexes

in all four languages, of *j into fortis and lenis reflexes in the first three languages, and of

*l into l and r without apparent conditions in Hawu and Bima. Suffice it to say that

although *t > d is found in both Hawu and Dhao, this does not appear to be true of all

dialects of Bimanese. Jonker (1896:4), speaking of the divergent Kòlo and Tòlo-Weri

dialects of Bimanese noted e.g. T-W mate ‘die’, implying (in the absence of further

evidence) that *t > d is a relatively recent change in standard Bimanese. The remaining

phonological innovations that are shared exclusively by Hawu and Bima also fail to bear

close scrutiny. Rather, parallel sound changes evidently have been unusually common in

this part of Indonesia, and there are no clear grounds for using such evidence to propose a

subgroup larger than that of Sumba-Hawu. Indeed, even within Sumba-Hawu caution

51

must be observed in using shared phonological innovations as subgrouping evidence,

since the highly distinctive *s > h change in Hawu and Kambera is known to be a late

change in both languages. Examples such as this are a reminder of the need for caution

in subgrouping by exclusively shared phonological innovations, since the same type of

change may occur independently in different languages or language groups, but appear to

be a historically a single innovation.

7. Beyond Sumba-Hawu. As noted earlier, there is some evidence that Sumba-Hawu

belongs to a larger grouping that also includes languages of western and central Flores.

Although the penultimate vowel correspondence in Manggarai (Mgg) wéla, Kambera

(Kmb) wàla, Hawu wila ‘flower’ is irregular and may indicate borrowing, the following

comparisons are suggestive of what more intensive work might reveal about lexical

isoglosses linking Sumba-Hawu with many of the languages of Flores: 1) Mgg ila ‘string

fish on a line’, Kmb ila ‘line for stringing fish’, 2) Mgg lalo, Kmb, Hawu lalu ‘orphan,

orphaned’, 3) Mgg luku ‘a fold, hem’, Kmb luku ‘to fold’, 4) Mgg lula ‘caulk, solder’,

Kmb lula ‘sticky, adhesive, thick, of fluids’, 5) Mgg lumuN ‘to like, long for’, Kambera

lumuNu ‘show interest in, care or worry about, take to heart’, 6) Mgg merik, Kmb miriku

‘small, fine’, 7) Mgg pu, Kmb pú ‘smacking sound of lips when spitting’, 8) Mgg puci

‘enter’, Kmb puhi ‘insert, as a pole in a posthole, ring on the finger’, 9) Mgg pupu ‘a

snake, the cobra: Naja sputatrix’, Kmb pupu ‘venomous snake, the green tree adder:

Lachesis gramineus’, Mgg wené, Hawu w

↔

ñi ‘areca palm’. In addition to lexical

evidence of this type the apparently sporadic prenasalization in PMP *taqun > Mgg

ntauN, Kmb ndauNu ‘year’, and perhaps the common semantic innovation in Mgg awaN,

Kmb awaNu ‘sky’ (< PMP *awan ‘cloud’) point to a similar conclusion.

There is thus substantial support for an exclusive Sumba-Hawu subgroup, and limited

support for a wider subgroup that includes Sumba-Hawu and the languages of western

and central Flores, but no clear support for ‘Bima-Sumba’. Why, then, did Esser propose

such a group? Esser worked in Sulawesi, and apparently had no firsthand experience

with languages of the Lesser Sundas. In the early twentieth century the reigning expert

on the languages of this region was J.C.G. Jonker (1857-1919), who published

extensively on Bimanese, Roti, Leti, and Timorese (Atoni), and it is clear that Esser

relied on Jonker for this part of his classification. In the Introduction to his Bimanese

grammar, which was published in 1896, Jonker made his views on the classification of

the languages of the western Lesser Sundas very clear. First, he states that if there is an

east-west division of ‘Malayo-Polynesian’ languages within Indonesia based on

differences in the genitive construction, as first argued by Brandes (1884), then Bimanese

belong in the western group. Second, he notes that the closest relationship of Bimanese

is with ‘Sumbasch’ (= Kambera), that it shares less similarity with Hawu, and less still

with Manggarai, although the latter language has many loanwords from Bimanese.

Jonker was a major figure among government language officers during the Dutch

colonial period in Indonesia. He had a thorough knowledge of Rotinese, and a good

knowledge of several other languages of the region. However, like most Dutch linguists

of his day he was not trained in historical linguistics, a field which at that time was

dominated in western Europe by German, French and Danish scholars. His views on

52

subgrouping were thus likely to be impressionistic rather than founded on a careful

sifting of evidence. In fact, without a reconstructed baseline at the time there was no way

to clearly distinguish innovations from retentions. In retrospect it is not difficult to see

how Jonker (or Esser) might have fallen into the trap of grouping Bimanese with the

languages of Sumba, and Hawu-Dhao, since these can be linked by some etymologies

that are extremely seductive, even today. Consider the following:

(1) PMP *baRiw ‘begin to spoil (of food left uneaten too long) : Tiruray warey ‘stale,

spoiled’, Tagabili bali ‘spoiled, soured, rotten’, Mukah Melanau bayew ‘old, as a dry

coconut, or unmarried girl past her prime’, Iban bari

?

‘musty, “gone off”, as rice’, Ngaju

Dayak bayo ‘spoiled’, Toba Batak bari ‘foul-tasting, as rice’, Sundanese bari ‘old, stale,

cold (of foods that have stood around a long time)’, Javanese wayu ‘old, stale, having sat

around too long’, Sasak bari ‘old food that has begun to stink’, Bimanese mbai ‘rotten’,

Manggarai wari ‘poisonous, dangerous’, Kambera mbai ‘spoiled, musty (as rice)’.

(2) PMP *ma-baRani ‘brave, bold’ > Kapampangan bayani ‘valiant’, ma-bayani

‘abundant’, Maranao ma-bagani ‘brave’, Malagasy ma-vany ‘contemptuous, audacious’,

Bimanese mbani, Kambera mbeni ‘brave, bold’.

(3) PMP *ma-baseq ‘wet’ > Itbayaten ma-vasa

?

, Kalagan ma-basa

?

, Botolan Sambal ma-

baha

?

, Lun Dayeh m

↔

-baa

?

, Malay basah, Bimanese mbeca, Kambera mbaha ‘wet’.

(4) PMP *ma-besuR ‘satiated’ > Kavalan ma-bisuR, Paiwan ma-vetu, Isneg ma-btúg,

Bare’e ma-bosu, Bimanese mbocu, Kambera mbíhu ‘satiated’.

(5) PMP *ma-bekuk or *ma-biNkuk ‘bent, curved’ > Iban beNkok, Balinese beNkuk,

Tae’ bakkuk ‘crooked, curved, bent’, Rembong wekuk ‘curved; bow (for shooting)’, or

Bintulu bikuk ‘bent, winding, crooked’, Ngaju Dayak biNkok ‘bent, curved, as a stick’,

Old Javanese a-miNkuk ‘to bend, curve (river)’, Manggarai wikuk ‘sit hunched over,

squat, crouch, Bimanese mbeko, Kambera mbeku ‘bent’.

In (1) Bimanese and Kambera show identical forms that differ strikingly from cognates in

other languages. These have arisen through three historical changes: merger of *-iw with

*-i, loss of *R, and prefixal vowel sycope in *ma-baRiw. PMP *-iw and *i evidently

merged in all CEMP languages. Loss of *R is not universal, but also occurs in Komodo,

some (but not all) languages of western and central Flores, and in many of the languages

of Esser’s ‘Ambon-Timor’ group, including Roti, Atoni, Tetun, Kemak, and Galoli. The

change *baRiw > mbai is thus easily explained as a product of convergence. Initially the

prenasalization in this form appears to be highly distinctive, but closer inspection of the

comparative evidence shows that this is simply a particular case of a type of sound

change common to many of the languages of eastern Indonesia.

PAN and PMP marked most stative verbs with the prefix *ma-, thus setting up a highly

disfavored phonotactic sequence in bases that began with *p or *b. Since homorganically

prenasalized clusters were common in PMP, vowel syncope in the sequence *mVp- or

*mVb- would eliminate the disfavored phoneme configuration at a minimal cost to

53

morpheme structure. Table 12 shows reflexes of *b- and *p- in Tetun of central Timor,

and then reflexes of the same consonants following the stative prefix *ma-:

TABLE 12: Reflexes of PMP *b-, *p-, and *ma-b-, *ma-p- in Tetun

PMP Tetun

babuy fafi pig

balu falu-k widowed

belaj falar spread out

beli foli-n price, cost

bities fitis calf of leg

buaq fua-n fruit

bulan fulan moon

---------------------------------------------------------------------

pa- ha- causative

panaq hana shoot an arrow

paNdan (pendan) hedan pandanus

pija hira how much/how many?

pitu hitu seven

punti hudi banana

puqun hun base of tree

----------------------------------------------------------------------

ma-panas manas hot, warm

hapejes (> ma-pejes) moras painful; sick

qapeju (> ma-peju) moru-k gall; bitter

ma-putiq muti white

ma-belaj belar broad, spread out

ma-belit belit viscous, sticky

ma-besuR bosu satiated

Although they are initially puzzling, the reflexes manas, moras, moru-k and muti become

intelligible under a hypothesis of prefixal vowel syncope in the sequences *ma-p- and

*ma-b-. Following syncope mp- was reduced to m-: *ma-panas > mpanas > manas. By

contrast, initial *mb- and all medial clusters were reduced by loss of the nasal, after

postnasal voicing: *kempuN > Tetun kabu-n ‘stomach, abdomen’, *kapas > kabas

‘cotton plant’ (cf. Manggarai kampas), *pintik ‘fillip, flick with finger’ > hidik ‘hit one

thing upon another’, *punti > hudi ‘banana’.

A wider sampling of languages shows that plain and prenasalized labial stops are

distinguished throughout much of the Lesser Sundas, and parts of the Moluccas: PMP

*ma-penuq > Bimanese

≡

ini, Keo, Ende mbenu, Ngadha, Sika, Kemak benu, Kodi

mbanuka, Kambera mbínu, Mambai beun, Lamaholot (Larantuka) menuh, Lamaholot

(Lamalera) menu-N ‘full (container)’, *ma-beRat > Kodi mboto, Anakalangu buatu,

Kambera mbotu, Lamaholot (Larantuka) ba

?

at (cp. *bulan > Kodi, Anakalangu wula,

Kambera wulaNu, Lamaholot wulan ‘moon’, *batu > Kodi, Anakalangu, Kambera watu,

54

Lamaholot wato ‘stone’, reflecting *b- rather than *mb-), *ma-panas > Ngadha bana,

Kodi banaho, Kambera mbanahu, Tetun manas, Kemak banas-an, Galoli banas, Kisar

mana-k ‘warm, hot’ (cp. *pitu > Kodi pitu, Kambera pihu, Tetun, Galoli hitu, Kemak icu,

Kisar wo-isu-k ‘seven’ for what appears to be the regular reflex of *p- as opposed to

*mp-), *ma-putiq > Helong, Atoni muti, Tetun muti-n, Kemak buci, Mambai, Galoli buti

‘white’. It follows, then, that vowel syncope in *ma-baRiw > Bimanese, Kambera mbai

‘spoiled, rotten’ has no more diagnostic value for the Bima-Sumba hypothesis than the

loss of *R or the merger of *-iw with *-i. Rather, a number of languages in eastern

Indonesia share some or all of these changes. The same arguments apply to items (2)-(5).

In this connection it is noteworthy that in most languages (Sika is an apparent exception)

syncope does not occur before bases that begin with a non-labial consonant, suggesting

that the initial motivation for this change was avoidance of dissimilar labial onsets in

successive syllables.11 In effect these developments exhibit a kind of consonant ‘grade’

reminiscent of the more familiar examples known from Oceanic languages, but one that

apparently began through regular or at least recurrent sound change with bases that

contained an initial labial obstruent. Preliminary inspection of the comparative evidence

suggests that if consonant grade distinctions were already found in PCMP, they would

have been limited to the labials, with prenasalization of non-labial base-initial consonants

developing at a later time in individual languages. Table 13 summarizes the oral and

nasal grade reflexes of PMP *p and *b word-initially in a sample of languages scattered

over eastern Indonesia:

TABLE 13: Oral and nasal grade reflexes of PMP *p/b- in languages of eastern Indonesia

PMP p- ma-p- b- ma-b-

‘Bima-Sumba’

Bimanese f/p mp w/b mb

Kambera p mb w mb

Manggarai p p w b~w

Ngadha p b v b

Hawu p/0 p w/b/≡b/≡

‘Ambon-Timor’

Atoni h m f f

Tetun h m f b

Kemak p b f f?

Sika p b w b

Buru p b f b

8. Conclusions. It should be clear from the foregoing data and discussion that the Bima-

Sumba hypothesis does not satisfy the basic criteria for acceptance as a valid linguistic

subgroup, namely unambiguous evidence of exclusively shared innovation that are not

attributable to borrowing or convergence. Despite the strong evidence for a Sumba-

Hawu subgroup, there is every indication that the identification of larger subgroups

11 Synchronic alternations such as Bimanese saNa ‘fork of a branch’ : ncaNa ‘bifurcate’, Kambera ka-haNa

‘forked’ : njaNa ‘branch’, or Bimanese sara ‘miss the mark’ : ncara (also sala) ‘wrong, in error’, Kambera

hala/njala ‘wrong, in error’ show that historically secondary nasal grade reflexes in language of eastern

Indonesia probably have multiple origins, a topic that cannot be taken up in the present context.

55

incorporating languages of the Lesser Sundas will prove difficult. In particular, it is

doubtful that Esser’s ‘Ambon-Timor’ group has any more validity than ‘Bima-Sumba’.

As noted in Blust (1993), CMP almost certainly was the product of rapid movement from

northeast Indonesia through the central and southern Moluccas and into the Lesser Sunda

islands, forming a dialect chain that favored the ready diffusion of innovations across

incipient language boundaries, and hence destroying any tree-like structure that may have

previously arisen. Although the evidence from some languages is ambiguous, syncope of

the vowel of *ma- ‘stative’ in labial-initial bases is scattered throughout the Central

Malayo-Polynesian languages, hence Esser’s ‘Bima-Sumba’ and ‘Ambon-Timor’ groups.

Whether this reflects a single innovation, or is the result of recurrent innovations remains

uncertain. In any case, it is clear that the most strikingly distinctive innovations shared

by Bimanese with Kambera, Hawu-Dhao, or Manggarai cannot be taken as evidence for

an exclusive ‘Bima-Sumba’ group, and with this conclusion comes the end to a myth that

has lasted far longer than one might have imagined possible.

APPENDIX 1

The following is a modified Swadesh 200-word list in PMP, Bimanese, Manggarai,

Kambera, and Hawu. Bimanese material is from fieldnotes collected in January, 1972

from Abdul Karim Sahidu, originally from the village of Maria, about 17 km. from Bima,

Sahidu (1978), and Ismail, Azis, Yakub, Taufik and Usman (1985), Manggarai material

is from Verheijen (1967), Kambera data from Onvlee (1984), and Hawu data from

Wijngaarden (1896), and Walker (1982). Some supplementary data for Hawu have been

taken from Fox (n.d.).

1) PMP i babaw (A) ma-qudip (A) amin (A) ka/ma (A)

di atas (B)

Bimanese vavo (A) mori (A) menana (B) la≡o (B)

Manggarai éta (B) mosé (A) tauN (C) agu (C)

Kambera díta (B) miripu (A) mbúlu/ dàNu (D)

ndàba (D)

Hawu ida (B) muri (A) hari hari (E) Na (E)

Tetun leten (C) moris (A) hotu hotu (F) ho (F)

English above alive all and/with

2) PMP qabu (A) i, di (A) likud (A) zaqat (A)

Bimanese kalu≡u (B) i (A) kontu (B) da≡ae (B)

Manggarai rawuk (C) oné (B) wekur (C?) da?at (A)

Kambera aü (A) la (C) kawiNgu (D) akatu (C)

Hawu awu (A) pa (D) d↔ni (E) apa (D)

Tetun falahok (D?) iha (E) kotuk (B?) aat

(A?)

English ash at back bad

3) PMP tian (A) i babaq (A) Raya (A) manu-

56

manuk(A)

Bimanese loko (B) ava mai (B) na?e (B) nasi (B)

Manggarai bara (C) wa (A) mésé (C) kaka lélap (C)

Kambera kambu (D) lumbu (C) bokulu (D) mahawuruNu(D)

Hawu elu (E) wawa (A) worena (E) dolila (E)

Tetun kabun (D) kraik (D) boot (F) manu (A)

English belly below big bird

4) PMP kaRat (A) ma-qitem (A) daRaq hiup (A)

Bimanese ha?a (A) me?e (B) ra?a (A) ufi (B)

Manggarai akit (B) miteN (A) dara (A) pur (C)

Kambera kati (C?) mítiNu (A) wai ria (A) ha pui (D)

Hawu hi≡e/hi≡i (D) guru/m↔di (A) ra (A) tiu (A?)

Tetun hanisi (E) metan (A) raan (A) hu (C)

English bite black blood blow

5) PMP tuqelaN (A) daqan (A) susu (A) ma-ñawa (A)

Bimanese pe?e (B) saNa (B) susu (A) sasi nava (A)

Manggarai toko (C) paNa (C) cucu (A) nai (B)

Kambera rí (D) lai (D) huhu (A) haNahu (C)

Hawu rui (D) kelai (D) huhu (A) heNaa (C?)

Tetun ruin (D) sanak (B) susun (A) nawan (A)

English bone branch breast breathe

6) PMP tunu (A) beli (A) anak (A) piliq (A)

Bimanese ka?a (B) veli (A) ana (A) kaale (B)

Manggarai tapa (C) weli (A) anak (A) pilé (A)

Kambera tunu (A) diNi (B) ana (A) pindi (C)

Hawu tunu (A) w↔li (A) ana (A) ≡ei/pili (A)

Tetun sunu (D) sosa (C) oan (B) kakan (D)

English burn buy child choose

7) PMP Rabun (A) ma-diNdiN (A) maRi (A) tanek/

zakan(A/B)

Bimanese apu (B) ≡usi (B) mai (A) mbako (C)

Manggarai rewuN (A?) menes (C) mai (A) teneN (D)

Kambera karumaNu (C) mariNu (A?) mai (A) manahu (E)

Hawu mer↔mu (D) meriNi (A?) d↔ka/mai (A) hogo (F)

Tetun kalohan (E) malirin (A?) mai (A) da?an (B)

English cloud cold come cook

8) PMP ihap (A) taNis (A) taRaq/ qalejaw (A)

tektek (A/B)

Bimanese réké (L) naNi (A) dompo (C) ai marai (B)

Manggarai bilaN (L?) rétaN (B) déntér (D) leso (A)

57

Kambera diha (B) hí (C) kadípu (E?) lodu (A)

Hawu ta-reke (L) taNi (A) ↔ta/↔te (F) loo (A)

Tetun sura (C) tanis (A) taa (A) loron (A)

English count cry cut day

9) PMP m-atay (A) kali (A) cemeD (A) asu (A)

daki (B)

Bimanese made (A) Nari (A) sampu (C) lako (B)

Manggarai mata (A) cacé (B) saki (B) acu (A)

Kambera meti (A) àki (C) marihaku (D) ahu (A)

Hawu made (A) kei (A) ra?i (B)Naka (C)

Tetun mate (A) ke?e (D) karebo (E) asu (A)

English die dig dirty dog

10) PMP hipi (A) inum (A) ma-RaNaw(A)dumpul (A)

Bimanese nifi (A) nono (B) maNo (A) dampa (A?)

Manggarai nipi (A) inuN (A) daNo (A?) dumpul (A)

Kambera maNgadipaNu (B) unuNu (A) màdu (B) kadumba (A?)

Hawu ni (A) Ninu (A) kemaNu (A) wada (B)

Tetun mehi (A) hemu (C) maran (C) lak meik (C)

English dream drink dry dull

11) PMP qabuk/qapuk (A/B) taliNa (A) taneq (A) kaen (A)

Bimanese kalu≡u (C) fiko (B) dana (A) Naha (A)

Manggarai kebok (D) tilu (C) tana (A) haN (A)

Kambera kaNabu (E) ka-hilu (C) tana (A) NaNu (A)

Hawu awu (A) wo-dilu (C) rai (B) Na?a/Na?e(A)

Tetun rai ahun (B) tilun (C) rai (B) ha (A)

English dust ear earth eat

12) PMP qateluR/qiteluR (A) siku (A) mata (A) nabuq (A)

Bimanese dolu (A) cihu (A) mada (A) ma≡u (A?)

Manggarai telo (A) ciku (A) mata (A) pa?u (B)

Kambera tílu (A) hiu (A) mata (A) kanabu (A)

Hawu d↔lu (A) wo-hiu (A) mada (A) me-nawu (A)

Tetun tolun (A) sikun (A) matan (A) monu (C)

English egg elbow eye fall

13) PMP ma-zauq (A) meñak (A) ama (A) ma-takut (A)

Bimanese do?o (A) apa (B) ama (A) dahu (A)

Manggarai olu (B) wusak (C) ema (A) rantaN (B)

Kambera marau (A) mina (A) ama (A) mandauta (A?)

Hawu ya?u (A) m↔ñi (A) ama (A) meda?u (A)

Tetun dook (A) mina (A) ama (A) ta?uk (A)

English far fat father fear

58

14) PMP bulu (A) hapuy (A) hikan (A) lima (A)

Bimanese kere (B) afi (A) uta (B) lima (A)

Manggarai wulu (A) api (A) ikaN (A) lima (A)

Kambera wulu (A) epi (A) iaNu (A) lima (A)

Hawu ra?u/rou/ru (C) ai (A) nadu?u (C) l↔mi (A)

Tetun fulun (A) ha?i (A?) ikan (A lima (A)

English feather fire fish five

15) PMP buNa (A) Rebek (A) qaqay (A) halas (A)

Bimanese vunta (B) Nemo (B) ei (B) vu≡a (B)

Manggarai wéla (C) lélap (C) wa?i (A?) puar (C)

Kambera wàla (C) hawuruNu (D) wihi (C) alahu (A)

Hawu wila (C) lila (E) y↔la (D) liha (E)

Tetun ai funan (A) semu (F) ain (A) alas (A)

English flower fly (v.) foot/leg forest

16) PMP epat (A) buaq (A) ma-penuq (A) beRay (A)

Bimanese upa (A) vua (A) ≡ini (B?) mbei (A)

Manggarai pat (A) wua (A) peno (A) sotor (B)

Kambera patu (A) wua (A) mbínu (A) wuaNu (C)

Hawu ↔pa (A) ≡ue/wue (A) tobo (C) wie (A)

Tetun haat (A) ai fuan (A) nakonu (D) fo (A?)

English four fruit full give

17) PMP ma-pia (A) baliji (A) mataq (A) tinaqi (A)

Bimanese taho (B) mpori (B) jao (L) loko (B)

Manggarai dia (C?) remaN (C) ta?a (B) tuka (C)

Kambera hàmu (D) rumba (D) múru (C) tanai (A)

Hawu ie (A) yu?u (E) meN↔ru (D) tenei (A)

Tetun di?ak (C) hae (F) matak (A) teen (A)

English good grass green guts

18) PMP buhek (A) (qa)lima (A) ia (A) qulu (A)

Bimanese hoNgo (B) rima (A) sia (A) tuta (B)

Manggarai wuk (A) limé (A) hia (A) ulu (A)

Kambera lúNgi (C) lima (A) nyu-na (B) tíku (C)

Hawu ra?u/rou/ru (D) ruai (B) noo (C?) k↔tu (D?)

Tetun fuuk (A) liman (A) nia (A) ulun (A)

English hair hand he/she head

19) PMP deNeR (A) ma-beRat (A) palu (A) gemgem (A)

Bimanese de?e (B) tani (B) lambo (B) nenti (B)

Manggarai deNé (A) mendo (C) tebol (C) pekeN (C)

Kambera roNu (A) mbotu (A) palu (A) yàpa (D)

Hawu r↔Ni (A) me-j↔ni (D) d↔tu (D) ↔mi (E)

Tetun rona (A) todan (E) tae (E) humur (F)

59

English hear heavy hit hold

20) PMP Rumaq (A) kua/kuja (A) maN-asu/ bana/

qanup (A/B) laki (A/B)

Bimanese uma (A) une (B) Ngalo (C) rahi (B)

Manggarai mbaru (B) co?o (C) wonok (D) rona (C)

Kambera uma (A) Ngiki (D) patamaNu (E) lei (B)

Hawu ↔mu (A) heNa (E) hue/hui (F) ihi ↔mu (D)

Tetun uma (A) nuu nebe (F) te?ur (G) la?en (B)

English house how hunt husband

21) PMP aku (A) ka/nu (A) di dalem (A) bunuq/

p-atay (A/B)

Bimanese ndai ku (A) pai (B) i ade (B) hade (B)

Manggarai aku (A) émé (C) délem (A) mbelé (C)

Kambera nyu-Nga (B) yàka (D) dalu (A) pa-meti (B)

Hawu du/ya (C) we (E) ara (A) pe-made (B)

Tetun ha?u (A) kalau (L) iha laran (A) hoo (D)

English I if in kill

22) PMP tuhud (A) taqu (A) danaw (A) tawa (A)

Bimanese tatu?u (A) ≡ade (B) ndano (A) hari (B)

Manggarai tu?us (A) Ngepuk (C) rana (A) tawa (A)

Kambera kambàku (B) piNu (D) tàma (B) riki (C)

Hawu ru tu (A) ee/ia (E) ------ ole (D)

Tetun tur (A) hatene (F) ue lihun (C) hanasa (E)

English knee know lake laugh

23) PMP dahun (A) ka-wiRi (A) inep (A) qatay (A)

Bimanese ro?o (A) ku?i (A) ndore (B) hodo (B)

Manggarai sauN (A) léo (B) berit (C) ati (A)

Kambera rau (A) ka-lai (C) taNgíluNu (D) eti (A)

Hawu ra?u/rou/ru (A) ke-riu (B?) pahoro ↔ni (E) ade (A)

Tetun tahan (B) karuk (D) toba (F) aten (A)

English leaf left (side) lie down liver

24) PMP anaduq (A) kutu (A) qulej (A) ma-Ruqanay

la-laki (A/B)

Bimanese naru (A) hudu (A) kako (B) mone (A)

Manggarai léwé (B) utu (A) mbolaN (C) rona (A)

Kambera ma-lai (C) wutu (A) kataru (D) mini (A)

Hawu merai (C) udu (A) ↔tu (E) mone (A)

Tetun naruk (A) utu (A) ular (A) mane (A)

English long louse maggot

man/male

60

25) PMP hadu (A) hesi/isi (A) bulan (A) ina (A)

Bimanese mboto (B) hi?i (B) vura (A) ina (A)

Manggarai do (C) ici (A) wulaN (A) endé (B)

Kambera daNu (D) toluNu (C) wulaNu (A) ina (A)

Hawu ae (E) hedai (D) w↔ru (A) ina (A)

Tetun ua?in (F) na?an (E) fulan (A) ina (A)

English many meat moon mother

26) PMP bukij/buled (A/B) baqbaq (A) kuhkuh (A) Najan (A)

Bimanese doro (C) asa (B) uhu (A) Nara (A)

Manggarai poco (D) mu?u (C) wuku (A) NasaN (A)

Kambera NgunduNu (E) Naru (D) wú (A) tamu (B)

Hawu lede (F) wu≡a (E) ku?u (A) Nara (A)

Tetun foho (G) ibun (F) kukun (A) naran (A)

English mountain mouth nail/claw name

27) PMP kepit (A) pusej (A) hazani (A) liqeR (A)

Bimanese tuka (B) voke (B) eni (A) vo?o (B)

Manggarai pimpet (C) putes (A) baliN (B) bokak (C)

Kambera kahukulu (D) puhu (A) ma-reni (A) Ngoru (D)

Hawu ≡↔ra (E) ↔hu (A) umu (C) lakoko (E)

Tetun kloot (F) husar (A) kreis (D) kakorok (F)

English narrow navel near neck

28) PMP zaRum (A) baqeRu (A) beRNi (A) qijuN/

qujuN(A)

Bimanese ndau (A) ≡ou (A) maNai (B) ilu (A)

Manggarai jaruN (A) weru (A) wié (C) isuN (A)

Kambera utu (B) mbaru (A) ruduNu (D) uru (A)

Hawu yi-yau (A) wiu (A) m↔a (E) hew↔Na (B)

Tetun daun (A) foun (A) kalan (F) inur (C?)

English needle new night nose

29) PMP qazi/diaq (A/B) ma-tuqah (A) esa/isa (A) buka (A)

Bimanese vati (C) tua (A) ica (A) heNga (B)

Manggarai toa (D) tu?a (A) ca (A) wuka (A)

Kambera ndia (B) kaweda (B) diha (A?) buNgahu (C)

Hawu ao (E) dui (C) ↔hi (A) boka (A)

Tetun la (F) tuan (A) ida (A?) loke (D)

English no/not old one to open

30) PMP tau (A) mula/ quzan (A) ka-labaw (A)

tanem (A/B)

Bimanese dou (A) Ngua (C) ura (A) karavo (A)

61

Manggarai ata (B) weri (D) usaN (A) belawo(A)

Kambera tau (A) tonduNu (E) uraNu (A) kalaü (A)

Hawu dou (A) h↔la (F) ayi (A?) keyoe (C)

Tetun ema (C) kuda (C) udan (A) laho (A)

English person to plant rain rat

31) PMP ma-iRaq (A) ka-wanan (A) ma-esak/ zalan (A)

ta-esak (A)

Bimanese kala (B) vana (A) ntasa (A) ncai/laluru (B)

Manggarai ndéréN (C) wanaN (A) té?é (B) salaN (A)

Kambera rara (D) ka-wana (A) kamàku (C) lariNu (C)

Hawu mea (A) ke-aNa (B) mami (D) yara (A)

Tetun mean (A) kuana (A) tasak (A) dalan (A)

English red right side ripe road

32) PMP qatep (A) Ramut (A) talih (A) ma-baRiw (A)

Bimanese ≡utu (B) amu (A) ai (B) mbai (A)

Manggarai épa (C) waké (B) tali/wasé (A) wau (B)

Kambera kawuku (D) amu (A) líku (C) mbai (A)

Hawu ------ amo (A) dari (A) apa (C)

Tetun kakuluk (E) amut (A) tali (A) dodok (D)

English roof/thatch root rope rotten

33) PMP qasiRa/ qenay (A) kaRi/ kaRaw (A)

timus (A/B) tutuR (A/B)

Bimanese sia (A) sarae (B) Ngahi (C?) kao (A)

Manggarai ci?é (C) laiN (C) curup (D) koér (B)

Kambera mehi (D) wara (D) kànja (E) paní (C)

Hawu meN↔hi (E) laha lae (E) ane (F) g↔ri (D)

Tetun masin (D) rai henek (F) ha?ak (G) koi (E)

English salt sand say scratch

34) PMP tasik (A) kita (A) tahiq/ ma-tazem (A)

zahit (A/B)

Bimanese moti (B) eda (A?) nda?u (C) leme/Naha (B)

Manggarai tacik (A) lélo (B) ja?ik (B?) harat (C)

Kambera tehiku (A) ita/Nita (A) utu (D) dàka (D)

Hawu lou (C) N↔di (A) yau (C) na?a (E)

Tetun tasi (A) hare (C) lita (E) meik (F)

English sea see sew sharp

35) PMP ma-babaq (A) huaji (A) kaka (A) ma-sakit (A)

Bimanese poro (B) ari (A) sa?e (B) supu (B)

Manggarai wokok (C) asé (A) ka?é (B?) beti (C)

Kambera kababa (A) eri (A) aya (C) hídu (D)

Hawu ≡a≡a (A) ari (A) a?a (A) ahe/p↔a (E)

62

Tetun betek (D) alin (A) biin/maun (D) moras (E)

English short y. sibling o. sibling sick

36) PMP tudan (A) kulit (A) laNit (A) ma-tiduR/

ma-tuduR (A)

Bimanese doho (B) huri (A) laNi (A) maru (B)

Manggarai lonto (C) loké (B) awaN/ toko (C)

laNit (B/A)

Kambera mandapu (D) manula (C) awaNu (B) mahuru (D)

Hawu y↔di (E) kuri (A) liru (C) b↔

yi (E)

Tetun tur (F) kulit (A) lalehan (D) dukur (F)

English sit skin sky sleep

37) PMP dikiq/kedi (A/B) hajek (A) qasu/ hulaR/

qebel (A/B) nipay (A/B)

Bimanese to?i (C) saNufi (B) o≡u (B?) sava (C)

Manggarai koé (D) isuk (C) nus (C) ular (A)

Kambera kudu (E) haNudu (D) kawú (D) ularu

(A)

Hawu iki (A?) tee (E) habu (E) do≡oho (D)

Tetun ki?ik (F) horon (F) dubun (F) samea (E)

English small to smell smoke snake

38) PMP luzaq (A) bekaq/belaq (A/B) peRes (A) suksuk (A)

tebek (B)

Bimanese katufe (B) bi?a (C) pua (A?) tuba (B)

Manggarai ipo (C) cahir (D) kejé (B) dus/jelok (C)

Kambera paNànji (D) wàka (A) pijaru (C) hunju (D)

Hawu pe-ilu (E) ≡↔ka (A) pe-h↔le (D) t↔≡o/t↔≡u (Β)

Tetun kaban (F) fotak (E) kumu (E) sona (E)

English spit split squeeze stab

39) PMP diRi/tuqud/kedeN bituqen (A) takaw (A) batu (A)

(A/B/C)

Bimanese kii (C) ntara (B) mpaNa (B) vadu (A)

Manggarai tu?u (B ) ntala (B) tako (A) watu (A)

Kambera landaNu (D) kandunu (C) búti (C) watu (A)

Hawu k↔i (C) moto (D) me-na?o (A) wadu (A)

Tetun harii (A) fitun (A) hana?o (A) fatuk (A)

English stand star steal stone

40) PMP sepsep (A) ma-emis (A) baReq (A) laNuy/

naNuy (A)

Bimanese hinti (B) maci (B) vinte (B) liva (B)

Manggarai mecol (C) mincé (C) bara (A) aNiN (C)

63

Kambera homu (D) làNga (D) wotuhu (C) Neni (A)

Hawu h↔pa (E) N↔ta (E) wihu (D) naNi (A)

Tetun susu (F) midar (F) bubu (E) nani (A)

English suck sweet swell swim

41) PMP ikuR (A) i-na (A) si-ida (A) ma-kapal (A)

Bimanese keto (B) aka (B) sia oho (B) te≡e (B)

Manggarai iko (A?) h-itu (C) isa (A) kimpur (C)

Kambera kiku (A) nina (A) nyu-da (A) tímbi (B?)

Hawu lai (C) nahare (D) ro (C) ≡ai (D)

Tetun ikun (A) ne?e ba (E) sia (D) ma?ar (E)

English tail that they thick

42) PMP ma-nipis (A) demdem (A) i-ni (A) i kahu (A)

Bimanese nipi (A) kananu (B) ake (B) ndai mu (B)

Manggarai mipis (A) nuk (C) ho?o (C) hau (A)

Kambera manipa (A?) pikiru (L) yia (D) nyu-mu (B)

Hawu m↔ni (A) oe (D) nae/nane (E) au (A)

Tetun ni?is (A?) hanoin (E) ne?e (F) o (A)

English thin think this thou

43) PMP telu (A) tudaq (A) hiket (A) dilaq (A)

Bimanese tolu (A) to≡a (B) iki (B)lera (A?)

Manggarai telu (A) boét (C) poNo (C) lema (B)

Kambera tílu (A) tuku (D) hondu (D) làma (B)

Hawu t↔lu (A) ≡igo (E) ≡oro (E) we?o (C)

Tetun tolu (A) tuda (A) kesi (F) nanan (D)

English three throw tie tongue

44) PMP ipen/nipen (A) kahiw (A) tuqu (A) biliN/

puter (A/B)

Bimanese voi (B) haju (A) poda (B) vari (C)

Manggarai Nis (C) haju (A) tu?u (A) seléko (D)

Kambera Nàndu (D) (piNi) ai (A) tuba (C) hambútiru (E)

Hawu Nutu (E) ayu (A) ------ ≡↔li (C)

Tetun nihan (A) ai (A) lolos (D) fali (C)

English tooth tree/wood true turn

45) PMP duha (A) um-utaq (A) lakaw/ ma-panas (A)

panaw (A/B)

Bimanese ua (A) lohi (B) lampa (C) pana (A)

Manggarai sua (A) roa (C) lako (A) kolaN (B)

Kambera dua (A) muta (A) laku (A) mbanahu (A)

Hawu ue (A) m↔du (A) bela (D) pana (A)

64

Tetun rua (A) muta (A) la?o (A) manas (A)

English two vomit walk warm/hot

46) PMP baseq (A) danum/ k-ita (A) k-ami

(A)

wahiR (A/B)

Bimanese vaca (A) oi (B) ndai ita (A) nami (A)

Manggarai waca (A) waé (B) ita (A) ami (A)

Kambera baha (A) wai (B) nyu-ta (A) nyu-ma (B)

Hawu ≡aha (A) ei (B) di (B) yi (C)

Tetun fasi (B?) ue (B) ita (A) ami (A)

English wash water we (incl.) we (excl.)

47) PMP ma-baseq (A) apa (A) p-ijan (A) i-nu (A)

Bimanese mbeca (A?) au (B) ≡une ai (B) i≡e (B)

Manggarai baca (A) apa (A) ca pisa (A?) baté (C)

Kambera mbaha (A) Ngara (C) piraNu (A) Ngí (D)

Hawu bo≡o (B) Na (D) p↔ri (A) mi (E)

Tetun bokon (C) sa (E) ba?i hira (A) iha ne?e be (F)

English wet what when where

48) PMP ma-putiq/ i sai (A) ma-labeR/ qasawa (A)

burak (A/B) belaj (A/B)

Bimanese bura (B) cou (B) paja (C) vei (B)

Manggarai bakok (C) cai (A) akék (D) wina (B)

Kambera bàrahu (D) Ngá (C) mbàlaru (B) paha (C)

Hawu pudi (A) naduu (D) b↔la (B) ihu ↔mu (D)

Tetun mutin (A) se (E) luan (E) feen (B?)

English white who wide wife

49) PMP haNin (A) panij/qelad (A/B) bahi/ quma (A)

b-in-ahi (A)

Bimanese aNi (A) kalete (C) sive (B) karavi

(B)

Manggarai buru (B) lebé (D) wina (A) gori (C)

Kambera Nilu (C) kàpa (E) kawini (A) yeli (D)

Hawu N↔lu (C) ↔la (B) b↔ni (A) y↔ga (E)

Tetun anin (A) liras (F) feto (C) knaar (F)

English wind wing woman work

50) PMP ma-huab (A) taqun (A) ma-kunij (A) kamu (A)

Bimanese mava (A) mba?a (B) monca (B) ndai oho (B)

Manggarai Noap (A) ntauN (A) té?é (C) méu (C)

Kambera mawa (A) ndauNu (A) iju (D) nyu-mu (A?)

Hawu ------ tou (A) kelara (E) mu (A)

Tetun maas (A) tinan (C) modok (F) emi (D)

65

English yawn year yellow yo

APPENDIX 2

Reflexes of PMP in Kambera, Hawu or both12

PMP PSH Kambera Hawu Gloss

01. *kaka *aka (?) ----- a?a eSb

02. *halas *alas alahu ----- forest

03. *qahelu *alu alu alu pestle

04. *ama *ama ama ↔ma F, FB

05. *Ramut *amo amu amo root

06. *anak *ana ana ana child

07 *anay *ane maN-aní (?) ----- termite

08. *qapuR *apu ----- ao lime for betel

09. *hapuy *api epi ai fire

10. *apuN *apuN apuNu ----- float

11. *huaji *ari eri ari y//sb

12. *qajeN *aruN aruNu ----- charcoal

13. *asu *asu ahu ----- dog

14. *qatay *ate eti ade liver

15. *qauR *au au ----- bamboo sp.

16. *qawa *awa awa ----- milkfish

17. *awaN *awaN awaNu ----- atmosphere

18. *qabu *awu aü awu ash

19. *kahiw *ayu ai ayu wood, tree

20. *babaq *baba ka-baba ≡a≡a short

21. *bahi *bai bai ----- female animal

22. *bayu *bai bai ----- pound rice

23. *bales *balas balahu ----- return in kind

24. *balik *bali weli ≡ale return

25. *baRu *baRu ka-baru/waru ----- hibiscus

26. *baseq *basa (b)aha ≡aha wash

27. *bekaq *b↔ka bàka, wàka ≡↔ka split

28. *beNkuq *b↔Nku bàNgu, wàNgu -----

bent, curved

29. *bikaq *bika bika ----- split

30. *b-in-ahi *bine wini b↔ni female

31. *betas *botas (?) botahu ----- break off

32. *bejbej *bubur búburu ----- bundle

33. *bukas *bukas bukahu boka to open

34. *bulat *bulak (?) bulaku ----- open eyes

35. *buNkuq *buNku buNgu ----- bent from age

36. *beRsay *buse (?) búhi ----- paddle

37. *buta *buta ----- ≡↔du blind

12Sets are listed in alphabetical order by the shape of the Proto-Sumba-Hawu etymon.

66

38. *dalem *dal↔dalu ara in, inside

39. *daya *daya dia ----- upstream

40. *dekit *d↔kit ka-dàkitu ----- stick to

41. *di atas *di atas dítahu ida above; high

42. *diNdiN *dindiN dindiNu didi wall

43. *duha *dua dua ue two

44. *qelaj *↔la ----- ↔la wing

45. *enem *↔n↔m nomu ↔na six

46. *epat *↔pat patu ↔pa four

47. *empu *↔pu àpu ↔pu grandmother

48. *hikan *iaN iyaNu ----- fish

49. *iluR *ilu ilu ilu saliva

50. *ina *ina ina ina mother

51. *hinuq *inu ----- inu beads

52. *inum *inuN unuNu N-inu drink

53. *hipun *ipuN ipuNu ----- small fish sp.

54. *qiris *iris irihu ----- slice

55. *isa *isa iha, diha ↔hi one

56. *isi *isi ihi ihi meat; contents

57. *hituq *itu etu, itu ----- catfish

58. *qihu *iu iu ----- shark

59. *zauq *jau rau ya?u far

60. *zalan *jalan ----- yara road

61 *hazani *jani (?) ka-reni, ma-reni ----- near

62. *zaRum *jaRuN (?) ----- yau needle; sew

63. *zi(R)uq *jiu ----- yiu bathe

64. *zukut *ju?ut ----- yu?u grass

65. *kempuN *kampu kambu ----- belly

66. *ka-mudehi *kamuri kamuri ----- rudder

67. *kaRaw *kao kau kao scratch an itch

68. *kaRat-i *kati kati ----- bite

69. *kahu *kau -kau au 2sg

70. *ka-wanan *kawana kawana ----- right side

71. *kawit *kawit kaitu/ketu kawa (?) hook

72. *kali *keli maNeli kei dig

73. *kedeN *k↔d↔N ----- k↔i stand

up

74. *kali-wati *k↔lati ----- kelate earthworm

75. *ku-labaw *k↔lawo kalaü ----- rat, mouse

76. *kali-maNaw *k↔limaNo kalimaNu ----- mangrove crab

77. *kamali *k↔mali ----- kemali house

78. *kutana *k↔tana katana ----- ask

79. *katadu *k↔taru kataru ----- caterpillar sp.

80. *kuliliN *kili (?) kili ----- surround

81. *kuhkuh *kuku13 wú kuu nail, claw

13 Reconstruction supported by Anakalangu kuku, Kodi ghughu ‘nail, claw’.

67

82. *kulit *kulit kalitu kuri skin

83. *kuluR *kulu kulu ----- breadfruit

84. *qudaN *kuraN kuraNu k↔ru

shrimp

85. *labiq *labi lebi ----- excess

86. *labuq *labu labu ----- fall; anchor

87. *laRiw *lai pa-lai pe-rai flee; run

88. *lakaw *lako laku lako go, walk

89. *lalej *lal↔----- lara housefly

90. *lasuq *lasu lahu ----- scrotum; penis

91. *lahud *laur lauru/luru/laura lou downriver; sea

92. *lawaq-an (?) *lawaN lawaNu -----

spiderweb

93. *layaR *layaR líru lai sail

94. *laqia *layia layia ----- ginger

95. *leku *l↔ku ----- l↔ku fold

96. *qali-matek *l↔mak↔t lamakatu ----- leech

97. *leNa *l↔Na làNa ----- sesame

98. *lepet *l↔p↔t (?) lípa l↔pa wrap; fold

99. *liaN *liaN liaNu lie cave, grotto

100. *lilin *liliN liliNu lili wax, candle

101. *lima *lima lima l↔mi five

102. *liuq *liu liu ----- circumvent

103. *qalejaw *loo lodu loo day, sun

104. *lulun *luluN ----- lule roll up (mat)

105. *lumut *lumut lumutu ----- moss, algae

106. *maRi *mai mai mai come

107. *malip14 *malip (?) ----- mari laugh

108. *manuk *manu manu manu chicken

109. *mata *mata mata mada eye

110. *ma-etaq *mata mata mada raw

111. *ma-atay *mate meti made dead; die

112. *ma-huab *mawap mawa ----- yawn

113. *ma-buhek *mawuk maüku mawo drunk

114. *ma-baRiw *mbai mbai ----- spoiled

115. *ma-baRani *mbani mbeni ≡ani brave

116. *ma-baqeRu *mbaRu mbaru wiu (?) new

117. *ma-baseq *mbasa mbaha ----- wet

118. *ma-belaN *mb↔laN mbàlaNu ----- spotted

119. *ma-belaj *mb↔lar mbàlaru, mbelaru b↔la broad, flat

120. *ma-besuR *mb↔su mbíhu ≡ehu satiated

121. *ma-beRat *mb↔t mbotu ----- heavy

122. *ma-iRaq *mea mí mea red

123. *ma-laNu *m↔lanu (?) ----- melanu dizzy

124. *minaNa *m↔naNa manaNa menaNa estuary

14 4malip ‘laugh’ is an innovation restricted to Central-Eastern Malayo-Polynesian languages.

68

125. *takaw *m↔–na?o (?) ----- mena?o steal

126. *nasuk *m↔nasu manahu ----- boil; cook

127. *ma-nipis *m↔nipi manipa m↔ni thin, tenuous

128. *ma-takut *m↔nta?ut mandauta meda?u fear, afraid

129. *ma-tuduR *m↔nturu (?) mahuru ----- sleep

130. *manuka15 *m↔nu?a manua ----- wounded

131. *ma-ñilu *m↔ñilu ma-yilu meñilu sour

132. *ma-Nuda *m↔-Nura (ma)-Nura (me)-N↔ru young; green

133. *ma-tuqah *m↔tua matua ----- mature, old

134. *miñak *miña mina m↔ñi fatty,

greasy

135. *ma-isa *misa ----- miha alone

136. *ma-qitem *mit↔mmítiNu m↔di black

137. *ma-panas *mpanas mbanahu, panahu pana warm, hot

138. *ma-pataq *mpata mbata pada broken

139. *ma-penuq *mp↔nu mbínu ----- full

140. *ma-putiq *mputi ----- pudi white

141. *ma-Ruqanay *muane (?) míni mone male

142. *muRmuR *mumu mumu ----- rinse mouth

143. *muntay *munte mundi ----- citrus

fruit

144. *ma-qudip *murip miripu muri living, alive

145. *musuq *musu muhu muhu enemy; war

146. *um-utaq *muta muta m↔du vomit

147. *nabuq *nabu ka-nabu menawu fall

148. *nanaq *nana ----- nana pus

149. *anaduq *nanu (?) ----- nano long

150. *naNuy *naNi Neni (< met.) naNe/naNi swim

151. *naNka *naNka naNga -----

jackfruit

152. *qunap-i *napi nepi nai fish scale

153. *nabek *naw↔----- nawa wave, surf

154. *niknik *nini (?) ----- nini gnat, midge16

155. *nipi *nipi ----- ni dream

156. *qanitu *nitu ----- nidu ghost

157. *niuR *niu (?) ----- ñiu coconut

158. *-nta *nta -nda ----- 1pi genitive

159. *taqun *ntauN ndauNu tou year

160. *qulun-an *nulaN nulaNu n↔lu headrest

161. *nusa *nusa nuha ----- island

162. *ñamñam *ñami ----- ñami tasty; chew17

163. *maN-kaen *NaN NaNu Naa/Nae eat

164. *NaNa *NaNa NaNa NaNa gape

15 5robably PCEMP; cp. POC *manuka ‘wound, ulcer’.

16 Kambera niniru ‘tiny mosquito’ appears to be related, but the final syllable is unexplained.

17 Cp. POC *ñami ‘taste, tasty’.

69

165. *Najan *NaraN18 Nara Nara name

166. *sa-Na-Ratus *Natu ha-Nahu ----- hundred

167. *sa-Na-puluq *N↔pulu ka-mbulu Nuru group of ten

168. *kita *Nita Nita N↔di see19

169. *paqit *pait paita ----- bitter

170. *pandak *pandak pandaku ----- short

171. *paniN *paniN pàniNu ani bait

172. *paniki *pani?i paní ni?i flying

fox

173. *panaw *pano panu ano leucoderma

174. *papan *papa ka-papa ai papa plank, board

175. *paku *pa?u paü ----- fern

176. *pajay *pare pari are rice

177. *paRih *paRi pai/pari ----- stingray

178. *pahuq *pau pau pau mango

179. *hapejes *p↔da ----- p↔a sting; pain

180. *qapeju *p↔du pídu ↔u, p↔u gall; bitter

181. *laRiw *p↔lai palai perai run, flee20

182. *pa-lihi *p↔lili (?) palili ----- taboo

183. *mula *p↔mula pamula ----- to plant

184. *paNdan *p↔ndaN pàndaNu ----- pandanus

185. *peñu *p↔ñu ----- ↔ñu turtle

186. *ma-pia *pia ----- ie good

187. *piliq *pili pili ≡ili/pili choose

188. *pintu *pintu pindu ----- door

189. *pipi *pipi pipi ----- cheek

190. *pija *pira pira p↔ri how much?

191. *p-ijan *piraN piraNu ----- when?

192. *picik *pisik ----- pihik splash

193. *pitu *pitu pihu pidu seven

194. *putput *puput puputu ----- blow on

195. *puki *pu?i ----- u?i vulva

196. *pusej *pus↔puhu ↔hu navel

197. *pusuq *pusu puhu uhu heart

198. *daki *ra?i rei ra?i dirt

199. *dapuR (?) *ra?u raü ra?u hearth

200. *datu *ratu ratu ratu clan priest

201. *dahun *rau rau/rú rou/ru leaf, hair

202. *daRaq *raya ria ra blood

203. *deNeR *r↔N↔y (?) reNu/roNu r↔Ni hear

204. *depa *r↔pa ràpa r↔pa fathom

18 Kambera Nara ‘name’shows unexplained loss of the final consonant, possibly because it was reanalyzed

as part of the possessive suffix in earlier *Naran-na. It is reconstructed here on the basis of the causative

form pa-NaraNu ‘give a name to’.

19 Also Kambera ita ‘see’.

20 Evidently from a causative form *pa-laRiw ‘cause to flee’. Also Kambera lai ‘run’, only in the

expression njara lai ‘racing horse’.

70

205. *ditaq *rita rita ----- Alstonia sp.

206. *duRi *rui rí rui bone

207. *rusrus *rurus ruruhu ----- slip off; drag

208. *duyuN *ruyuN riNu -----

dugong

209. *Rusuk *Rusu (?) ----- ruhu rib

210. *salaq *sala/nsala hala, njala hala fault, error

211. *saNa *saNa/nsaNa haNa, njaNa haNa bifurcation

212. *sakay *sa?e hei ha?e climb

213. *sebu *s↔bu hàbu h↔≡u douse fire

214. *sumaNed *s↔maN↔hamaNu hemaNa soul

215. *sendiR *s↔ndi hàndi ----- prop up

216. *sepaq *s↔pa hàpa h↔pa chew betel

217. *seksek *s↔s↔k hàhaku ----- stuff, cram in

218. *cilak *silak hilaku ----- glow, shine

219. *sipit *sipit hipitu ----- pinch

220. *sisuq *sisu híhu ----- edible snail

221. *siku *siu hiu wo-hiu elbow

222. *siwa *siwa hiwa heo nine

223. *suluq *sulu hulu huru torch

224. *suduk *suru huru huru spoon, ladle

225. *susu *susu huhu huhu breast

226. *suksuk *susuk huhuku ----- penetrate

227. *taqi *tai tai dei feces

228. *talih *tali ----- dari rope

229. *tamiN *tamiN temiNu tami

shield

230. *taneq *tana tana ----- earth

231. *tanem *tan↔m taniNu tana bury

232. *taNis *taNis (?) ----- taNi cry, weep

233. *taRum *tao (?) ----- dao indigo

234. *tada *tara tara dara cockspur

235. *tasik *tasik tehiku dahi sea

236. *tau *tau tau dou person

237. *taRutuN *tetu (?) tetu ---- porcupine fish

238. *tebuh *t↔bu tíbu ↔bu sugarcane

239. *tebek *t↔bu (?) ----- t↔≡u stab

240. *teduN *t↔duN ----- t↔u hat

241. *teka *t↔ka tàka d↔ka arrive

242. *tulali *t↔lali taleli ----- flute

243. *telen *t↔l↔N (?) ----- d↔la/d↔le to swallow

244. *taliNa *t↔liNa taliNa ----- ear

245. *telu *t↔lu tílu t↔lu three

246. *qateluR *t↔lu tílu d↔lu egg

247. *qati-mela *t↔m↔la----- tem↔la flea

248. *timeRaq *t↔maRa (?) ----- tem↔ra tin, lead

71

249. *tamiaN *t↔miaN tamiaNu temie bamboo sp.

250. *tinaqi *t↔nai tanai tenei guts

251. *tenun *t↔nuN tínuNu ----- weave

252. *teNaq *t↔Na ----- t↔Na middle

253. *tiNadaq *t↔Nara taNara teNara look

upward

254. *tahep-i *t↔pi tàpi ----- winnow

255. *tepiR *t↔piR ----- d↔pi mat

256. *teriN *t↔riN tariNu ----- bamboo sp.

257. *tumah *tima (?) ka-tima ----- body

louse

258. *timuR *timiR (?) timiru ----- east monsoon

259. *qatimun *timun ----- dimu cucurbit

260. *tiRem *tio (?) tiu ----- oyster; mussel

261. *tuqud *tu (?) ----- tu knee

262. *tuba *tua tuwa ----- derris root

263. *tuak *tua (?) ----- due palm wine

264. *tumbuq *tumbu tumbu du≡u grow

265. *tumbuk *tumbuk tumbuku ----- hit, pound

266. *tuna *tuna tuna d↔no eel

267. *tunu *tunu tunu tunu burn

268. *tuktuk *tutuk tútuku tutu beat; peck

269. *quban *uaN uwaNu ----- gray hair

270. *quay *ui (?) iwi g-ui rattan

271. *bulat *wulak (?) wulaku ----- open eyes

272. *hulaR *ulaR ularu ----- snake

273. *qulin *uliN uliNu ----- rudder

274. *Rumaq *uma uma ↔mu house

275. *umpu *umpu umbu ----- grandparent

276. *qumay *ume umi ----- unicorn fish

277. *qunej *un↔unu ----- pith

278. *añem *uñaN (<M) unaNu ↔ñu plait, weave

279. *upas *upas upahu ----- poison

280. *quzan *uraN uraNu ↔

yi rain

281. *uRat *urat ura/uratu ----- lines on palm

282. *kuden *ur↔N wuruNu ↔ru cooking pot

283. *qutan *utaN utaNu ----- bush; forest

284. *kutu *utu wutu udu louse

285. *baRa *wa ----- wa lungs

286. *bahaq *wa ----- wa flood

287. *wahiR *wai wai ei water

288. *walu *walu walu aru eight

289. *wani *wani ----- oni honeybee

290. *baNa *waNa waNa ----- open (mouth)

291. *badas *wara wara ----- gravel, sand

292. *habaRat *waRat waratu wa west(wind)

72

293. *batad *watar wataru ----- millet; maize

294. *batu *watu watu wadu stone

295. *bahuq *wau wau wou stench

296. *babaq *wawa wawa wawa beneath

297. *babuy *wawi wei wawi pig

298. *beNkas *w↔kas wàkahu, wàNgahu ----- untie, undo

299. *belaj *w↔las (?) wàlahu ----- spread out

300. *beli *w↔li wíli w↔li buy; price

301. *beluk *w↔luk wàluku ----- bend; bulge

302. *betak *w↔ta ----- w↔ta split

303. *beRay *w↔ye (?) ----- wie give

304. *biRaq *wia wí ----- itching taro

305. *binehiq *wini wini wini seed

306. *buni *wini (?) wini ----- ringworm

307. *bities *wisi (?) wihi ----- calf; leg

308. *bisul *wisu ----- wihu pimple, boil

309. *kuRita *wita (?) wita ----- octopus

310. *bibiR *wiwi wiwi ----- lip

311. *buaq *wua wua wo/wue/≡ue fruit

312. *buqaya *wuaya wúya woe crocodile

313. *bulan *wulaN wulaNu ↔ru

moon

314. *buliR *wuli wili wuri ear of grain

315. *bulu *wulu wulu wuru hair, feathers

316. *buni *wuni ----- wuni hide, conceal

317. *bujaq *wura wura ----- foam

318. *busuR *wusu ----- wuhu hunting bow

319. *bubu *wuwu ----- wuwu fish trap

APPENDIX 3

The following is a partial list of cognate sets that appear to be confined to languages of

the proposed Sumba-Hawu group, here represented by its two best-described languages,

Kambera and Hawu. Sets are ordered alphabetically by the PSH form.

PMP PSH Kambera Hawu English

01. *hadu *ae aï ae much

02. *qaRus *aRis arihu ri current21

03. *busuR *d↔ka dàka r↔ka bow; weapon

04. *iro iru ↔ri pull, drag

05. *daNdaN *jiru diru me-ñiru warm oneself

21 Possibly from PMP *qaRus with irregular *u > i. If cognate the Hawu form shows unexplained loss of

the initial vowel.

73

by a fire (?)22

06. *k↔bia kabia kebe to bleat

07. *NarNar *k↔pipi kapipi kepipi sherd,

fragment

08. *kutana *k↔rai karai kerei to ask

09. *Rusuk *k↔rasakaraha keraha rib

10. *RantaN *k↔ruku karuku keruku basket

11. *wasay *k↔taka kataka ketaka axe

12. *tiktik *k↔t↔kukatiku ked↔ku tap, rap

13. *k↔wuru kawuru kewuru

whisper

14. *lamug *kiwu kiu kiwu mix

15. *ki *la la la toward

16. *daqan *lai lai ke-lai branch

17. *lamat (?) lamata lama k.o. vegetable

18. *paNdan *lanta landa lata pandanus

19. *qatuR *laRa lara raa regulate (?)

20. *qajeN *latu ka-latu ladu charcoal

21. *lau ka-lau rou milt

22. *batu *l↔misi lamihi lamuhi23 pit, seed of

fruit

23. *lepi(q) *l↔pa lípa l↔pa fold24

24. *tudaq *liNgu líNgu ligu throw

25. *mudehi *liu liu ke-riu behind

26. *wahiR *luku lúku loko river

27. *ma-esak *mami memi mami ripe25

28. *pa-qudip *maNgaN maNgaNu maga look after,

care for

29. *qaniNu *mawo maü mawo shadow

30. *belaq *mbeNu mbuNu ≡eNu piece,

classifier

31. *belbel *mb↔Na mbàNa ≡eNa26 mute; stupid

32. *liNaw *mbula mbula b↔lu forget

33. *lebleb *mbulu ta-mbulu bolo sink

34. *nabuq *mbunu mbunu bunu fall

35. *beRNi *m↔dasmàdahu m↔a night

36. *ma-lemaq *m↔ja màja m↔je27 soft, weak

37. *ma-tudan *m↔j↔jiN mandidiNu me-y↔i sit

38. *m↔ka màka m↔ka can, able

39. *pasaqan *m↔lai malai merei carry on

shoulder pole

22 Presumably from earlier *jiru, with prefixation and nasal substitution in the Hawu form.

23 Expected **lemihi.

24 Presumably distinct from e.g. Malay lipat ‘fold’, which should correspond to Kmb lípatu, Hawu lípa.

25 Possibly a reflex of PMP *me-emis ‘sweet’, but with semantic change.

26 Expected **b↔Na.

27 Expected **m↔ja.

74

40. *ma-niwaN *m↔lakaN malakaNu melaka thin, of people

41. *belaq *m↔ndua man-dua me-due half

42. *ma-talaw *m↔NgiNgit maNgiNgitu me♥i♥i afraid, timid

43. *m↔Ni màNi m↔Ni luck

44. *mera (?) mera mira equal, similar

45. *ma-lemaq *momo múmu momo soft

46. *bituqen *moto mutu moto star

47. *gatel *mp↔di mbàdi p↔i itch, itchy

48. *punti *muku muku mu?u banana

49. *gurgur/ *ndoro nduru oro thunder

*kudug

50. *iriN *ndutu ndutu dutu follow

51. *nusi nuhi nuhi GGP, GGC

52. *suksuk *nusuk nuhuku nuhu insert, put in

53. *haNin *N↔lu Nilu N↔lu wind

54. *alap *N↔ndi (?) Nàndi N↔du bring

55. *kepit *N↔pit Nàpitu Napi pinch

56. *biriN *Nidi 0Niri Nidi side, edge

57. *Nidis (?) Nidihu Niu to sob

58. *ipen *0Niti (?) Niti Nutu tooth

59. *tutur *pane paní ane say, speak

60. *be(N)kaR *p↔ku ka-pàku p↔ku to break

61. *p↔Nga paNga p↔ga step,

step over

62. *butbut *p↔pu (?) pàpu pu pick, pluck

63. *palu *p↔sa (?) pàha p↔hi hammer

64. *pitik (?) pitiku pudu bore into (?)

65. *bataN *pola (?) pola la trunk, stem

66. *qalad *pundit pundi pudi fence

67. *tuRun *puru puru puru descend

68. *terter/tirtir *r↔Ngu ràNgu ke-r↔gutremble, shake

69. *ma-diNdiN *riNu (?) riNu me-riNi cold

70. *lalaw *risi rihi rihi more, excess

71. *kidkid *rondaN (?) rondaNu rodo saw, file

72. *kusu *rusuk ruhuku roho/rohe rub

73. *samba hamba haba bucket

74. *malat (?) *s↔mala hamala hemala sword

75. *tunu *s↔N↔y (?) hoNu h↔Ni roast, bake

76. *hidem *s↔Nod haNondu h↔No keep

still,

silent

77. *baqak *s↔po hàpu h↔po break off

78. *bitik *siRu (?) hiru hi snare trap

79. *sope ka-hopi hope small basket

80. *kelas *sosi (?) hehi, híhi hohi to peel

75

81. *beNap *talu (?) Na-taru ma-dalu

surprised,

amazed

82. *kilala *tanda tanda tada know (a

person)

83. *takeb *taNa taNa ke-taNa cover,

lid

84. *t↔mbak (?) tumbaku t↔bo butt with

horns

85. *sauq *t↔naNga tanaNga tenaga anchor

86. *sepet? *titiN títiNu titi dam in

a river

87. *ma-qudip *tolis túlihu toli live

88. *tudaq *tuku tuku tuku throw

89. *bunuq/ *tukuN tukuNu tuku

hit, strike

*sambak

90. *quntu *uli (?) uli ↔li tusk, canine

tooth

91. *nunuk *w↔Nga wàNga w↔ga banyan

92. *buaq *w↔ñu (?) winu w↔ñi areca palm

93. *hiNus *wira wira w↔ri snot

94. *woka woka woka digging stick

95. *butuq *wotak wota woakapenis; testicles

96 *bulat *wole woli ke-wore round

97. *kiday *wuku wúku mata ke-wuku eyebrow

REFERENCES

Blust, Robert. 1976. A rediscovered Austronesian comparative paradigm. Oceanic

Linguistics 16:1-51.

Blust, Robert. 1983/84. More on the position of the languages of eastern Indonesia.

Oceanic Linguistics 22-23:1-28

Blust, Robert. 1991. The Greater Central Philippines hypothesis. Oceanic Linguistics

30:73-129.

Blust, Robert. 1993. Central and Central-Eastern Malayo-Polynesian. Oceanic

Linguistics 32:241-93.

Blust, Robert. 1996a. Some remarks on the linguistic position of Thao. Oceanic

Linguistics 35:272-94.

76

Blust, Robert. 1996b. The Neogrammarian hypothesis and pandemic irregularity. In

Mark Durie and Malcolm Ross, eds., The comparative method reviewed: regularity and

irregularity in language change. New York and Oxford: Oxford University Press.

Blust, Robert. 1997. Ablaut in western Borneo. Diachronica 14:1-30.

Blust, Robert. 2000. Why lexicostatistics doesn’t work: the ‘universal constant’

hypothesis and the Austronesian languages. In Colin Renfrew, April McMahon and

Larry Trask, eds., Time depth in historical linguistics, vol. 1:311-32. Cambridge: The

McDonald Institute for Archaeological Research.

Blust, Robert. 2004. Austronesian nasal substitution: a survey. Oceanic Linguistics

43:73-148.

Blust, Robert. 2007. Disyllabic attractors and anti-antigemination in Austronesian

sound change. Phonology 24:1-36.

Blust, Robert. N.d. Austronesian comparative dictionary. Available online at

ftp://ling.lll.hawaii.edu/pub/acd/.

Brandes, J.L.A. 1884. Bijdrage tot de vergelijkende klankleer der westerse afdeeling

van de Maleish-Polynesisch taalfamilie. Utrecht: P.W. van de Weijer.

Cense, A.A. 1979. Maksassaars-Nederlands woordenboek. Koninklijk Instituut voor

Taal-, Land- en Volkenkunde. The Hague: Nijhoff.

Chrétien, C. Douglas. 1965. The statistical structure of the Proto-Austronesian morph.

Lingua 14:243-70.

Donselaar, W.M. 1972. Aanteekeningen over het eiland Savoe. Mededeelingen van

wege het Nederlandsch Zendelingsgenootschap 16:281-332.

Dyen, Isidore. 1965. A lexicostatistical classification of the Austronesian languages.

Indiana University Publications in Anthropology and Linguistics, Memoir 19.

Supplement to IJAL 31.1. Baltimore: The Waverly Press.

Esser, S.J. 1938. Talen. In Atlas van tropisch Nederland. Het Koninklijk

Aardrijkskundig Genootschap in Samenwerking met den Topografischen Dienst in

Nederlandsch-Indië. The Hague: Nijhoff.

Fox, James J. 1977. Harvest of the palm: ecological change in eastern Indonesia.

Cambridge, MA and London: Harvard University Press.

Fox, James J. n.d. Unpublished Swadesh 200-word lists for 29 languages of the Lesser

Sunda chain.

77

Greenberg, Joseph H. 1970. Some generalizations concerning glottalic stops, especially

implosives. IJAL 36:123-45.

Grimes, Charles E. to appear. Hawu and Dhao in eastern Indonesia: revisiting their

relationship. In Michael Ewing and Marian Klamer, eds., Typological and areal

analyses: contributions from east Nusantara. Leiden: KITLV Press.

Grimes, Barbara F., ed. 2000. Ethnologue: Languages of the world. 14th ed. 2 vols.

Dallas, Tex.: SIL International.

Ismail, Mansyur, Muhidin Azis, M. Saleh Yakub, M. Taufik H, and M. Kasim Usman.

1985. Kamus Bima-Indonesia. Jakarta: Pusat Pembinaan dan Pengembangan Bahasa,

Departement Pendidikan dan Kebudayaan.

Jonker, J.C.G. 1893-1896. Bimaneesch-Hollandsch woordenboek, Bimaneesche

Teksten, Bimaneesche spraakkunst. Batavia: Verhandelingen van het Bataviaasch

Genootschap van Kunsten en Wetenschappen 47.1-2.

Jonker, J.C.G. 1903. Iets over de taal van Dao. In Album Kern (Opstellen geschreven

ter eere van Dr. H. Kern):85-89. Leiden: Brill.

Kern, Hendrik. 1892. Sawuneesche bijdragen: Volzinnen, samenspraken en

woordenlijst, met eene grammatische inleiding. BKI 41:157-96, 513-53.

Klamer, Marian. 1994. Kambera: A language of eastern Indonesia. Holland Institute of

Generative Linguistics. Dordrecht: ICG.

Lansing, J. Stephen, Murray P. Cox, Sean S. Downey, Brandon M. Gabler, Brian

Hallmark, Tatiana M. Karafet, Peter Norquest, John W. Schoenfelder, Herawati Sudoyo,

Joseph C. Watkins, and Michael F. Hammer. 2007. Coevolution of languages and genes

on the island of Sumba, eastern Indonesia. Proceedings of the National Academy of

Sciences 104.41:16022-16026.

Lebar, Frank M., editor and compiler. 1972. Ethnic groups of insular Southeast Asia,

vol. 1: Indonesia, Andaman Islands, and Madagascar. New Haven, Conn.: Human

Relations Area Files Press.

Needham, Rodney. 1980. Principles and variations in the structure of Sumbanese

society. In James J. Fox, ed., The flow of life: Essays on Eastern Indonesia:21-47.

Harvard University Press.

Onvlee, L. 1984. Kamberaas (Oost-Soembaas)-Nederlands woordenboek. Koninklijk

Instituut voor Taal-, Land- en Volkenkunde. Dordrecht: Foris Publications.

Owens, Melanie. 2000. Agreement in Bimanese. Unpublished M.A. thesis.

Christchurch, New Zealand: Department of Linguistics, University of Canterbury.

78

Ruhlen, Merritt. 1987. A guide to the world’s languages, vol. 1: Classification.

Stanford, California: Stanford University Press.

Sahidu, Abdul Karim. 1978. Kamus Bima-Indonesia. Typescript.

Schulte-Nordholt, H.G. 1972. Sumbawanese-Bimanese-Dompu. In Frank M. Lebar,

editor and compiler, Ethnic groups of insular Southeast Asia, vol. 1: Indonesia, Andaman

Islands, and Madagascar:69-73. New Haven: Human Relations Area Files Press.

Sneddon, J.N. 1989. The north Sulawesi microgroups: in search of higher level

connections. In Studies in Sulawesi linguistics, part I, ed. by James N. Sneddon:83-107.

NUSA: Linguistic studies of Indonesian and other languages in Indonesia, vol. 31.

Jakarta: Lembaga Bahasa Universitas Katolik Indonesia Atma Jaya.

Sneddon, J.N, and Hunggu Tajuddin Usup. 1986. Shared sound changes in the

Gorontalic language group: implications for subgrouping. Bijdragen tot de Taal-, Land-

en Volkenkunde 142(4):407-26.

Verheijen, Jilis A.J. 1967. Kamus Manggarai I: Manggarai-Indonesia. Koninklijk

Instituut voor Taal-, Land- en Volkenkunde van Nederlands-Indië. The Hague: Nijhoff.

Verheijen, J.A.J., and Charles E. Grimes. 1995. Manggarai. In D.T. Tryon, ed.,

Comparative Austronesian Dictionary: an introduction to Austronesian studies, Part I,

Fascicle I:585-92. Berlin: Mouton de Gruyter.

Walker, Alan T. 1982. A grammar of Sawu. NUSA 13. Jakarta: Universitas Atma Jaya.

Wijngaarden, J.K. 1896. Sawuneesche woordenlijst. Koninklijk Instituut voor Taal-,

Land- en Volkenkunde van Nederlands-Indië. The Hague: Nijhoff.

79