1 Innovative numerals in Malayo- Polynesian languages outside of Oceania Antoinette Schapper a & Harald Hammarström b Leiden University a , Universität zu Köln a , Radboud University b & Max Planck Institute for Evolutionary Anthropology b In this paper we seek to draw attention to Malayo- Polynesian languages outside of the Oceanic subgroup with innovative bases and complex numerals involving various additive, subtractive and multiplicative procedures. We highlight that the number of languages showing such innovations is more than previously recognised in the literature. Finally, we observe that the concentration of complex numeral innovations in the region of eastern Indonesia suggests Papuan influence, either through contact or substrate. However, we also note that socio-cultural factors, in the form of numeral taboos and conventionalised counting practices, may have played a role in driving innovations in numerals. 1. INTRODUCTION 1 There has been much discussion of the developments in innovative numeral formations in Oceanic languages. Galis (1960) observed that many AN languages to the north of New Guinea have exchanged the ancestral decimal system for various quinary systems. More recently, Dunn et al. (2008:739) observed that the decimal system in their sample of 22 western Oceanic languages was not very stable, having changed in almost half the languages to quinary systems. Blust (2009:274) suggests that the emergence of such different numeral systems in Oceanic languages is due to intensive contact and trading between Austronesian and Papuan language-speaking peoples. Smith (1988:51-53) similarly observed that counting systems were not particularly stable due to trading and exchange relations between AN and Papuan language speaking peoples in Morobe province of Papua New Guinea. However, he noted that the influence was not one way but that Austronesian languages could 1 We thank David Mead for his insights into South Sulawesi languages, and Marian Klamer and Leif Asplund for discussions on Sumba languages. We also thank David Kamholz for his insights into Cenderawasih languages and for help with access to the typescript Starrenburg 1915. Schapper further thanks Emilie T.B. Wellfelt for discussions on the cultural significance of numerals in Indonesia and Timor-Leste. Schapper’s research was conducted as part of an ESF-EuroCORES (EuroBABEL) research project with financial support from the Netherlands Organisation for Scientific Research.
56
Embed
Innovative numerals in Malayo- Polynesian languages ...harald/hhpub/schapper... · “additive numeral”: a numeral where the relation between components parts of a complex numeral
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Innovative numerals in Malayo-
Polynesian languages outside of Oceania
Antoinette Schappera & Harald Hammarströmb
Leiden Universitya, Universität zu Kölna, Radboud Universityb & Max
Planck Institute for Evolutionary Anthropologyb
In this paper we seek to draw attention to Malayo-
Polynesian languages outside of the Oceanic subgroup
with innovative bases and complex numerals involving
various additive, subtractive and multiplicative
procedures. We highlight that the number of languages
showing such innovations is more than previously
recognised in the literature. Finally, we observe that the
concentration of complex numeral innovations in the
region of eastern Indonesia suggests Papuan influence,
either through contact or substrate. However, we also
note that socio-cultural factors, in the form of numeral
taboos and conventionalised counting practices, may
have played a role in driving innovations in numerals.
1. INTRODUCTION1
There has been much discussion of the developments in innovative
numeral formations in Oceanic languages. Galis (1960) observed that
many AN languages to the north of New Guinea have exchanged the
ancestral decimal system for various quinary systems. More recently,
Dunn et al. (2008:739) observed that the decimal system in their sample of
22 western Oceanic languages was not very stable, having changed in
almost half the languages to quinary systems. Blust (2009:274) suggests
that the emergence of such different numeral systems in Oceanic
languages is due to intensive contact and trading between Austronesian
and Papuan language-speaking peoples. Smith (1988:51-53) similarly
observed that counting systems were not particularly stable due to trading
and exchange relations between AN and Papuan language speaking
peoples in Morobe province of Papua New Guinea. However, he noted
that the influence was not one way but that Austronesian languages could
1 We thank David Mead for his insights into South Sulawesi languages, and Marian
Klamer and Leif Asplund for discussions on Sumba languages. We also thank David
Kamholz for his insights into Cenderawasih languages and for help with access to the
typescript Starrenburg 1915. Schapper further thanks Emilie T.B. Wellfelt for discussions
on the cultural significance of numerals in Indonesia and Timor-Leste. Schapper’s
research was conducted as part of an ESF-EuroCORES (EuroBABEL) research project
with financial support from the Netherlands Organisation for Scientific Research.
2
not only lose their inherited decimal system but that Papuan languages can
also acquire it.
In this paper, we turn our attention to the comparatively little remarked
upon innovative formation of numerals in Malayo-Polynesian (MP)
languages outside of the Oceanic (OC) subgroup (cf., e.g., Ossart 2004).
Based on a survey of the numerals in 470 languages, this paper reports on
the results of the first systematic investigation of innovative complex
numerals in non-OC MP languages. The precise aims of our study are
threefold:
(i) to describe the variety of innovations of complex numerals
(e.g., 10-1 = 9) and of non-decimal numeral bases (e.g.,
base-5, base-20);
(ii) to draw attention to the concentration and diversity of such
innovative numeral formations in MP languages of eastern
Indonesia and East Timor, and;
(iii) give preliminarily suggestions as to the reasons for the
geographical skewing of such innovative numeral formation
in non-OC MP languages.
The paper is structured as follows. Section 2 provides an overview of
the terminology that we use to describe the different types of patterns of
numeral formation that are observed. Section 3 examines the variety of the
innovative ways of forming numerals in non-OC MP numerals. In
particular, we highlight the very limited number of numeral innovations in
WMP area (the Philippines, western Indonesia and mainland South-East
Asia) MP languages. This is contrasted with CEMP area (eastern
Indonesia and East Timor) MP languages which harbour at least a dozen
distinct innovations.2 In section 4 we seek to identify causal factors that
may have played a role in driving the multitude of numeral innovations in
the CEMP area. We observe that the areal concentration of complex
numeral innovations suggests influence from Papuan languages. We
further note that socio-cultural factors, in the form of numeral taboos and
conventionalised counting practices, are likely to have contributed to
numeral innovation. Section 5 concludes the discussion.
2. TERMINOLOGICAL PRELIMINIARIES
Numerals are ‘spoken normed expressions that are used to denote the
exact number of objects for an open class of objects in an open class of
social situations with the whole speech community in question’
(Hammarström 2010: 11). A numeral system is thus the arrangement of
individual numeral expressions together in a language.
2 Given the problematic nature of the WMP and CEMP nodes in the AN tree (see, e.g.,
Adelaar 2005), we use the terms “WMP area” and “CEMP area” to refer to broad
geographical regions in which MP languages are spoken and not to genealogical
groupings.
3
For the purposes of the present paper, a numeral system may be
classified as follows:
Base-5 (or quinary): if more than half the expressions 6-9 are
formed as 5+1, .., 5+4 respectively
Base-10 (or decimal): if more than half the expressions 20-99
are formed as x*10+y where x, y range from 1..9
Base-20 (or vigesimal): if more than half the expressions 20-99
are formed as x*20+y where x ranges from 1..9, and y
from 0..19
A numeral system may be both quinary and decimal or both quinary and
vigesimal (in fact, all bona fide attested quinary systems are also either
decimal or vigesimal – see Hammarström 2010). So, for instance, Pazeh
has a mixed-base numeral system in which numerals ‘six’ to ‘nine’ are
formed with a quinary base and higher numerals with a decimal base. By
contrast, Saisiyat has only a decimal base; ‘six’ does not constitute a base
in the language even though it is used in forming the numeral ‘seven’. The
fact that Saisiyat ‘six’ is limited to building only one other higher numeral
means that it does not meet the requirements for basehood as given above.
TABLE 1 ABOUT HERE.
Whilst Saisiyat ‘six’ does not constitute a base in the language, its use
in the numeral ‘seven’ draws attention to another kind of numeral
formation with which we are also concerned in this paper. We are not only
interested in the numeral bases in a language, but more broadly the
internal composition of numerals, that is, if and how numerals are made up
out of other numeral expressions. We call a monomorphemic numeral a
“simplex numeral”, and a numeral composed of several numeral
expressions a “complex numeral”. To describe (i) the arithmetical relation
between component elements in a complex numeral, and (ii) the role of
component elements in arithmetical operations3, the following terms are
used:
“additive numeral”: a numeral where the relation between
components parts of a complex numeral is
one of addition. The component parts are
“augend” and “addend”. So, for example,
in the equation 6+1 = 7, the augend is 6
and the addend is 1.
“subtractive numeral”: a numeral where the relation between
component parts of a complex numeral is
3 It is of course possible for arithmetical operations to be used in conjunction with one
another, e.g., 3x20 + 5+2 for ‘67’. Since these can still be accurately characterised with a
combination of the three basic operations (additive, subtractive and, multiplicative), we
restrict ourselves to these terms.
4
one of subtraction. The component parts
are “subtrahend” and “minuend”. So, for
example, in the equation 10−2 = 8, the
subtrahend is 2 and the minuend is 10.
“multiplicative numeral”: a numeral where the relation between
components parts of a complex numeral is
one of multiplication. The component parts
are “multiplier” and “multiplicand”. So,
for example, in the equation 3x2 = 6, the
multiplier is 3 and the multiplicand is 2.
Our analysis of numerals breaks down each number expression into
morphemes. The meaning, if known, of a morpheme can be inferred from
the meaning of it in isolation or inferred from the mathematical equation
the number expression constitutes.
Throughout this paper we rely on the definitions made in this section.
We repeatedly make use of the terms presented here and the reader is
referred to this section for clarification of any terminology.
3. DATA
A decimal counting system can be reconstructed for proto-Austronesian
(Blust 2009a: 268-274). This system is found spread throughout the
Austronesian world with easily recognisable cognates and it can be
reconstructed to various lower nodes of the Austronesian tree, such as
proto-Oceanic. This can be seen by comparing the reconstructed PAN and
POC numeral forms given in Table 2.
In the following subsections, we will see that multiple Austronesian
languages outside of the Oceanic subgroup have replaced these simplex
etymological numerals with innovative complex numerals. The majority
of our discussion deals with innovations in numerals ‘six’ to ‘nine’.
However, in cases of base changes (e.g., decimal > vigesmal) we also
discuss the expression of the numerals ‘ten’ and ‘hundred’. Unless
otherwise stated, however, the reader should understand there has been no
base change and that, for instance, *Ratus for ‘hundred’ is retained.
All data is cited in a unified IPA transcription from the earliest known
attestations to avoid interference from any post-historical changes.
However, unless otherwise noted, for all languages cited, all later
attestations (including own fieldwork by the second author on Bedoanas,
Erokwanas, Yaur, Yeresiam, Yeretuar and Wandamen in 2010) agree with
the earliest sources except for transcriptional matters irrelevant for the
present paper.
TABLE 2 ABOUT HERE.
3.1. Sumba
In three languages of western Sumba we find innovative numerals for
‘eight’ and ‘nine’. They are Lamboya, Kodi and Weyewa, and their
5
numerals are set out in Table 3. The data are from Wielenga (1917: 67)
and Leif Asplund (p.c., 2011), and [~] separates different or alternative
forms.
TABLE 3 ABOUT HERE.
The innovative ‘eight’ numerals in all three languages involve reflexes
of *Sepat ‘four’ and are taken to be a complex multiplicative numeral 4x2.
The forms for ‘eight’ appear to go back to a complex word composed of a
sequence of morphemes etymologically related to the causative morpheme
pV- followed by (n)do ‘two’ and pata ‘four’, for instance: Lamboya po-
do-pata ‘CAUS/VBZ-two-four’ ‘to make two fours’ (cf. Wielenga 1917: 67-
69; who translates Weyewa pondopata ‘eight’ as ‘two times four’).
‘Nine’ in Lamboya and Kodi are subtractive numerals involving a
reflex of *esa/isa ‘one’. Wielenga (1917: 67) suggests the etymology of
Kodi ɓanda iha ‘nine’ to be ‘the one not counted’. We interpret this to
mean that the word is composed of three morphemes: ba-nda-iha ‘COMP-
NEG-count’, a subtractive numeral ‘nine’ which literally means ‘[ten] one
not counted’. Note that today’s Kodi has a complementiser ba and a
The subgrouping of western Sumba languages is not well-understood.
As such, we adopt a conservative approach and posit two separate
innovations here. The first occurred in a hypothetical common ancestor of
Weyewa, Lamboya and Kodi and replaced a reflex of PAN *walu ‘eight’
with the multiplicative numeral. The second occurred in the common
ancestor of only Lamboya and Kodi and replaced *Siwa ‘nine’ with the
subtractive numeral.
3.2. Flores-Lembata
Table 4 presents an overview of the languages of Flores and Lembata
which show numeral innovations.
Numerals innovations in Flores are limited to a group of six
neighbouring languages in the centre of the island. In these, identical
complex numerals for ‘six’ to ‘nine’ have been innovated using several
different procedures: ‘six’ (5+1) and ‘seven’ (5+2) are quinary additive
numerals; ‘eight’ (2x4) is quaternary multiplicative, and; ‘nine’ ([10]-1) is
a subtractive numeral. The forms for ‘nine’ feature reflexes of PAN *isa
‘one’ and a negator *ta4, possibly followed by a reflex of an existential
verb ‘to be’, as in, for example, Rongga ta-ra-esa < ‘NEG-BE-one ‘(ten)
not/without one’.
Ende, Keo, Lio, Ngadha and Nage subgroup together (cf. Blust 2008b:
452). While Rongga’s affiliation is not discussed in the literature (e.g.,
4 Blust (1993) reconstructs a negator *ta ‘no, not’ to Proto-Central Malayo-Polynesian
(PCMP). Whilst the existence of the CMP subgroup is the subject of debate (see, e.g.
Donohue & Grimes 2008, Blust 2009b, Schapper 2011), it is clear that a negator *ta is
reconstructable to smaller sub-groupings in the eastern Indonesian area.
6
Blust 2008a) it has been suspected to subgroup with Ngadha rather than
Manggarai, as previously assumed (Arka 2009:90). It is likely that these
Flores languages form a low-level subgroup and as we such we treat the
shared numeral innovations as the result of a single innovation in a
common ancestor.
7
TABLE 4 ABOUT HERE
The numerals of Kedang, spoken Lembata island to the east of Flores, show a different
set of numeral innovations. Kedang ‘eight’ is formed by a quaternary multiplicative, as in
the central Flores languages. However, whilst the formatives in the Kedang complex
numeral, butu ‘four’ and rai ‘two’, appear to be related to those in the Flores languages’
quaternary multiplicative, there is an interesting difference: in Kedang the ‘four’ element
precedes the ‘two’ element, whereas in the Flores instances the ‘four’ element always
follows the ‘two’ element (e.g., Nganda rua butu ‘eight’ < rua ‘two’ plus butu ‘four’. In
contrast to the subtractive pattern in central Flores, Kedang ‘nine’ is an additive numeral
(5+4). Kedang is the only language on Lembata island that has innovative ‘eight’ and
‘nine’. The Lamaholot varieties spoken elsewhere on Lembata retain the PAN decimal
system. The Kedang numeral innovations thus appear to have occurred in that language
independent of other Austronesian languages in the area.
3.3. Timor
In three languages of Timor we find innovative additive base-five numerals for ‘six’ to
‘nine’ (Table 5). All three retain a dedicated lexeme for ‘ten’. The languages are
Tokodede spoken in North-Central Timor, Mambae spoken in Central Timor-Leste and
Naueti spoken in South-East Timor-Leste.
TABLE 5 ABOUT HERE
In Tokodede ‘six’ to ‘nine’ are formed with a conjunction, wou ‘and’ plus a digit ‘one’
to ‘four’. No numeral ‘five’ is present, but its value is merely understood from context,
i.e., ‘six’ is simply ‘plus one’, ‘seven’ ‘plus two’ and so forth. Mambae has a similar
pattern: ‘six’ to ‘nine’ are formed with a conjunction, nai, plus a digit ‘one’ to ‘four’. Lim
‘five’ is only optionally present. So, for instance, ‘seven’ can be expressed either as lim
nai rua ‘five and two’ or as nai rua ‘and two’. In Naueti kailima ‘five’ is always present
in additive numerals for ‘six’ to ‘nine’. The additive procedure is expressed by the
morpheme resi ‘plus, more’.
The numeral ‘six’ is noteworthy in all three languages because of the extra nasal
segment [n] they include. This reflects the PAN numeral ligature *ŋa (Blust 2009a: 269),
which appears to have been used exclusively to link units of ‘one’ in complex numerals.5
As a result, the Naueti numeral kailima resin ‘five plus’ does not appear so anomalous for
the absence of the numeral ‘one’, since the numeral ‘one’ can be inferred from the
presence of the ligature (resin < *resi-na).6
These additive five numerals in Timor appear to have been innovated independently.
This is almost certainly the case for Naueti as against Tokodede and Mambae. Naueti is
isolated from the other languages with innovative base-five numerals, and intervening
Austronesian languages have regular reflexes of inherited AN numerals for ‘six’ to
‘nine’.
5 Cross-linguistically, the use of ligatures for units of one is very common, e.g., French vingt-et-un ‘21’,
literally ‘twenty and one’, but vingt-deux ’22’, literally ‘twenty - two’. 6 Arnaud & Campagnolo (1998:V2:n868) have resina as opposed to resin here.
8
3.4. Western-central Maluku
Several closely related languages in the Western-Central subgroup of central Maluku7
have innovative subtractive numerals for ‘eight’ and ‘nine’. The forms are given in Table
6.8
TABLE 6 ABOUT HERE
Collins (1981:36) argues that the innovative numerals are reconstructable to the
common ancestor of the three languages. The numerals were in the proto-language
formed by combining a prefix *ta- signalling the substrative procedure with a reflex of
*Dua ‘two’ in ‘eight’ and *isa ‘one’ in ‘nine’. The proposed proto-forms are set out in
(1).
Proto-Buru-Sula-Taliabo numerals according to Collins (1981:36)
(1) 8: *ta-rua ‘minus two’ ~ *walu
9: *ta-sia ‘minus one’ ~ *siwe
Since Collins set-up a Sula-Taliabo subgroup to the exclusion of Buru, the cross-
cutting *ta-rua isogloss, common to Sula and Buru but not Taliabo, presents a
problematic innovation. To explain it Collins proposes that the Sula-Taliabo-Buru proto-
language also had etymological parallel forms and that the daughter languages chose to
retain only one form each; thus, the retention of the etymological walu in Taliabo. A
methodology that reconstructs parallel forms followed by ad hoc retentions can explain
any data and thus not count as an adequate explanation. Collins (1981)'s Sula-Taliabo
subgroup is argued on the strength of three phonological innovations r > h, t > c /_# and
ʔ > h. However, Collins (1981)'s own description of these innovations exposes their
weakness as evidence for a Sula-Taliabo subgroup. Firstly, the reflex of r in Buru is
uncertain, and may include Buru as well, according to Collins (1981:33-34)9. Secondly, t
> c is not observed in Sula (Sula shows t > Ø) – an intermediate c is only posited to
explain a vowel change (1981:32). But, as far as we can tell, this vowel change could just
as well have another another origin than an intermediate c. Thirdly, actually only Taliabo
shows ʔ > h, while Sula and Buru have ʔ > Ø, and Collins (1981:35) posits that Sula once
did undergo ʔ > h followed by a subsequent h > Ø. Clearly, a preferable solution is to
reject the Sula-Taliabo subgroup and posit a subgroup for Sula and Buru based on the
*ta-rua innovation and common phonological innovations such as ʔ > Ø and y > Ø. This
solution is also preferable to a borrowing scenario since the innovation extends to all
dialects of Sulaic and Buruic and, in any case, it would be odd to borrow only the
numeral ‘eight’.
7 Collins (1981, 1983:20) subclassifies the West Central Maluku languages Ambelau, Buru, Sula and
Taliabo. Collins (1989) explores some Taliabo dialects, of which Kadai is an outlier. Grimes (2000, 2009)
adds Hukumina and Lisela and Grimes and Grimes (1984) separate Mangole from Sula. All divisions are
argued for by phonological innovations. 8 Ambelau as recorded by Wallace (1869:623) is not part of the numeral innovations. 9 An anonymous reviewer points to unpublished work establishing that the reflex is
indeed /h/ “in nearly all well-established etymologies”.
9
The etymology of initial *ta- morpheme is equivocal. Collins (1981:43) suggests that
*ta- could be related to Soboyo ta- ‘towards’, but we find the semantic link between
subtraction and allative motion to be weak and not supported cross-linguistically (Hanke
2005). Similarly, we conclude that it is unlikely to have meant ‘ten’. Reflexes of ‘ten’ in
these languages (e.g., Buru polo) have no commonality with the form *ta-. Furthermore,
it is cross-linguistically usual for subtractive numerals to have some lexeme overtly
expressing the subtraction (Greenberg 1978: 259). Most notably, *ta- is identical to the
morpheme found in the Flores-Lembata languages’ subtractive numeral for ‘nine’ and
may be also a reflex of an earlier negator (cf. innovative negators in Buru moo ‘no, not’,
and Taliabo daaŋ ‘no, not’).
3.5. Aru
Complex numerals for ‘seven’ and ‘eight’ are found in the languages of Aru. The data
from seven Aru languages is set out in Table 7. The innovations are found in all Aru
languages for which we have data and appear to be reconstructable to their immediate
common ancestor, Proto-Aru (PARU).
TABLE 7 ABOUT HERE
The PARU numeral ‘eight’ is an innovative multiplicative numeral composed of *kawa
‘four’ and *rua ‘two’. This compound is still plainly evident in many of the modern Aru
language. For instance, Kola kafarua ‘eight’ is clearly composed from kafa ‘four’ and rua
‘two’.
The PARU numeral ‘seven’ is an innovative additive numeral: the innovative PARU
*dubu ‘six’ was compounded with another morpheme *sam apparently denoting the
additive operation ‘plus 1’. The components of this compound numeral are still readily
apparent in modern Ujir dubusam ‘seven’ and Dobel dubujam ‘seven’, while in other
languages the numeral forms have been reduced through the loss of a medial syllable
taken away (from ten)’ (Mills 1975: 229; Adelaar 1985: 137-138).14 Two languages in
particular stand out. The first is Minangkabau salapan ‘eight’ which apparently reflects a
compound *sa- ‘one’ + alap-an ‘take’ for ‘nine’ and not ‘eight’. Explaining this, Blust
(1981: 467 fn. 5) suggests that Proto-Malayic had *sǝmbilan and *salapan as synonyms
for ‘nine’, originally complex numerals based on the two different ‘take’ verbs. In
Minangkabau, he proposes, salapan shifted to mean ‘eight’ after the compound forms
became intransparent to speakers. The dual reconstruction of *sǝmbilan and *salapan is
supported by Seraway Middle-Malay which retains reflexes of both. Sundanese and
Makassarese have salapan and salapaŋ for ‘nine’ respectively. These numerals (along
with several others in each language) were presumably borrowed from Malay at a time
when the doublet for ‘nine’ was still present.15
TABLE 19 ABOUT HERE
Languages of the South Sulawesi subgroup all reflect innovative subtractive numerals
for ‘eight’ and ‘nine’ respectively. A selection of the numerals in six languages in the
subgroup is provided in Table 20.
TABLE 20 ABOUT HERE
The South Sulawesi proto-forms from which the innovative subtractive numerals are
thought to be descended are *karua(a) ‘eight’ and *kasera(a) ‘nine’. Both of the proto-
numerals follow the pattern *ka- + 'one / two' (+ *-a).16 The reconstruction of *kasera(a)
‘nine’ is somewhat problematic, because some languages of the subgroup do not reflect
*sera ‘one’ in their subtractive compound. Sirk (1989:62) explains the discrepancies in
the form of numeral ‘one’ in the innovative subtractive numerals:
‘..., it is likely that at the PSS stage there existed *kasera(a) 'nine' besides *sera ‘one’, but
somewhat later, when *sera got lost and the composition of the reflex of *kasera(a) became
14 One Malayic language, Banjarese, does not reflect these, having replaced reflexes of the proto-Malayic
‘eight’ and ‘nine’, with Javanese (Adelaar 1985: 138). 15 Under the wide ranging influence of Malay, one or more of these numerals have been borrowed into
several other Austronesian groups, for instance, the Tamanic (and Makasarese languages to name just a
few of many. However, since the borrowing occurred presumably after the complex origins of the
numerals had been subtracted, we don’t count them here as innovations proper. 16 David Mead (pers. comm.) notes that, “given that the Seko forms end in a long vowel, it is probable that
at the level of their common ancestor you would have to speak of a confix *ka- ... -aq, since the historical
source of long vowels in Seko is always -Vq. However, outside_ of Seko, *ka- ... -a would indeed be
correct.”
16
unclear to speakers, it was discarded in some dialects/languages in favour of a new derivative
from ‘one’: kamesaʔa, kamesa, kaassa, or the like. Such derivatives may have been formed by
analogy on the model of the word for 'eight', which had remained analysable in most
languages/dialects (true, the Seko word for 'eight' [karo'a] may cast doubt on such an
explanation.) The other possibility, which apparently conforms better to Seko data, is that
*kasera(a) and *kamesa(a) already existed side by side in different dialects at the PSS stage.’
Outside of Bugis, the Makassar languages, Campalagian and Mamuju, the forms mesa,
mesaʔ, meesa and meesaʔ are the usual responses for ‘one’. The discrepancies might also
be explained as the result of the diffusion of subtractive pattern throughout the subgroup
(Charles E. Grimes pers. comm.). Ideally, in that case, we would see proximate languages
outside the South Sulawesi subgroup also with the subtractive numerals, however, we do
not.17
One language of the subgroup, Makassarese, is exceptional in not reflecting the
innovative subtractive numerals. Makassar ‘seven’ and ‘nine’ are borrowings from
Malay. The formation of ‘eight’ is, however, notable, being an innovative additive
1 1 (e)sa esa sa haʔesa əsa esa 1 ɦudeʔ 2 2 ɹua zua zua ʔesa rua rua ɗua 2 sue 3 3 telu telu tela ʔesa tedu təlu telu 3 tælu 4 4 wutu vutu wutu ʔesa wutu sutu wutu 4 ɦapaʔ 5 5 lima lima lima ʔesa dima lima lima 5 leme 6 5+1 lima esa lima esa lima sa ʔesa dima ʔesa lima əsa lima esa 6 ɦænæng 7 5+2 lima ɹua lima rua lima zua ʔesa dima rua lima rua lima zua 7 pitu 8 2x4 ɹua mbhutu rua butu rua butu ʔesa rua mbutu rua mbutu zua butu 4x2 butu rai 9 [10]-1 tara esa ter esa tra sa ʔesa tera ʔesa təra əsa tea esa 5+4 leme ɦapaʔ 10 10 sambulu habulu sabulu ha mbudu sambulu sa bulu 10 pulu
36
Table 5: Timor languages with innovative numeral formation Tokodede
(Likisa)
[tkd]
Mambae
(Ainaro)
[mgm]
Naueti
[nxa]
Analysis Expression Expression Expression
1 1 iso id se 2 2 ru ru ~ rua kairua † 3 3 telo teul ~ tel kaitelu 4 4 pat fat ~ pat kahaa 5 5 lim lim kailima 6 5+1 wou niso (lim) nain ide kailima resin 7 5+2 wou ru (lim) nai rua kailima resi kairua 8 5+3 wou telo (lim) nai telu kailima resi kaitelu 9 5+4 wou pat (lim) nai pata kailima resi kahaa 10 10 sagulu sakul ~ sagul welisé
† Naueti numerals ‘two’ to ‘nine’ contain a fossilised classifier kai < PMP *kahiw ‘wood’
37
Table 6: Western-central Maluku languages with innovative numeral formation Buru
1 1 ot set etu ʔetu ~ je et ôt eti 2 2 rua rua rua Ro ru rua ru 3 3 las lati lasi laj laes lat la 4 4 kafa ka ka ʔawa kau ka kau 5 5 lima lima lima lima lim lêma lim 6 6 dum dubu dubu dubu dum dum dum 7 6+1 dubam dubusam dubem dubujam dubam dubám dobam 8 4x2 kafarua karua karua ʔaro karu kɔrua karu 9 9† tera tera tera jera sêr sêra ser 10 10 fuh uisia uraɸa ia wur urɸaiɸ urɸaɸaj urweu † PARU *tera ‘nine’ may also have been an innovative complex numeral. It is difficult to overlook the similarity between the forms of
Arunese ‘nine’ and the clearly subtractive numerals for ‘nine’ in the Flores and the Western-Central Maluku languages already
39
discussed in this paper. However, the absence of an identifiable morpheme denoting ‘one’ makes the case for seeing PARU ‘nine’
as a historically complex subtractive numeral slim.
40
Table 8: Western Cenderawasih languages with innovative numeral formation
† Fabritius (1855) gives utin for Tandia ‘twenty’. However, this lexeme is clearly related to the lexeme ‘hundred’ in other Cenderawasih
languages. No later sources verify utin for Tandia ‘twenty’, and we regard it here as a mistaken attribution of ‘hundred’ to ‘twenty’ in
the language.
‡ Dalrymple & Mofu (2012:21) note that they were unable to elicit the Dusner numeral ‘forty’.
‼ Empty cells in the table indicate that the numeral was not given in the source.
Table 15: Languages of the Onin group with innovative numeral formation Onin
[oni]
Sekar
[skz]
Uruangnirin [urn]
Analysis Expression Expression Expression
1 1 sa sa sa 2 2 nuwa nowa nua 3 3 teni tεni teni 4 4 fāt fāt fat 5 5 nima nima nima 6 6 nem nεm nem 7 [6]+1 tara sa tara sa taraŋ sa 8 [6]+2 tara nuwa taras nowa teri nua 9 10-1 sa puti sa puti sa puti 10 10x1 pusua pusua puca 20 10x2 puti nua puti nua ‡
‡ No form given in the source.
48
Table 16: Languages of Arguni-Bedoanas-Erokwanas group with innovative numeral formation Arguni
[agf]
Goras Erokwanas
[erw]
Fior Bedoanas
[bed]
Analysis Expression Expression Expression
1 1 sia sa ~ sia sia 2 2 ru ru ru 3 3 taur taur taur 4 4 fat vat fat 5 5 rim rim rim 6 6 anεm anjam anεm 7 6+1 ? nĕmbatu nambátu nĕmbatu 8 4x2 butεrua navu narwu 9 9 nεswε naswa nεswε 10 10 samburé sambura samburε 20 20x1 sinon sia sinon sa sinjon sa 40 20x2 sinon ru sinon ru sinjon ru 100 100 ratisa rati sa rati sε
49
Table 17: Languages of north Bomberai with innovative numeral formation Irarutu
[irh]†
Kuri
[nbn]
Analysis Expression Analysis Expression
1 1 eso 1 eso
2 2 rivu 2 ru
3 3 toru 3 tor
4 4 gegete 4 gegete
5 5 frada vida 5 fradĕβi
6 [5]+1 teresu 5+1 fra defi freso
7 [5]+2 tereru 5+2 fra defi freru
8 [5]+3 tereturu 5+3 fra defi fretor
9 [5]+4 teregite 5+4 fra defi fregégete
10 5x2 fradaru 5x2 fra dru
20 20 matuténi 20x1 tmatu tri eso
40 20x2 matuténi rivu 20x2 ‡
† Referred to as “Kaitero”.
‡ No form given in the source.
50
Table 18: Kowiai [kwh] numerals ‘one’ to ‘forty’ Analysis Expression
1 1 samosi
2 2 rueti
3 3 towru
4 4 fāt
5 5 rimi
6 5+1 rim samosi
7 5+2 rim rueti
8 5+3 rim towru
9 5+4 rim fāt
10 10 wutsja
20 10x2 seümbut rueti 40 10x4 seümbut fat
100 100 ratsja
51
Table 19: Numerals in Malayic languages (from Adelaar 1985:136)
† tujuh < ‘point’ from telunjuk ‘pointing finger’
Standard Malay Minangkabau Seraway Middle-Malay Iban Jakarta Malay Sundanese
1 1 seua mesa meesaʔ mesa mesa mesaʔ 1 sere 2 2 dua daddua dua dua dua duwa 2 ruwa 3 3 tǝllu tallu tallu tallu tallu italu 3 tallu 4 4 ǝppa appaʔ appaʔ appaʔ appaʔ upaʔ 4 appa 5 5 lima lima lima lima lima lima 5 lima 6 6 ǝnnǝng unun anan unun unung unung 6 annaŋ 7 7 pitu pitu pitu pitu pitu pitu 7 tuju 8 10-2 arua karua karua karua karua karoaʔa† 7+1 sagantuju 9 10-1 asera kasera kassera kasera kamesa kamesaʔa 9 salapaŋ 10 10 pulo sapulo sangpulo saʔpulo sappulo sappuloo 10 sampulo
† Grimes & Grimes (1987:131) give sakkupaʔang for Seko ‘eight’. We have been unable to find this form in any other source on Seko (e.g., Laskowske 2007,
Sirk 1989, Mills 1975) and therefore we discard it here.
53
Table 21: Ilongot [ilk] Analysis Expression
1 1 sit 2 2 dewa 3 3 teɣo 4 4 opat 5 5 tambiaŋ 6 5+1 tambiaŋ no sit 7 5+2 tambiaŋ no dewa 8 5+3 tambiaŋ no teɣo 9 5+4 tambiaŋ no opat 10 10 (na)puló
54
Table 22: Summary of innovations in non-OC MP languages