Numbers yet again - Re: New Book from SIL PNG

Malcolm Ross Malcolm.Ross at ANU.EDU.AU
Sat Jun 30 06:45:27 UTC 2007

I haven't been following this discussion, but perhaps I can offer  
some clarification with regard to comments I have just seen. If I'm  
going over previously trodden ground, please excuse me.

> >(Arop Sissano - N New Guinea).
> >This information, that Arop and Sissano only had 1 and 2, is an
> >illegal inference from the fact that only 1, 2 were published in  
> >a wordlist in a book by Churchill. That he only listed 1 and 2 is  
> >understandable as he was doing a comparative study, and there is  
> >no statement there to say that there were only two numbers.

No, Chan has it right. Sissano counts 1, 2, 2_1, 2+2, 2+2+1 (at  
least, it did when I collected data from native speakers in the  
1970s). Arop and Sissano are dialects of the same language, spoken in  
the large village on Sissano Lagoon that was destroyed by a tsunami  
some years ago.

> The only data I currently have is from Eugene Chan’s list of  
> numbers at
> It shows 1 and 2 used for all numbers from 1-10. Chan was careful,  
> and didn’t infer or claim numbering sequences that he hadn’t seen  
> in authoritative sources. Many of his numbering systems stop, at,  
> say, 6.
> The related Sera, next door, certainly does have morphemes for 3  
> and 5, but similar morphemes for 1 & 2.
> >Anyone seeing the actual number morphemes, which are not
> >Austronesian and not obviously cognate with any close Papuan
> > language, in Arop and Sissano would have to ask if they are  
> really linear Austronesian descendants!?
> They are probably not linear Austronesian descendants!?. Should  
> they be?

What is a 'linear Austronesian descendant'? If you mean that these  
languages represent continuity from generation to generation, rather  
than masses of Austronesian borrowings into a Papuan language, then,  
yes, they are clearly Austronesian, but have undergone a striking set  
of changes (especially Sera). According to Laycock, if I remember  
correctly, speakers of Arop claimed to be descended from Papuan  
speakers from somewhere inland (Olo, I think) who had been given  
asylum at Sissano and acquired the language, retaining just a few Olo  

> The two morphemes, /pontanen/ and /entin/, certainly don’t look  
> like conventional An numbers, but they could, just possibly, be An  
> hand parts, like finger, or thumb, or even ordinal numbers – first  
> and second – where the use of visible hand tallying made the  
> voicing of numbers almost redundant. Or they could even be special  
> names for counting certain objects. I will need to wait for more  
> information, before I get myself into even more illegalities.

As far as I know, these are not inherited Austronesian words. At some  
point, shared ancestors of Arop/Sissano and Sera speakers lost the  
quinary system of their earlier ancestors and replaced it with a  
Papuan-like system. I don't know where the words came from. They  
don't crop up elsewhere in wordlists for these languages.

- Malcolm Ross


Malcolm D. Ross
Professor, Department of Linguistics
Research School of Pacific and Asian Studies
Building No. 9, The Australian National University
CANBERRA A.C.T. 0200, Australia

ANU CRICOS Provider Number is 00120C

On 30/06/2007, at 2:37 PM, Richard Parker wrote:

> >>  The old documents preserve a lot. I know of only one An language
> >>that had only 2 words, 1 and 2, for their entire counting system
> >Also, what seems not be widely known, body-tally systems are  
> >attested in the torres straits and mainland australia. There are  
> >some refs (though Bill McGregor should have a bigger database of  
> the Australian cases):
> >
> Many thanks for your paper (and your others) – very useful.
> >>  eg: many Vanuatu languages still retain hand+1,2,3,4 for 6-9,
> >> so those  islands must have been first settled before the full
> >> An decimal system was conceived. It's possible, even, to detect  
> >> sequential waves of Vanuatu settlement as the number systems
> >> grow more sophisticated.
> >> There's an alternative, of course, that they were too stupid,  
> >>or too conservative, to accept a simple new system brought in by  
> >> Austronesian-speakers ready-equipped with the PAn decimal >>system.
> >This account is wrong in two places. The Vanuatu lgs could not have
> >_retained_ an old hand+1,2,3,4-system if they are Oceanic (or you'd
> >have to revise the POc-numeral reconstruction considerably).
> You said it! Something really should be done about large
> quantities of linguistically illegal number systems scattered
> around Oceania.
> I was using retained in the ‘normal’ way, not realising its
> implications as a linguistic technical term.
> But most Vanuatu languages use recognisable POc morphemes to
> construct 6-9. (I have next to no data on Vanuatu numbers above
> 10). The number systems noticeably ‘degrade’ from north to south,
> until, in New Caledonia, POc morphemes are almost unrecognisable.
> POc: *sa-kai, *ta-sa, *tai, *kai  *rua  *tolu  *pat, *pati, *pani   
> *lima,  *onom,  *pitu,  *walu,  *siwa,  *sa (nga) puluq
> Now, are the following examples retentions of older systems in the  
> normal sense, or innovations in the narrow linguistic sense?
> Motlav (Banks Islands, N Vanuatu): Bi-twagh, Bo-yo, Be-tel, Be-Bet,  
> teBe-lem, leBe-te, liBi-yo, leBi-tel, leBe-Bet, songwul
> Katbol (Malekula): sapm, i-ru, i-tl, i-Bat, i-lim, sout, so-ru, se- 
> tl, se-Bat, langal
> Iaai (Loyalty Islands): xacha, lo, kun, wak, thabung, thabung ke  
> nua xacha, thabung ke nua lo, thabung ke nua kun, thabung ke nua  
> wak, li benita
> Orowe (New Caledonia): rrake, keehru, kerrere, kevwe, keni, keni me  
> rrake, keni me keehru, keni me kerrere, keni me kevwe, keni me keni.
> (4 types of accented e omitted, for clarity)
> It looks quite obvious (to me) that the Oroweans have a less
> developed system, where 10 = 5 & 5, than the Motlavians, who use a  
> single, freestanding word, and all of them have systems that are  
> less developed than proto-Oceanic, which has a single ‘meaning-
> free’ word for all the 1-10 digits.
> How can comparative theory linguistics accommodate this paradox?
> >  Second, languages switch back and forth between decimal, quinary
> > and vigesimal systems with little correlation to stupidity. See a
> > recent OL article by Bender and Beller called classifiers and
> > counting systems or similar).
> I’ve read most of their articles available on the web, and can
> readily accept that people could retain a traditional system,
> possibly with substantial numeral classifiers, for counting things  
> that are culturally important to them, alongside a new (usually
> decimal) system for new things, just as they still do on this
> island, where Spanish applies to some things, Surigaonon to
> others, and Americano when you’re not quite sure.
> And we English count dozens of eggs, and scores of years. The  
> special counting system for tennis is said to be preserved in a  
> small temple at Wimbledon.
> I’ve yet to see any proven demonstration (although plenty of
> inferences) that whole groups of people have actually changed
> their systems back to something ‘more primitive’.
> Who would want to order thabung ke nua lo (say,cigarettes)in Iaai,  
> when he could just say *pitu in his ancestral proto-Oceanic?
> I’m actively hunting for examples of change of system and
> morphemes,together, either way, but, so far, I’ve only come across  
> loans of some isolated words, as in Swahili, where Arabic names
> have only been adopted for 6,7, and 9: moja, mbili, tatu, nne,
> tano, sita, saba, nane, tisa, kumi, but Bantu retained (sorry –
> kept) for everything else.
> Some of it is truly baffling:
> The Papuan languages of Halmahera seem to use one (and only one)
> An word in their 1-10 words - /siwo/ for nine, yet their close An
> neighbours, Patani and Sawai, use /fapolo/ and /popet/.
> This kind of 9 morpheme ‘1 before 10’ is rare, and usually seems
> to be symptomatic of a ‘suppressed’ base 4 system, where 8=2x4, as  
> in some groups in Flores and Sumba, others in Aru Island, Enggano,  
> Wuvulu, and almost half of the Taiwanese languages.
> The only straightforward An base 4 systems I can find are Biem and  
> Wogeo in the New Guinea Schouten family.
> (In none of the above do I have any data for numbers above 10, so
> I can’t yet tell if they go on from 4-8 to call 12 a dozen. I’m
> hoping somebody can help me out on this).
> The construction of 9 in Nghada (Flores) - /ta esa/ is almost the
> same as Taokas (Taiwan) /tanaso/. Both have an combination phrase
> 8 involving 4 morphemes, /zua butu/ and / mahalpat/. They’re 2200
> miles apart, and there’s nothing comparable on the straight line
> between them.
> Here is a comparativist’s view of it:
> ‘An interesting set consists of Thao tanacu , Favorlang tannacho ,
>  Taokas tanaso '9', which point to an earlier *[st]a[nng]aCu. The
> first syllable might reflect *sa- 'one', in which case we are
> perhaps dealing with a subtractive form.’
> Laurent Sagart ‘The Higher Phylogeny Of Austronesian And The  
> Position Of Tai-Kadai’
> It actually seems to be the start of a new (suppressed) base 4
> count, from two 4s, and maybe the reconstruction puts the 1 at the  
> wrong end.
> >Yes there are such cases. When I said there is no other etymology
> >for 5 than 'hand', I should have said 'hand' or 'some part of the  
> >hand'.
> >But I didn't, so you can have the 10 dollars if you want.
> No way would I claim a prize in such a specious way. But it is  
> worth noting that ‘whole hand’ is not always the morpheme.
> Regards
> Richard
> _______________________________________________
> An-lang mailing list
> An-lang at

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>
-------------- next part --------------
An-lang mailing list
An-lang at

More information about the An-lang mailing list