Re: Sound changes causing divergence of ordinals from cardinals

The only natlang I can think of with two entirely different sets of roots is
Japanese.  The different sets of roots don't carry the cardinal-ordinal
distinction, but it *is* the only natlang I know about that has two
unrelated sets of roots in the first place.

For those of you unfamiliar with Japanese:  one set of roots is borrowed
from various dialects of Chinese.  These roots are combined with suffixes in
order to count objects -- the suffixes change based on the shape of the
object being counted.  Therefore:

"nihon" = two long, thin objects (pens, trees, neckties);
"nimai" = two thin, flat objects (pieces of paper, tickets, etc.);
"nisatsu" = two volumes (books, newspapers, magazines).

The second set of roots is a native Japanese set, holdovers from before the
Chinese equivalents were borrowed.  These words describe objects that don't
fit into any appropriate shape categories.

Do any other natlangs do anything "bizarre" like (or unlike) that?


>From: John Vertical <johnvertical@...>
>Reply-To: Constructed Languages List <CONLANG@...>
>To: CONLANG@listserv.brown.edu
>Subject: Re: Sound changes causing divergence of ordinals from cardinals
>Date: Fri, 6 Jan 2006 00:47:38 +0200
>
>Neat stuff. Still, this does nothing to extend the cardinal/ordinal
>distinction to other numerals! They'll definitely get messed up too in all
>these sound shifts, but it would be unlikely that one series would turn out
>resembling the ordinals more than the cardinals, let alone contrasting the
>two as equally possible roots.
>
>Of course, we can just derive those later, but where's the fun in that? ;)
>
>
>Alex Fink wrote:
>>Nik Taylor wrote:
>> >Alex Fink wrote:
>> > > (2) Primary stress is penultimate (unless there's only one vowel).
>>Vowels
>> > > under the primary stress shift: non-high vowels rise (/a e 2 Q o/ >
>>/e i y o
>> > > u/), while high vowels break (/i y u/ > /j@ H@ w@/).
>> >
>> >(3) Intervocalic voiceless stops become fricatives (/x/ further develops
>> >to /h/ and then 0)
>>
>>Weird that original /h/ sticks around, then.
>
>Indeed, better delete them intervocally too.
>
>OK, with lots of adjacent vowels now...
>(4) Sequences of adjacent vowels with the same roundedbackness develop into
>long vowels. Schwa matches with all series. The vowel is high (or low) iff
>all the originals vowels also are, else mid.
>
>And to stop this from being just "big" shifts, I think we should also allow
>one "small" (ie. affecting only a few of the words herein) sound shift per
>turn:
>(4.5) Immediately before stress, /tS dZ/ lenite to /S Z/ unless preceded by
>an alveolar continuant.
>
>Value Cardinal        Ordinal
>1     'j@wQ           i'we:
>2     'nero           na'rue
>3     'mjo:           mi'ue
>4     'to:            'to:e
>5     'kin            'kinke
>6     'jesu           ja'swe:
>7     'Sj@gu          Si'gwe:
>8     'dZuz2          Zo'ze:
>9     'hj@ja          hi'je:
>10    'neo            na'ue
>11    'hero           ha'rue
>12    'utsu           ot'swe:
>20    naro'no:        naro'no:e
>21    naronQ'H@wQ     naronQy'we:
>22    narono:'noro    narono:nQ'rue
>30    mio'no:         mionQ'ue
>31    naro'no:ro    narono:'rue
>32    naro'no:tsu     narono:'tswe:
>33    miono:'mjo:    miono:mi'ue
>
>I also postulate words for 100, 1000 and 10000:
>100 'tSifu Se'pwe: (from tSepu tSepuke)
>1000 'wegor wa'gurke (from wagor wagorke)
>10000 Sun'tSyfu tSuntS2'pw@ke (from tSuntSepu tSuntSepuke )
>
>These do not compound with each other or smaller numerals, so there's no
>need to individually process thru numbers larger than 100.
>
>Still needs to somehow perturb the still somewhat regular -e ending of the
>ordinals and the matching initial consonants. And all those medial glides
>can't be healthy either...
>
>John Vertical

Replies