On Sat, 24 May 2003 16:01:29 +0200, Miguel Carrasquer <
mcv@...>
wrote:
>4) the split-ergative model. Pure ergative languages are rare. Most
>"ergative" languages in fact show a mix of ergative and accusative. I
>mentioned the case of modern Indo-Aryan languages, which are partially
>ergative (in the past tense) but otherwise accusativic. Another
>common split-ergative system depends on degrees of animacy. The more
>"animate", the more likely a word-class is to behave according to the
>accusative model, while word-classes of lesser animacy tend to use an
>ergative system. The categories genrerally look like this (from high
>to low "animacy"):
>
>1. first person pronouns
>2. second person pronouns
>3. third person pronouns
>4. nouns denoting humans
>5. nouns denoting animals
>6. nouns denoting things
I forgot to add an important aspect: it's not only a question of
"animacy" but also of "determinacy". Determinate nouns and pronouns
[personal pronouns are always determinate] are higher up the scale
than indeterminate ones.
Some other caveats: while (1/2) personal pronouns are more likely to
show an accusative pattern even in otherwise ergative languages, they
can be irregular in showing a _marked_ subject case (derived from the
ergative). In Sumerian for instance, personal pronouns are
accusativic, unlike normal nouns which are ergativic, but the subject
case on pronouns is marked by -e, the ergative marker (g~a-e "I", za-e
"you").
Finally, to give an idea of my own ideas on the subject (and object),
I would have to say that in this particular matter, I would tend to
side with Glen (despite his kindergarten debating techniques). I
think the animate nominative marker *-z and the inanimate
nominative/accusative marker *-d are best explained as coming from
definite markers derived from the demonstrative pronoun (*so and
*to-), added to the unextended root. *so and *to-, in turn, can be
derived from a single demonstrative pronoun **t-, with ergative **tu
and absolutive **ta. When added to a nominal root, these became *-zW
and *-d according to the soundlaw which voices asyllabic endings at
word-end (except at sentence-end, i.e. in [most] verbal endings).
The marked accusative ending *-m _does_ perhaps come from a
genitive/possessive (cf. the Eskimo ergative/genitive marker *-m ~ *-p
and the Kartvelian ergative marker *-m(a)). In the personal pronouns
*-me, *-mé occurs as a genitive/accusative marker (Gen. *méme, *téwe,
*séwe < **mú-ma, **tú-ma, **sú-ma; Acc. *mé, *twé, *swé, with accent
shift from **me-mé, **te-wé, **se-wé; plural Acc. *n.smé, *usmé, *smé
< **m(W)es(W)-mé, **(s)Wes(W)-mé, **sWes(W)-mé).
The most likely model to explain this shift from genitive to
accusative in PIE (and Uralic), if the ancestral (Proto-Nostratic)
language was indeed ergativic, is the antipassive model. Presumably
the (transitive) present tense shifted from a plain ergative to an
antipassive construction, producing a system where both transitive and
intransitive verbs always had a subject in the absolutive (optionally
extended with definite markers **-tu > *-z and **-ta > *-d), and an
object (if any) in the *-m(a)-genitive. The person markers in this
antipassive present are absolutive:
*-mu *-mu-ána -> *-m(W)-i *-m(W)én-i, *-m(W)én
*-tu *-tu-ána -> *-s(W)-i *-t(W)én-i, *-t(W)ér
*-0 + *-ta *-0-ána + *-ta -> *-t-i *-ént-i, *-ér
The (transitive) past presumably remained initially ergativic, and had
ergative person markers:
*-mu *-mu-átu -> *-m(W) *-m(W)és(W)
*-tu *-tu-á[tu] -> *-s(W) *-t(W)é
*-0 + *-tu *-0-átu + *-tu -> *-s(W) *-és(W) > *-ér-sW
Note the lack of distinction between abs. and erg. in the 1/2 sg.
Noteworthy is also the "ergativic" 2pl. in *-té for expected *-tés(W)
[analogical *-tes(W) does occur in Latin -tis and Armenian -yk`].
This is due to what I call the TATU-rule, a kind of haplology that
affected sequences of morphemes containing *t, in different ways.
The 2pl. verbal ending which can be analyzed as 2nd person marker *tu
- plural *átu: **-tu-átu > **-tu-á > **-tWé > *-té is one example.
Another is the 2pl. personal pronoun nom. [2p **tu - plural **atu]
**tu-átu > **uátu > **úsW > *yú(:)s.
The ablative and instrumental plural show something similar [nominal
oblique **abhi - abl/ins **ât/**át - plural article **atu]:
**-abhí-ât-atu > **-abhíâtu > *-bhíosW > *-bhiós; **-abhi-át-atu >
**-abhiátu > **-bhísW > *-bhis.
Finally, the lack of *-s on the nominative of demonstrative *so can
perhaps be explained in the same way [dem.pronoun *t + ergative *u -
thematic vowel *á - anaph.pronoun *i - nominative ending =
pronoun+ergative *tu]: **tu-á i-tu > **tuátu > **tuáu > *sWo(u) > *so,
*so(:)u.
=======================
Miguel Carrasquer Vidal
mcv@...