Personal Pronouns / Ergativity

[Jens Elmegaard Rasmussen]

On Mon, 14 Jun 1999, Carol F. Justus wrote:

> Dear Jens and Wolfgang,

> The issues of diathesis, ergativity, and reflexive pronouns under
> discussion here are indeed complex and in need of clarification. I comment
> on one really.

I try and read really.

[...]

> Jens' response [to a correct report given by Wolfgang Schulze of what
> is perhaps the most widespread view about an IE passive - Jens]:

> [Jens:]

>> I strongly disagree with the statement that PIE had no passive. In
>> fact it had several. One of the basic functions of the middle voice
>> was patently to express the passive use of transitive verbs.

> Let's stop here for a moment.

> [General references to people telling us what passive is in other
> languages, esp.:]

> In 1991 Klaiman (Grammatical Voice? Oxford UP?) then typologically
> distinguished between the role of a passive as derived voice in
> nominative-accusative languages and the role of the middle in active
> languages which have 'basic voice', not 'derived voice'. The role of a
> middle in active langauges was to move members of one verb class into
> the other of two verb classes, as active languages typically have two
> distinct verb classes: verbs such as 'sing, run, dance' (active) and
> 'lie, sit, stand' (stative). Such verbs are essentially intransitive.
> Devices also created transitive verbs. [...]

> Gamkrelidze and Ivanov have suggested that PIE was active in type, citing
> the two verb classes in Hittite, the -mi and -hi conjugations, among other
> things. If active in type refers to having primarily intransitive verbs and
> derivational strategies to make them transitive, then PIE had no passive.
> It takes a transitive verb to be passivized.

Well, the mi- and hi-conjugations plainly are NOT the two "basic series
"they are cracked up to be, esp. in Ivanov's 1965 monograph. For one
thing, there is an Anatolian mediopassive which is not a member of the
hi-conjugation, so the hi-conjugation is not the flimsy "inactive" common
precursor of the middle and the perfect of the other languages. On the
contrary: The Anatolian hi-conjugation is the IE perfect, period. The
stroy is a simple one which has been told - in perhaps not so simple terms
- by Eichner and others many times. Anatolian has given up the IE
distinction of verbal aspect stems, having then only one verbal stem per
lexical verb. The actual stem surviving from PIE may be a "present stem"
(as in es- 'be', kuen- 'kill'), an aorist stem (as in mer- 'vanish', sanh-
'seek', te- 'say', ganes- 'recognize') or a perfect stem (ishai- 'bind',
ispai- 'eat one's fill', probably ar-/er- 'come' and, for semantic
reasons, quite possibly sakk-/sekk- 'know'). All of these stems formed
what could become a preterite, and so IE imperfects, aorists and perfects
live on as Hitt. preterites, just as they do in Classical Sanskrit. On the
analogy of the imperfect which had a present beside it from of old,
preterites based on aorists and perfects created presents by adding the
present ending enlargement -i. The 1sg prs. based on a perfect was *-ha-i,
Old Hitt. -hhe > -hhi; the prt. -ha is retained in Luvian and Lycian,
while Hittite adjusted the expected *-hha to the ending of the
mi-conjugation -un (from syllabic *-m) thus creating -hhun. Thus, the
preterite of the hi-conjugation basically continues the IE perfect (with
some help from its closest friends, note esp. the 3sg -s from the s-aorist
necessitated because of the lack of consonantal material in the 3sg
perfect itself). The bulk of Hitt. hi-verbs, however, are intruders:
Practically every verb that had the vocalism *-o- in IE has ended up being
a Hitt. hi-verb, a fact that stronly indicates that the old stock of
hi-verbs had that vocalism already, a demand met most easily if they were
perfects. We find IE causative-iteratives, intensives, reduplicated
aorists (!) in the hi-conjug., and also some for which the vocalism has
not been precisely *-o-, but apparently close enough to make the verb
share the fate of the perfect. Thus the denom. type newahh- is hi,
although the vocalism was IE *-a- (from pre-IE laryngeal-coloured *-e-),
and so is da:- 'take' based on the middle voice of *deH3- 'give' and so
fairly plainly continuing *d at 3- with syllabified laryngeal. That such a
crude and primitive analogy could work so well and not represent the truth
would be beyond my comprehension and incompatible with normal standards of
common sense.

That, however, does not in itself exclude an old "active" typology for a
prestage of IE, only now you have to look elsewhere for its traces, and
that would of course be in the dichotomy prs. (prs./aor.) : midd./pf.
(which then would show up as -mi vs. -hi in Hitt., if only indirectly).
But I cannot see this looks too good either: While medium (or perfect)
tantum verbs are indeed not very action-related, the opposite does not
hold for the active. Naturally *H1ey- 'go' and *{gwh}en- 'kill' are good
active verbs, but why is *H1es- 'be' and *k^{th}ey- 'dwell' members of the
same club? My answer to that is simple: Because club membership is not at
all based on an active/inactive parameter: They are active because they
are not the mediopassive of anything. Has anyone in his right mind ever
claimed that *{gwh}e'n-mi 'I kill' was intransitive? And what about 'eat'
and 'millk'? And the many root aorists that are plainly underived too? Was
{kw}e'r-t 'made' once intransitive? On what basis is THAT claimed to be
known - or am I missing a point since it looks so silly? Media tantum like
*k^e'y-or 'lies', "e:'s-or" 'sits' (reduplicated? simply not knowable),
*we's-tor 'is dressed' and some others may be the medium part of an older
two-voice system preserved with most other verbs. That would of course
mean that the old function of the middle voice was not exclusively
passive, for here it is not: you are not lying by being kept in a lying
position (but _things_ are), this looks more like a reflexive which was
then one of the meanings of the middle voice very early on (which nobody
has ever denied, I guess). I would deny, however, that the reflexive (and
reflexive-like) function was the ONLY function of the middle at any time,
for there are too many things that demand its having a true passive force
(by the unsurprising definitions you quoted) in a very remote period when
some very old phonetic changes were operative.

   Thus, since there was no ban on transitive verbs in PIE which obviously
had plenty of them, there was not THAT ban on a passive category. What a
relief.

> Along with primarily
> intransitive verbs and the lack of a passive, PIE had no transitive verb of
> possession, no 'have'. 'Have' entered dialectally by different paths. There
> are now two recent statements on the path of 'have' into attested IE
> languages, one in the second volume of the recently published Lehmann Fs.
> (Journal of IE Studies Monograph 31) and another in the Proceedings of the
> Xth UCLA IE Conference (JIES Monograph 32).

I do not see the relevance of the question of how speakers of IE said
'have' for the discussion of the existence of a category expressing the
passive of transitive verbs. I see the relevance of 'have' for a somewhat
different discussion which seems not to have been addressed: How could a
stative derivative with suffix *-eH1- from 'take' come to mean 'keep,
hold, have' as it obviously did in so many instances? And if it is an
areal thing as you claim (and I find no particular reason to believe), how
could it happen just once so that there was a place it could spread from?
This is the semantic nuance you rather expect for the perfect: 'I have
taken and now hold' would be a very good way to say 'I have' in IE terms
(cf. Eng. "I've got"), but that is not the way it is said, except perhaps
in Slavic where OCS imamI can indeed be derived from an IE perfect. It
seems the "stative" derivative verbs in *-eH1- still hold their secrets,
and so I cannot exclude that they will eventually turn out to support the
active/inactive idea - but neither can I exclude that they turn out to
join all the other alleged indications and be simply irrelevant.

[...]

What you say about my description of the causative and factitive verbs
that lost their transitive value (factitives then becoming ingressive)
simply repeats what I said with a use of English modality as if it were an
objection. If stripped of the rhetorics it appears to contain no added
information and no expression of disagreement.

[...]

> Coming back then to Jens:

>> Alongside this, there was the "stative" morpheme //-eH1-//;
>> this was stative with intransitive verbs, as aorist *sed-eH1- 'enter a
>> sitting position', prs. *sed-H1-ye'- 'be in a sitting position, sit',

> This stative was not productive in PIE (see the issue of 'have' in the
> Lehmann Fs.). The root 'sit' in fact is most interesting here. Hittite
> 'sit' is derived as middle of 'be': eszi 'is', esa(ri) 'sits'. [...]

If the stative formation was not productive in IE it is even better, for
then it was a fossil which makes it even older. The stative has now been
discovered in Indic, cf. Ilya Yakubovich in the 10th UCLA Conference
volume, picking up an idea of Jamison's which she had given up herself. I
may add to this the striking stem-formation of some of the forms of the
root kas'- 'see', prs. ca's.t.e, aor. akhyat, formed from an apparent
root-form "khya:-", as prs. khya:yate 'sees': this is simply the same stem
formation as Lat. videt, i.e. from *-eH1-ye-ti/-tor.

A plain middle voice of *H1es- would not give Skt. a:'ste, Gk. e:^stai
with a long vowel as opposed to a'sti, esti(n) 'is'. They may be related,
but not in exactly this way.

> This is not to deny that even Hittite had passive uses of medio-passives.
> The point is to look at what is productive in the system. [...]

I hope most of us are too sophisticated to need that kind of lecturing.

> Coming back then to Jens' causative suffixes in Sanskrit (-aya-) and
> Hittite (-nu-), I would suggest that they were two independent innovations
> after the breakup.

But you have *-eye- in Hittite too (la:ki 'put lying', wasse- 'dress' frpm
*logh-e'ye- and *wos-e'ye- resp., in the former with the normal passage
into the hi-conjugation of a verb with o-vocalism, while the latter has
resisted this analogical pressure); and you also have de-adjectival
factitives in -nu- in Sanskrit (dabhno'ti, identical with tepnuzzi;
dr.s.-n.o'-ti : Gk. thrasu's), as well as both of them elsewhere in
diverse IE languages. Moreover, there is no obvious way they could have
been created secondarily from inherited material. Both _must_ have been
PIE, and theories incompatible with this inference just are no good.

It seems to me that there is a basic error inherent in the frequent
"explanation" of mysterious categories and forms as "late", "secondary" or
"einzelsprachlich". If a morphological type is too young to belong to the
protolanguage it must have been formed from material the protolanguage,
indeed the particular poststage of it had, and then it should be easier,
not harder, for us to discover its origins, for in that case the timespan
to be bridged is shorter than in the case of very old forms. This error is
very often committed when dealing with categories that have become
productive, such as the thematic verbs or the s-aorist. They became
productive, oh yes, and so all their examples cannot go back to the
protolanguage, but some MUST, otherwise there would have been no nucleus
for the expansion.

[...]

> This is not meant as a statement about what changes can or do take place.
> This is part of a larger attempt to arrive at a plausible solution to
> what did take place in one well-attested language family.

My posting is meant as both.

Jens