Download - Before PIE_ Motives for Thematism.pdf
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
1/26
Folia
Linguistica Historica XVI /7-2 pp. 3-27
Societas Linguistica Europaea
BEFOREPIE:
MOTIVES
FOR
THEMATISM
N.E. COLLINGE
0.
Studies
ofProto-Indo-European morphology and morphosyntax
have
lagged
behind
those
making
progress on its
phonology.
The latter enjoyed
the limelight in the
nineteenth
Centuryand a
little
later;and
more recently
waves
of the
laryngeal
theory an d varieties of the glottalic theory have
maintained the
impetus
not to m ention
more
idiosyncratic ideas of diph-
thonga l bases andfreelyvarying
con sonant-features).
Still,in the last quar-
ter-century m uch solid w ork
has
been done
on
PIE s
gram matical
history,
although the current fashion seems to turn to such morphophonological
matters
s
suffixal shape, paradigm dynamics, and the like.
In
pursuit
of
genetic connections there
has
been also
an
urge
to em-
brace
super-divergence. Those attracted
by
Pedersen s
notion
(1924;
see
also 1962: 338)
of a
nostratic
Ursprache
have used macro-derivation
to
link PIE
with
Dravidian, Kartvelian, Afro-Asiatic
and
Semitic
- and
even
that domain seems undulyparochial
to
some
who
would cast
the net
even
wider like Greenberg; see Matisoff 1990 on this zeal for megalo-com-
parison ).
To this end many tradition al findings have been exploited, those
especially of phonology and lexical semantics. Shapes-with-meanings al-
ready
assigned to PIE
have been recognised elsewhere.
But PIE is no
more than a diagrammatic pro-language, summarising the reconciliation
of
various
reflexes
seen or deducted in the
evidential
languages; and that
summary cannot long precede their first diaspora. When morphosyntax
has been brou ght in, nuclear sentence structures in IE their transitivity
patterns)
have been linked with
active or ergative modes of
grammar
- a
tactic
of
long stand ing
-
wherever these
are found.
Even
an
improve-
ment-succession has been suggested, from active via ergative to accusative
type. But the unreal and achronic nature of PIE defeats this approach,
apart from
the
doubtful Status
of
individual proposals.
Yet what if we can get back to some firmer internal history of PIE
grammar? There should be traceable lines of cognitive development pre-
ceding and helping to shape the latest form of its morphosyntax. This
pre-morphosyntactic
era may both constrain
megalo-comparison
and tidy
up bitsof PIE morphology
which
haveresisted
attempts
to fit them
into
the
overall
picture sofar.Here willbeexamined
three
similar, but distinct
in
time
s
well
s
function), marking-procedures in a
proto-language
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
2/26
which was still morphologically immature. That is not yet PIE. As a name
for
it when it is struggling to denote half-cognitive,
half-grammatical
cate-
gories let us use pre-morphological PIE - in
brief,
PPIE. This title is
not to be confused with Rix s pre-Proto-IE (1988: 101) which indicates
all pre-diaspora PIE; it is nearer to Georgiev s
(1984)
PIE I, orFrh-
indoeuropisch.
Considering arou nd twenty millennia, Georgiev
saw no
morphologyat all emerging in this first phase (aspect, personal pronouns
and unmarked locative being the first fruits of its successor, PIE II, when
grammar takes root
and
produces other pronouns, too,
and precase
par-
ticles (this isM ittelindoeuropisch . This timetable is possible and ignores
larger genetic speculations. But
unhappily
it also ignores th e three mech-
anisms examined
below,
though their traces need explanation
and
their
relative chronology consideration. And in these mechanisms the common
element of
form
isthematism.
PPIE
deployed thematism in various ways both before and
after
its
Speakers became sensitive to a
fundamental
communicative distinction:
that between entities (which in a given
context
may or may not
exist
fo r
Speaker
and hearer) and events or predications (which may or may not
be
true s
to fact). This is the essential basis for the evolving syntactic
dyarchy,
of
N(oun)
and
V(erb).
The
physical sign
was the
suffixation
of
a
lexical root
by
means
of a
vowel
of
middle height (front
or
back, these
later alternating). Theroot became a stem ; it received further, andlater,
affixes to convey the mittelindoe uropisc h categories noted above and
then others. Possibly the
early
pronoun alone overlaps in emergence with
the thematic vowel. Thematism does not occur sos to set up the stem;
it m erely joins w ith th e instantiation of that structure by unextended con-
sonant-final,
or (other)
vowel-final,
forms. Hence its name is unf ortunate:
thetic would
avoid misleading suggestions
s to its role. But the
term
is
traditional
and
will
be
used
here;
of
itself
it
explains nothing. ( Ha lf-the-
matics ,like Latin fer-t,where syncope or similar interference has obscured
the issue, are here
ignored.)
Each of the three types of thematism is merkmaltragend(shaped by
th e
added vowel)
and
markiert (carrying
a
positive
function
which
th e
unextended form
lacks).
In
order
they are:
(I) basic root plus front vowel: (N
(II) entity-root plus bac k vowel:
(III) event-root plus alternatin g vowel:
V
e
/o.
(I) belongs to an epoch
well predating
the inflectional and
derivational
phase
of PIE; it is e ssentially ofPPIE. (The other types reflect cognitive
revelations possibly peculiar toPPIEand early PIE; there is no need to
relyon finding them in, say, Kartvelian.) Indeed,
PPIE s
choice of e
s
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
3/26
the
sign
of (I) is idiosyncratic, and probably a bar to megalocomparison
even at the beginnings of speech.
There are problems with (II) and (III) if we accept notions of early
active
or
ergative
in Indo-European.
Form-function relations ought
to
leave
plenty of vestiges; and the
chronology
becomes awkward - because
if thematism
has
that sort
of
role
it
must
be
part
of
early
PIE and not
PPIE
at all. The evidence noted below suggests otherwise: and the only
language-group linked with PIE (even nostratically) which seems to be
ergative, southern Causasian,
is
really pseudo-ergative (for this diagnosis
see Trask 1979). And activity remains in more than vestigial form only in
tongues
s
distant
s
Guarani
or
Eastern Porno. Comments
on
these
syn-
tactic are
offered below,
in
connection with
the
possible origins
of
this
or
that shape
of
words.
The
scepticism here voiced
s to
their
relevance
to (P)IE history receives some support from Matsumoto (1993), although
there
an
active stage
is
accepted.
If,
however, thematism
is
divorced
from
any such mode
of
grammar,
the
Charge
of
anachronism need
no
longer
be feared.
To pass
to the
mechanisms themselves,
in
their apparent relative
chronological order:
1. (N V)+e
An
arbitrary starting point may be the pre-verbal base *g
h
en +
e
set
up by Kurylowicz in 1964 (62-63) sunderlying the
later
PIE perfect. This
was conceived
to be a
root (conveying action) enlarged
by a suffix (of
state
or
undergoing)
to
produce
an
item
of
no m inal Status;
in
later terms,
an
adjective
or
participle. Watkins
(1969: 105-118)
charted
its
evolution
into the basic member of the PIE set of verbal forms which displayed
sensitivity to
time, voice, person
and
number.
The
original form
is ,
seem-
ingly, metanalysed:
the
structure
l*g
h
en
+
a
plus (N-nominative) zero/
passes
to /*g
h
en-
plus zero plus (verbal) e/ ,
the final
element assuming
the
signalling
of
activity/predication.
This result, no longer n om inal, is then
seen s
that non-person, non-number
form of the
verb which
is
sub-
sequ ently called third person Singular
and
taken
to be the
platform
on
which m ore complex
forms are
differentially built
(on
this concept, known
s Watkins
law,
see Collinge 1985: 239-240). In succession, the
precise
order being here irrelevant, the personal m arkers, the
here
and now*
suffix
-l/,and the middle voice sign +oa ll fall into
place.
The early +e shrivels
into
an
adventitious
and
otiose
piece
of
morphology. This history
is
possible. It may miss
much.
Kurylowicz glossed bis
launch-pad
s
(he is (being)/hasbeen) killed .
This verbal adjective
was
supposed
to be
reverbalised
by
accent shift
or
metaphonic or apophonic
ablaut:
so *g*
h
onel*g
h
onel*g
h
n?lo, plus af-
fixes. It
become
the PIE
perfect swell s
itsoldest form of
medio-passive.
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
4/26
Yet only a
passive sense
is first
offered
for it; and
such suggested perfects
s
*iouge
are
active.
The
function
of
e
is
likened
to
that
of
later
H
but the latter is passive-intransitiverather than middle (cf. Skt.
hata-
and
gata-).
The path which takes *g *
h
ene+i to active Gk.
theinei 'kills'
is
equally obscure.
Let us rather shift our gaze to an actually documented (unaccented)
e.One
proposed category exploiting that shape
is the
famous
injunc-
tive ,
aPIE
feature much
in favour
again,
after its first
proposal
by
Brug-
mann(1880-2)whose examples of its simplest form were Skt.
bharat,
sthf,
Gk.phere,
st f .
It is
only
the final
person-marker that
is clear
(-f
> -0
in
Greek); unless crasis
is
proved,
stht,
stt
do not
show
e.
And in
most examples quoted by adherents, aspect is already present v ia apo-
phonyor
reduplication (so,
in
Greek,leipe,mimne
- cf.
West1989).Hence
the thematic
vowel looks
to be a vestige of some earlier positive
Signal,
now
reduced to being a s tem-formant only. As to the injunctive in general,
Hoffmann
(1967: 140) assigns
to it the
indication
of
iterative action
or
general validity (re-cast by Dunkel (1992: 200) s the eternally valid').
More interesting is H offmann s
claim,
based on the absence of expected
augments
in
Vedic, that here
we
have
a
tenseless, moodless
and
non-re-
portive verbal category, which merely
'mentions'
erwhnt)
the
action.
That,sWest says(1989:135), looks
like
'aprimitive featureof the
Indo-
European verb .
If
so it is worth seeking out in its pre-personal stage
(and preablautal, pre-reduplicative stages, too). Dunkel
first
uses
the
sup-
posed
injunctive to
explain (hoping
to
improve
on
Szem erenyi s
(1979)
metathesised
phraseswhich
may
underline
initial/? Variation in
some
Greek lexemes)
th e
left-hand members
of
compounds like
Av. vanat-
p^sana,
Ved.Bharad-vja- or Gk. heleptolis *helet-polis). With perhaps
more cogency, he then (203-206) analysestypes like Ved.
Trasa-dsya-,
Gk.
Mene-laos
s
having
s
the first
element
imperative
forms. (French
porte-monnaile , Eng.
forget-me-not are
adduced
s
evidential
of the
process.)
Further, fo r Dunkel these forms are of uncertain person s be-
tween second and
third
- which putsusintoan erawhen even that dis-
tinction
was not yet cristallised into a formal morphological Opposition.
If an imperative did stand there, a solution becomes possible which
is more compatible with an early stage of
cognition,
when in Speakers
urgency and imprecision went band in
band.
W hy should th e activity b e
so economically but clearly specified at all? I t may have been to expostu-
late,
in a
peremptory
and
jussive fashion, towards
an
entity
(that
it
appear
or
that
it
act)
or
towards
an
event (that
it
happen).
In
other words, that
N or V be
manifested. Such
a
jussive, arising from
otherwise
thwarted
needs,is a likely
enough
origin at the envisagedpointin the
evolution
of
the languageifnot of all speech.Thereafter,whenPIE hasarrived,the
form willbe 'imperative'of anevent onceV is
morphologically
furnished,
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
5/26
and 'vocative' of an
entity once
N is
similarly paradigm atised.
It is not a
real player
in the
emergent drama
of
nominal inflection (when argument-
role
and spatial relation are formalised),
but
a vestigial element parked
there, and
feit
to be worth keeping. So much is indicated by:
(1) the betra ying absence of crasis
or
contractual lengthening in the
o-stem
vocatives: hence
f-,
added
to the
basic roo t, versus e.g. stem
extension dative
- i < *-o +ei;
(2) the form s persistence (even, spottily, into m od ern Greek );
(3) the quite different
processes
invoked to sup ply the same signal
else-
where: e.g. zero
affix
(Skt.
rjan \
truncation (Skt.
tanu\
see Beekes
1985: 101;
for
j-loss
see
Winter
1969);
accent
shift
(Gk.
g fmai .
This makes understandable Beekes'
paradoxical
remark (1985: 106) on
'the
ending
of the
vocative (which
hadno
ending
in PIE)'.
There
is,
however, no need to follow him (101) in equating the expostulatory suffix
e
with thefree-standing ex clama tory particle.The formerwas aprimeval
addition when N and V differed
only
crudely. Through
history
its am-
bivalencebetween them abides. It makes possible, for instance, the famous
Carpel
jest
of
Trimalchio (Pertonus,
Satyricon 36)
who,
in
calling
so
upon
bis
carefully
named servant,
fuses
entity with event:
eodem
uerbo
et
uocat
et imperat .
Winter
rightly sees
(1969: 212) that imperative and vocative
aremirror imagesof each other; he deplores the absence of proof the reof,
but what further proof is needed? An addressee is for both a prime re-
quirement. That thus
the second
person
acquiresa
high
degree of
cen-
trality is a phenomenon encountered elsewhere: in most Algonquian
lan-
guages
a
verbal construct
has to
have
an
initial
k- if
'you'
is
included
in
any nuclear
role,
agent or patient. So we seem to be justifying at last
Somm er s opinion
(1937:
187-195) that here
is a
pre-inflectional element.
In V,sin N, the function broke loose from the simple form, and m anifold
shapes of PIE imperatives then appear. No
later
development quite dims
the
light
ofPPIE 4 e,
despite
the
inclusion
of
third person
forms in the
'mood'(nottomention first personhortatorymarkers). As a directappeal
to the hearer,
s
existent or active, e so began and so continued.
2.
N + o
Therehasbeen unease in recent yearsstowhat element is the head
of a noun-phrase. Deictics,
some
prono uns am ong them, have won many
votes
s candidates fo r
that Status, critical
in any
head-driven grammar.
Hence
the DPtheo ry (replacing theNP),and theneoCho mskyan DP-node
(cf. Webelhuth 1995: 89 fn 25;398). This
reassessment
affects such ana-
lyses
s
thatofNichols(1986),which
sets
apart those languages (the
ma-
jority)which signal m odification within the phrase by a head-marker, those
which
put the
marker
on the
dependent item,
and
some which
do
both
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
6/26
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
7/26
thematic
and
athematic declensions. Besides,
the
System
is recent
s
being
post-Anatolian
(pace Kammenhuber (1985: 449-450)
on
Luwian
possible
/-feminines);
Hittite had nominals which were lexically female
(cf.
Brosman 1984:
346 and
1994),
but not
morphologically. Gender
I,
however, is the prior dichotomy into plus/minus
animate.
Here a formal
plus marker is expected; but the facts are discouraging: (1) V-governing
+S is
added
to all
animate (non-neuter)
subjects - allowing for
Szeme-
renyi s law
(see Collinge 1985: 237-238)
- and
does
not tie up
notably
with
+o-; nor is its
presence
in
athematics
the
result
of
subsequent dif-
fusion, s
far s one can see. (2) Neuters are the institutionalised
Version
of
inanimates,
and are
never animate
in the
early
evidence.
But
their dis-
tribution is unhe lpful: absent in diphthongal stems and in the minority in
/-stems and
M-stems, they
are
(equally strangely) present among
the ap-
parent founder-members
of the
the m atic declension (cf. Brosm an 1979:
61). They occur the re f ractiona lly more frequently than
in
Hittite high-
vowel
stems. It is usual to accept
s
original in PIE *yiigo-,
*pedo-,
*wergo-
(and possibly
*Hwerso- ,
*dno- *dro- , *petro-/ptero- . It is bey ond belief
that athematic inanimates
are all secondary;but
conversely thematic
in-
animates seem
so
numerous
in
Hittite (about 170, according
to
Brosman
1979),
and
them atic neuters
so
solidly testified
s
original,
sto
cast doubt
on
any proposed stage (even Ur-PPIE ) wherein existed
only
non-neuter,
animate, nominalswith o-. Of course, the
later
proliferation of o-stem
membership (especially
in
Hittite
and
Indic,
less
markedly
in
Greek
and
Latin) - wherein neuters are very com mon (Brosm an 1984: 357) - does
cloud
the
issue somewhat;
and
animacy
and sex are
always shifting affairs
in
grammar
(e.g. Greek
loved to use feminine
thematics
for
inanimate
objects).
But the
evidence
does
not add up to
indicate that gender,
of
either sort, was the reason behind thematism in N.
2.2. N-thematism, unlike verbal
e
/o
discussed
below,
does
not
spread
acrosssub-categories; nouns do not have e.g. voice or tense inherently.
Perhaps, therefore,
its
con tribution
was
rather
to
some sem antic property,
peculiar to that lexical
class.
But one may pause to wonder whether
4
pre-
morphosyntactic
might mean
pre-dating accusative-type
syntax
5
.
That
is,
whether thematism
is
among
the
vestiges
of an
earlier
active or ergative
type of Operation. That ergativity was a stage of PIE has been suggested
intermittently
from
Uhlenbeck (1901),
via
Pedersen (1907, 1933)
and
Vail-
lant
(1936), down
to
recent analysts like Schmalstieg
(1980,
1986, 1988)
or
Kortlandt
(1983).
Beekes
(1985:
192) derives
the
entire
o-stem
para-
digm
from the
single
os
ending (for such
paradigm-creation see Plank
1991),
s this
is for
him
the ergative
marker
in the PIE
hysterodynamic
inflection
(but
see
M atsumoto 1993
on
this).
If
here
we
have
a
declension
which is s a
whole inimical
to entities which do not readily
serve
s
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
8/26
10
agents, then indeed
One would
not expect neuters in the
o-stems
(Beekes
1985: 194). Yet neute rs and o ther inan im ate s occur the re
freely.
And in
only
ergativoid nominals
are at
home there, what
is the function of the
+
s marker? Again,
if
nouns
not so
marked
are
doing
some
other
Job
than the ergative, why keep their +o? (Such paradigm
bu i ld-up
elsewhere
tends to be either retricted to levelling,
s
when Latin replaces gen. sg.
- with -/f around 25 BC to level it with -ins,-ium,-f; or is stepped ,
s
Tocharian
first
establishes an oblique case and then adds further suf-
fixes to it to provide an array of spatial cases.)
Now sporadic recourse to
ergativity
is comm on enough. Celtic has an
Innovation
w hich signals
alike
the
intransitive sub ject
(S) and the
transitive
object (P): Irishleabhare
it
is a book agrees with
bua ilim
e beat
him .
The Hindi typemaine kitb
likhl
is a neo-ergative construct (on which
see Bubenik 1993)
deriving from
Sanskrit participal
maya pustakarii likhi-
tam wherein by me the book is
written
becomes
(erg.)
wrote the book
(absol.) Garrett proposes (1990) that
a
Common Anatolian ergative
Operation
arose
similarly
from transitive
V
with null subject plus neuter
N (instrumental ablative), the latter s ending -anz a)) com ing ultimately
from *-anti.
This
split
usage with neuters, like Hittite clitics conveying
only
S, is
post-PIE.
All these examples show ho w ergativity can pop up
at any time; that rules them out s strong proof of
early
or
inevitable
stages
of evolving
grammars.
For
ergativity
s
itself something led
up to,
note Anderson 1977.
PIE to-
seems
to fuse S and P in
e.g. Skt.
gata-
gone
but hata- killed ;
but, pace Garrett 1990: 263, this means only that
it
signals resultative
Status
(like potential result in Greek -tos; and co-oc-
curring with
an
agent marker
in
Skt, k-ta-va n)t-).
As for the
deceptive
English ergativoid + ee, ma rking S/P versus A in escapeelemployee versus
employer, this attaches widely to any
V-concerned
N. So in referee, am-
putee\
even
in V-less
lexemes like
refiigee.
Only special pairing preserves
-erl-ee w here
-ed
is available, notably when three entities are involved (s
in mortgaging). To all
this scepticism
on the historical
relevance
of
erga-
tivity one may addVillar s
doubts
(1983,
1984)
on it s a PIE
mechanism,
considering
its
forms.
Objections arise
to the
supposition
of a PIE
active epoch, too.
It
might
be reflected in case (so Matsumoto 1993). But that early IE behaviour
was of the
active/stative sort
describedby
Klimov
(1977) has
been rather
less frequently
urged, despite
the
active
-*
ergative
-*
accusative develop-
ment
noted above. (In the
Klimovian
model, nouns in pre-passive con-
structions
or
functioning s patients
are
Stative,
and
they become active
when they
lead
intransitive predications or act
agents.)
The tension be-
tween
activity and state is probably universal; it is certainly endemic in
IE. Aspect
enshrines resultative state
(perfectivity), and
even
the
mixed
idea
of
on-going action
(imperfectivity) s
against simple action.
The
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
9/26
11
middle voice
is
often belived
to
relate
at
least partially
to
stativity:
cf.
e.g.
Jasanoff
1978:
119
Rix 1988: 104,
166-167
sees them
s
partners
syn-
cretising inlatePIE - or tochangeofstate, sbyHar t 1988: 88). Indeed,
stativity was positively lexicalised by denominatives in
+e s c))-
see
Watkins 1971: 90-91; Jasanoff 1978: 125). But it is not clear which is the
unmarked
member of the pair inIE; and so relation to the Klimovmodel
is hard to state. In any case, n on e of this really ma tters because the t iming
iswrong.
Active-type
grammar has its own prerequisites: it needs enough
morphology
already inplace to signal argumen t-relation , N-V concord or
government, and verbal diathesis. But the
pre-morphological
nature, and
the still
evolving
morphophonology, of thematism force our eyes back to
amoreprimitive era
PPIE)
inwhichamorelikelyreason for this ma rkin g
is
a fundamental
cognitive appreciation
of sub-types of
items here
of N).
The recourse is not so much systemic
s
epistemic.
Twoother facts about
the o-stems
should
be
borne
in
mind.
1) It has
long been noted that between the IE languages there iscon siderable dis-
agreement s
to
which lexical items belong
in
that declension. This dis-
agreement
widens
s
the
declension becomes more populous,
and the
lex-
emic matching
is
particularly sparse between Anatolian
and the
rest
of
IE. 2)
Those
lexemes which
are
safely
deduced
to be
original
o-stems
are
also very
few
cf. e.g. Brosm an 1972: 62).
In
seekingPPIEmotivation
for marking off these words, one must allow for their being a minority
of oddities with local control on their selection.
There
is
another possible explanation
for
thembased
on an
earlykind
of sentence-architecture.
The PIE
pattern
of
case-marked roles
in a
t ran-
sitivity
System may have
succeeded
one in which, above all, one N was
limelighted s
the
topic
of the
communicat ion.Sen sitivity
to the
perspec-
tive
of the
event
sets
up
Topics and Com men ts) instead
of, or in
interplay
with, Subjects. Note here
Li
1976,
and
especially therein Lehmann
and
Li - Thompson.) Given an epoch when one of these stood out s the
essentials ignandum,+ could
be a
vestige
of the
signalling.
A
top icalising
PPIE or
just
a
topic-prominent PPIE;
see
Lehmann 1976: 456)
is
sup-
ported
by the
absence
of any
constraint among
the thematics
arising
from
inanimacy. A Speakermay set up
anything
s hisTopic. That the original
thematics were
so
few,
and
neuters made
up so
high
a
proportion
of
them
evenso,doesnot then matter. Moreover,
diffusion
ofo-stem membership
is
just
s
plausible
from a
starting point
in
topicalisation.
On the
other
band,
topic-marking implies
a
choice between multiple arguments
s
to
which is to be the topic without disruption of the event itself or the
various roles).Then
the
distinction between e.g.
a topicalised agent/actor
and a topicalised
patient/outer party will
need
signalling s
in
Tagalog
both byparticle and byverb affix). But no IE or PIE mechanism of that
sort
hasbeen identified;
wordorder
andpossible
intonational devices
are
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
10/26
12
irrelevant to our search, even if the latter were accessible. Voice might
have
been
so
used;
but the
most likely nuclear Opposition would have
demanded an active/passive contrast, while the obvious and agreed de-
duction from IE comparison is that the middle preceded the passive and
was the source (apart from some endings of basically Stative sense) of its
forms.This, again, is a false trail to the explanation of thematism in
PPIE.
2.3. If we return to the notion of
early
cognitive demands, various par-
ticularities suggest themselves.
If an
entity needs
to
receive special sig-
nalling it is presumably difficult to establish, in a sentence-usable value,
in its own right. It may be interpretable in speech only if :
(1) some other entity can be presumed s its anchor; or
(2) its pu rpo rt is determinable from the speech-context; or
(3) it is delineated by, or
cognitively
derived from, some basic root in
the
same semantic
field.
These
adjustments
may be
termed those
of (1)
(cor)relativity;
(2)
con-
text-sensitivity; (3) secondary definition. C omm on to them is the inderlying
sense
of
bafflement,
and its
dispersal,
among early communicators.
In this connection (especially in considering the first and second con-
ditions) it is
useful
to recall a group of nom inals noted by Schmidt
(1986:
96). They are not quite homogeneous, but in all three sorts they combine
a lack of objective Status with a readiness to serve
s
agents. They are:
kinship terms, names and pronouns
(let
us say, the KNP set). For first
noticing their like behaviour credit
is
usually given
to
McLendon (1978,
reporting
on
Eastern
Porno, an
active-type Hokan language
of
northern
California; the example is also used by Mallison - Blake 1981: 52). Does
0-marking
tie in
with
any of
these explicative mechanisms?
2.3.1.With a strictly correlative solution the kinship terms fit well.
Uncle
presupposes nephewlniece and imp lies relative ages;
cousin
is translative
between
2 +
persons;
no
term
is
absolute. Formally
the
picture
is
cloudier.
PIEkinship terms can be thematic: Hitt. alias
father
(cf.
annas mother )
agrees
with atta in Hellenic, Germanic and Slavic traces; Hitt.
huhhaS
grandfather
goes with Lat. auos\
and
son
is often
thematic
s Gk.
hu y)os,
Celt. *makkos.
But
w-stems
are
rife
and not all
attributable
to
analogy
or
attraction: Lat.
nurusmay
derive from
PIE
snoru- son s
wife
(so Szemerenyi
1977: 68),
the
underlying
form
being sunu i suyu
and
swe
are
equally original. Besides,
the
need
for a
clear relational marker
was
satisfied
not by
thematism
but by
creating
the
kinship marker par
excellence
in
+ t)er
(on
which,
and the
similar
+w y)
y
see
Benveniste
1973: 171, 205-206; also Gamkrelidze - Ivanov 1984: 761-772).
Names
are not
correlative:
one is not
John simply
in
relation
to
another s being Charles or
M ary.
Relativity may be
imposed,
by reference
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
11/26
13
to historical or literary patterns
( he
played
Hai
to my
Falstaff
etc.) or
in
legalistic
formulae
(s
in the
Roman
w ife s
marital declaration
ubi tu
Gaius ego Gaia)\ it is not inherent. Nor is
there evidence that thematism
was essential in (P)PIE names, even s basic s those of gods or rivers;
and we lack a credible history of IE onomastic Formation in general.
Pronouns
agree
with
other nominals
s to
some endings
in the
o-stem
declension.
So in the
nominative plu ral
in -01, the
ablative singular
in -:d
(a significantly
rare
use of the plain
voiced
stop, or
glottalic ejective,
s
a PIE grammatical
affix),
an d perhaps in the genitive singular (Beekes
(1985:
186) derives
the
Greek
H os-o from a
post-Mycenaean pronominal
source).
Bu t
these agreements
are too
sporadic
to
offset
the
fact
that
while pronouns are often opposed in deixis (this versus that, o r hiciste-ille,
in relation to the
speaker s
location) they are not correlative: that is, in
he buys these and sells those
th e
verbs entail each other
but the
pronouns
do not
(cf. Fillmore 1977:
72-73).
A more serious objection still is the existence among the thematics of
items
like
th e
widespread,
and formally and
sem antically consistent,
lexeme
*wlk
w
o-/luk
w
o-. It is
hard
to
conceive
any
sense
in
w hich
the wolf
is necessarily (cor)relative to anything
eise.
Yet it seems to be a founder
member
of the
paradigm.
2.3.2.The KNP set is more homogeneous in its
sensitivity
to the context
(s with type (2) above). For
them,
context presents
itself
s a
series
of
concentric
cricles. The
outermost
is the world of
discourse,
in
which
names
operate.
Then comes the social circle
within
which the kinship
terms make
sense,
with
a
smaller circle
of
local
or
familir referents.
A
yet
smaller
circle,
of
immediate speech
Situation,
gives meaning
to the
deictics; whilethe narrowest band of alll ts T and
*you
be understood,
this being
each successive
utterance.
Yet
again
w e
face patchiness
of
for-
mal
incidence in the various rings. It is quite understandable to desire to
mark
those lexemes which lack
a
context-free
In terp reta tion ; even e.g.
*woik o- dwelling may
convey
no clear
idea
of its
part
in
life,
let alone
its local shape, except
s
an element of the speech Situation. B ut whereas
an
approaching stranger, encountering
the
utterance give
them to
H ilary s
cousin,
can
attach
no
specific
content
to any of the
three
KNP
terms,
in
give
them to H ilary s horse at least the
last word conveys
a
clear
sense
and possible
Identification.
Yet the PIE word for
horse
is another
early
o-stem,
and defeats the contextual solution.
2.3.3.Let ustest
type (3), secondary
deflnition. As is
well known,
in PIE
an added - o- m ay signify a
derivation
N -> N or
A (dj):
a
Standard ex ample
is*roteH2 wheeF
(Lat.
rota)
* *rotH2-o- Vheeled
(Skt.
ratha-
chariot ).
Genitives in
-o
are relevant,
given
the adjectival
value
of the genitive (and
vice
versa).
Even
the use of
o
s a
linking
device in
compounds with
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
12/26
14
many stem-types may be a later extension of this sign of modification.
It
is therefore attractive to suppose that already in
PPIE
some roots were
thus
given
an
immediate extension
of
meaning;
to put it
more crudely,
some lexemes were usable only
if
the y could
be
interpreted s sub-types,
or special m anifestation s, of
existing
roots which had cognitive acceptance.
Thus *yug+o is the agricultural application of *Vyug ( join * yoke ),
and *woik'+ o
understood only
in the light of *Vwk' ( dwelP -> dwelling ).
Not many entities
would
seem so defeating s to need this epistemic
lo-
cating; hence the
low
n um ber of original N
-l
o lexemes. And the choice
would
depend
on the
speech group; hence
the
sparseness
of
interlingua l
cognates in that declension. To mark urgency at the N
=
V stage +e
was used; with separated N, recourse to the parallel back
vowel
is very
na tural. Patchiness
of
adoption produces
*nerin
Greek
((a)ner-) but *nero
in
Indic
(Skt.
nara-) \ man
and
manlike
were rivals, and perhaps the
basic form seemed overstrong, so that Latin has it only onomastically, in
Ner-.Patchy K NP incidence
is
reasonable too,
and
o- mark ing
may
Start
so.
But
a serious problem arises, none the
less.
One is obliged to find,
or
cogently
conjecture, an
underlying root
(N or V) in all cases.
This
is
not
straightforward even with
KNP items;
with objective items,
s
ma n y
undoubtedly are, the challenge is greater. For example, under the wolf
word there mustlurk some root
-
presumably *VJk
w
/luk
-
with
a
credible
m eaning which wolves can live up to. Perhaps it is *howP, an obvious pro-
pensity of the species; the activity is denoted by ul/ol in onomatopoetic
words elsewhere. This
is
just
the
sort
of
challenge
to fire the
ingenuity
of IE scholars.
[2.3.4.
Let us
take
a
speculative instance.
One can
easily belive that
the
horse
was noted
s
having
a characteristically
large
and expressive eye.
Then we require a root
*Vek
w
,
balancing the known *Vk
w
so common
in semantically related form s. Now,
in
establishing
the
dorsal stops
of
PIE,
some award original Status
to a
unisegmental labiovelar (the phonic
ele-
ments being later separated out in e.g. Anatolian or Tocharian B); others
think
that *k
w
is a
secondary contraction
of
bisegmental
kw or
k w
in
some languages (or in some words in those languages,
s
Sanskrit has
sacate
but asva-). The Indo-Iranian evidence suggests PIE ek wo-\ but
Celtic (OWel. epo) and Germanic (OE eoh) and Latin (s?quo- is like
s quor)
justify
an
Urform with
-k
w
-.
That some degree
of
co-existence
(and squinting between them ) is to be recognised isclear from Gk. hippos,
wherein
(apart
from theoddly high first vowel and the dialectal aspiration)
the -pp-
needs
explanation. (So it does in Boetian Thio-ppastos. In each
case
the
usual etymology
is from forms with medial *-k'w-.Then Greek
should lose the labial with or without retention of syllabic weight, if the
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
13/26
15
historyof other - C w > - sequences is our guide (so
monos/monos
from
- H H > -
tettareslpisures from
-tw-).
The
expected
form
is
then
*ikos or
*/:/coy
or
ikkosj of
which
only the
last
is found
(and only
in Etymologicum
Magnum
474. 12). The labiality m ust come from somewhere
eise,
and the obvious
source is the
rival -k
w
-
which before back vowels gives -/?- in Greek. Con-
versely,
if -k
w
~
is credited to the etymon the gemination needs excuse,
and
this
appears to be the
influence
of
bisegmental -k w-. Formal rivalry
seems to occur, then (see Szemerenyi 1990: 69 on the debated relation
of these consonantal combinations). If so, a root *Vek
makes
sensible
Szem erenyi s
etymology
ofblep (Laconian - glep) s from
*g
w
l-ek
w
-
cast
(my)
glance
at
(1969:
236-238;
1974:
145-146).
( Bu t
the
e -alternant
is
hardly a retrograde format ion ,s he says. If it reflects th e PIE me ta -
phonic stage it should be o (s it occurs in the noun element of the un-
derlying
phrase); if the verb is a late u nilingua l creation, a clearcut Greek
equation of back vowel
with
n o u n bu t front with verb is not there; cf.
nouns like belos,
tekos
with verbs like boulomai,lou.And at
least
one
would look
for a
pairing
of
blep
and
*blopos.)
If the
horse were termed
th e looking or
big-eyed
anim al, then *ek
w
o-/ek w o- is a ready form ation.
Of course, one m ight object th at dogs, too, have struc k hum ans, and no
doubt th e
Indo-Europeans,
with their
large
and
soulful
eyes. Well, th e
same etym on (but w ith zero grade and a nasal extension) is at hand: *k
w
o-
n/k wo-n (s
was
suggested
by
Cohen (1988-91)
to the
surprise
of
some:
see Polome 1994: 195). Greek ku-on-, Skt. svan-, Lith.
su-
all fit; Latin
would have *guon-(i)- *coni- (which
was
revocalised
to
canis
to
avoid
collision with
the
other
*coni~
word
ash,
dust
- cf. Gk.
konis
-
which
went
to cinis\ this seems more credible than that the -0- derived analogi-
cally
to
catulus
pup , s Szemerenyi (1987: 878) thinks). A corresponding
bisegmental *ok
w-
should accom pany the com m on *ok
w
- in this m eaning;
note
its
traces
in
Attic Greek
triottis
(-0CC-),
prospon
(-o:C-),
Homeric
eis
Spa\
2.4. After all, however, thematic neuters import problematic features of
their own:
2.4.1.
PPIE Speakers would have had no difficulty in apostrophising in-
animates if,
s
is probable, their W el tanschauung was animistic. Nor
does
it ever trouble Speakers, to judge from
Philoctetes
8 toxon phflon (So-
phocles,Philoc. 1128), or Lear s blow, winds, and crackyour cheeks (III 2),
or a
contemporar/s
come
on,
car
t
start, will
youl But ,
while
inanimates
which are admitted to the m asu line- fem inine set do possess a vocative in
-e
(s
Roman poets
address
their opus
s l ibelle - cf.
M artial
4.
89),
an-
imate referents which happen to have neuter forms never do. Even s
lates Plautus females with neuter names have o- less vocatives like mea
Gymnasium (several
examples are in the scene at Cistellaria 51-110).
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
14/26
16
There is no warrant for
recognising
an e-vocative in Indo-Irania n or An a-
tolian, where
vowel
histories obscure what is probably the
Stern-final
-o -
of the class. In other words, where the expoctulatory suffix is employed,
a
vocative
in
-e entails
a nominative in
-os.After all,
it is
understandable
to call on (and appropriately mark) only an entity which is capable of
appearing and/or postively acting. Thus this original cognitive linking, once
established, becomesset s hard s apieceofmorphological granite.
2.4.2.It is harder, however, even to a pproach an answer sto why neuters
in
the o-stem
paradigm,
and
there
alone,
possess
a final
-m
in the lexical
(or non-oblique)
case.
At first they comprise only inanimates; these are
unlikely actors or instigators at the Start, and
long
remain equally
unlikely
s subjects
of
sentences
(s
obviously
in
Hittite
and
Greek). Therefore
they do not need a nominative m arke r to separate their
non-existent
agent
role from
their nominal patientive one. They do not need an accusative
sign
either;
the
basic lexical shape willserve
fo r
nu clear grammatical pur-
poses.
And,
in all the
other declensions,
so it
does (the traditional Nom-
Voc-Acc lay-out inmanualsisqu ite otiose- only Accexists).Certainly,
Hittite thematic adjectives do sporadically lose their final
-n
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
15/26
17
3. V + elo
The third strategy of them atism appears at a late stage ofPPIE, over-
lapping
with PIE.
At that epoch one expects increasingly
grammaticaP
tasks marked
minimally
(or not at all): the PIE endingless locative belongs
there.
Suffixing by mid
vowels
now formed
stems
from
event-roots. This
was not, however, to establish
(s
with N) a separate pre-morphological
sub-class.
Some verbs
remained a thematic
throughout (if
defective );
others were
only
partially thematised, and none entirely so. The stative-
resultative perfect was never so marked. There was a corresponding
functional spottiness:
by the time of PIE
formal Opposition
may
mark
no
functional difference (s when aorists are both thematic and athematic;
root and sigmatic aorists are the latter) while conversely clear Opposition
of aspectual values may be obscured by unwelcome likeness of form (cf.
the Sanskrit non-durative past asicat poured and the du rative past
atudat
was
striking ,
both thematic).
Also, there has by now arrived an awareness of the usefulness of a
back/front contrast in mid vow els, mo rphoph onically emp loyable (this con-
trast may have been produ ced by differential placing of word-accent; bu t,
for
the
doubts,
see
Szemerenyi
1990: 125). This tactic
waslong-lived: cf.
English
sing/sang
etc.,
or
the phonologically alike mod ern Ru ssian
elo
con-
trast (see
Itkin
1994,
for
example).
In PIE it
might signal
the
derivation
V -* V ,
where
V may be
intensive
or
frequentative
(so Gk. pher >
phore
etc.)
or be
realised
s
a
deverbative
N
(s
Gk.
nomos nomos
-
nem, Lat. toga - tegoetc.). Anom alies are rife: Greek has
e
in N belos
y
tekos) and
o
in V
lou, oiomai)\
while PIE first person endings in dual
and
plural
are
-wes
or -wos,
-mes
or
-mos.
The
root-syllable o
in the
perfect
is
less
predictable
once
the
other
signs,
like reduplication,
are in
place:
cf. Gk.
leloipa
bu t
pepheuga.
It is
possible that
the odd
Variation
in
e.g. Gk. leg-o-men ,
leg-e-te
arises from
Manczak s rule (1960)
that before
a sonant
post-accentual
PIEebecameo \but the colour contrast is ruined.
(For similar wobbliness
in
apophony, i.e.
eC/0C, note
highly variable
guna/zero
re la t ion between presents
and aorists, or the
u n u s u a l
vrddhi/guna Opposition
for
number
in
proterodynamic
Narten
presents
or Insler aorists (Narten 1968; Insler 1972).)
Vowel suffixation had
obvious euphonic merits
in an
evolving mor-
phology if
consonantal endings needed
to be
attached
to consonant-fmal
roots.Yet it
occurs equally
after
resonants
and
vowels
(s in
Greek con-
tract
verbs). So it must
have some
early more
semantic
role. Was
this
the
marking
of one
polar term
in an old
active System
a
la Klimov
(1977)?
The
difference between e.g.
Gk. -men and bain-o-men
Ve
are going is
then unclear (time reference apart);
and if the
latter means really
we are
actively
moving
(essentially
the view of
Eichner 1975:
77 fn 3, on
limited
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
16/26
18
evidence),
how
does Homeric e
d ienai made a
move
fit in?
True,
medio-passive perfects
are
athematic
an d
resultative-stative
in
sense,
so
that thematism seems to be the sign of
/+
active/. B ut Renou (1925) was
sure,
from
Vedic, that
/+
active/
was the
unmarked term,
in
both
senses;
and that thematism denoted / +
Stative/
gains some support from the en-
duringuseof that value and the
re-elaboration
of positive indicators (such
s IE
+e(s(c))
- for
Stative-inchoative, identified
by
Watkins,
e.g. 1971:
86-89). But the polar terms are not easy to apply to e.g.Gk.
eimi be
buteimi
(will)
go ,
where
form and function
disagree.
And
subject-activity
verbs
are
thematic
in
some dialects
of
Greek
and not in
others
(s
Attic
agre
capture
versus Aeolic
agremi)
or
differently
in
dialects
of the
same
group (s Attic didsi he gives versus Miletian
didoi,
or Attic histanai
to stand
versus Euboean kathist ffn)
or
even between tenses (cf.Attic
present
tithesi
with imperfect
ethei,
not *etithe). On all this consult Tbcker
1990:
73 fn 83. In
effect,
the
reason behind
e/o
root-suffixation
in only
some verbal forms remains to be found.
For Kurylowicz (1956: 74) its starting point was in aorist formations,
already
so fashioned; but that
simply
shifts
the question elsewhere. The
origin for Watkins
(1969:
65) was themiddle voice; and Hart (1990:
448,
462)
supposes it, at least inClassI
presents,
to come
from
the
middle
3
sg.ending
-o,
although its position within the verbal construct seems to
rule
that out.
In any
case,
to
locate
its
start-point
is not
enough: what
was it meant to do? To that question many answers have been offered;
their variety is illustrated in Figure 1:
name date Suggestion
Thieme
Meillet
Renou
Knobloch
(so Kortlandt 1983)
Risch (so Rix 1986)
Eichner
Schmalstieg (cf.idem
1980,
1986, 1988)
Shields(cf.idem 1992)
9 9
93
93
953
965
975
978
989
partial
marker
of
middle voice ( une
tendance vers la
voix
moyenne )
(1)
component
of
morphs
+
sko/
+ no
(2 )
subjunctive
(from
athematics)
marker
of indeterminate dimension
( eventueP ... independant de toute
categorie
verbale )
marks
that verb has object (so does
nominal thematism)
marker ofsubjunctive (itself source of
indicative)
partial marker
of
continuous action
concordal marker of ergative-type actor
(=
pronoun)
marker of non-present (for
fuller
-yo-)
Note:seebelow on
Valliant (1936, 1937)
and Kurytowicz
(1956).
Figure
1.
Functional
diagnoses of (P)PIE
verbal thematism
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
17/26
19
It is
relevant that
Plank (1980) has
demonstrated that
confusion be-
tween
the
marking
of
subject
and
object
in a
transitive
System
(or, pre-
sumably,S /P and A in an ergative one) is a relatively tolerable amb iguity.
This ruling could commend the solution of
both
Knobloch and Schmal-
stieg:but
they
are
intended
to
preclude each other. Again,
transitivity has
numerous
diagnostics
and in its
career
has had
many unexpected
Signals
(see
Hopper
- Thompson 1980). In IE, Toc harian was described by inter
(1980) s so using (a)
stem-initial
consonant palatalisation (in the pre-
terite), and (b) a nasal formant (in the present - he notes that all Latin
nasal-infixed verbs are transitive with the solitary exception of the -cum-
bere
dervatives),
and (c)
initial
syllable accent
(in
second
subjunctive
and
causative fo rm ations). Jam ison
(1993)
thoug ht Sanskrit-ay-
on
old
-0-pre-
sents to be transitiviser. These
fmdings
prom ote the notion that
thematism
is an object-sign; by their wide differences they reduce its inevitability.
Indeed, long ago
Kurylowicz
provided (1956: 74 fn 47) the simplest
counter-argument to the Knob loch-Kortlandt position, em ploy ing IE data
where the arguments o f bis opponents rested first on K abardian and then
on Hungarian. He noted that there are plenty of clearly transitive verbs
in (P)IE which stay athematic: he listed many from the gveda atti,
degdhi,
vesti
...
and could have
added
all the Greek
second
wave group,
tithemi, ollttmi
etc.). Presumed
early PIE
ergativity enters here. To the
suspicions aired above one may add disbelief that so
careful
a marking
of
the so
careful
a
marking
of the
S-V-O
(or
A-V-P)
syntagm
should
so
soon fade to a scarcely visible and purely paradigmatic affair. Such fading
is
more serious than the common
loss within
a single category of some
indistinct or arbitrary Opposition.
(In passing, Vaillant
is
omitted from Figure
l
because
his
1936
paper
was not really concerned with verbal thematism
s
such, while his 1937
article sought to relate
Class
I and VI thematic verbs to Hittite
-M
and
-m i conjugations.)
Kurylowicz once suggested (1956: 73) that one function of thematism
s
in
*leik
w
elo-
was to show the subju nctive mood.
This
idea can be
traced
to
Meillet
(1931), and
perhaps arose independently
in
Renou (1932)
s
a development of his eventueP. It was revived and revised by Risch in
1965; he sou ght to establish the pr ior ity of the subjunc tives a mood over
the indicative (cf.
Rix 1986). Within
PPIE this construct may be
called
the
pre-subjunctive - s pre-modaP might mislead, and such demo-
dalised derivations
s
Tocharian
B
imperfect
< PIE
optative
are not
directly relevant, although they do demonstrate possibilities. Now, many
languages
are less
sensitive
to
(for example) time than
to a
fundamental
differentiation of
verbal predications
s between the
actual
and the
con-
tingent. Thus
inHopi the
verbal diatheses
are basic,
habitual (consu etudi-
nal),
and
contingent;
and the
contingent
value
often contextually entails
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
18/26
20
/+ fu ture/ . Tagalog meshes the parameters of aspect and reality,
s
in
Figure
2:
(Tagalog)
aspect
p u nc tu a l
durat ive
reality
actual
(past) event
(present) event
process
cont ingent
hypothet ical event
possible
[
f u t u r e
process
Figure
2.
The
Latin
ideal
second person subjunctive falls
in
here: cLdelem
haud
ferme midien inuenias uirum (Terence, Andria 460)
you
hardly ever
find
a man
faithful
to a wom an . I t is
therefore quite thinkable that
the e/o
suffix
was an early
marker
of a
deep
but
simple distinction. Delicacy
s
to persons, number and so on came later (cf. Schlerath Rittner 1985,
passim); and the markers were placed fur ther from the root,
s
being
syntactical-inflecting
whereas thematism
was
concerned with assessment
of
th e
event s
degree of reality in the context (and so, more adjacent to
lexical values). Merely to
mention
th e
verb s
action may be a
doubtful
thing; but to assess the eventsnot actual bu t'happenable' seems natural
enough. If anything clear emerges from sociolinguistic enquiry it is that
the schema if p then
q',
or possibly q, given p', is well established even
in
mentally naive speech com mu nities.
One may
comm end again Gonda s
old formula(1956: 69-70) concerning a
'process...
not yet having a higher
degree
of being than mental existence... visualisation ...' (That Gonda
actually contrasts this value with contingency is due to
bis reserving
th e
latter term
for the
precise pragmatic force
of
what
is
said, doubt s
op-
posed
to
fear, etc.)
The next thing was for the marked form to be used to convey the
unmarked
sense.This
sort of linguistic Inflation is awkward but common.
It was aided in this case by the convenience of the linking vowel in con-
sonant-bound sequences
like
*ya n)g+te.
The
sense
of
contingency (ap-
plicable in respect of wanting or urging or supposing the event) abides
s
the
subjunctive;
it s
early date
and
tem poral insensitivity perm it
it s
having s fur ther
affixes
both
old secondary*
endings
and new primary*
ones (with
-/). The
mood value
was so
attached
to
-e/o-, however, that
once thematic forms had faded into simple declaratives they had the con-
tingent
function
re-imposed upon them,
by
repeated thematism:
so, in
Greek,
s
-men
had been
ex tended
to
-o-men,
faded bain-o-men spawned
*bain-o-o-men > bainmen.Some find it hard to accept that these pre-
subjunctives
could
come to have a non-contingent (faded)
role
at all, s
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
19/26
21
doesHart
(1990:
446-447)
on the ground
that e.g.
the
future
is not em-
ployable in apresent,
non-modal,
sense whereasthe converse is
f requent ,
s inEng./ am
leaving
tomorrow).But the
future
is soused, being poised
between tenseand mood; so acharacter in Terence
(Phormio
801) defends
a
previous
Statement by sayingsie erit\non temeredico it
is
so; Fm sure
of
it ; and
when Seneca declares that glory alwaysaccompanies courage
he uses comitabitur
Epi.
Mor. 79.13).There is an extra implicature that
corroboration is still to seek;but then the present used of a
future
event
also carries
an
implicature, that
the
process
has
sgood sbegun.
The
pre-subjunctive
split
into sub-categories
of
which future-marking
w as
only
one (to
invert
the
sequence proposed
in
Hahn
1953).
One
cannot
imagine
that the thematic
form
was the original, unlessby supposingGk.
i-men kei-tai
or
Skt. as-ti, se-te
to
have been produced
by
subtraction.
Hencesome additional task must have been imposed
on the
basic verb,
and yet the new
form assigned
to
that task
w as
able
to
lose
it s
force
by
attrition
or Inflation. (So Trkish verbal
affix
-mi - loses it s
inferential
value inwritten, third person, past tense constructs;seeLewis 1967: 122,
162versus 101, 140.)
It is harder to decide which of two
ultimately
contrasting systemic
terms is
prior .
The
formal
origin and the system s inventionmay be at
variance. Within the IE modal System, selsewhere, the athematic shape
is in
place
first.
While
it
represents
the
indicative non-contingent) value,
historical
an d
systemic priority
are
congruent. Even when,
in
some verbs,
the thematic shape has been misapplied so
s
to convey the unmarked
term,
the
thematism
i s
repeated
to
make
a
physical sign
of the
plus term;
so
that
the
indicative
form is
still
th e
earlier
of the
two.
Yet the
value
indicative
in a
mood-system becomes
the
ground-term only after
the
emergence of a thematic shapewith a marked modal value;so, in a sense,
the indicative does derive
from
the subjunctive. The nearest parallel in
outer history which comes to mind is in the
evolving
collegiate structure
of
the
University
of
Durham founded1832).From 1833
to
1845
the
entire
body of members were the university. In 1846, with the founding of Hat-
field Hall (later College), the prior group
only
then became the senior
College University College)
in the new
dispensation.)
4.This
paper
1
has made several interlinked suggestions. One is that a
periodwe may call PPIEw as characterised by a series of cognitive
leaps.
These
were increasingly sophisticated
and
abstract;
and
they ended
in a
morphological overlap
with
the organised paradigmatic marking,in PIE,
ofconventionalised categories.Another is that,
within
an era of imprecise
but
permansive
pre-morphology,
tworevelations
co-occurred.
The first
1. Of which an earlier
version
w as given at the PercevalMaitlandLaurenceseminar
in Cambridge, 25 May 1994.
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
20/26
22
was a philosophical grasp of the diffe renc e between entity and predication;
the
second
was a phonological appreciation of the balanced nature of
front
and back vocalism, and of its capacity of signalling that items are
related but opposed. (This use of metaphony went along with an equally
patchy reliance on thegrades of apoph ony to pe rform slightly mo re reg-
ularised tasks.) This picture ofhistorical stages issummarised inFigure3:
stages ofPPIE
1.
early -*
2. middle
-
3.
late
-
(post-Anatolian?)
whatis to b e
signalled
expostulation
secondary
definition
contigency
of
event
onwhat
element
N
= V
N
V
with
wha t
vowel
W
[o]
[e/o]
Figure 3.
The late overlap withPIE when it had lost the Anatolian group of
lan-
guages
(which show no sign of a subjunctive) is balanced by an early
possible overlap with an epoch of nostratic adherence, even with a quite
embryonicspeech stage.To face this latter n otion is to reject the im itation
of
reconstruction
to
(s
K urylowicz
(1964:
58)
proposed)
'stages
bordering
the historical reality*. We should pursue thematism rather
further
back
than that,
and
recognise
it
s
one
process
of
shaping which
had
successive
varieties of form and ofm eaning.To say so is simplyto put two and two
together;to be sure that the semantic details
proposed
above arecorrect
may
be tomaketheanswerfive. But
Indo-European ists
shouldnot neglect
this area ofenquiry, and can no doubt get the sum right in the end.
Neville Collinge
31
StansgateAv
CAMBRIDGE
CB2 2QZ
UNITED K I N G D O M
REFERENCES
Anderson, S.R.
1977 On mechanisms by which languages become ergative , in: C.N. Li (ed.),
317-363.
Beeks, R.S.R
1985 T he origfns of the Indo European nominal inflection. Innsbruck: Innsb. Bei-
trge
zur
Sprachwissenschaft
46.
Benveniste,E.
1973 Indo European languageand
society.
[Iranslationby E. Palmer of
(1969)
Le
vocabularie des
institutions indoeurop
ens.]London: Faber &
Faber.
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
21/26
23
Brosman, P.W.Jr.
1979 The Hitt ite neu ter
-stems ,KZ 93: 54- 63.
1984 The
IE
cognates
of the
Hit t i teai
and
au
stems ,
JIES
12:
345-365.
1994 The PIE
r/i/-, /uu
and
f^
-sterns , Folia
Linguistica
Historica XV: 3-19.
Brugmann,
K. H. Osthoff
1880 Morphologishe
Untersuchungen.
Vol.3. Leipzig: THibner.
Bubenik,
V.
1993
Mo rphological and
syntactic change
in late
Middle
Indo-Aryan ,
JIES
21 :
259-281.
Burrow,
T .
1955 Th e
Sanskrit
language. London: Faber
&
Faber.
Chomsky, A.N.
1982 Lectures
on
govemment
and
binding.
Dordrecht:
Foris.
1995 Bare phrasestructure , in: G. Webelhuth(ed.),385-439.
Cohen, G.L.
1988
Etymology and linguistic
principles.
(2vols.)Rolla, MO : Universityof
Missouri.
Cole,
P.
J.M. Sadock (eds.)
1977 Syntax an d
Sematics
8: Grammatical
relations.
New York:
Academic
Press.
Collinge, N.E.
1985 Th e laws of Indo European. A msterdam P hiladelph ia: Benjamins.
1993 Review
of
Plank 1991.
Studies
in
Language 17:
264-269.
Cresswell, M.L.
1994 Language in the
world.
Cambridge: Cambridge University
Press.
Dixon,
R.M.W. (ed.)
1976
Grammatical categories
in
Australianlanguages.
Canberra:Australien Institute
of
Aboriginal Studies.
Dunkel, G.E.
1992
Two old
problems inGreek: and5 \ Glotta 70: 197-
223.
Eichner, H.
1975 Die Vorgeschichte
des
hettitischen Verbalsystems ,
in: H.
Rix(ed.), 71-104.
Etter, A. (ed.)
1986
O o pe ro si .
(Fs.
Risch.)
Berlin
New
York:
\\folter
de
Gruyter.
Fillmore, CJ.
1977 The case forcase
reopened ,
in: P.Cole J.M. Sadock (eds.),59-81.
Fisiak, J. (ed.)
1980 Historical morpho logy. The Hague: Mouton.
Gamkrelidze, TV. V.V. Ivanov
1984 Indoevropejskij jazyk i indOeiropeicy.
(2
vols.)
Tbilisi:
Tbilisi State University.
[Translated
by J.
Nichols
(1995) s
Indo European
and the
Indo Europeans
(Trends in Ling uistic s 80.) Berlin New York: Mouton de
Gruyter.]
Garett, A.
1990 The origin
of NP
split ergativity ,Language
66:
261-296.
Georgiev, V.l.
1984 Die drei Hauptperioden
des
indoeurop ischen ,
D iachronica 1:65-87.
Gonda,J.
1956 Th e
character
of the Indo European moods.
Wiesbaden: Harrassowitz.
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
22/26
24
Hahn,E.A.
1953 Subjunctive
an d Optative:
Their origin s
futures.
New York: American Philo-
logical Association.
Hart, G.R.
1988
Anatolian
evidence
and the
origins
of the Indo-European medio-passive ,
BS OAS 51:
69-95.
1990 'Class l present', subjunctiveand m iddle voice in Indo-European , BS OAS
53: 446-468.
Heesterman,J.C. G.H. Schokker VI. Su bra ma n ia m(eds.)
1968 Pratidnam. (Fs. Kuiper.)The Hague:
Mouton.
Heinz,A M. Karas*
(eds.)
1965
Symbolae linguisticae.
(Fs.
Kurytowicz.) Wroclaw Warszawa
Krakow:
Polska Akademie Nauk.
Hoffman,
K.
1967 Der Injunktiv im
Veda.
Heidelberg: Winter.
Hopper,PJ. S.A. Thompson
1980
Transitivity
in g ra mma rand discourse ,
Language
56 :251-299.
Insler,S.
1972 On proterodynamicroot present inflection ,
M SS
30:55-64. (Butsee also
Language
48(1972): 557.)
Itkin, I.B.
1994 Jee raz o Ceredovanij e/ v sovremennom russkom jazyke , V Ja 1994. 1:
126-133.
Jamison,
S.
1983
Function and
form
in the
-aya-
formations of the Rig
veda
an d
Atharva Veda.
Gttingen:
KZ
Ergnzungsheft
31.
Jasanoff, J.
1978 S tative and middle in Indo-European. Innsbruck: In stitu t fr Sprachwissen-
schaft
der Universitt Innsbruck.
Kammenhuber , A .
1985 Zu m M od us i n j u n k t i v u nd z u m Dre i -genu s -Sy s tem im U r indoger -
manischen ,
in: U.
Pieper
G.
Stickel
(eds.),435-466.
Klimov,
G.A.
1977
Tipologija jazykov aktivnogo
stroja.
Moskva:
AN SSSR.
Knobloch,
J.
1953
La
voyelle themat ique
-e/o-:
serait-elle
un
indice d'objet indoeuropen? ,
Lingua3: 407-420.
Kortlandt, F.
1983
Proto-Indo-European
verbal
syntax , JIES 11:
307-324.
Kurytowicz,J.
1956
L apoph on ie en
indO-europoen.
Wroctaw:
Polska
Akademia
N a u k .
1964
The
inflectional categories
of
Indo-European. Heidelberg: Winter.
Lehman, W.P.
1976 From topic
to
subject
in
Indo-European ,
in:
C.N.
Li
(ed.),445-456.
Lewis, G.L.
1967 Turkish g rammar. Oxford: Clarendon
Press.
Li,
C.N. (ed.)
1976
Subject
and
topic. New
York: Academic
Press.
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
23/26
25
1977 Mechanisms
of
syntactic change.Austin,
TX:
University of TexasPress.
Mallinson,
G. BJ.
Blake
1981
Language typology.
Amsterdam
New
York
Oxford:
North
Holland.
Manczak,W .
1960
Origine de Tapophonie e/o en
indo-europen ,
Lingua 9:277-287.
Matisoff, J.A.
1990
On
megalocomparison ,
Landtage 66:106-120.
Matsumoto, K.
1993
Problem
of ergativity in Indo-European ,JIES 21:303-329.
McLendon, S.
1978 Ergativity,
case,
and
transitivity
in
Eastern Porno ,U L
44:
1-9.
Meillet,
A.
1925 Review of Fs. Wackernagel (1923).
BSL
25: 46-49.
1931
Caractere
s^condaire
du
type
thomat ique
indo-europe en ,
BSL 32:
193-203.
M hlestein,H E. Risch (eds.)
1979
Colloguium
Mycenaeum. Neuchatel: Universite* de Neuchatel.
Nar ten ,J.
1968 Zum
'proterodynamischen' Wurzelpr sens , in:
J.C.
Heesterman
et al.
(eds.), 9-19.
Nichols, J.
1986 Head-marking
and
dependent -marking grammar ,Language
62:56-119.
Pedersen, H.
1907
Neues
und nachtr gliches , KZ40:129-217.
1924 [1962J Sprogsvidenskaben i
det
nittende
Aarhundrede:
metoder og
resultater.Copen-
hagen: G yldendalske Boghan del. [Translation by J.W . Spargo, The
discovery
of
language.
Bloomin gton, In.: In dian a UniversityPress.]
1933 Zur
Frage
nach
der
Urverwandtschaft
des
indoeurop
ischer mit dem
ugro-
fionischen , Mem. Soc. Finno-Ougrienne
67:
308-326.
Pieper,
U G. Stickel(eds.)
1985 Studia
linguistica diachronica
et
synchronica.
Berlin: Walter
de
Gruyter.
Plank, F.
1980 Enconding grammat ica l relations: Acceptable and unacceptable non-dis-
tinctness ,
in: J. Fisiak
(ed.),289-325.
1991
Paradigms:The economyof inflection.
Berlin
New
York: Mouton
de
Gruy-
ter.
Plank,F. (ed.)
1979
Ergativity. New
York: Academic
Press.
,E.G.
1994 Reviewof Cohen
1988-91.
JIES 22:195-196.
Puhvel,J.
(ed.)
1969 Substance and
structure
of language.Berkeley
Los Angeles:
University
of
California Press.
Renou,L.
1925 Le type vedique
tudati .
Mtlanges Vendryes. Paris: Champion,309-316.
1932
Aproposdu
subjonctifvedique ,
BSL 33:
5-30.
Risch,
E.
1965 Zum Problem
de r
thematischen
Konjugation ,
in :A.
Heinz
M .
Karas*
(eds.):
232-242.
(Also
inKleine Schriften
(1981):
701-709.)
Brought to you by | University of Sussex
Authenticated | 139 184 30 135
Download Date | 8/23/12 10:50 PM
-
8/11/2019 Before PIE_ Motives for Thematism.pdf
24/26
26
Rix,
H.
1986
Zur Entstehung des indogemianishen Modussystems.
Innsbruck: Innsbrucker
Institut f r Sprachwissenschaft.
1988 The
Proto-Indo-European
middle: Content , forms
an d
origin ,MS S 49:101-
119.
Rix, H. (ed.)
1975 Flexion
un d
Wonbildung. (Vth Fachtagungde r indogermanischen Gesells-
chaft.) Wiesbaden: Reichert.
Schlerath, B . V. Riltne r
(eds.)
1985
GrammatischeKategorien:Funktionund Geschichte.
Wiesbaden: Reichert. (=
Vllth Fachtagungde r indogermanischen Gesellschaft.)
Schmalstieg,
WR.
1978 The m iddle
voice re-visited ,
Folia
Linguistica
12:
349-366.
1980 Indo-European linguistics.U niversity Park PA: Pennsylvania State University
Press.
1986 The ergative
function
of the Proto-indo-European
genitive ,
(Remarks on
Kortlandt 1983.)
JIES
14:
161-172.
1988 The ergative syntax
of the
Indo-European middle
aorist ,Lingua
Posnanien-
sis31:11-22.
Schmidt,K.H.
1986 Zu r Vo rgesch ichte
des
prdikat iven Syntagmas
in
indogermanischen ,
in :
A.
Etter (ed.),90-103.
Shields,
K.C. Jr.
1989 The
origins
of the thematic vowel , IF 94: 7- 20.
1992
A history of Indo-European
verb mor