Topic markers on direct objects

Fri Aug 5 18:27:24 UTC 2005

Dear Alice, dear colleagues,

as Alice has posted her response to my modest contribution to the 
O-Topic issue to this list, I'd like to briefly answer using the same 
way. Basically, Alice is right when claiming that

> Udi does not have a particle or affixal topic marker, let alone one 
> used exclusively on direct objects. 

and

> Second, the clitics that Wolfgang refers to as “floating agreement 
> clitics” do not mark new topics, but focus. 

However, I think that, here, we face a definitory problem. In my eyes, 
Focus and Topic do not constitute distinct categorial domains (at least 
in Udi), but are structurally coupled forming a functional cluster of 
pragmatic marking. Accordingly, a language may (!) use a single strategy 
to encode the pragmatic 'interest' of the speaker in multiple facettes. 
It would then show up as some marker to indicate topic features, in case 
it is occurs with topic sensitive referents. In this sense, the Udi 
dative (in the Nizh dialect) or Dative2 (in the Vartashen dialect) serve 
to encode just what Alice describes:

> (...) use of the dative case for a direct object in Udi is also not 
> topic marking, but indicates instead definite direct objects (...)  It 
> is true that definiteness is related to topicality, but a definite 
> direct object need not be topic. 

I'm not sure whether the distinction '(in)definitenes' vs. degrees of 
topicality always is useful. Let me quote from a private mail that I 
have sent to Claire Bowern on the same topic (concerning Turkish). I 
wrote: "I'm not quite sure whether there really is a pronounced 
difference between the notion of (in)definiteness and grades of 
topicality. As far as I know, the Turkish 'accusative' (-yI) is used to 
mark referents that have either been mentioned previously in the 
conversation or are otherwise clearly identified by the context (to use 
Underhill's wordings in his Turkish Grammar).  In my eyes, the first 
feature clearly refers to Given Topic, which (in Turkish) remains 
unmarked in S/A function but which is marked for the -yI form in O 
function. The second feature refers to what I had in mind when talking 
about frames, scripts, and knowledge states (> typicality = what can be 
expected to be relevant or typical in a given situation, frame etc.). As 
far as I know, differences in knowledge states are directly linked to 
what is linguistically called '(in)definiteness'. Personally, I define 
TOPIC as a pragmatic function used to relate a bit of information to the 
knowledge state of the speaker/hearer (be it textual or contextual). And 
this is exactly what Turkish -yI (according to my data and informants) 
does: It subcategorizes the TOPIC function according to the notion of 
giveness (> known) vs. non-giveness (> 'new')".

I have the impression that Udi works at least in parts in analogy with 
Turkish. However contrary to Turkish (but with amazing parallels in some 
Northwestern Iranian languages), a functional blend occurs in case a 
referential entity (liable for topicalization) and a focusing strategy 
collide: The 'normal' (prototypical) interpretation of this blend would 
be that of newness and relevance especially in case the referent is 
marked for a function typical for the newness domain (e.g. the 
O-function). This does not exclude that with other types of 
constituents, the 'Focus' function prevails. Maybe that - as [for the 
first time] shown by Alice in her wonderful book on endoclitics - the 
Focus function of Udi floating agreement clitics represents the original 
functional domain - however, in contemporary Udi the constructional type 
'Referent:O [zero-marked] + AGR' very often reflects features of  'New 
Topic'.

As for the reverse issue, namely the use of the dative(2) to encode - as 
I claim - a Given Topic in O-function, I stick to what I have quoted 
above. Note that I use the term Given Topic also in the context of 
typicality: Accordingly, a referent functions as a Given Topic also if 
it relates to a certain (typically expectable) knowledge state, as in 
the following Udi phrase: A person says, that "if we know beforehand 
that the general director will come....":

mu"t'la"q'    c'oy-a        xam-p-sun-e                     laze^m
absolutely  face-DAT  shaving-LV-MASD-3sg  necessary
'...it is absolutely necessary to shave one's face.' [Nizh dialect]

Here, the dative marked term c'oya is (in my terms) a Given Topic 
because it refers to a referent in a typical and known situation (of 
shaving). The story goes on:

te-ne-sa        s^o-t'in           yaxun       mo"hk'a"m  
dava-ne         b-esa
NEG-3sg-if   he-SA-ERG  we:COM  strong           quarrel-3sg  do-PRES
'Otherwise, he strongly quarrels with us.'

Here, the term 'dava' (quarrel) introduces a New Topic of high pragmatic 
relevgance (zero-marked O + AGR). There are many instances in Udi tales, 
where a referent in O-function is marked for the dative(2) even though 
it is not 'definite' in the strict sense of the term, compare (taken 
from the same tale):

zu   ta-ze^-sa             k'oy-a            s'um-a          
u-ze^-k-sa             o%s'a%   c^'e-ze^-sa
I     go-1sg-$:PRES  house-DAT  bread-DAT  eat-1sg-$-PRES  then          
go=out-1sg-$:PRES
'I go home, have supper and then I go away.'

Here, the speaker surely does not have in mind a specific (definite) 
kind of 'bread' (> meal), but again a typical situation of having 
supper. All this does not exclude that the dative(2) is also used to 
encode 'true' definite referents, but that's probably not the primary 
(original) function of the case metaphorized from the allative (note 
that in Old Udi (the language of a Palimpsest recently found in the Mt. 
Sinai monastery, 6/7th century AD), the dative(2) competes with a 'true' 
definite article: Here, definiteness is normally encoded with the help 
of an article, whereas the dative(2) functions independently (again (in 
my eyes) to encode a Given Topic (in the broader sense))....

In sum, I think that Alice looks at the same phenomenon from just a 
different perspective. A (massive) corpus-based analysis of dative(2) 
and zero-O + AGR constructions will perhaps tell more about which 
perspective makes more sense with respect to the Udi patterns (I 
currently prepare such an analysis for my Function Grammar of Udi 
project). For the time being, I have the impression that both 
perspectives can be taken. All depends from which theoretical approach 
is chosen. Still, one point should be stressed: Without Alice's 
pioneering work (her Endoclitics book), such a discussion wouldn't have 
been possible. Thanks for this!

Very best wishes and many thanks for the stimulating comments,
Wolfgang

-- 
#############################
Prof. Dr. Wolfgang Schulze
Institut für Allgemeine und Typologische Sprachwissenschaft  (IATS)
[General Linguistics and Language Typology]
Department für Kommunikation und Sprachen / F 13.14
Ludwig-Maximilians-Universität München
Geschwister-Scholl-Platz 1
D-80539 München
Tel.:     ++49-(0)89-2180 2486 (secretary)
             ++49-(0)89-2180 5343 (office)
Fax:     ++49-(0)89-2180 5345
E-mail: W.Schulze at lrz.uni-muenchen.de
Web: http://www.ats.lmu.de/index.php
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20050805/8bf3764b/attachment.htm>