frequency and agreement

Wolfgang Schulze W.Schulze at LRZ.UNI-MUENCHEN.DE
Mon Apr 3 12:36:00 UTC 2000

Andrew asked:

> Is anyone aware of work that has been done on grammatical agreement and
> frequency?  For example, if a verb agrees with an object (variably) by
> (variably)adding some sort of verbal inflection, at what point is there
> generally considered to be "agreement."  Is it at 50%, 75%, 100% of the
> time?
> The particular phenomenon that I am working with is indirect object
>   clitic-doubling in Spanish.  This has been called agreement by many
>   working in formalist and functionalist frameworks.  I am interested in
>   at what point clitic-doubling ceases to be doubling and begins to be
>   grammatical agreement.

In this context, statistics appears to be a rather problematic
indicator, at least to me. The grammaticalization of clitics (or
what-so-ever) as agreement markers has a high functional value (e.g.,
focus, anaphoric or cataphoric (pragmatic) reference (clefting),
assignment of syntactic functions (grounding)...). In case a language
system is 'on the way' to develop agreement techniques we should expect
that the occurence or 'missing' of agreement elements correponds to the
activation degree of such functional properties. Consider the following
example from Udi (South East Caucasian):

(1)    chowal-ay tur-e sa cac baIq'-ec-e
        sparrow-gen foot-dat one thorn stick-intrans-perf
        'In the sparrow's foot there was stuck a thorn.'

(2)    chowal sa karna-n-a-ne tac-i
        sparrow:abs one old=woman-sa-dat-3sg:s go:past-aor
        'The sparrow went ('flew') to an old woman.'

(3)    chowal t'ia ar-i p-i-ne s^o-t'-u
        sparrow:abs there come:past-aor say-aor-3sg:a dist-sa-dat
        'The sparrow came there [and] said to her...'

In (1) AGR is mssing, in (2) it is present, and in (3) it is missing
with the first verb, but present with the second one. The distribution
of both patterns is conditioned by functional aspects: In (1) we have a
stative construction without a distinct focus constituent, in (2) the
'old woman' is focused (constituent focusing) [note that we have
floating agreement clitics in Udi!], in (3) _ari_ and _pine_ are chained
which conditions that only the last part of such a chain is marked by a
clitic. If we count all occurences of agreement elements in an Udi text,
we would arrive at perhaps 60% (depending on the kind of text you
choose, some text go up to 95%). This figure would tell us that
agreement techniques in Udi have strong functional properties in a
synchronic perspective (which can also be inferred from the 'floating'
    The problem Andrew addresses is associated with the presence or
absence of pronominal referents (that is agreement is trigerred by
personal pronouns or something like that). Only in such instances we can
describe 'doubling' techniques, scugh as Udi

(4)    zu kala-zu
        I big-1sg:s
        'I [am] big (foc).'

(5)    zu-zu kala
        I-1sg:s big
        'I (foc) am big'

It sounds logical (to me) to call such a technique an 'agreement'
technique if the pronominal constituent can be deleted, cf.

(6)    kala-zu
        'I am big' (no constituent focus here)

In such cases, the clitic does not have an overt (textual) antecedent
but serves to identify a given actant on the verb. But I do not think
that this assumption is adequate in the opposite sense.
    A language has (pronoun based) agrement if the clitic in question
can appear without its antecedent (but not necessarily does).
    A language has a system of pronominal doubling, if the antecedent
can appear without the clitic (but not necessarily does).
    [Note that I superficially use 'antecedent' for any kind of
pronominal trigger ('ante-cedent' (anaphoric) as well as 'post-cedent'

Prof. Dr. Wolfgang Schulze
Institut fuer Allgemeine und Indogermanische Sprachwissenschaft
Ludwig-Maximilians-Universitaet München
Geschwister-Scholl-Platz 1
D-80539 München
Tel.: +89-21805343 / Fax: +89-21805345
Email: W.Schulze at

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Funknet mailing list