Transitivity

[Wolfgang Schulze]

At 20:27 05.01.99 +1000, Stuart Robinson wrote:

>I know the literature on transitivity is enormous, but I was wondering
>whether people on the list could recommend a couple of good works on the
>issue of where the prototypical transitive scenario originates--in
>particular, the issue of whether it is innate or learned. I know that some
>authors (e.g., Hopper & Thompson) feel that transitivity derives its
>prominence from its association with other things (such as grounding) while
>others think that it is somehow basic (perhaps innate?). I would assume
>that if you thought transitivity was derived, you would have to think that
>it is learned, but that if you thought it was basic, you would lean towards
>nativism of some sort. But I would be interested in hearing what other
>people think. Thanks in advance.

I think most people from outside the MIT-Orthodoxy would agree upon the
claim that linguistic transitivity is somehow derived from other properties
of human cognition/behavior. In order to specify the 'source(s)' of
transitivity and to describe the derivational procedures or emergent
activities that result in the linguistic phenomenon of transitivity it
seems to be useful first to explore the possible links between anything
'transitive-like' outside the language architecture and some elements of
the architecture itself. Traditionally, transitivity is related either to
the lexicon (valency patterns etc.) or to (morpho)syntax (sentence patterns
etc.). If you claim that transitivity is represented in the lexicon than
the source of the phenomenon should be seeked for within the domain of
conceptualisation. In case you prefer a syntactic reading of transitivity,
then a potential candidate for its source would be something like the
architecture of information processing and communication.

However, these two types of approaching the underlying conditions to not
exclude each other. I strongly believe that the mental lexicon represents
nothing but a 'generification' of external stimuli that are mentally
construed as events. Basically, I assume that human cognition accommodates
such single external stimuli in terms of  'permanent objects' (Piaget):
Such constructed 'permanent events' have two basic aspects: First, they get
a relational reading that allows to construe referential 'names' for the
paradigm of now permanent object that are 'normally' experienced as being
part of a given 'permanent event'. Hence, people construe (in the process
of language acquisition) a typology of generic 'objects' that are embedded
in a standard event. Quite parallel, the relation itself is "freed" from
its standard objects and acquires a 'verbal' reading. In my opinion all
verbs are 'generic' in this sense, too. The conceptualisation of these
emergent structures (as (referential) objects and verbs) maintain much of
their basic conditional (event based) nature. However, (more or less
generic) constructions of elements involved in an event are associated with
a much high degree of stability in time (and [partly] in space), whereas
the relational structure between them (in a given or generic event) is much
more unstable: Remember that any relational structure with in an 'event'
can only be experienced by checking the complex degree of change that a
given 'object' undergoes ['states' have to be judged a little bit different].

Second, the construction of 'permanent events' (and - subsequently - of the
elements involved in such events) seems to be based of the cognitive
procedure of splitting up an event cluster into a sequence of 'elements',
if any kind of communication is intended with respect to the event. This
procedure naturally is related to the fact that communication based on
language has to respect the prerogatives established by the communicative
technique itself: Language allows the clustering of event experience to a
much lesser degree than non-verbal cognitive activities [though clustering
itself still is a very important and universal option in language]. The
serialization (or unpacking) of 'permanent events' (together with their
instantiation in a given context) takes place according to a set of basic
assumptions (or 'cognitive hypotheses') about how events are structured.
One of the most prominent hypothesis sure is that 'omne quod movetur ab
alio movetur'. This hypothesis allows to separate 'states' from
'non-states' (though the fixing of what is a state heavily depends on the
acquired world knowledge). The above mentioned Peripathetic principle ends
up in a second hypothesis metaphorized from the first one, namely that
there is no 'reason without effect' or no 'effect without reason' (the
famous C(<'motor')->E(<'motum') vector). I don't think that the C->E vector
[you can call it the transitivity vector, if you want] itself is innate;
rather it seems that it is learnt in the process of the assimilation and
accommodation of event experience. More precisely: The experience of
'motion' as a basic factor of events is metaphorized to the extent that the
habitus and general knowledge system of a given speech community has
sanctioned it.

The C->E vector naturally is not the only (metaphorical) basis for
transitivity. The structural coupling of the C->E vector with the
communicative information flow (itself emergent from the serialization of
clustered events) allows to focus on any component of the vector: For
instance, the "C" domain can be regarded as more central than the "E"
domain (in my terms C->e) or vice versa (c->E). This may effect the
internal structure of an event representation as well as its embedding in a
given co-text (pivot etc.). Or, potential representatives (or 'actants') of
one of the domains (or even of both) are graded in a prototypical sense
('heavy actants' are the most prototypical, and 'light actants' are the
least prototypical (and marked) representatives). Or, the C->E vector
allows to canonically mask one of the domains (C->zero;zero->E). These and
many other procedures are emergent aspects of the activity of the
cognitive-communicative interface [we should bear in mind that
communication is a cognitive parameter too, though strongly 'autonomized']
and lead to the particularization of the universals of how human beings
construe event: These particularizations are based on the coupling of world
and communicative knowledge and a given language system. Moreover, language
is a system to pass tradition (or a given habitus) together with its
actualization, but it is also a traditional system itself (language is
passed by language). The resulting diachronic features of a language system
are very often anachronistic with respect to how people get used to
communicate their way of construing events; hence, the linguistic templates
of how to encode the C->E vector together with its particularizations
acquire a pseudo-autonomous character that has its own history and that is
processed according to the prerogatives of the paradigmatic procedures of
human cognition.

To sum up: In my mind we cannot claim that 'transitivity' as a linguistic
'category' (or so) is either learnt or native. Transitivity is an emergent
property of language that results from the complex interaction of
cognition, communication, and habitus. The fact that all languages seem to
have something like transitive 'properties' does not hint at the
universality of this 'category' but rather at the universality of the
interaction of the above mentioned three constituents.

For those who want to learn a little more about the claims I have made
here: I have elaborated the claims made here in the first volume of my series

"Person, Klasse, Kongruenz: Fragmente einer Kategorialtypologie des
einfachen Satzes in den ostkauaksischen Sprachen. Vol. 1 (in two parts):
Die Grundlagen". München/Newcastle: Lincom 1998 [this volume unfortunately
is in German (and has many misprints, sorry for that!)].

This volume concentrates on what I call a "Grammar of Scenes and Scenarios"
(GSS) both with respect to its theoretical foundations and its
(typological) architecture [East Caucasian is chosen as a field of
evaluation (in the forthcoming volumes) and does not play a prominent role
in this first volume].

Wolfgang

Prof. Dr. Wolfgang Schulze
Institut für Allgemeine und Indogermanische Sprachwissenschaft
Universität München
Geschwister-Scholl-Platz 1
D-80539 München
Tel.: +89-21802486 (secr.)
      +89-21802485 (office)
http://www.lrz-muenchen.de/~wschulze/