Wed Jan 13 19:39:27 UTC 1999


I was waiting for someone to make some of the points that I thought
could (& should) have been made in the discussion of transitivity.
But given that the discussion seems to have petered out, maybe this is
a good point to try & make them; under two headings.


One of the great contributions of the Hopper & Thompsopn (1980) article
was to gather under one roof all the phenomena associated with transitivity,
and underscore the strong--and multiple--typologicaland textual associations
they display . But focusing on strong associations tends, on occasion, to
distract our attention from recognizing a simple caveat--that an association,
be it even a strong one, need not necessarily mean identity.


The semantic core of transitivity resides in the framing of the prototype
transitive event (a) acting, volitional CAUSER/AGENT
                 (b) passive, affected CAUSEE/PATIENT
                 (c) fast changing, telic EVENT/VERB
The cognitive framing of such events in humans harkens back to the pre-
linguistic sensory-motor developmental stage in the first 9 months of life,
where the sensory-motor foundations of our experiential universe--external,
internal & social--are laid out. At this developmental stage, human
discourse is largely mono-propositional (single states/events) rather
than multi-propositional (coherence over multiple states/events). This is
also a stage that, in principle, is not all that distinct from the perceptual/
cognitive organization of pre-human species.

I think we can take it for granted that there is a strong ADAPTIVE motivation
for why higher vertebrates concentrate so many resources on the processing of
prototype transitive events. All three key features of such events reside
at the very heart of the daily survival of a moving, self-feeding, social,
interacting, prey-or-predator species. The fact that the mammalian perceptual
system is so biased toward changes/telicity, self-propelled entities/animates,
affected objects/foods/prey etc. is exhaustively well established. This is
true in all major perceptual modalities, and simply gets extended into the
higher cognitive processes that are but evolutionary extensions, elaborations
and extra complications of the early perceptual modalities.


The cognitive "framing" of semantically-transitive events (a. above) is
fundamentally of the type Ron Langacker has been talking about in his
extensive work. Now, while the communicative-pragmatics of transitivity
interacts--extensively--with the cognitive-semantic framing, the two not
identical. To begin with, it is not the NATURE of the event itself that
needs to be at issue here, but often rather the WIDER framing of the
event within the more extended communicative goals of the speaker/hearer.
So that the very same transitive event may be framed with different
ATTENTIONAL FOCUSING. Thus, in the active-direct framing of the event:
          (1) Mary demolished the house
the topical/attention focus is stronger on 'Mary' than on the house,
a fact that has been demonstrated by both discourse-heuristic measurements
and, more recently, by attentional-cognitive measurements (Russ Tomlin's work,
among others). In the "classical" promotional passive, on the other hand,
as in:
          (2) The house was demolished by Mary
the topical/attentional focus is on the patient, again a fact that has been
sufficiently demonstrated by both text-based & cognitive measures.

Now, there is absolutely no evidence that the three semantic core-features
of transitivity need have changed even one iota between the active-direct (1)
and the passive (2). So there is, in principle, a healthy DISSOCIATION
between the semantics & pragmatics of transitivity. Having noted that, one
could of course see that there **could** be a CARRYOVER--thus a strong
association--between the pragmatics & semantics of de-transitivization.
Thus, for example, in (3) below, semantic AGENCY is presumsably deficient:
          (3) The house was demolished in the earthquake
And indeed, there are some syntactic constructions--middle voice, in
particular--in which the pragmatics & semantics of de-transitivization are
almost obligatorily associated, as in:
          (4)  (Mary broke the glass)
               The glass broke
               The glass is broken
               The glass is breakable
               The glass breaks real easy

So, the semantics & pragmatics of TRANSITIVE EVENT or DE-TRANSITIVE EVENT
can often coincide. But the fact that they CAN also be fully dissociated
from e.o. suggests that, at least in principle, one ought to consider them
as two distinct 'cycles' of cognitive framing, each one with its own
specific properties, & each driven by its own distinct functional-adaptive

One could go on and make the very same argumnent for the anti-passive,
citing both AP constructions that allow a healthy dissociation between the
pragmatics of de-transitivity (DE-TOPICALIZING THE PATIENT) and its event
semantics, but also AP constructions that exhibit strong association
between the two.

One of the most strinking facets of STRONG ASSOCIATION between the semantics
& pragmatics of transitivity is observed in the **text distribution** of
transtive & intransitive constructions, where the ACTIVE-DIRECT is
overwhelming in text-frequency, and both PASS & AP relatively rare. But
again, strong frequency association in behavior (SUBJ = AGT;  OBJ = PAT)
should in no way be interpreted as **identity** (Ed Keenan made this
unfortunate slip when he listed AGT as one of the properties of SUBJ in
his 1976 paper).

In passing, one may as well note that  the distinctness of event-semantics
from communicative pragmatics was recognized IMPLICITLY in Chomsky's 1965
(Aspects) model, where 'deep structure' was seen as fully isomorphic with
propositional (EVENT) semantics. While the un-mentioned discourse pragmatics
was provided for, albeit only **tacitly**, by various "markers" that
triggered 'transformations'. In this version of 'the model', such 'markers'
were built into the DS tree and treated, with the normal disregard for
functional correlates, as purely syntactic.


There is a wealth of evidence from the study of vertebrate & mammalian &
primate & human perception suggesting that the bias towards ACTING AGENTS,
FAST CHANGES & SALIENT OBJECTS is genetically wired-in. It remains an open
question whether the cognitive mechanisms responsible for the strong
PRAGMATIC-COMMUNICATIVE bias we display, in our everyday communication,
towards talking more about agents (i.e. investing more time & attention in
them) is the very same as, or is distinct from, the neurologically/
evolutionarily much older mechanisms of event perception/cognition. There
is very clearly a strong assoiciation between the two. But there is just
as clearly a strong disssociation. Thus, the bias toward action/change/motion
is much more automatic/unattended/subconscious. We simply can't help it
(just observe your cat, dog, horse or child...).

On the other hand, communicative/textual evidence suggests, at least
tentatively, that some more sophisticated CONTEXT-SCANNING CHOICES may
be involved in the discourse-pragmatics of transitivity & de-transitivi-
zation. So, while we know relatively little so far about the neurology
associated with grammatical constructions, even Russ Tomlin's cuing
experiments suggest that attention can be manipulated by changing contexts.
I.e. that we pay attention to the communicative context. The process itself
is largely automatic, but it is not necessarily the same process that impells
us to pay attention to transitive events (over stasis)--even in the total
absence of communicative intent.

Finally, given both the order-of-magnitude jump in complexity between
event-semantics & discourse-pragmatics, their clear inclusion relation
(discourse subsumes event-clauses, but not vice-versa), and lastly,
their rather distinct evolutionary history (all vertebrates engage in event
perception; but only the most complex social communicating species indulge
in multi-propositional coherence...), one strongly suspects that the
relevant processing mechanisms--while strongly connected--are not identical.

Cheers,   T. Givon

More information about the Funknet mailing list