TRANSITIVITY

[Stuart Robinson]

I would like to respond to the comments made by Givon regarding
transitivity, because I think what he says is important.

A. Cause & Effect
        One point Givon makes is that "an association, be it even a  strong
one, need not necessarily mean identity". I think this point is
fundamental, because essentially the same caveat applies to the overarching
claim of Hopper & Thompson  (1980)--viz.,that transitivity derives its
prominence from foregrounding. Setting aside the question of just how
strong the correlation between the two really is, the problem is that
correlation does not  necessarily imply causation. Perhaps both
transitivity and foregrounding  correlate because they are associated with
something else--say, a deep-seated semantic prototype. It's like the
researchers who were studying ulcer development and found that there was a
high correlation between developing ulcers and having a neat, well-trimmed
moustache! Obviously, having a neat, well-trimmed moustache didn't cause
ulcers, but there was some sort of relationship between the two
nonetheless. Alternatively, think of the relationship between thunder and
lightning. Even though the two correlate perfectly, one cannot be said to
cause the other. Rather, they are both effects of  the same underlying
cause--an electromagnetic event.
        I think DeLancey makes essentially this same point in his 1984
paper, "Transitivity in grammar and cognition". He doesn't think that
transitivity derives its prominence from foregrounding. Rather, he thinks
that:

"the various transitivity parameters cohere in the way that they do because
they code aspects of a coherent semantic prototype, and that transitivity
in morphosyntax is associated with foregrounding in discourse because
events which approximate the transitive prototype are more like to be of
interest and thus inherently more likely to constitute foregrounded
information" (p. 55).

B. The Nature of O
        So, if we concentrate on trying to characterise this semantic
prototype, we have to think about Givon's take on it. He claims that one
feature of the prototype for a transitive event is a "passive, affected
CAUSEE/PATIENT". I think that Givon has said elsewhere that  "the
prototypical transitive object-patient is not a human, but primarily a dumb
inanimate" (1990: 630). But how does this claim relate to Hopper &
Thompson's claim that a highly individuated O is a highly transitive
feature? There is a tendency across languages to single out animate,
definite O's for special treatment (e.g., Spanish, Hindi, etc.), which
Hopper & Thomspon see as being some sort of underscoring of the high
transitivity of such clauses, but this doesn't jibe very well with
classical markedness theory, which claims that special marking should be
given to configurations that are unusual or unexpected. I think Croft
(1988) and Aristar (to
appear:http://linguistlist.org/aristar/transitivity.html) both tackle this
question, but they come up with different solutions. Croft basically
rejects the claim that the most natural O is highly individuated while
Aristar takes it on board and adopts a middle-of-the-road position--viz.,
that the most natural O is neither low or high in individuation, but rather
somewhere in between. It seems to me that part of the problem is that
highly individuated O's are likely to be topical and topical O's tend to
get passivised. But this seems to be an area of considerable controversy.

This topic doesn't seem to have generated a great deal of interest on the
list but hopefully the discussion won't peter out entirely. I am quite
interested in hearing the opinion of others on this matter and would like
to take this opportunity to the various people who have responded to my
original posting.

Sincerely,
Stuart Robinson

______________________

Stuart P. Robinson (Stuart.Robinson at anu.edu.au)
Linguistics Department, Australian National University
Canberra ACT 0200, Australia
PHONE: 61-2-6249-0703 |||| FAX: 61-2-6279-8214