Database Reference
In-Depth Information
corresponding passive construction 9.1.7. Active traverses the NAG branches
top down and from left to right. Passive traverses them top down and from
right to left.
This alternative is reflected by the different numbering of arcs in the NAGs
for active and passive. The numbering is purely descriptive, and forces the
navigation to conform to a certain order for realizing a given content, repre-
sented by the SRG, in a certain language. There is another possibility, however,
namely to let the navigation find the way through the graph on its own, based
on the principle of least effort , and not on numbers which record the orders
actually observed in a certain natural language example.
In DBS, the principle of least effort is applied (i) to traversing a branch com-
pletely or exiting early, (ii) to follow or not follow the time-linear order inher-
ent in the content, and (iii) to not reversing or reversing the direction of the
navigation. In short, a time-linear direction continuing in the same direction as
long as possible constitutes a lesser effort than a multiple visit, an anti-time-
linear direction, or a voluntary reversal of direction.
An example of exiting a branch early, requiring a multiple visit for a com-
plete traversal, is unbounded dependency (9.4.2). An example of an anti-time-
linear navigation is the traversal of an extraprositional coordination from right
to left (9.2.4). A voluntary change of direction arises in the choice of passive
over active.
Grammatical constructions which follow the principle of least effort are
regarded as the linguistically unmarked 29 case in DBS, while constructions
which run against least effort are regarded as marked. In this sense, realiz-
ing or interpreting an active, for example, requires less effort than realizing or
interpreting the corresponding passive.
The reason that passive and the other marked traversals are used in com-
munication anyway is due to other principles which override least effort. For
example, theme-rheme (topic-comment), WH question formation, etc., induce
word order requirements of their own which are often in marked contrast to
the order corresponding to least effort.
For the linguist, the empirical challenge presented by a natural language ex-
ample is the SRG, the signature, and the proplet representation; for this, the
correct semantic details must be discovered, preferably for several different
languages in parallel. The possible NAGs, in contrast, follow automatically.
For surface realization, the correct NAG must be selected. Also, the traversal
steps which realize a surface must be indicated. They provide the input to LA-
speak. Experience at the CLUE has shown that these tasks are easily acquired.
Search WWH ::




Custom Search