Empirical Issues in Syntax and Semantics 9 (EISS 9 ... - CSSP - CNRS
Empirical Issues in Syntax and Semantics 9 (EISS 9 ... - CSSP - CNRS
Empirical Issues in Syntax and Semantics 9 (EISS 9 ... - CSSP - CNRS
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
sends<br />
⎡<br />
⎤<br />
causation<br />
EFFECTOR 8<br />
THEME 9<br />
GOAL 10<br />
[ ]<br />
activity<br />
CAUSE<br />
EFFECTOR 8<br />
⎡ ⎤<br />
change-of-loc<br />
⎢ ⎢ ⎥<br />
⎣EFFECT<br />
⎣THEME 9 ⎥<br />
⎦⎦<br />
DESTINATION 10<br />
DO construction<br />
⎡<br />
causation<br />
EFFECTOR 1<br />
THEME 2<br />
GOAL 3<br />
[ ]<br />
activity<br />
CAUSE<br />
EFFECTOR 1<br />
⎡<br />
change-of-poss<br />
⎢ ⎢<br />
⎣EFFECT<br />
⎣<br />
⎤<br />
⎤<br />
⎥<br />
THEME 2 ⎥<br />
⎦⎦<br />
RECIPIENT 3<br />
Figure 14: Lexical frame <strong>and</strong> construction frame of sends <strong>and</strong> the DO construction<br />
NP<br />
[I= 8]<br />
VP<br />
S<br />
V [E= 0] NP [I= 3] NP<br />
[I= 2]<br />
sends<br />
⎡<br />
⎤<br />
causation<br />
EFFECTOR 8<br />
THEME 2<br />
GOAL 3<br />
[ ]<br />
0<br />
activity<br />
CAUSE<br />
EFFECTOR 8<br />
⎧<br />
⎡ ⎤ ⎡ ⎤⎫<br />
⎪⎨ change-of-loc change-of-poss ⎪⎬<br />
⎢<br />
⎢ ⎥ ⎢ ⎥<br />
EFFECT<br />
⎣<br />
⎣THEME 2⎦,<br />
⎣THEME 2 ⎦ ⎥<br />
⎦<br />
⎪⎩ DESTINATION 3 RECIPIENT 3 ⎪⎭<br />
Figure 15: Anchored tree for sends with the DO construction<br />
DO construction anchored with sends) is, roughly, a causation with effects along different dimensions<br />
or ‘scales’: there is a change of location of the theme <strong>and</strong> at the same time the theme<br />
undergoes also a change of possession.<br />
There are different ways to avoid the mismatch between the two frames. The possibility we<br />
choose <strong>in</strong> this paper is to use set-valued attributes <strong>and</strong> to assume a special set unification for<br />
these. In our case, the attribute EFFECT would have a set of changes as value. When unify<strong>in</strong>g<br />
two such sets, the follow<strong>in</strong>g strategy is adopted: for two elements belong<strong>in</strong>g to the respective<br />
sets, if they are of the same type or one is of a subtype of the other, they must unify <strong>and</strong> the result<br />
is part of the result<strong>in</strong>g set. Otherwise, we take the two elements to describe different aspects that<br />
should be considered as a conjunction. We therefore add each of them to the result<strong>in</strong>g set of<br />
frames. In our example, this would lead to the anchored tree <strong>in</strong> Fig. 15. Note that, <strong>in</strong> order to<br />
obta<strong>in</strong> the <strong>in</strong>tended identifications between participants of events, we need the top level roles<br />
here. They make sure the dest<strong>in</strong>ation of the change of location is identified with the recipient of<br />
the change of possession s<strong>in</strong>ce both are co-<strong>in</strong>dexed to the GOAL roles of their frames. 12<br />
6. Conclusion<br />
LTAG is a lexicalized tree grammar formalism with an extended doma<strong>in</strong> of locality <strong>and</strong> rich<br />
possibilities for factoriz<strong>in</strong>g syntactic <strong>and</strong> semantic <strong>in</strong>formation on a metagrammatical level. In<br />
12 An alternative approach, which does not require set-valued attributes, would be to treat the different changes as<br />
two different perspectives on the effect of the causation event, represented by two different attributes of the frame.<br />
But the details <strong>and</strong> the consequences of this solution have to be left to future research.<br />
182