Generating Referring Expressions in a Multimodal Context: An empirically oriented approach
Ielka van der Sluis, Emiel Krahmer · 2001
Figure heads are words without autonomous content, such as the copula and the infinitival to. They pose a challenge for the head-driven semantics of HPSG-94, since the latter requires the CONTENTINUCLEUS value of a head-complement combination to be identical to the one of its head daughter. As a consequence, if the head is semantically vacuous, the combination is —erroneously— predicted to be vacuous as well. In order to repair this, HPSG-94 stipulates that the CONTENT value of a vacuous word is identified with the one of its complement. Technically, this gives the right result, but it leaves a number of issues unresolved. First, if the copula is vacuous, then what happens to the semantic contribution of its tense? Second, what are the criteria for identifying vacuous words? Third, are there any alternatives for the figure head treatment, and if so, what is the kind of evidence that can be used to motivate a choice? In dealing with these questions, I will integrate a DRT style treatment of the tenses in the HPSG framework (section 3), I will define a number of criteria for identifying vacuous verbs (section 4) and I will provide some empirical evidence against the figure head treatment, replacing it with an alternative analysis in which vacuous verbs have no CONTENT feature at all (section 5).