15.1.4 Ambiguities versus Disjunction

Much of natural language processing can be understood as solving combinatoric problems. The reason is that natural language is full of ambiguities which one may express by disjunctive logic formulas.

As an example, we consider agreement information for the the German adjective schön (nice). Its inflection schoenen can have be used with a set of possible gender-case-number information. The possibilities can be described by the following propositional formula:

$\begin{array}{lc@{\wedge}c@{\wedge}c@{\qquad}l} & masc & dat & sg & \mbox{dem schönen Mann} \\ \vee& masc & acc & sg & \mbox{den schönen Mann}\\ \vee& masc & nom & pl & \mbox{die schönen Männer} \\ \vee& masc & gen & pl & \mbox{der schönen Männer} \\ \vee& masc & dat & pl & \mbox{den schönen Männern} \\ \vee& masc & acc & pl & \mbox{die schönen Männer} \\ \vee& fem & gen & sg & \mbox{der schönen Frau} \\ \vee& fem & dat & sg & \mbox{der schönen Frau} \\ \vee& fem & nom & pl & \mbox{die schönen Frauen} \\ \vee& fem & gen & pl & \mbox{der schönen Frauen} \\ \vee& fem & dat & pl & \mbox{den schönen Frauen} \\ \vee& fem & acc & pl & \mbox{die schönen Frauen} \end{array}$

The next question is how to express disjunctive information in a concurrent constraint language such as Oz.