In recent years, the notion of ambiguity has come under close scrutiny. This was mainly caused by developments in automatic natural language processing, where it is becoming more common to use formal representations of meaning that are themselves ambiguous. Such ambiguous representations are called underspecified because they do not contain sufficient information to determine their truth conditions uniquely. In accordance with this new, 'underspecifying' perspective on ambiguity, the papers in this volume deal not only with traditional questions of disambiguation, but also with the theoretical underpinnings of underspecified meaning representations and with the possibilities of using these representations in logical inference.