Int. J. Human—Computer Studies (1997) 47, 531— 563 Attribute grammars as a robust technical basis for a human– computer interaction general purpose architecture GIOVANNI ADORNI AND AGOSTINO POGGI Department of Computer Engineering, University of Parma, Viale delle Scienze, 43100 Parma, Italy GIACOMO FERRARI Department of Humanist Sciences, University of Turin, Via Galileo Ferraris 109, 13100 Vercelli, Italy (Received 23 May 1995 and accepted in revised form 30 April 1997) In this paper, we present a Natural Language Interface (NLI) that combines some of the advantages of using general purpose grammars with some of the features found in semantic grammars. This is achieved by an Attribute Grammar so as to create an intermediate representation which is interpreted in terms of domain-specific routines. NLI has been used as a component of a Human—Computer Interaction (HCI) system, working on different domains. After a discussion of the theory underlying the system, two examples of the use of NLI on practical domains are described: the first is in the field of three-dimensional scene generation, the second in the field of road map car planning. Some evaluation issues are also included in this paper.  1997 Academic Press Limited 1. Introduction Since the early 1970s Natural Language processing has been confronted with two different tasks: the challenge of setting up more and more sophisticated computational models of human language and communication which led to the building of complex systems that integrate an increasing number of interacting components and the pressure to use natural language processing techniques to perform real work which has made many researchers look for engineered versions of the most firmly grounded tools. The architecture of early human—computer interaction systems such a LUNAR, (Woods, Kaplan & Nash-Webber, 1972) and LIFER (Hendrix, 1977) was very similar to that of a compiler, but they relied on robust and extended natural language grammars, based on quickly stabilized implementation techniques. Later, the attempts at integrating complex reference resolution and focus shifting (see, e.g. Grosz, 1977; Sidner, 1983; Grosz & Sidner, 1986) or inference of user intentions (see, e.g. Allen, 1983; Allen & Litman, 1986) or user modelling into a human—computer interaction system, tended towards an architecture where the trace of the previous interaction and a strong knowledge component were to substitute for the simple compiler structure. In more recent times, even though experiments with dialogue systems have in no way exhausted even a fraction of the theoretical or technical issues in this domain, a new 531 1071-5819/97/100531#33$25.00/0/hc 970146  1997 Academic Press Limited