Introduction

This chapter examines some of the special requirements of a knowledge representation formalism that arise from the planning of linguistic actions. Utterance planning requires the ability to reason about a wide variety of intensional concepts that include knowledge per se, mutual knowledge, belief, and intention. Intensional concepts can be represented in intensional logic by operators that apply to both individuals and sentences. What makes intensional operators different from ordinary extensional ones such as conjunction and disjunction is that one cannot substitute terms that have the same truth-value within the scope of one of these operators without sometimes changing the truth-value of the entire sentence. For example, suppose that John knows Mary's phone number. Suppose that unbeknown to John, Mary lives with Bill — and therefore Bill's phone number is the same as Mary's. It does not follow from these premises that John knows what Bill's phone number is.

The planning of linguistic actions requires reasoning about several different types of intensional operators. In this research we shall be concerned with the operators Know (and occasionally the related operator Believe), Mutually-Know, Knowref (knowing the denotation of a description), Intend (intending to make a proposition true) and Intend-To-Do (intending to perform a particular action).

Introduction

This chapter discusses the problems of planning surface linguistic actions, including surface speech acts, concept activation actions, and focusing actions. What distinguishes these surface linguistic acts from the illocutionary acts considered in Chapter 5 is that they correspond directly to parts of the utterance that are produced by the planning agent. An agent intends to convey a proposition by performing an illocutionary act. There may be many choices available to him for the purpose of conveying the proposition with the intended illocutionary force. For example, he may make a direct request by using an imperative, or perform the act of requesting indirectly by asking a question. He usually has many options available to him for referring to objects in the world.

A surface linguistic act, on the other hand, represents a particular linguistic realization of the intended illocutionary act. Planning a surface speech act entails making choices about the many options that are left open by a high-level specification of an illocutionary act. In addition, the surface speech act must satisfy a multitude of constraints imposed by the grammar of the language. The domain of reasoning done by the planner includes actions along with their preconditions and effects. The grammatical constraints lie outside this domain of actions and goals (excluding, of course, the implicit goal of producing coherent English), and are therefore most suitably specified within a different system.

Introduction

The planning of natural-language utterances builds on contributions from a number of disciplines. The construction of the multiagent planning system is relevant to artificial intelligence research on planning and knowledge representation. The axiomatization of illocutionary acts discussed in Chapter 5 relies on results in speech act theory and the philosophy of language. Constructing a grammar of English builds on the study of syntax in linguistics and of semantics in both linguistics and philosophy. A complete survey of the relevant literature would go far beyond the scope of this book. This chapter is included to give the reader an overview of some of the most important research that is pertinent to utterance planning.

Language generation

It was quite a long time before the problem of language generation began to receive the attention it deserves. Beginning in about 1982, there has been a virtual explosion in the quantity of research being done in this field, and a complete review of all of it could well fill a book (see Bole and McDonald, forthcoming). This chapter presents an overview of some of the earlier work that provides a foundation for the research that follows.

Several early language-generation systems, (e.g. Friedman, 1969), were designed more for the purpose of testing grammars than for communication.

This book is based on research I did in the Stanford University Computer Science Department for the degree of Doctor of Philosophy. I express my sincere gratitude to my dissertation reading committee: Terry Winograd, Gary Hendrix, Doug Lenat and Nils Nilsson. Their discussion and comments contributed greatly to the research reported here. Barbara Grosz's thoughtful comments on my thesis contributed significantly to the quality of the research. I also thank Phil Cohen and Bonnie Webber for providing detailed comments on the first draft of this book and for providing many useful suggestions, and Aravind Joshi for his efforts in editing the Cambridge University Press Studies in Natural Language Processing series.

This research was supported by the Office of Naval Research under contract N00014-80-C-0296, and by the National Science Foundation under grant MCS-8115105. The preparation of this book was in part made possible by a gift from the System Development Foundation to SRI International as part of a coordinated research effort with the Center for the Study of Language and Information at Stanford University.

This book would be totally unreadable were it not for the efforts of SRI International Senior Technical Editor Savel Kliachko, who transformed my muddled ramblings into golden prose.

Toward a theory of language generation and communication

A primary goal of natural-language generation research in artificial intelligence is to design a system that is capable of producing utterances with the same fluency as that of a human speaker. One could imagine a “Turing Test” of sorts in which a person was presented with a dialogue between a human and a computer and, on the basis of the naturalness of its use of the English language, asked to identify which participant was the computer. Unfortunately, no natural-language generation system yet developed can pass the test for an extended dialogue.

A language-generation system capable of passing this test would obviously have a great deal of syntactic competence. It would be capable of using correctly and appropriately such syntactic devices as conjunction and ellipsis; it would be competent at fitting its utterances into a discourse, using pronominal references where appropriate, choosing syntactic structures consistent with the changing focus, and giving an overall feeling of coherence to the discourse. The system would have a large knowledge base of basic concepts and commonsense knowledge so that it could converse about any situation that arose naturally in its domain.

However, even if a language-generation system met all the above criteria, it might still not be able to pass our “Turing Test” because to know only about the syntactic and semantic rules of the language is not enough.

Introduction

This chapter deals with the design and implementation of a planning system called KAMP (an acronym for Knowledge And Modalities Planner) that is capable of planning to influence another agent's knowledge and intentions. The motivation for the development of such a planning system is the production of natural-language utterances. However, a planner with such capabilities is useful in any domain in which information-gathering actions play an important role, even though the domain does not necessarily involve planning speech acts or coordinating actions among multiple agents.

One could imagine, for example, a police crime laboratory to which officers bring for analysis substances found at the scene of a crime. The system's goal is to identify the unknown substance. The planner would know of certain laboratory operations that agents would be capable of performing — in effect actions that would produce knowledge about what the substance is or is not. A plan would consist of a sequence of such information-gathering actions, and the result of executing the entire plan would be that the agent performing the actions knows the identity of the mystery substance. Since the primary motivation for KAMP is a linguistic one, most of the examples will be taken from utterance planning; the reader should note, however, that the mechanisms proposed are general and appear to have interesting applications in other areas as well.

With Radulphus Brito and his contemporaries, modistic grammar had reached its apogee, and shortly afterward, metatheoretical objections brought its development to a halt. The objections came from Latin Averroists, such as Johannes Aurifaber (a follower of John of Jandun), and from nominalists such as Ockham. The Modistae, as moderate realists, had assumed that reality had a definite structure that was mirrored in cognition and in language; the objectors challenged this assumption, arguing that words were purely arbitrary representations of thoughts, and hence that the modistic theory of the cognitive and ontological basis of language was untenable.

The fall of modism

The debate over objections to modism seems to have broken out first, not at Paris, but at Erfurt, which by the early 1300s was the site of an intellectually active group of schools, even if not yet a university. Our first record of it comes from a public discussion presided over by Johannes Aurifaber in October of 1333 (or possibly 1332); an anonymous but detailed report survives and has been edited by Pinborg (1967:215-32). The nominal topic of the discussion is whether nominativo hie magister (‘in the nominative hie magister’)2 is a sentence. Aurifaber quickly gives the standard answer that it is if one supplies a verb – nominativo ‘hie magister’ ponitur or the like – and moves on ‘omissis aliis’ to the rather different issue of whether there are such things as modes of signifying.

The authorship of most medieval grammatical treatises is uncertain; quite often nothing is known about the author of a treatise except what can be inferred from internal evidence. This appendix comprises a set of notes giving information about particularly obscure authors, explaining how I have dealt with treatises that have been attributed to more than one author in the modern literature, and, where applicable, presenting additional evidence bearing on questions of authorship. In general, I have referred to authors by the names by which they are known to modern scholars, even when such names represent uncertain inferences. When a treatise has long been attributed to an author but the weight of the evidence is now against that attribution, I have prefixed Tseudo-’ to the traditional author's name. In a couple of cases I have used names assigned by Pinborg, consisting of ‘Anonymus’ followed by the present location of the manuscript (which is not necessarily where the medieval author was active).

Anonymus Norimbergensis

This is Pinborg's name for the otherwise unknown author or authors of two series of Quaestiones super Priscianum minor em (treating books XVII and XVIII respectively) that are found in Nuremberg Stadtbibliothek manuscript Cent.V.21, fol. 31r-34v and 35r-53r. The two series are written in the same hand and represent the same doctrinal milieu (that of Radulphus Brito, Thomas of Erfurt, and the other late Parisian Modistae); it is not known whether they are in fact the work of a single author.

The significance of the treatises was pointed out by Pinborg (1980b: 204–6), who presented a transcription and translation of one interesting passage.

The speculative grammarians of the High Middle Ages are remembered today for two major achievements; they formulated the theory of modi significandi, which comprises an elaborate attempt to explain linguistic structure in terms of the structure of cognition and of reality, and they developed an elaborate theory of syncax from which some concepts, such as government and dependency, have survived to the present day. The former has been studied extensively in modern times, while the latter has been neglected.

This book deals with the origins and development of the theories of syntactic structure used by a group of grammarians and logicians who flourished at Paris between about 1270 and 1310 and who were later called ‘Modistae’ because of their emphasis on modi significandi. I am focusing on roughly the period from Martin of Dacia (c. 1270), who was, as far as we know, the first to construct a fully modistic theory of syntax, to Radulphus Brito (c. 1300), the last major contributor to modistic theory before the rise of nominalism diverted grammarians’ attention to methodological matters.

The primary goal of my study is exegesis. I am concerned more with explicating the conceptual content of the medieval theory than with presenting its complete history. In so doing, I am consciously writing for two audiences – linguists who may know little about medieval philosophy, and medievalists who may know little about linguistics – and I hope readers in each group will forgive me for having done certain things purely for the benefit of the other.

The true test of any linguistic theory is, of course, its actual use in the analysis of language. As the Modistae applied their model of sentence structure to the various sentence-types of Latin, they found, naturally enough, that certain structures posed problems, and they argued for or against modifications in the theoretical framework in much the same manner as their present-day counterparts.

Among the important issues in modistic syntax were constituency versus dependency (an issue raised in the analysis of coordinate structures and embedded sentences), grammatical relations in impersonal clauses, and the relation of anaphoric pronouns to their antecedents. In this chapter I shall present samples of the modistic discussion of these issues, drawn mainly from the work of Radulphus Brito. I make no attempt to trace the complete history of each question, but only to give a sampling of the arguments.

Conjunctions and constituent structure

One of the basic presuppositions of modistic syntax is that all grammatical relations link individual words, not groups of words. That is, the modistic model is a dependency grammar, not a constituency grammar – and, like all dependency grammars, it has difficulty dealing with coordinating conjunctions.

Consider, for example, the sentence Socrates et Plato currunt, ‘Socrates and Plato are running’: what is its subject?

Modistic syntactic theory represents in large part the continued development of earlier ideas. Its background begins with the late Roman grammarians who wrote down descriptions of the classical Latin language for posterity – writers such as Donatus, Charisius, Diomedes, and Servius in the fourth century, Phocas and Pompeius in the fifth, and Priscian in the early sixth. The grammarians of the Carolingian Renaissance, such as Alcuin of York (fl. 781-96), Sedulius Scottus (fl. 848-58), and Remigius of Auxerre (c. 900), made use of the writings of many of their Roman predecessors and kept essentially the same descriptive framework.

By the eleventh or twelfth century, however, the range of Roman grammarians whose works were used in the schools of northern Europe had narrowed to two: Donatus and Priscian. The Ars grammatica of Donatus – the first part of which, the Ars minor, is phrased in a catechism-like question-and-answer format – says nothing significant about syntax. The only substantial Roman source for medieval syntactic theory is therefore Priscian, whose eighteen-book Institutiones grammaticae constitute the most voluminous, most thorough, and most disorganized of the surviving Roman grammars. The first sixteen books, referred to in the Middle Ages as Priscianus maior, discuss the individual parts of speech, and the last two (Priscianus minor) are explicitly devoted to syntax, though syntactic information is scattered through the other books as well.

This book is a revised version of a Ph.D. dissertation written at Yale under the direction of Rulon Wells, to whom I shall always be grateful for showing me the value, and beginning to teach me the art, of asking interesting questions about anything and everything.

It must also serve as a memorial to the late Professor Jan Pinborg, Director of the Institute of Medieval Greek and Latin Philology at the University of Copenhagen, who took a great interest in it and provided me with much help while it was being written, but unfortunately did not live to see it published. His last communication to me, just a few weeks before he died, was a detailed set of comments on my dissertation; they were invaluable in preparing the present version.

This work would not have been possible without the aid of National Science Foundation Grant Number BNS-81-05359, which enabled me to travel to Europe, examine manuscripts, confer with scholars, and assemble a microfilm collection. Many people offered assistance and advice along the way; some who deserve special thanks are R. H. Robins (University of London), Bruce Barker-Benfleld (Oxford), W. Keith Percival (Kansas), Irene Rosier (CNRS, France), C. H. Kneepkens (Nijmegen), K. M. Fredborg (Copenhagen); the two additional readers of the original dissertation at Yale, Edward Stankiewicz and C. J. Herington; and P. H. Matthews, who served as reader for the Cambridge University Press. All conclusions and opinions expressed here are of course my own and do not necessarily reflect the views of any of the people acknowledged here, nor of the National Science Foundation or any other governmental agency.

The Modistae were a group of grammarians and logicians,1 active principally at Paris during the second half of the thirteenth century, who held that in addition to having a pronunciation and a meaning, every word has a set of properties called modi significant (modes of signifying) that determine how the particular meaning is encoded into the language. For example, the verb currere, ‘to run’, and the noun cursus, ‘a run’, signify the same thing – the act of running; the difference between them is that they signify it in different ways (under different modes). Currere signifies running conceived of as an act or change in progress {per modum fluxus et'fieri); cursus signifies it as an enduring entity {per modum habitus et permanentiae). Modes of signifying account not only for the differences between the various parts of speech, but also for all the other properties of a word that reside neither in its meaning (in the strict sense) nor in its pronunciation – including such things as gender, number, case, and form (simple or complex).

The Modistae

Only a small part of the vast literature of medieval grammar has been printed or even read through in modern times. Nonetheless, by studying many unedited manuscripts as well as all available printed editions, Pinborg (1967) has been able to trace the main outlines of the development of modistic grammar, and subsequent research has contributed many details to the picture. In this section I shall summarize and update Pinborg's results, with emphasis on the medieval grammarians whose works constitute the data for the present study.