Context modeling and the generation of spoken discourse (original) (raw)
Automatic referent resolution of deictic and anaphoric expressions
Computational Linguistics, 1995
Deictic and anaphoric expressions frequently cause problems for natural language analysis. In this paper we present a single model that accounts for referent resolution of deictic and anaphoric expressions in a research prototype of a multimodal user interface called EDWARD. The linguistic expressions are keyed in by a user and are possibly accompanied by pointing gestures. The proposed model for reference resolution elaborates on notions of context factors and salience and integrates both linguistic and perceptual context effects. The model is contrasted with two alternative referent resolution models, namely, a simplistic one and the more sophisticated model proposed by . Based on empirical and analytical grounds, we conclude that the model we propose is preferable from a computational and engineering point of view.
Oxford Handbooks Online, 2005
Communication via a natural language requires two fundamental skills, producing 'text' (written or spoken) and understanding it. Th is chapter introduces newcomers to the fi eld of computational approaches to the former-natural language generation (henceforth NLG)-showing some of the theoretical and practical problems that linguists, computer scientists, and psychologists have encountered when trying to explain how language works in machines or in our minds. ¹ Authors in alphabetical order . G I: W N L G (C, L, S D)
Natural language production as a process of decision-making under constraints
1980
A theory of the llnguistic conponent of the language production process has been developed, and has been used as the basis of a computer program ("MUNIBLE") that produces texts for five experimental "micro-speakers". The theory is based on a view of production as a goal-directed, decision-making process. The speakers goals ("messages") are interpreted as a program of decisions to be made, ordered according to their relative importance and dependencies. An explicit representation of the results of earlier decisions is used to constrain later ones. Goals may be described to the linguistic component using any representation the designer chooses provided they can be enumerated as a hierarchy of relations in a fixed vocabulary. A dictionary and set of interface ftnctions must be written to interpret the vocabulary for the linguistic component. Pending decisions and the results of earlier decisions are represented as a surface-level, syntactic constituent siructure. Texts are produced by realizing each goal-relation in turn according to the decisionprocedure given by its dictionary entry. Decisions specify a sequence of linguistic actions that may be described as the instantiation and traversal of a phrasal schema selected from the grammar. The relation's arguments are mapped into the leaves of the phrase which is then traversed interpreted top-down and left-to-right by a uniform, data-directed controller that produces the text by executing the actions specified by the phrase. All decisions are local to the linguistic context at the position of the controller. Decisions must be indelible, i.e. not retractable-the controller does not backup. Because of these properties, the process is subject to structurally-induced linguistic constrains that govern when a decision can be made, the kinds of information it may appeal to, and what it may effect. Given certain assumptions about how the speaker structures its input goals, the process can be guarenteed to run at a bounded rate and in linear time with the size of the input. The use of an explicit linguistic structure as the representation of given and planned decisions, makes it possible to write general purpose rules and heuristics that apply directly in terms of the representational structure and need know nothing about its semantic content. An elaborate grammar has been written including rules for word order and morphology, agreement, subordination, thematic transformations, ellipsis, wh-movement, and subsequent reference.
The Realis Model of Human Interpreters and Its Application in Computational Linguistics
2010
As we strive for sophisticated machine translation and reliable information extraction, we have launched a subproject pertaining to modelling human interpreters. The model is based on ReALIS, a new “postMontagovian” discourse-semantic theory concerning the formal interpretation of sentences constituting coherent discourses, with a lifelong model of lexical, interpersonal and cultural / encyclopedic knowledge of interpreters in its center including their reciprocal knowledge on each other. After the introduction of ReALIS, we provide linguistic data in order to show that intelligent language processing requires a realistic model of human interpreters. Then we put down some principles of the implementation (in progress) and demonstrate how to apply our model in computational linguistics. 1 REALIS: THE THEORY IN THE BACKGROUND ReALIS, REciprocal And Lifelong Interpretation System, is a new “post-Montagovian” (Kamp et al. 2005) theory concerning the formal interpretation of sentences co...
Semantics in Generative Grammar
Language, 1999
1 1.2 Frege on compositionality 2 1.3 Tutorial on sets and functions 3 1.3.1 Sets 4 1.3.2 Questions and answers about the abstraction notation for sets 5 1.3.3 Functions 10 2 Executing the Fregean Program 13 2.1 First example of a Fregean interpretation 13 2.1.1 Applying the semantics to an example 16 2.1.2 Deriving truth-conditions in an extensional semantics 2.1.3 Object language and metalanguage 2.2 Sets and their characteristic functions 2.3 Adding transitive verbs: semantic types and denotation domains 2.4 Schonfinkelization 2.5 Defining functions in the i\.-notation 3 Semantics and Syntax 3.1 Type-driven interpretation 3.2 The structure of the input to semantic interpretation 3.3 Well-formedness and interpretability 3.4 The a-Criterion 3.5 Argument structure and linking 4 More of English: Nonverbal Predicates, Modifiers, Definite Descriptions 4.1 Semantically vacuous words 4.2 Nonverbal predicates VI Contents 4.3 Predicates as restrictive modifiers 63 4.3.1 A new composition rule 65 4.3.2 Modification as functional application 66 4.3.3 Evidence from nonintersective adjectives? 4.4 The definite article 73 4.4.1 A lexical entry inspired by Frege 73 4.4.2 Partial denotations and the distinction between presupposition and assertion 75 4.4.3 Uniqueness and utterance context 80 4.4.4 Presupposition failure versus uninterpretability 81 4.5 Modifiers in definite descriptions Contents 6.5.1 A little history 6.5.2 Relational and Sch O nfinkeled denotations for determine r s vii 147 6.6 Forma l properties of relational determiner meanings 6.7 Presuppositional quantifier phrases 6.7. 1 "Both" and "neither" 6.7.2 Presuppositionality and the relational theory 6.7.3 Other examples of presupposing DPs 157 6.8 Presuppositional quantifier phrases: controversial cases 6.8.1 Strawson's reconstruction of Aristotelian logic 6.8.2 Are all determiners presuppositional? 6.8.3 Nonextensional interpretation 6.8.4 Nonpresuppositional behavior in weak determiners 7 Quantification and Grammar 7.1 The problem of quantifiers in ob j ect position 7.2 Repairing the type mismatch in situ 7.2.1 An example of a "flexible types" approach 7.2.2 Excursion: flexible types for connectives 7.3 Repairing the type mismatch by movement 7. 4 Excursion: quantifiers in natural language and predicate logic 7.4.1 Separating quantifiers from variable binding 7.4.2 l-place and 2-place quantifiers 7.5 Choosing between quantifier movement and in situ interpretation : three standard arguments 7.5.1 Scope ambiguity and "inverse " scope 7.5.2 Antecedent-contained deletion 7. 5. 3 Quantifiers that bind pronouns 8 Syntactic and Semantic Constraints on Quantifier Movement 8.1 Which DPs may move, and which ones must? 8.2 How much moves along? And how far can you move? 8.3 What are potential landing sites for moving quantifiers ? 8.4 Quantifying into VP 8.4.1 Quantifiers taking narrow scope with respect to auxiliary negation , 8.4.2 Quantifying into W, VP-internal subjects , and flexible types 8.5 Quantifying into PP, AP, and NP 221 8.5.1 A problem of undergeneration 221 8.5.2 PP-internal subjects 8.5. 3 Subjects in all lexically headed XPs? 228 viii Contents 8.6 Quantifying into DP 230 8.6.1 Readings that can only be represented by DP adjunction? 232 8.6.2 Indirect evidence for DP adjunction : a problem with free IP adjunction? 233 8.6.3 Summary 234 9 Bound and Referential Pronouns and Ellipsis 239 9.1 Referential pronouns as free variables 239 9.1.1 Deictic versus anaphoric, referential versus bound-variable pronouns 23 9 9.1 .2 Utterance contexts and variable assignments 9.2 Co-reference Or binding ? 9.3 Pronouns in the theory of ellipsis 9.3.1 Background: the LF Identity Condition on ellipsis 9. 3.2 Referential pronouns and ellipsis 9.3.3 The "sloppy identity" puzzle and its solution
2004
We present the rags (Reference Architecture for Generation Systems) framework: a specification of an abstract Natural Language Generation (NLG) system architecture to support sharing, re-use, comparison and evaluation of NLG technologies. We argue that the evidence from a survey of actual NLG systems calls for a different emphasis in a reference proposal from that seen in similar initiatives in information extraction and multimedia interfaces. We introduce the framework itself, in particular the two-level data model that allows us to support the complex data requirements of NLG systems in a flexible and coherent fashion, and describe our efforts to validate the framework through a range of implementations. * This is a revised and updated version of the paper "A Reference Architecture for Generation Systems" which appeared (in error) in Natural Language Engineering 10(3/4) the Special Issue on Software Architectures for Language Engineering. This version should be cited in preference to the earlier one. 1 This survey drew on Reiter's original (Reiter 1994) formulation of the model. The later (Reiter and Dale 2000) formulation uses slightly different terminology, which we also use here, but for our purposes is otherwise not significantly different. 2 The systems surveyed were: