Showing papers on "Natural language published in 1979"

Trainable grammars for speech recognition

Abstract: Algorithms which are based on modeling speech as a finite‐state, hidden Markov process have been very successful in recent years. This paper presents a generalization of these algorithms to certain denumerable‐state, hidden Markov processes. This algorithm permits automatic training of the stochastic analog of an arbitrary context free grammar. In particular, in contrast to many grammatical inference methods, the new algorithm allows the grammar to have an arbitrary degree of ambiguity. Since natural language is often syntactically ambiguous, it is necessary for the grammatical inference algorithm to allow for this ambiguity. Furthermore, allowing ambiguity in the grammar allows errors in the recognition process to be explicitly modeled in the grammar rather than added as an extra component.

Towards a Computational Theory of Definite Anaphora Comprehension in English Discourse

Abstract: : This report investigates the process of focussing as a description and explanation of the comprehension of certain anaphoric expressions in English discourse. The investigation centers on the interpretation of definite anaphora, that is, on the personal pronouns, and noun phrases used with a definite article the, this, or that. Focussing is formalized as a process in which a speaker centers attention on a particular aspect of the discourse. An algorithmic description specifies what the speaker can focus on and how the speaker may change the focus of the discourse as the discourse unfolds. The algorithm allows for a simple focussing mechanism to be constructed: an element in focus, an ordered collection of alternate foci, and a stack of old foci. The data structure for the element in focus is a representation which encodes a limited set of associations between it and other elements from the discourse as well as from general knowledge. This report also establishes other constraints which are needed for the successful comprehension of anaphoric expressions. The focussing mechanism is designed to take advantage of syntactic and semantic information encoded as constraints on the choice of anaphora interpretation. These constraints are due to the work of language researchers; and the focussing mechanism provides a principled means for choosing when to apply the constraints in the comprehension process.

Language and Organization

Abstract: This paper argues that insight into organization functioning is related to the fit between language of description and type of organizational phenomenon. Natural language may be more powerful than mathematical language for understanding and describing many organization processes.

Mothers' speech to children and syntactic development: some simple relationships.

Abstract: This study investigated the relationships between children's linguistic environments and their language acquisition. Speech samples taken from seven firstborn children and their mothers when the children were 1; 6 and 2; 3 were analysed within a number of semantic and syntactic categories to determine correlations between mothers' speech and subsequent language development. Several characteristics of mothers' speech (e.g. utterance length, use of pronouns) significantly predicted later child speech. The significant correlations suggested that mothers' choice of simple constructions facilitated language growth. Further, they showed that the motherese code differed from adult-adult speech in ways which aided language development. Differences between our study and previous investigations of environmental effects on language development probably resulted from the failure of earlier investigations to take into account children's level of language competence at the time when environmental effects were assessed.

n-Gram Statistics for Natural Language Understanding and Text Processing

Abstract: n-gram (n = 1 to 5) statistics and other properties of the English language were derived for applications in natural language understanding and text processing. They were computed from a well-known corpus composed of 1 million word samples. Similar properties were also derived from the most frequent 1000 words of three other corpuses. The positional distributions of n-grams obtained in the present study are discussed. Statistical studies on word length and trends of n-gram frequencies versus vocabulary are presented. In addition to a survey of n-gram statistics found in the literature, a collection of n-gram statistics obtained by other researchers is reviewed and compared.

Syntax and semantics in a word-sequence recognition system

Abstract: A recognizer of meaningful sequences of isolated words spoken in the Italian language is presented. The system has been designed in a modular fashion and the current version has been obtained by adding the syntactic and semantic levels to the isolated-word recognizer previously developed by the same authors. The lexical, syntactic and semantic levels are integrated into a hierarchical knowledge source represented by a transition network grammar. The parsing is directed by the control unit and is performed by means of a combined strategy of bottom-up and top-down techniques, using a modified version of the Earley algorithm. Lexical hypotheses are verified at the phonemic level assigning them a score based on a distance measure evaluated by an error-correcting parsing algorithm. The protocol used in the preliminary experi - ments is a robot command language.

On branching quantifiers in English

Abstract: One of Hintikka's aims, in the paper Hintikka (1974),3 was to show that there are simple sentences of English which contain essential uses of branching quantification. If he is correct, it is a discovery with significant implications for linguistics, for the philosophy of natural language, and perhaps even for mathematical logic. Philosophically, it would influence our views of the ontological commitment inherent in specific natural language constructions, since branching quantification is a way of hiding quantification over various kinds of abstract abstract objects (functions from individuals to individuals, sets of individuals, etc.). Linguistically, the discovery of branching quantification would force us to re-examine, and perhaps re-interpret, Frege's principle of compositionality according to which the meaning of a given expression is determined by the meanings of its constituent phrases. For example, the meaning of a branching quantifier expression of logic like:

Skimming stories in real time: an experiment in integrated understanding.

Abstract: : This dissertation describes a new method of automated text analysis. FRUMP (Fast Reading Understanding and Memory Program) is a working natural language processing system that has been implemented to demonstrate the viability of this new approach. The system skims news stories directly from the United Press International news wire and produces a summary of what it understands. FRUMP is able to correctly process news articles it has never before seen. (Author)

Perception, Representation, and the Forms of Action: Towards an Historical Epistemology

Abstract: Among the things which are generally taken to change, historically, are ideas, theories, social systems, technologies, customs, beliefs. Biological evolutionary changes or developments are often distinguished from and sometimes compared with these historical changes: Thus, species-change, or the evolution of particular organs or traits, or even geological change are taken to be processes of natural transformation, as distinct from those post-natural or cultural changes which may be characterized as historical, and which involve human action and human history distinctively. Thus, for example, cultural evolution is contrasted with natural or biological evolution, the ‘noosphere’ is contrasted with the ‘biosphere’, and transhistorical species-specific traints, such as erect posture, or speech are contrasted with culturally variant features, such as particular natural languages, or even more differentially, styles or customs or political systems.

Solving mechanics problems using meta-level inference

Abstract: In this paper we shall describe a program (MECHO), written in Prolog[14], which solves a wide range of mechanics problems from statements in both predicate calculus and English. Mecho uses the technique of meta-level inference to control search in natural language understanding, common sense inference, model formation and algebraic manipulation. We argue that this is a powerful technique for controlling search while retaining the modularity of declarative knowledge representations.

Paraphrasing Using Given and New Information in a Question-Answer System

Abstract: The design and implementation of a paraphrase component for a natural language question-answer system (CO-OP) is presented. A major point made is the role of given and new information in formulating a paraphrase that differs in a meaningful way from the user's question. A description is also given of the transformational grammar used by the paraphraser to generate questions.

Towards more “natural” interactive systems

Abstract: It is argued that to obtain the maximum benefit from interactive computer systems principles of program and dialogue design are needed. It is unlikely that natural languages such as English will provide a suitable basis for designing a man-computer dialogue. The objective should be to model the task domain in a way that will be comprehensible to the user and to provide an explicit “image” of the “underlying processes”. Design principles are discussed both for general purpose programming languages and “bespoke” languages intended as tools for a specific purpose. It is concluded that AI research may eventually provide intelligent tools with the inferential powers necessary for a genuine dialogue, but that for the time being it is better to make the underlying mechanisms and their limitations as explicit as possible.

Is Problem-Solving Language?

Abstract: Opinion has shifted during the last decade about the ability of chimpanzees to learn language. Recent projects have reversed earlier failures to establish communication between man and chimpanzee by bypassing the troublesome (for chimpanzees) vocal medium of language. Nevertheless, just what a chimpanzee can learn about language remains controversial. Through the media of American Sign Language (e.g., Gardner amp; Gardner 1969) and the artificial visual languages invented by Premack (1970a) and by Rumbaugh and Glasersfeld (1973), chimpanzees have been taught to produce and comprehend far greater vocabularies than were thought possible following the unsuccessful attempts of the Hayeses (1951), the Kelloggs [1933 (1967); this volume], and others to communicate with chimpanzees via vocal languages. What is at issue is whether a chimpanzee can master relationships between words—in particular, relationships as expressed in sentences.

The structure and organization of a semantic net for comprehension and inference

Abstract: We have developed a network representation for propositional knowledge that we believe to be capable of encoding any proposition expressible in natural language. The representation can be regarded as a computer-oriented logic with associative access paths from concepts to propositions. Its syntax is closely modeled on predicate calculus but includes constructs for expressing some kinds of vague and uncertain knowledge. The representation allows the encoding and efficient use of caselike semantic constraints on predicate arguments for the purpose of language comprehension: these constraints are simply implications of the predicates concerned. Our approach to language comprehension is based on nonprimitive representations. We argue that primitive representations of simple propositions are often extremely complex, and offer no real advantages. We have demonstrated these ideas with a mini-implementation capable of mapping certain kinds of declarative sentences into the network representation. The implementation emphasizes the proper handling of iterated adjectival modifiers, especially comparative modifiers. More recently, we have worked on the problem of rapid access to the facts that are relevant to a query. Our solution involves the use of back-link structures from concepts to propositions, called “topic access skeletons,” which conform with general topic hierarchies in memory. For example, the proposition “Clyde is grey” is classified under the “coloring” topic for Clyde, which is subsumed under the “appearance” topic, and in turn under the “external quality” topic, and finally under the “physical quality” topic for Clyde. The form of a query (or of an assertion) can be used to determine what concepts in memory should be accessed as starting points, and what paths in the associated access skeletons should be followed in order to access the relevant information. We have demonstrated the feasibility of building such hierarchies, inserting information into them automatically, and accessing the inserted information with a second experimental implementation. The hierarchic organization appears capable of providing order-of-magnitude improvements in question-answering efficiency, with only a doubling in storage costs.

Programming in natural language: “NLC” as a prototype

Abstract: The state of the art in computational linguistics has progressed to the point where it is now possible to process simple programs written in natural language. This report describes a natural language programming system called NLC which enables a computer user to type English commands into a display terminal and watch them executed on example data shown on the screen. The system is designed to process data stored in matrices or tables, and any problem which can be represented in such structures can be handled if the total storage requirements are not excessive.

Centered logic: the role of entity centered sentence representation in natural language inferencing

Abstract: We will briefly describe the role of entity centered structure (ECS) of sentences in natural language inferencing. The basic structure of sentences in discourse, generally singles out an entity, to be called center, among all those which are the arguments of the main predicate. ECS makes n-ary predicates look like monadic by temporarily masking their structure, thereby affecting the relative ease with which certain inferences are made and information is retrieved. This short paper deals with a preliminary formulation of a system designed to capture these ideas and contains several examples of how some natural language inferences can be represented in the system. Formal properties of the system are under investigation.

Research in Natural Language Understanding

Abstract: : The goals of the project are to develop techniques required for fluent and effective communication between a decision maker and an intelligent computerized display system in the context of complex decision tasks such as military command and control. This problem is approached as a natural language understanding problem, since most of the techniques required would still be necessary for an artificial language designed specifically for the task. Characteristics that are considered important for such communication are the ability for the user to omit details that can be inferred by the system and to express requests in a form that 'comes naturally' without extensive forethought or problem solving. These characteristics lead to the necessity for a language structure that mirrors the user's conceptual model of the task and the equivalents of anaphoric reference, ellipsis, and context-dependent interpretation of requests. these in turn lead to requirements for handling large data bases of general world knowledge to support the necessary inferences. The project is seeking to develop techniques for representing and using real world knowledge in this context, and for combining it efficiently with syntactic and semantic knowledge. This report discusses aspects of research to date and a general approach to definite anaphoric reference and near-deterministic parsing strategies.

Language Distance and the Magnitude of the Language Learning Task

Abstract: A learner brings to the classroom many characteristics which are relevant to predictions about his career as a language learner. These characteristics are the product of his membership of a community; he shares its language and its attitudes to, beliefs about, motivations for and traditions in, language learning in general, and in the learning of specific second languages. And he possesses particular features of personality as formed by his personal history of maturation and experience. I am concerned in this paper with the role of only two of these characteristics, both related to the community he belongs to: the nature of his mother tongue and any other languages known to him, and the beliefs current in his community, which he presumably shares, as to the nature, extent and probable success in the learning task which lies ahead of him.

Experience with the evaluation of natural language question answerers

Abstract: Research in natural language processing could be facilitated by thorough and critical evaluations of natural language systems. Two measurements, conceptual and linguistic completeness, are defined and discussed in this paper. Testing done on two natural language question answerers demonstrated that the conceptual coverage of such systems should be extended to better satisfy the needs and expectations of users.

Strategies for Requesting in Spanish and English: Structural Similarities and Pragmatic Differences.

Abstract: This paper reports on recent work in the acquisition of pragmatic competence in a second language. It centers on the speech act of requesting and the semantic strategies for conveying that speech act. First, a contrastive analysis of Spanish and English strategies is undertaken to show that basically the same strategies are available in both languages. The paper goes on to describe an experiment in which native speakers of each language were asked to judge the relative politeness of each strategy. It concludes with an examination of the frequency of use of the various strategies for conveying requests among a population of bilingual children. The findings show that, while basically the Same request strategies are available to speakers of Spanish and English, the use of those strategies differs markedly. More polite strategies are employed in speaking Spanish, while more neutral strategies are used in English.

Utterance and objective: issues in natural language communication

Abstract: Communication in natural language requires a combination of language-specific and general common-sense reasoning capabilities, the ability to represent and reason about the beliefs, goals, and plans of multiple agents, and the recognition that utterances are multifaceted. This paper evaluates the capabilities of natural language processing systems against these requirements and identifies crucial areas for future research in language processing, commonsense reasoning, and their coordination.

Visual, analog representations for natural languages understanding

Abstract: In order for a natural language system to truly "know what it is talking about," it must have a connection to the real-world correlates of language. For language describing physical objects and their relations in a scene, a visual analog representation of the scene can provide a useful target structure to be shared by a language understanding system and a computer vision system. This paper discusses the generation of visual analog representations from input English sentences. It also describes the operation of a LISP program which generates such a representation from simple English sentences describing a scene. A sequence of sentences can result in a fairly elaborate model. The program can then answer questions about relationships between the objects, even though the relationships in question may not have been explicit in the original scene description. Results suggest that the direct testing of visual analog representations may be an important way to bypass long chains of reasoning and to thus avoid (he combinational problems inherent in such reasoning methods.

Conditionals, Generic Quantifiers, and Other Applications of Subgames

Abstract: In examining the interrelations of use and meaning, one of the most promising testing grounds is constituted by the theory of conditional sentences in natural languages. On this ground the differences between different approaches to meaning and those between the several uses of “use” have clashed dramatically, and yet left many of the principal problems unresolved. The truth-functional analysis of “if—then” sentences is as interesting an example of an approach to meaning by means of recursive truth-characterizations as one can hope to find. Yet it has run into a sharp criticism from those philosophers of language whose paradigm of meaning-giving use is usage, i.e., intralinguistic use. These philosophers are sometimes misleadingly called ordinary-language philosophers. However, they have likewise failed to solve many of the most interesting questions concerning the actual behavior of conditionals in natural languages. The initial problems we shall be dealing with in this work are cases in point. Hence the field is wide open for new approaches.

Structure in a Manual Communication System Developed Without a Conventional Language Model: Language Without a Helping Hand

Abstract: Publisher Summary This chapter presents the human language learning flexibility with respect to one particular learning condition found in all natural language learning situations but in none of the feral situations: the role of linguistic input. The role of linguistic input in accounting for language acquisition has been minimized on the grounds that the speech the child hears, as it resembles adult-to-adult talk, is too unruly for the young child to abstract language organization from it. This chapter examines the ways in which semantic relations are conveyed in the communication systems developed by six deaf children of hearing parents. These children use different signs for communication. Deictic signs are typically pointing gestures. These pointings maintained a constant kinesic form in all contexts and are used to single out objects, people, places, and the like in the surroundings. In contrast, characterizing signs are stylized pantomimes whose iconic forms vary with the intended meaning of each sign. This chapter investigates the particular semantic relations the deaf child conveys in his spontaneously generated communication system. In principle, semantic relations can be conveyed in a single sign or word unit. Children who can't hear produce two basic classes of phrase types: actions and attributes. An action phrase is used to request the execution of an action, or to comment on an action that is being, has been, will be, or can be executed. In contrast, an attribute phrase is one which is used to comment on the perceptual characteristics of an object.

Word expert parsing

Abstract: An approach to natural language meaning-based parsing in which the unit linguistic knowledge is the word rather than the rewrite rule is described. In the Word Expert Parser, knowledge about language is distributed across a population of procedural experts, each representing a word of the language, and each an expert at diagnosing that word a intended usage in context. The parser is structured around a coroutine control environment in which the generator-like word experts ask questions and exchange information in coming to collective agreement on sentence meaning. The word Expert theory is advanced as a better cognitive model of human language expertise than the traditional rule-based approach. The technical discussion is organized around examples taken from the prototype LISP system which implements parts of the theory.