Analogical Modeling of Language

Hardcover, xi + 212 pages, NOT ex-library. Gently age-toned endpapers, else clean and bright throughout, with unmarked text, free of inscriptions and stamps. Firm secure binding. Published without a dust jacket. -- One important purpose of this book is to compare two completely different approaches to describing language. The first of these approaches, commonly called structuralist, is the traditional method for describing behavior which can be broadly characterized as a system of classification. However, a number of conceptual and empirical problem arise when we try to describe language in terms of rules. In order to eliminate these difficulties, this book introduces a new way of accounting for language behavior, one that can be called analogical and one that will change the way we think about language. Linguistics without rules: There is no need on the part of the analyst or the computer to discover any rules to predict language behavior. Explicit definition of analogy: A well-defined analogical model eliminates the imprecise and impressionistic appeals to 'analogy' characteristic of language studies in the past. Grammar leakage: 'All grammars leak', but until now there has been no explicit way to account for the leakage that occurs at 'rule boundaries'. Probabilistic language behavior: The analogical approach eliminates the problems that language variationists have run up against in trying to use probabilistic rules to account for linguistic variation. Historical drift: Cases of probabilistic historical change can now be handled in a systematic and predictable way. Natural statistics: All the statistical properties necessary for discrete multivariate analysis can be derived, but without any statistical calculation. Massively parallel processing: The algorithm for determining the analogical prediction works most efficiently and appropriately under massively parallel processing. An alternative to connectionist models: An analogical model avoids the problems with Rumelhart and McClelland's 'parallel distributed processing' and other activation models of language (such as inordinately long computing times, the plethora of fudge factors, and the inability to learn given probabilities, avoid specific outcomes, and adjust to alternative rules of usage). -- Contents: Introduction: - Structuralist Versus Analogical Descriptions - An Overview of the Book / 1 Problems with Rule Descriptions of Language: - Three Basic Types of Language Behavior - Empirical Problems - Conceptual Problems / 2 An Analogical Approach: - An Overview of the Analogical Model - Constructing the Analogical Set for a Given Context - Exponential Effect in a Saturated Deterministic Field - Predicting the Three Basic Behaviors Analogically / 3 Three Examples from English: - Specifying the Data Set - Categorical Behavior: The Indefinite Article - Exceptional/Regular Behavior: Spelling Initial /h/ - Idiosyncratic Behavior: Voicing Onset Time / 4 Predicting Non-Deterministic Language Variation: - Learning Probabilities - McClelland and Rumelhart's Connectionist Approach - Imperfect Memory and Multivariate Predictions - Rule Approaches to Language Variation - Sociolinguistic Variation in Arabic / 5 Analogical Predictions of the Past Tense in Finnish: - An Overview of the Verb Types - Past Tense in Standard Finnish - Variation in the Past Tense - Properties of an Analogically Predicted Past Tense Form - Predicting the Past Tense for Infrequent Verbs - Predicting the Past Tense for Verbs in the Data Set - Historical Drift of the Past Tense - A Note on the Historical Development / 6 Conclusion: - Massively Parallel Processing of Analogical Sets - A Final Analogy; References; Appendix: 1 Symbols for the Phonemes of English 2 Data Sets: - Indefinite Article in English - Spelling of Initial /h/ in English - Categorical Determination of /b,p/ in English - Terms of Address in Colloquial Egyptian Arabic - Past Tense in Finnish 3 Pascal Program.…