Emmon Bach

On raising : one rule of English grammar and its theoretical implications

For some time it has been generally accepted by students of English grammar that a rule of Raising exists and that it functions to produce derived main clause subjects. Following Rosenbaums work, it has also been widely accepted that this rule functions in a specified class of cases to derive main clause objects. However, in recent work, Chomsky has rejected the view that there is any Raising rule that produces derived main clause objects. According to his latest position, only the derived subject function of the rule is an actual feature of English grammar.On Raising is highly critical and is devoted chiefly to supporting the claim that English does contain a rule of Raising -- a rule that has the function of taking the complement subject noun phrase in certain complement constructions and reassigning it as a constituent of the main clause. The author presents something on the order of two dozen arguments that Raising produces derived objects. In the course of this discussion, he also considers various other theoretical and descriptive consequences of, and questions raised by, the existence of Raising.

Crossed and nested dependencies in German and Dutch : A psycholinguistic study

Abstract The clause-final verbal clusters in Standard Dutch and German differ strikingly in the kinds of dependencies they normally permit between verbs and their arguments, with Dutch preferring crossed dependencies and German nested. This study investigates the consequences of these differences for the psycholinguistic processing complexity of sentences containing either crossed or nested dependencies. German and Dutch subjects performed two tasks-ratings of comprehensibility and a test of successful comprehension-on matched sets of sentences which varied in complexity from a simple sentence to one containing three levels of embedding. The results show no difference between Dutch and German for sentences within the normal range (up to one level of embedding), but with a significant preference emerging for the Dutch crossed order for the more complex strings. We argue that this rules out the push-down stack as the universal basis for the human parsing mechanism.

Natural Language Metaphysics

Publisher Summary The chapter presents a serious account of the semantics of natural language. Linguistics, like any other field of inquiry, lives off of puzzles. Linguists, like other seekers after understanding, usually follow the maxim: Divide and conquer! The general framework for descriptions of natural language draws upon two traditions: that of generative theory as developed in the past several decades under the leadership of Noam Chomsky and others; that of model-theoretic semantics as inspired especially by Richard Montague. The chapter is concerned with asking about such a program or theory with respect to certain phenomena and puzzles that seem to go beyond pure semantics as usually conceived. It is possible and desirable to draw a sharp line between “constructional” (or “structural”) semantics and lexical semantics. There are two parts of the enterprise of doing the semantics of natural languages where metaphysical questions rear their—ugly or beautiful?—heads: in making decisions about the general structure and content of our models and their relation to the things in our syntaxes, and at points where it seems that we have to “go inside” the meanings of particular lexical items to state compositional rules of the semantics.

Some Generalizations of Categorial Grammars

The last decade of work in syntax has seen a marked demotion in the importance of transformations in the classical sense. We see this not only in the various explicit proposals to do away with transformations altogether but also in recent developments in transformational theories themselves. With the reduction of the transformational component to the single rule “move alpha” almost all of the work of generating “all and only” the syntactic structures of a language must be borne by other devices and principles.

The NP-S analysis of relative clauses and compositional semantics

ConclusionsWe have sketched how it is possible to give an analysis for adjoined relative clauses which is consistent with the compositionality principle and have shown that the technique which seems necessary for this analysis can be used to provide a compositional semantics for the NP-S analysis of English relative clauses.It is unlikely that anyone working within the framework of a compositional theory would choose the NP-S analysis for English, since it is clearly much less elegant and simple, in some intuitive sense, than the alternative Nom-S analysis. Our results seem to indicate, however, that such an analysis cannot be ruled out in principle, since any constraint on the theory that would exclude the NP-S analysis would seem to exclude the Hittite analysis as well. So the arguments for the Nom-S analysis in the English case must be based on other grounds, or the Hittite analysis is also incorrect and should be ruled out, or the happy discovery of some as yet unknown principles will allow one but not the other.

Purpose Clauses and Control

My concern in this paper is the syntax and semantics of purpose-clauses in English. The relation of this problem to the question of the nature of syntactic representation is that the topics I will discuss raise interesting questions about what one might not want to include as explicit parts of either a syntactic or a semantic representation.

Categorial Grammars as Theories of Language

In recent years, there has been a growing interest in categorial grammar as a framework for formulating empirical theories about natural language. This conference bears witness to that revival of interest. How well does this framework fare when used in this way? And how well do particular theories in what we might call the family of categorial theories fare when they are put up against the test of natural language description and explanation? I say ‘family’ of theories, for there have been a number of different developments, all of which take off from the fundamental idea of a categorial grammar as it was first introduced by Ajdukiewicz and later modified and studied by Bar-Hillel, Curry, and Lambek. In this paper I would like to discuss these questions, considering a number of different hypotheses that have been put forward within the broad framework that we may call ‘extended categorial grammar’ and making a few comparisons with other theories. In my remarks, I will take as a general framework the program and set of assumptions that have been called ‘extended Montague grammar’ and in particular a slightly modified version of Montague’s ‘Universal Grammar’ (UG: Paper 7 in Montague, 1974). From this point of view, the syntax of a language is looked at as a kind of algebra. Then, the empirical problem of categorial grammar can be seen as part of a general program that tries to answer these questions: (A) What is the set of primitive and derived categories that we need to describe and explain natural languages in their syntax and semantics (and phonology, etc.)? (B) What are the operations that we need to describe and explain natural languages (in the syntax, semantics, phonology, morphology, etc.)? (C) What are the relations that we need in order to hook up with each other the various categories and operations mentioned or alluded to in (A) and (B)?

Structural Linguistics and the Philosophy of Science

In 195 7 American linguistics seemed to have reached a plateaur of achievement and acceptance on which its practitioners could pause in retrospective pride. That year saw the publication, under the aegis of the American Council of Learned Societies, of a sampling of papers, edited by Martin Joos (Reading,r in Linguistics) and documenting, according to its subtitle, &dquo;The development of descriptive linguistics in America since 1925.&dquo; In a period of about five years around that date H. A. Gleason, Jr., Charles F. Hockett and Archibald A. Hill published textbooks. which summarized and extended current views on linguistics; while another leading linguist, Kenneth L. Pike, summed up his considerable experience in a preliminary version of a major work

Eventualities, Grammar, and Language Diversity

Languages differ widely in what is put into their grammars. Typological studies must take this into account. In the realm of “eventology” I compare a number of North American languages from the point of view of showing how a basic and probably universal classification of events, processes, states can enter into the grammar of words and phrases in different ways. The principal exhibits are from languages in the two branches of Wakashan, which are typologically very similar in gross characteristics and whose differences are thus especially interesting for questions of language diversity. The discussion is set out against a background of issues revolving around language diversity.

Gekreuzte und geschachtelte Abhängigkeiten im Deutschen und Niederländischen: Eine psycholinguistische Studie

Im Niederlandischen und im Deutschen unterscheiden sich die satzfinalen Verbcluster auffallig in den Abhangigkeiten, die zwischen den Verben und ihren Argumenten erlaubt sind. Wahrend das Niederlandische gekreuzte Abhangigkeiten bevorzugt, zieht das Deutsche geschachtelte vor. Diese Studie erforscht Konsequenzen dieser Unterschiede fur die psycholinguistische Verarbeitungskomplexitat von Satzen, die entweder gekreuzte oder geschachtelte Anordnungen involvieren. Zwei Aufgaben wurden mit deutschen und niederlandischen Versuchspersonen durchgefuhrt: Bewertung der Verstandlichkeit und ein Test, in dem das Verstehen uberpruft wurde. In jedem Fall waren die Satze mit denen der jeweiligen anderen Sprache vergleichbar. Die Komplexitat reichte vom zusammengesetzten Satz bis zu Satzen mit drei Einbettungsebenen. Die Ergebnisse zeigen fur Satze bis zu einem Einbettungsgrad keinen Unterschied zwischen Niederlandisch und Deutsch; ein deutlicher Vorteil ergibt sich jedoch durch die gekreuzte Anordnung des Niederlandischen bei komplexeren Satzen. Wir mochten behaupten, das dies den Kellerspeicher (‚pushdown stack‘) als Universale des menschlichen Sprachverarbeitungsmechanismus unmoglich macht.