Formal Grammar

Theory and Variation across English and Norwegian

Terje Lohndal

• Publisher: Routledge
• Year: 2017
• ISBN:
  9781351971904
• Edition: 1
• Topic: Lingue e linguistica
• Subtopic: Linguistica

This volume draws together fourteen previously published papers which explore the nature of mental grammar through a formal, generative approach. The book begins by outlining the development of formal grammar in the last fifty years, with a particular focus on the work of Noam Chomsky, and moves into an examination of a diverse set of phenomena in various languages that shed light on theory and model construction. Many of the papers focus on comparisons between English and Norwegian, highlighting the importance of comparative approaches to the study of language. With a comprehensive collection of papers that demonstrate the richness of formal approaches, this volume is key reading for students and scholars interested in the study of grammar.

Part A

Transformational Constraints

1 Brief Overview of the History of Generative Syntax^*

with Howard Lasnik

1.1 Background and Outline

Scientific grammar goes back to the Middle Ages, and specifically the study, by Modistic philosophers, of language as a phenomenon independent of thought. In a sense, the tradition is even older, dating back to Classical Antiquity, and spanning several cultures—after all, every traditional writing system presupposes some serious linguistic theorizing. In the humanistic Renaissance, philosophers started worrying, also, about the relation between language and mind, and as the Ages of Exploration and Reason came to be, about the problem of creativity and what it reveals about the natural world—where according to Descartes it effectively constituted a “second substance”. By the time Darwin began to revolutionize our thinking about human nature, philology was a profession in its own right, so much so that the discovery of the Indo-European ancestry and how it gave rise to hundreds of different languages served as a central inspiration in Darwin’s evolutionary theory. Many of the theoretical insights of linguistics in the twentieth century date back to this modern tradition, particularly as coupled together with late nineteenth and early twentieth century developments in mathematical logic and philosophy more generally.

Saussure (1916) initiated contemporary structural linguistics by emphasizing how language should be conceived as separate from what it is used for and by concentrating on how language is, not how it changes. Bloomfield (1933), Wells (1947) and Harris (1951) developed structuralism further and Noam Chomsky’s work developed, in particular, in immediate reaction to Harris’s program. A fundamental difference between structuralism and generative grammar stems from the fact that Chomsky focused on those aspects of structure that make the system recursive, whereas structuralism left those for the realm of what we nowadays call performance. Structuralism in fact focused on finite levels of language, such as morphophonemics, where notions like “linguistic feature” or the paradigmatic inventory underlying phonemics came to be understood (see again especially Harris 1951). But it was the syntax put to the side at the time that especially interested Chomsky, particularly since it was taken to address a key element in the problem of linguistic creativity. For this purpose, Chomsky borrowed from the axiomatic-deductive method in mathematical logic, developed a generation earlier in its computational formulation—more concretely via Davis (1958; which had circulated as a draft much prior to its publication date). Chomsky systematized and generalized Emil Post’s version of “recursive function theory” (see Post 1944), and eventually came to propose formal devices of his own (“transformations”; see the following).

Aside from these theoretical considerations pertaining to the precise structure of language and its implications, generative grammar from Chomsky’s perspective always had a conceptual angle that informs the enterprise to this day: Syntax is seen as a natural system, somehow rooted in human psychology and biology. This point of view constituted the bulk of Chomsky’s reaction to behaviorism, his later exploration of complex forms of biology, and, more generally, his insistence over six decades on approaching linguistic structure with the same sorts of tools and attitudes that one should assume for an intricate biological phenomenon, like adaptive immunity.

All of Chomsky’s work has centered on two fundamental questions:

1. (1) What is the correct characterization of someone who speaks a language? What kind of capacity is “knowledge of language”?
2. (2) How does this capacity arise in the individual? What aspects of it are acquired by exposure to relevant information (“learned”), and what aspects are present in advance of any experience (“wired in”)?

Chomsky’s earliest work, in the 1950s, raised and focused on question (1), since explicit and comprehensive answers to that question had never been provided before. Chomsky’s answer posited a computational system in the human mind that provides statements of the basic phrase structure patterns of languages (phrase structure rules) and more complex operations for manipulating these basic phrase structures (transformations). This framework, and its direct descendants, fall under the general title Transformational Generative Grammar (generative meaning explicit, in the sense of mathematics).

In the 1960s, the research began to shift more toward question (2), and Chomsky called the theory that was developed the Standard Theory. Chomsky coined the term explanatory adequacy for putative answers to that question. A theory of language, regarded as one component of a theory of the human mind, must make available grammars for all possible human languages. To attain explanatory adequacy, the theory must in addition show how the learner selects the correct grammar from among all the available ones, based on restricted data. The theories of the 1950s and early 1960s made an infinite number of grammars available, so the explanatory problem was severe.

Through the late 1960s and 1970s, to enhance explanatory adequacy, theorists proposed more and more constraints on the notion “possible human grammar”. Ross (1967) was a particularly influential and pioneering study looking at locality restrictions (den Dikken and Lahne 2013). These moves were explicitly motivated by considerations of explanatory adequacy, though general considerations of simplicity also played a role. This culminated in the Principles and Parameters framework (Bošković 2013) and, more specifically, in the Government and Binding approach that Chomsky (1981) proposes. The latter led to a wide range of cross-linguistic research since a core part of the program involved comparative syntax and used comparative data to help refine theoretical definitions of terms like government and binding.

At the same time as these developments took place, a number of researchers departed from Chomsky’s specific approach. Generative Semantics, in particular, was a very prominent theory in the late 1960s; today some Generative Semantics ideas have returned, as we discuss in the following. In the early 1980s, nontransformational theories such as Lexical-Functional Grammar (Kaplan and Bresnan 1982; Sells 2013), Generalized Phrase Structure Grammar (Gazdar et al. 1985; Blevins and Sag 2013) and Tree Adjoining Grammar (Joshi, Levy and Takahashi 1975, Joshi 1985; see Frank 2013) were also developed. We say a bit more about these in the following and contextualize them to make it clear why they emerged and what the main differences are between these theories and the more mainstream Chomskyan theories.

In the late 1980s, Chomsky started to explore what has become known as the Minimalist Program, with its emphasis on simplicity in theorizing and on moving beyond explanatory adequacy in the sense of asking why the language faculty has the properties it does. This approach is most explicitly outlined in Chomsky (1995b). Recent and ongoing work by Chomsky (2000, 2001, 2004, 2007, 2008) and many others continues to develop this framework.

This chapter is organized as follows. Section 1.2 discusses the earliest generative approaches, namely, those explicated in Syntactic Structures (1957) and Aspects of the Theory of Syntax (1965). We examine some relevant differences between these two theories, and we discuss some general properties of transformations. Section 1.3 discusses the syntax/semantics interface in early generative grammar and beyond, whereas Section 1.4 is an overview of how phrase structure has developed from the early days of generative grammar until today. In section 1.5, we discuss the role in the evolving theories of rules and filters versus principles. Section 1.6 is concerned with derivations and the derivation versus representation issue. In Principles and Parameters theory, Chomsky explicitly introduced economy principles for the first time, and we give a summary of some of these in section 1.7. A few concluding remarks are provided in section 1.8.

1.2 The Earliest Generative Approaches: Syntactic Structures and Aspects

Chomsky’s earliest work developed in reaction to the structuralist work mentioned in section 1.1. As a student of Zellig Harris, Chomsky was very familiar with Harris’s program and he developed his own work in reaction to Harris (1951). Harris had one sentence transform into another. This approach was therefore not able to give any systematic explanation for the more abstract kind of phenomena Chomsky started to deal with in The Logical Structure of Linguistic Theory (LSLT 1955) and Syntactic Structures. In order to deal with these phenomena, it is necessary to relate abstract structures to abstract structures. Let us now look at some of the characteristics of Chomsky’s earliest work.

Infinity and structure are the fundamental characteristics of human language, and they can both be captured, in part, by way of a context-free phrase structure (PS) grammar. One such device (a Σ, F grammar in Chomsky’s terminology) consists of

1. (3)
   1. A designated initial symbol (or a set thereof) (Σ);
   2. Rewrite rules (F), which consist of a single symbol on the left, followed by an arrow, followed by at least one symbol.

A derivation consists of a series of lines such that the first line is one of the designated initial symbols, and to proceed from one line to the next we replace one symbol by the sequence of symbols it can be rewritten as, until there are no more symbols that can be rewritten. For instance, given

1. (4)
   1. Designated initial symbol (Σ): S
   2. Rewrite Rules (F):
      • S → NP VP
      • NP → N
      • VP → V
      • N → John
      • V → laughs

we can obtain a derivation as in (5):

1. (5) Line 1: S
   • Line 2: NP VP
   • Line 3: N VP
   • Line 4: N V
   • Line 5: John V
   • Line 6: John laughs

Chomsky (1965) called rules like the last two in (4), which rewrite a particular nonterminal symbol as a single terminal symbol, lexical insertion rules—a distinction not made in the theories of Chomsky (1955, 1957).

PS grammars capture constituent structure by introducing nonterminal (unpronounced) symbols. Given (5), we can connect each ...

Table of contents

• Cover
• Title
• Copyright
• Contents
• Acknowledgments
• Original Publication Details
• Introduction
• Part A Transformational Constraints
• Part B The Syntax–Semantics Interface
• Part C Multilingualism and Formal Grammar
• Index