The Discourse Function of Inversion in English

Betty Birner

• Publisher: Taylor & Francis
• Year: 2014
• Print ISBN: 9780815325567
• eBook ISBN: 9781136753978
• Edition: 1
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

First published in 1997. This dissertation presents a discourse-functional account of English inversion, based on an empirical study of natural language data. The central finding is that inversion is subject to a pragmatic constraint on the information status of its constituents; specifically, the information represented by the preposed constituent must be at least as familiar within the discourse as is that represented by the postposed constituent.

1
Introduction

It has long been a primary goal of linguistic research to make explicit the implicit competence possessed by a native speaker of a language, but only in recent years has the relationship between form and function in discourse been recognized as an integral part of this competence. If we hope to adequately model a speaker's linguistic knowledge, it is insufficient merely to investigate the rules governing the grarnmaticality of linguistic forms; researchers must likewise investigate the circumstances under which these forms may be used, and the influence of context not only on the acceptability or unacceptability of a given syntactic form, but also on the interpretation induced by the use of that form. Without an understanding of the appropriate use of linguistic signs, it will be impossible to fully understand natural-language discourse, and equally impossible to model it computationally

It is well documented that a speaker may exploit the interaction between discourse context and propositional meaning to give rise to conversational implicatures; that these implicatures are not tied to the form of the utterance is evidenced by their NONDETACHABILITY (Grice 1975). A speaker may also, however, exploit the interaction between the discourse context and the syntactic form itself for the purpose of structuring the information represented by the utterance either internally or with respect to other information believed to be (or desired to be, or soon to be) in the hearer's discourse model. That is, while natural languages provide their speakers with a variety of syntactic forms for conveying a single semantic proposition, the speaker's selection from among the available options may serve an information-packaging function (Chafe 1976, Vallduví 1992, inter alia). A speaker's use of a syntactic construction with a particular discourse function, then, will license the hearer to infer that the relationship between the information represented by the utterance and other relevant information in the discourse is appropriate to this discourse function. If such a relationship does not hold, the utterance is infelicitous.

Given that the syntactic options available to a speaker are language-specific, we may assume that the mapping of discourse functions to syntactic constructions is similarly language-specific; it has moreover been demonstrated that similar syntactic forms may be used in different languages or dialects for different pragmatic purposes (e.g., Prince 1981a, 1986). Thus, constraints on the appropriate use of syntactic forms are both conventional and (at least to some degree) arbitrary, and represent an important aspect of a native speaker's linguistic competence

One of the syntactic options available to speakers of English is INVERSION, exemplified in 1:

1. • (a) Labor savings are achieved because the crew is put to better use than cleaning belts manually; <also eliminated> is <the expense of buying costly chemicals>.
     [WOODEXTRA, August 1988]
   • (b) George, can you do me a favor? <Up in my room, on the nightstand,> is <a pinkish-reddish envelope that has to go out immediately>.
     [T.L., 7/24/89, on the telephone]
   • (c) Expressions like "geezer," "codger," "fuddy-duddy," "old goat" obviously disparage the old person, <Not so obvious> are <the euphemisms "umpteen years young" and "Golden Ager>." Meant to be flattering, such terms assume that old age is repugnant.
     ["Let's drop the language of ageism," Chicago Tribune, 3/15/90, sec. 1, p. 25]
   • (d) Usually on a Saturday Route 111 is buzzing with shoppers pillaging the malls hacked from the former fields of corn, rye, tomatoes, cabbage, and strawberries. <Across the highway, the four concrete lanes and the median divider of aluminum battered by many forgotten accidents,> stands <a low building faced in dark clinker brick that in the years since Harry watched its shell being slapped together of plywood has been a succession of unsuccessful restaurants and now serves as the Chuck Wagon, specializing in barbecued take-outs>. The Chuck Wagon too seems quiet today.
     [Updike 1981:7]
   • (e) You can drive as fast as you like in the outside lane on a West German highway and may feel like the king of the road—until you look in the rear mirror. <Zooming in on you like a guided missile> comes <a rival contender, bullying you to get out of the way>. "Cars are used as weapons," says Herbert Schnoor, Interior Minister of the state of North Rhine-Westphalia. "The road has become an arena."
     ["Should Germany brake the world's fastest highways?", Chicago Tribune, 8/8/89, sec. 5, p. 1]

What the italicized sentences in 1 have in common is that the NP of which some location, status, or activity is being predicated (i.e., the logical subject) appears to the right of the verb, with some other constituent appearing to the left of the verb, where the subject would canonically appear. These preposed and postposed constituents appear in angled brackets.

While many attempts have been made to ascribe a function to inversion, I will show in chapter 3 that no function yet proposed can adequately account for the distribution of inversions occurring naturally in discourse. Instead, I will demonstrate in chapter 4 that there is a strong correlation between the position of a constituent in an inversion and the familiarity within the discourse of the entity represented by that constituent. It is my goal to show that inversion is basically an information-packaging mechanism, allowing the presentation of relatively familiar information before a comparatively unfamiliar logical subject. In this way, inversion serves in essence what Green (1980) has called a CONNECTIVE FUNCTION, relating the information represented by the inversion to the surrounding context in a coherent manner. The primary advantage of this approach over previous analyses is its emphasis on relative, rather than absolute, information status, and on different types of givenness (Prince 1981b, 1992). Moreover, an analysis of naturally-occurring tokens of inversion indicates that evoked and inferrable information are treated alike with respect to inversion.

While the corpus also shows a significant asymmetry in terms of the morphological definiteness of the preposed and postposed constituents, this asymmetry is due to the fact that while inversion is sensitive to discourse-familiarity, definiteness is sensitive to hearer-familiarity (Prince 1992).¹ The results of this study further suggest that the restriction of the verb in an inversion to either be or an unaccusative verb of existence or appearance (Firbas 1966b, Levin 1991, inter alia) is due to pragmatic factors; specifically, the verb must not represent new information in the discourse. Finally, a comparison of the results of this research with those reported in Prince 1992 indicates that discourse-familiarity correlates not with subjecthood but rather with relative sentence position.

1.1 Methodology

Researchers in linguistics have typically adopted one of two possible avenues for investigating native-speaker competence, basing their conclusions either on intuitions (their own or informants') regarding constructed data, or on an examination of naturally-occurring data. Both approaches, however, are imperfect. Intuitions reported by informants (or obtained by the researchers' introspection) may reflect what they believe they would say, or what they have been taught is correct, rather than what they actually produce in natural settings; such processing factors as sentence length and complexity may also interfere with one's ability to accurately report on one's own linguistic competence. A large corpus of naturally-occurring data culled from a variety of linguistic contexts provides the researcher with language as it is actually used; however, naturally-occurring data will inevitably contain performance errors and therefore cannot be considered a perfect reflection of a speaker's competence. The best approach may be a combination of the two, with intuition suggesting directions for research and serving as a check on the naturally-occurring data collected.

The present study is based in part on an examination of a large corpus of non-randomly collected naturally-occurring tokens of English inversion. The corpus contains 1778 tokens, 1661 of which are from written sources and 107 of which are from spoken sources (with the sources of the remaining 10 being indeterminate). The preponderance of written tokens reflects in large part the relative difficulty of collecting spoken tokens, particularly when one is trying to obtain for study not only the inversion itself but also the context in which it appeared. Although inversions do occur in conversation, by the time the inversion is recognized it is often too late to recall the entire sentence verbatim, much less the preceding context. Thus, while it is entirely possible that inversion is less common in spoken than in written English, the proportions in the corpus should not in themselves be taken as evidence for this.²

The method of collection of the tokens, as noted above, was nonrandom; the corpus used represents a joint effort by Beth Levin, Gregory Ward and myself, and includes tokens noted by us and donated to us by others. The corpus, therefore, reflects the particular instances of inversion that happened to be both encountered and noticed by its contributors, and hence is not necessarily a representative sample. For this reason, it would be inappropriate to draw conclusions regarding proportions of various types or characteristics of inversion based on their proportions in the corpus. However, conclusions can be drawn concerning what occurs (and, much more tentatively, what does not seem to occur) in inversion in English based on a corpus of this size. In the absence of any evidence to the contrary, for example, it seems reasonable to assume that if some inversion type is both well represented in the corpus and intuitively grammatical, then it is in fact grammatical. This type of argument will be used in chapter 3 to counter several previously made claims regarding what is and is not possible in inversion. Additionally, given the size of the corpus and the variety of sources drawn from, strong patterns evident in such factors as type of verb and relative familiarity of elements can be considered highly suggestive of patterns in English inversion in general. Thus, while a 3:2 ratio in the corpus for some factor cannot and will not be taken as evidence for a 3:2 ratio for this factor among English inversions in general, a 95% or 100% occurrence in the corpus of some characteristic can and will be taken as highly suggestive that this characteristic holds for most or all acceptable inversions in English, respectively. Strong patterns and correlations in the corpus will be offered as supporting evidence for claims made in chapters 4 and 5.

1.2 Coding Factors

Each token in the corpus was coded for the following characteristics:

• Canonical word order acceptability.
• Source: Written or spoken.
• Phrasal categories of the proposed and postposed constituents.
• Relative information status: The status, in terms of assumed familiarity (Prince 1981b), of the information represented by the preposed and postposed constituents.
• New topic: Whether the topic of the next clause is the same as, a subset of, or inferentially related to, the information represented by either the preposed or postposed constituent.
• Open proposition: Whether an 'open proposition' (Prince 1986) roughly consisting of the information represented by the preposed constituent and the verb may be assumed to be salient in the discourse at the time the inversion is uttered. (See chapter 5 for discussion.)
• Verb: The verbal material appearing between the preposed and postposed constituents—i.e., the main verb or, in cases where the main verb is preposed (VP inversion), auxiliary be.
• Transitivity of verb.
• Active vs. passive verb.
• Locativity of verb: Whether the verb denotes or suggests spatial or temporal location, motion or direction. (See chapter 2 for discussion.)
• Verb agreement: Whether the verb agrees with the postposed constituent.
• Verb tense.
• Morphological definiteness or indefiniteness of the postposed constituent and any NPs appearing in the preposed constituent.
• Embeddedness: Whether the inversion is embedded within a higher clause.
• Polarity.
• Locativity of preposed constituent: Whether the preposed constituent denotes or suggests spatial or temporal location, motion or direction. (See chapter 2 for discussion.)
• Formal weight: Whether the preposed or postposed constituent seems clearly 'heavier', or longer and/or more syntactically complex (Hartvigson and Jakobsen 1974).

The relevance of many of these coding factors will become apparent in later chapters. It should be noted that for some of these characteristics, such as topichood, an objective decision could not always be made. Details of the coding process are discussed at length in Appendix A, where the results for all coding factors are presented.

Analysis of the data was facilitated by the use of a database program, PC-File version 5.01, which provided quick access to, and counts of, tokens with selected characteristics and combinations of characteristics. Most of the statistical information presented in this work is the result of calculations performed by PC-File.

1.3 Organization

This book is organized as follows: In chapter 2, inversion is defined and differentiated from superficially similar constructions, including preposing, subject-auxiliary inversion, quotation inversion, and there-insertion. The different syntactic types of inversion are listed and discussed in turn. In chapter 3, previous accounts of inversion are discussed, with a strong emphasis on functional accounts, and various difficulties with these analyses are presented. Chapter 4 presents and defends the major claims of this work regarding the discourse function of inversion; these claims are situated as an outgrowth of recent work on the functions of syntax, and their implications for current theory are discussed. Chapter 5 presents an investigation into the classes of verbs that appear in various inversion types, and examines pragmatic constraints on these verbs; the chapter ends with an examination of the relationship between intonation and the discourse function of inversion. Chapter 6 contains a summary, a brief discussion of implications for the study of related constructions, and suggestions for further research.

1.4 Notational Conventions

Whenever possible, the examples provided in this work represent naturally-occurring instances of inversion culled from the corpus. Such tokens are presented with attribution, given in square brackets at the end of the token. Where the source is a book, the author, year, and page number are given; full bibliographic information is given in "Sources for Tokens" at the end of the book. Where no attribution is presented, the example has been const...

Table of contents

• Cover
• Title
• Copyright
• Dedication
• Contents
• List of Illustrations
• Preface
• Acknowledgments
• Half Title
• 1 Introduction
• 2 Definition and Preliminary Taxonomy
• 3 Previous Analyses
• 4 The Function of Inversion
• 5 Verbs and Related Issues
• 6 Summary and Conclusions
• Appendix: Coding Factors and Results
• Notes
• Sources for Tokens
• References
• Index