Formalism and Functionalism in Linguistics: The Engineer and the Collector 9781138316119, 9780429455858

Table of contents :
Cover
Half Title
Title
Copyright
Contents
Acknowledgments
Introduction
1 Defining ‘Formalism’ and ‘Functionalism’
1.1 Starting Out
1.2 Defining the Terms
1.3 The Engineer and the Collector
1.4 Two Examples
1.5 On the Architecture of Formalism Versus Functionalism
References
2 Background to the Current Debate
2.1 Is There a Starting Point?
2.2 Before the Twentieth Century
2.3 Twentieth-Century Formalist Linguistics to 1950
2.4 Twentieth-Century Functionalism from the Prague Circle to Pike
2.5 Why the Background Matters
References
3 Contemporary Formalist Linguistics
3.1 The Internal Structure of Modern Formalism
3.2 What Makes Modern Formalism Formalist?
3.2.1 Early Transformational Grammar
3.2.2 Principles and Parameters
3.2.3 The Minimalist Program
3.2.4 The Common Formalist Character of Generative Grammar
3.3 Diversity Within Contemporary Formalism
References
4 Contemporary Functionalist Linguistics
4.1 The Internal Structure of Modern Functionalism
4.2 The Scope of Modern Functionalism
4.2.1 Systemic Functional Linguistics
4.2.2 Langacker’s Cognitive Grammar
4.2.3 Talmy Givón
4.3 Taxonomies of Modern Functionalist Linguistics
4.4 Graphic Style in Modern Formalism and Functionalism
References
5 Formalism and Functionalism in Action
5.1 Applying Formalism and Functionalism
5.2 In the Analysis of Syntactic Phenomena
5.2.1 Word Order
5.2.2 Transitivity
5.3 In Conceptualization and Study of Child Language Learning
5.4 In Debate About the Origin of Human Language
5.5 Gaps, Overlap, and Combat Between Formalism and Functionalism
References
6 Formalism and Functionalism Juxtaposed
6.1 Formalism and Functionalism in Silhouette
6.2 Face to Face
6.2.1 In Conversation
6.2.2 How Formal Is Functionalism?
6.2.3 How Functionalist Is Formalism?
6.3 Side by Side
6.3.1 Explanation
6.3.2 (Non)complementarity
6.4 Back to Back
References
7 Conclusion
7.1 Engineers and Collectors, Revisited
7.2 Portrait of a Linguistic Mindset
7.3 Formalism and Functionalism Face the World Online
References
Appendix: Characteristics of Formalism Versus Functionalism
Index

Citation preview

Formalism and Functionalism in Linguistics

This volume is a concise introduction to the lively ongoing debate between formalist and functionalist approaches to the study of language. The book grounds its comparisons between the two in both historical and contemporary contexts where, broadly speaking, formalists’focus on structural relationships and idealized linguistic data contrasts with functionalists’ commitment to analyzing real language used as a communicative tool. The book highlights key sub-varieties, proponents, and critiques of each respective approach. It concludes by comparing formalist versus functionalist contributions in three domains of linguistic research: in the analysis of specific grammatical constructions; in the study of language acquisition; and in interdisciplinary research on the origins of language. Taken together, the volume opens insight into an important tension in linguistic theory and provides students and scholars with a more nuanced understanding of the structure of the discipline of modern linguistics. Margaret Thomas is Professor of Linguistics in the Department of Slavic and Eastern Languages and Literatures at Boston College, USA.

Formalism and Functionalism in Linguistics

The Engineer and the Collector Margaret Thomas

First published 2020 by Routledge 52 Vanderbilt Avenue, New York, NY 10017 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2020 Taylor & Francis The right of Margaret Thomas to be identified as author of this work has been asserted by her in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data A catalog record for this book has been requested ISBN: 978-1-138-31611-9 hbk ISBN: 978-0-429-45585-8 ebk Typeset in Times New Roman by Apex CoVantage, LLC

Contents

Acknowledgmentsviii Introduction

1

1

Defining ‘Formalism’ and ‘Functionalism’ 1.1 Starting Out 4 1.2  Defining the Terms  5 1.3  The Engineer and the Collector  6 1.4 Two Examples 8 1.5  On the Architecture of Formalism Versus Functionalism 12 References 13

4

2

Background to the Current Debate 2.1  Is There a Starting Point?  15 2.2  Before the Twentieth Century  16 2.3  Twentieth-Century Formalist Linguistics to 1950  20 2.4 Twentieth-Century Functionalism from the Prague Circle to Pike  23 2.5  Why the Background Matters  28 References 28

15

3

Contemporary Formalist Linguistics 3.1  The Internal Structure of Modern Formalism  32 3.2  What Makes Modern Formalism Formalist?  32 3.2.1  Early Transformational Grammar  33 3.2.2  Principles and Parameters  36 3.2.3  The Minimalist Program  38

32

vi  Contents 3.2.4 The Common Formalist Character of Generative Grammar  40 3.3  Diversity Within Contemporary Formalism  40 References 44 4

Contemporary Functionalist Linguistics 4.1  The Internal Structure of Modern Functionalism  45 4.2  The Scope of Modern Functionalism  45 4.2.1  Systemic Functional Linguistics  46 4.2.2  Langacker’s Cognitive Grammar  48 4.2.3  Talmy Givón  50 4.3  Taxonomies of Modern Functionalist Linguistics  51 4.4  Graphic Style in Modern Formalism and Functionalism  55 References  58

45

5

Formalism and Functionalism in Action 5.1  Applying Formalism and Functionalism  60 5.2  In the Analysis of Syntactic Phenomena  60 5.2.1  Word Order  60 5.2.2  Transitivity  64 5.3  In Conceptualization and Study of Child Language Learning 67 5.4  In Debate About the Origin of Human Language  72 5.5  Gaps, Overlap, and Combat Between Formalism and Functionalism  78 References 79

60

6

Formalism and Functionalism Juxtaposed 6.1  Formalism and Functionalism in Silhouette  83 6.2  Face to Face  83 6.2.1 In Conversation 83 6.2.2  How Formal Is Functionalism?  85 6.2.3  How Functionalist Is Formalism?  89 6.3  Side by Side  92 6.3.1 Explanation 92 6.3.2 (Non)complementarity 96 6.4  Back to Back  99 References  101

83

Contents vii 7 Conclusion 7.1  Engineers and Collectors, Revisited  105 7.2  Portrait of a Linguistic Mindset  105 7.3  Formalism and Functionalism Face the World Online  107 References  110

105

Appendix: Characteristics of Formalism Versus Functionalism111 Index113

Acknowledgments

I am grateful to Boston College students of linguistics who enrolled in the inaugural version of the course Formalism and functionalism. They worked through this material with me, adding a lot to my understanding. I also thank my Research Assistants Camila Loforte, whose diligence and resourcefulness improved the text on every level and Seung Hwan Kim, whose help with matters large and small was invaluable at the last stages of completion of this project. Editor Elysse Preposi was a pleasure to work with, as was the rest of the Routledge production team. Finally, I thank the anonymous reviewers of the manuscript for their thoughtful critique.

Introduction

Readers of this book probably do not need to be introduced to the observation that human language displays intricate patterns, on many levels of organization. They probably already recognize that languages both resemble and differ from each other, and, moreover, that language is characteristic of our species, so that without it human culture would be unimaginable. It is likely that readers assume from the start that language is a complex multidimensional phenomenon. Many, but perhaps not all, identify linguistics as the scientific study of language. On the other hand, there is much that readers of this book may not share as students of language. Some of that diversity of outlook derives from sub-specialization, because (for example) historical linguistics does not overlap very much with speech pathology, nor does dialectology with neurolinguistics. Some of it derives from exposure to different language data: one linguist works within the tradition of Germanic language scholarship, while another does field research in Melanesia. Moreover, linguists and linguists-in-training adopt different stances along theoretical continua. Following in the style of their education, and according to their personal intellectual dispositions, they find their own sweet spots within all manner of controversies. The diversity of sub-specialization, exposure, and stance (among other factors) keeps modern linguistics lively. This book addresses one particularly lively dimension of difference among contemporary linguists, the upshot of which is under-appreciated. That difference comes to the surface in debate between ‘formalist’ and ‘functionalist’ approaches to the study of language. Not all linguistic research can be identified with either formalism or functionalism, nor does every individual proposal or claim fall exhaustively within one of the two camps. Nevertheless, achieving a clear understanding of what is at stake in the tension between formalism and functionalism is invaluable. In particular, it helps make sense of the competing claims that students of linguistics encounter and enlarges their sense of the scope of the discipline. It is not my goal to judge the outcome of

2  Introduction competition between formalism and functionalism or to try to resolve the tension between them. Rather, my hope is that by learning to recognize the traits of formalism versus functionalism, readers will be better able to assess (in)consistency in a body of work and better able to tailor their own analyses of language data to whatever theoretical commitments they make. This book is organized as follows. Chapter 1 points out some defining features of formalism and of functionalism, illustrating them with a look at two recent books which introduce the public to modern linguistic research, one from each side of the debate. Chapter 2 provides some historical context to modern formalism and functionalism. Chapters 3 and 4 are sibling chapters, with the first surveying contemporary formalist linguistics from the late 1950s onward and the second depicting representative schools and scholars of contemporary functionalist linguistics over the same interval. Chapter 5 depicts formalism and functionalism in action, showing how each has been brought to bear on three issues: the analysis of particular grammatical phenomena (word order; transitivity); research on child language learning; and the fraught question of the origin of human language. Chapter 6 juxtaposes formalism against functionalism and vice versa from several standpoints, analyzing different positions the two have taken up relative to each other. Chapter 7 concludes. In the interest of conciseness, I have excluded two plausible topics from this text. One is research on phonology, as almost all the examples and discussion included here address morphosyntax. Several commentators have remarked that there is more coherence across formalist and functionalist work in phonology than in other domains (Haspelmath 2000; Curnow 2002; Carnie & Mendoza-Denton 2003). I have therefore opted to focus on issues where the two approaches are in sharper contrast. Another neglected topic is the phenomena of grammaticalization, the process by which, over time, a referential lexical item shifts to take on a functional (or grammatical) role. There are two reasons why, reluctantly, I  excluded discussion of grammaticalization: first, because it has a very high profile in functionalist accounts of language change, while it is more peripheral to formalism and addressed in a more diffuse manner. This imbalance complicates comparison. Second, Fischer (2007; see also Fischer, Rosenbach & Stein 2000) gives a rich book-length treatment of grammaticalization, managing to balance formalist and functionalist perspectives. Fischer’s text should be fully accessible to readers of the current book, and I recommend it as follow-up reading. I would also recommend Newmeyer (1998), which offers both depth and scope of treatment of formalism versus functionalism from the perspective of a committed but open minded formalist.

Introduction 3

References Carnie, Andrew and Norma Mendoza-Denton. (2003). ‘Functionalism is/n’t formalism: An interactive review of Darnell et al. (1999)’. Journal of Linguistics 39: 373–389. Curnow, Timothy J. (2002). [Review of the book Functionalism and formalism in linguistics]. Studies in Language 26(2): 505–512. Fischer, Olga. (2007). Morphosyntactic change: Functional and formal perspectives. Oxford: Oxford University Press. Fischer, Olga, Anette Rosenbach, and Dieter Stein (Eds.). (2000). Pathways of change: Grammaticalization in English. Amsterdam/Philadelphia: John Benjamins Press. Haspelmath, Martin. (2000). ‘Why can’t we talk to each other?’ [Review of the book Language form and language function]. Lingua 110: 235–255. Newmeyer, Frederick J. (1998). Language form and language function. Cambridge, MA: MIT Press.

1 D  efining ‘Formalism’ and ‘Functionalism’

1.1  Starting Out The terms ‘formalism’ and ‘functionalism’ name two contrasting stances that some language scholars assume, sometimes explicitly, sometimes only implicitly. Distinctively formalist versus functionalist approaches are recognizable in (for example) how linguists gather, analyze, and account for language data, how they design their research programs, and how they define the scope of linguistic inquiry. Employment of these two labels is relatively new, as is the notion that they are rivals within the same conceptual space. Nevertheless, each has a line of descent in the history of linguistics up to the present day. Chapter 1 provides some preliminary descriptions of what the two terms stand for. The remainder of this text illustrates, elaborates on, and sometimes challenges those descriptions. One might start by pointing out what ‘formalism’ and ‘functionalism’ are not. They are not necessarily self-consciously designated categories with relatively clear boundaries, like ‘Darwinian’ and ‘anti-Darwinian’. Transporting the issue into a different domain by way of illustration, ‘formalism’ and ‘functionalism’ have some of the properties of informal labels like ‘progressive’ or ‘centrist’ as those terms are used in political discussions. One could, for instance, dispute whether a particular policy counts as truly ‘progressive’ despite its supporters’ claim to that characterization. One could also argue whether another policy is or isn’t ‘centrist’ regardless of whether proponents position it on a continuum between ‘progressive’ and ‘centrist’; and many policies may not be recognizably either progressive or centrist. Nevertheless, these categories still serve a purpose, as do ‘formalism’ and ‘functionalism’. They help organize a complex, multi-dimensional, conceptual landscape and sharpen our perceptual and analytic powers. A second important caveat, and a second way in which formalism and functionalism differ from categories like ‘Darwinian’ versus ‘antiDarwinian’, is that they are not necessarily symmetrical relative to each other. Formalism brings into view language phenomena which

Defining ‘Formalism’ and ‘Functionalism’ 5 functionalism discounts or ignores and vice versa. The two approaches do sometimes address the same issue only to arrive at complementary conclusions. However, there are also instances where formalism and functionalism each independently pursues an objective, employing assumptions and conceptual tools that the other would consider alien or extraneous. In this sense, formalism and functionalism do not always compete side by side. Sometimes their contributions to our understanding of human language can be coordinated. At other times they are simply incommensurable.

1.2  Defining the Terms Curnow (2002: 506) quipped that formalists are scholars who are sure they know what ‘functionalism’ is, whereas functionalists are those who are confident that they can readily define ‘formalism’. The remark is light-hearted, but it calls attention to the difficulty each side may encounter in identifying a coherent core within its own practices and intellectual commitments. We will start by taking a step away from language to consider the sense of the words ‘formalism’ and ‘functionalism’ in other disciplines. Formalism in general—in art, architecture, literary criticism, logic, philosophy—emphasizes the structure and organization of the object of analysis, sometimes conceived as extracted out of its overlying material presentation. Formal logic translates a sentence into a skeleton of basic symbolic elements, removing whatever cannot be so represented. Formal literary criticism highlights the linguistic material and component parts of a text rather than its reception or historical role. Formal(ist) architecture is characterized by monumental symmetry, precise alignment of iterated geometric shapes, nested angles, subdued colors. Although the term takes on different senses in different domains, those approaches or styles labeled ‘formal’ typically look inward to the shape and organization of a phenomenon, often by abstracting away from its surface appearance and its relationships with neighboring phenomena. ‘Functionalism’, on the other hand, emphasizes the roles that phenomena play in their larger contexts, sublimating structure to function. In the first half of the twentieth century, functionalism in sociology turned the study of families, cultural practices, and institutions toward investigation of how these elements of a society contributed (or not) to the community’s integrity and stability overall. In architecture, the American Louis Sullivan (1856–1924) famously asserted that ‘form ever follows function’ (1896: 406), hyperbolically depicting that claim as ‘the pervading law of all things organic and inorganic’. Sullivan’s aphorism, conventionally cited as ‘form follows function’, was adopted by designers who deliberately shaped buildings according to their local environments and the uses to which people put them. For example, rather than flattening a hillside to accommodate the footprint of a

6  Defining ‘Formalism’ and ‘Functionalism’ library, functionalist architects may build the library into the hill; they may round the corners of an office tower on a busy street to facilitate the flow of pedestrians around it; they may extend the overhang above an entrance to a house to protect owners from the rain as they fumble for their keys, even as that extension interrupts the harmony of the building’s facade. In these ways, functionalism not only incorporates but prizes asymmetry and even idiosyncrasy as a response to external, contextually imposed demands. Overall, then, it may be fair to say that for formalists, form transcends function, or at least that they prioritize form over content or context. For functionalists, form derives from function in the sense that the role a phenomenon fills within its environment determines its internal structure. In linguistics, formalists analyze language as a system shaped by formal, structural rules. Formalists take for granted that close articulation of those rules (or constraints, principles, etc.) is the central task of language scholars. Fischer (2007: 54) wrote that for formalists ‘the system of grammar [is] more important as an object of study than the actual language data’ (emphasis in the original). Therefore, formalists characteristically look inward to explain linguistic phenomena: in some instances, this means that they explain a newly-observed fact of language according to how it fits within the configuration of already-established formal linguistic facts. In other instances, formalists look inward to explain facts according to their status in language change or acquisition. Functionalists, on the other hand, prioritize language data and prize detailed, contextualized, records of its use. They characteristically look outside language for explanations, under the assumption that languages are what they are because of the exigencies of human communication and cognition, or because of the external cultural environment in which language is used. To functionalists, the central task is to define how the shape of language data is connected to its communicative purposes and to human cognitive resources.

1.3  The Engineer and the Collector From these bare-bones definitions of our two key terms, it is clear that each is not the simple converse of the other. Although in some instances formalism and functionalism do look like paired opposites, it is more accurate to conceive of them as heterogeneous assemblies of traits which are independent of each other and only sometimes match up as mirror images across the formalist/functionalist divide. At the risk of over-simplification, the Appendix to this book compiles some of those traits, contrasting in one column the purported features of (at least some versions of) formalism to the purported features of (at least some versions of) functionalism in the opposing column. It is important to grasp, first, that the contents of the Appendix are not a summary, but only a collection of entry points; and, second, that the entries in the two lists characterize the two terms rather than define them, since not all specimens of formalism or functionalism exhibit all traits. An analogy

Defining ‘Formalism’ and ‘Functionalism’ 7 may help. Consider two lists that informally assemble the characteristics of animals versus plants: for animals, ‘can move’, ‘eats and digests’, ‘reproduces sexually’; for plants, ‘green color’, ‘has roots’, ‘requires sun’. The lists would have to be read with the understanding that, in fact, not every animal reproduces sexually, while many plants do so; and that there certainly exist both plants that are not green (e.g., Monotropa uniflora, also know as ‘ghost pipe’) and animals that are green (e.g., parrots). Analogously, formalists characteristically embrace abstraction, while functionalists are characteristically at home carrying out deep analyses of particulars and specific cases. But sometimes the converse relationships may hold, and sometimes neither abstraction nor a focus on particulars is relevant to work that is otherwise recognizably formalist or functionalist. The contents of the Appendix will be filled out, explored, and sometimes disrupted, in the remainder of this text. In the meantime, it may be useful to fuse each of these sets of traits into two images, each one representing metaphorically the approach to analysis of language data taken by formalists versus functionalists. Representing formalism, picture an engineer’s framework for a building silhouetted against the sky, comprised of idealized geometrical shapes iteratively embedded within each other. That framework is the bare skeleton that underlies a structure, stripped of its recognizable accoutrements such as plaster, paint, woodwork, and all manner of surface treatment. It is modular in the sense that its sub-parts can be separated, analyzed as independent units, and assembled piecemeal to comprise the whole. A viewer is struck by the symmetry of multiple parallel lines and angles and by the abstractness of the total ensemble relative to the daily experience of encountering walls, doors, or stairways. Likewise, formalist linguists foreground the skeleton of a language, the patterned substructures of its parts, and the immaterial, sometimes only hypothesized, relationships that hold among those parts in a manner that looks beyond how we encounter them in their normal working environments. Representing functionalism, picture what was called in the eighteenth century a ‘cabinet of curiosity’, a miniaturized precursor to the modern museum of natural history. Returning travelers would assemble for display their collected mementos from abroad, sometimes variously comprising a shell, a rock, a taxidermy specimen, a human artifact, the fragment of a bone, a preserved flower or insect. Typically, each sample sits in its own box or enclosed niche or on a shelf or drawer comprised of multiple small spaces each dedicated to a single item. Because the logic behind the juxtaposition of items is not necessarily obvious, this style of exhibition discourages facile generalizations. Instead, it seems to communicate that every specimen is worthy of attention and holds a unique place in the total collection. A viewer moves freely among them, speculating on the role played by each item in the human or natural ecology that produced it. Likewise, functionalist linguists observe and curate language phenomena, inspecting them for evidence of

8  Defining ‘Formalism’ and ‘Functionalism’ how they interact and making sense of their material properties with reference to the totality of the superstructure which, together, they constitute. In the terms of this metaphor, a formalist scholar of language resembles a linguistic engineer, who designs models of the inner framework that supports human language. A functionalist language scholar resembles a linguistic collector or curator, who scrutinizes samples of human languages with the goal of understanding how they accomplish their communicative roles. The metaphorical images of engineer versus collector probably overstate differences between formalists and functionalists. However, these images have the virtue of not forcing the one into a mirror-image mold of the other, since what an engineer accomplishes is not undone, nor is it performed in reverse, by what a collector accomplishes. The two images have the additional virtue of problematizing any facile initiative to combine formalism with functionalism. Although the work of an engineer and of a collector are not incompatible per se, to harmonize the two in any meaningful way without undermining their differences would require cautious and respectful treatment of them both and deep understanding of the attributes of each one.

1.4  Two Examples Two recent non-technical books that introduce linguistics to the public exemplify contrasting formalist versus functionalist approaches in modern study of language. The two texts differ starkly—in their goals, their rhetoric, and their style of argumentation. One is short, one long. One author writes from the center of modern formalism; the other author’s functionalism is eccentric. Both books are mostly taken up with exposition of their authors’ positions, with only one giving much attention to the opposing point of view. However, they are comparable in the sense that each takes a thoroughgoing, uncompromising stance that sharply contrasts with the stance taken by the other. Surveying them side by side provides a useful microcosm of how one variety of modern formalism differs from one variety of modern functionalism, with the two situated intellectually so that the distance between them is large. Andrea Moro’s Impossible Languages (2016) represents a formalist point of view. Affiliated with the Institute for Advanced Study at the University of Pavia, Moro (b. 1962) is a linguist and neuroscientist whose work has been consistently informed by generative theory (e.g., Chomsky 1995). He begins by fearlessly announcing that ‘To define the class of possible languages: this is the ultimate aim of linguistics’ (p. 1). Moro argues that there are biologically based limits to the structure of a human language, which make sense of the notion of an ‘impossible’ language. An impossible language would be one that could be shown not to share the ‘unique, intricate set of principles with relatively little variation, producing a surprising constellation of properties’ (p. 5) asserted to be attested in all languages. Moro’s book is an extended argument that, in fact, impossible languages

Defining ‘Formalism’ and ‘Functionalism’ 9 do not exist because those ‘unique, intricate’ principles necessarily constrain the structure of every natural human language. Moreover, principles of language structure cannot, according to Moro, be discovered on the basis of simple observation of the surface forms of sounds and words. Rather, they are derived by theory-driven analysis of what child language learners know about the structure of language, despite the absence or inadequacy of evidence in their environment that could lead them to that knowledge, that is, despite ‘the poverty of the stimulus’ (Chomsky 1980). Moro’s approach to the study of language is consummately formalist. A brief aside on page 2 spells this out. He acknowledges that alongside their formal and physical properties, ‘human languages must be endowed with other general properties that make them not only possible but usable’. He specifies that those properties include the capacity to carry meaning and serve communicative ends—then he pivots to remark flatly that ‘These properties are trivial’ (p. 2). With that in place, Moro’s text goes on to exhibit many of the hallmarks of formalism. For example, the linguistic data Moro cites is highly stylized, in the sense that he engineers it to demonstrate specific language properties rather than to depict actual speakers’ actual usage (e.g., Who do you think that Mary wants to describe themselves?, p. 23; This fact that John runs surprises me, p. 29). Moro identifies these data dichotomously as either syntactically well-formed or ill-formed, without reference to any context of their use in speech or writing. It is also significant that Moro consistently makes bold synthetic claims, then seeks support for those claims in highly abstract analyses of language data. For instance, he argues that the complex syntactic properties of all languages can be attributed to the operation of two functions, namely recursion and locality (pp. 34ff). Later, Moro introduces experimental evidence that the acoustic waves generated by speech correlate with the electromagnetic waves generated by silent reading. On this basis, he speculates that recursion and locality can be united in being derived from a code in which sound waves and brain waves converge, amounting to an unexpected ‘unification of the physical properties of language with its formal ones’ (p. 96). Moro’s idealization of linguistic data and his facility with abstraction are recognizably formalist in spirit. The specifically generativist orientation within the formalism of Impossible languages is evident as well, for example, in that Moro presupposes the existence of a cognitive faculty specific to human language (p.  59) and in that he consistently compares the work of linguists, their methodology, and their results to those in the natural sciences, including chemistry (p. 17), immunology (p. 19), mathematics (p. 27), neurobiology (pp. 45–59), and physics (pp. 63–70). It is also characteristic of generativist formalism to privilege the ease and speed with which children come into language as an especially consequential source of insight into the nature of human language (pp. 13–15)—or, in Moro’s terms, into the non-existence of impossible languages.

10  Defining ‘Formalism’ and ‘Functionalism’ In these ways, Moro’s approach and the focus of his project itself—to define what would count as an impossible language—are demonstrably formalist, the work of a linguist-engineer. Moro’s book sharply contrasts with Daniel L. Everett’s 2012 Language, the Cultural Tool. Everett (b. 1951) is a controversial American anthropological linguist whose research on Amazonian languages, especially Pirahã (spoken by a few hundred people in the center of the Brazilian Amazon basin) has attracted attention in both academic and popular media. As with Moro, Everett’s title establishes his theme: ‘language is a tool for us and designed by us. And we set the rules for the interpretation and use of language’ (p. 124; emphasis in the original). Not all functionalists would agree with the precise stance he takes, but Everett is still as recognizable as a functionalist as Moro is as a formalist. Everett devotes more attention to the opposing point of view than does Moro, so that Language, the Cultural Tool is about three times the length of Impossible Languages. Everett’s book has met with more public attention than Moro’s in the non-specialist press (Radford 2012; Tomlinson 2012; Williams 2012) and from linguists (McWhorter 2012; Enfield 2013). It ranges widely over many topics in linguistics and linguistic anthropology of high interest to the general public, including the Sapir-Whorf Hypothesis; color and kinship terms; child language learning; animal communication; garden path sentences; the foibles of prescriptive grammar; creole languages; differences between icon, index, and symbol; rare articulatory gestures like linguolabials, ejectives, or implosives; and so forth. Everett exuberantly displays this diverse collection of language concepts, claims, and practices in the service of illustrating his theme of the mutual influence between language and culture. In this, Everett’s emphasis is less on language as a general human phenomenon as it is on specific languages and their associations with specific groups: kinship terms and incest taboos in Hawaiian (pp. 246–248); invention versus cultural borrowing in Sequoya’s creation of the Cherokee syllabary (pp. 273–276); default feminine gender marking on nouns in the Amazonian language Banawá in the context of segregation by sex in Banawá culture (pp. 207–210). The centerpiece of Everett’s collection of linguistic marvels, however, is Pirahã language and culture. His presentation is based on deep personal experience with speakers of Pirahã, gleaned from years of living among them. Throughout the book, he uses the features of Pirahã to illustrate the nexus of language, cognition, and culture. Everett reports, for instance, that the Pirahã language lacks words that mark temporal and aspectual differences: verbs don’t encode either the time reference of an event (ate dinner; will eat dinner) or its character as ongoing or completed (was eating; has eaten). He sees this as directly reflecting Pirahã speakers’ radical here-andnow cognitive orientation, which entails a lack of conceptualization of the passage of time. They live in ‘a society in which members sleep, eat, hunt, fish, and gather without regard to time of day, day of the week, week of

Defining ‘Formalism’ and ‘Functionalism’ 11 the month, or month of the year’ (p. 269). Likewise, Everett comments on the extensive practices of reciprocal physical grooming among the Pirahã, which co-exist in their culture with a relative dearth of phatic utterances, that is, the ‘linguistic grooming’ that (for example) speakers of English engage in with small-talk exchanges like ‘How are you?’/‘Fine, and you?’ or ‘Bye now’/‘See you later’ (pp. 236–239). Everett interprets these observations as evidence for how linguistic form conforms to culturally-embedded linguistic function. His treatment of data is similarly functionalist, on several levels. He provides long passages of transliterated Pirahã reproducing connected speech (pp. 278–279; 295–297), introduced with the specifics of the who, when, where, and why of the attested language. Everett’s ‘bottom-up’ inductivist stance is even inscribed in the famous story of how he first approached the Pirahã people, adverted to in Language, the Cultural Tool and fully developed in Everett (2008). Everett arrived in Brazil as a missionary whose goal was to learn Pirahã for the purpose of Christian evangelization. He had had some training in applied linguistics to prepare him to translate the Bible. As time passed, however, Everett came to appreciate and value Pirahã culture, so that his orientation shifted from religious conversion to the effort to understand both their language and their culture. He then acquired the intellectual tools to do advanced linguistic research by completing a doctorate in linguistics at the University of Campinas in São Paulo. In Everett’s biography as in his intellectual priorities, the data came first and then the theory followed to help make sense of the data, rather than vice versa. To sharpen the contrast between Moro and Everett, it may help to assess their treatments of a topic that comes up in both books: localization of language in the brain. Both scholars give compressed versions of the famous story of French surgeon and physical anthropologist Paul Broca’s 1861 autopsy of a man with aphasia, which led Broca to identify the physical locus of human language in the third frontal convolution of the left hemisphere (‘Broca’s area’: Moro 2016: 48–49; Everett 2012: 74–75). Both also caution readers against too literal a version of the claim that grammar resides exhaustively in a single identifiable location. But Moro and Everett present the complexities that attend localization of language in the brain quite differently. Moro suggests that rather than being the seat of the language faculty, Broca’s area may resemble a hub airport in the world of air travel: not a destination in itself, but the crucial place where many lines cross in a multiplex system. Moro narrates a series of neurolinguistic experiments which compared blood flow to Broca’s area when subjects tried to learn one of two facets of an invented language: rules that were recursive, and which resemble those of natural languages, versus rules that were linear and non-recursive (e.g., ‘form a question by inverting the order of words in a sentence’), which presumably resemble hypothesized impossible languages (pp. 55–59). The results showed increasing involvement in Broca’s

12  Defining ‘Formalism’ and ‘Functionalism’ area as subjects learned recursive rules and decreasing involvement as they learned linear, ‘impossible’, rules. Although Moro warns that his results do not resolve the exact role of Broca’s area, he asserts that those results do show that ‘the distinction between possible and impossible languages is embodied in the brain’, a finding which denies ‘the alleged arbitrary, cultural, and conventional character of human languages’ (p. 59). In these ways, Moro introduces the classic notion of brain localization; retreats from its strong version; then reports an empirical study that assumes a nuanced relevance for Broca’s area. That study employs recognizably formalist methods and constructs. Its results confirm Moro’s formalist commitments. Everett likewise adheres to his functionalist commitments. He frames brain localization research in a more skeptical light, emphasizing that ‘the evidence is far from clear’ (p. 73) and labeling the reputation of Broca’s area as ‘part of urban and academic folklore’ (p. 74). Everett lists five objections to investing Broca’s area with a unique linguistic role but stops short of denying that it has any independent existence or relevance to language: ‘everyone agrees that there are regions of the brain involved in language’ (p. 75). He goes on to cite several studies showing that when blind persons read Braille, they activate parts of the brain associated with the processing of visual stimuli (p. 76). He interprets this finding as evidence that, first, the external fact of being blind bears on brain function, and second, that language processing is not confined to identified language centers. Everett also refers to research which reported that children struggling to learn to read grow new brain material in as little as six months of intensive training (p. 78), a finding he takes to suggest that the structure of the brain is susceptible to environmental experience. He concludes that ‘brain specialization and anatomy can be influenced by culture’ (p. 78)—a functionalist’s implicit rejection of Moro’s assertion that the brain determines language. It is salient that neither Moro nor Everett adverts to the research that the other party brings to light.

1.5 On the Architecture of Formalism Versus Functionalism It is worth noting from the outset that modern formalism is strongly centripetal, if not isomorphic, while modern functionalism is much more internally diverse, so that no single example can stand for the whole. Our initial case study foreshadows this characteristic difference in the internal structure of the two approaches, a matter to which Chapters 3 and 4 will return. Contemporary generative grammar, Moro’s foundation, is the dominant present-day version of formalism. Although Moro doesn’t represent the whole of formalism, he writes from inside its modern center. In contrast, Everett writes

Defining ‘Formalism’ and ‘Functionalism’ 13 from the edge of one of several varieties of functionalism. His insistence that language is a cultural tool seems to presuppose, rather than directly assert, that ‘form follows function’ in the sense that what language is a tool for is culturally embedded communication. In commentary on his 2012 book that addresses this point, Everett labeled his style of linguistics as compatible with functionalism but distinct from it (2013: 647). Consulting the composite characterization of functionalism in the Appendix shows that Everett’s assumptions and working habits are strongly functionalist: inductive; restrained in the search for generalizations; attentive to speakers’ performance rather than their internal states and to language differences across sub-groups of speakers. Moro, the linguistengineer, and Everett, the linguist-collector, do not hold comparable positions within their respective formalist versus functionalist domains, nor do those domains have commensurable internal architecture. Still, we can start from here.

References Chomsky, Noam. (1980). Rules and representations. New York: Columbia University Press. Chomsky, Noam. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Curnow, Timothy J. (2002). [Review of the book Functionalism and formalism in linguistics]. Studies in Language 26(2): 505–512. Enfield, N. J. (2013). ‘What I’m reading: Language, culture, and mind: Trends and standards in the latest pendulum swing’ [Review of the book Language: The cultural tool]. Journal of the Royal Anthropological Institute 19: 155–169. Everett, Daniel L. (2008) Don’t sleep, there are snakes: Life and language in the Amazonian jungle. New York: Pantheon. Everett, Daniel L. (2012). Language: The cultural tool. New York: Random House. Everett, Daniel L. (2013). ‘The state of whose art? Reply to Nick Enfield’s review of Language: The cultural tool’. Journal of the Royal Anthropological Institute 19: 645–648. Fischer, Olga. (2007). Morphosyntactic change: Functional and formal perspectives. Oxford: Oxford University Press. McWhorter, John. (2012, 8 April). ‘Repeat after me’ [Review of the book Language: The Cultural Tool]. The New York Times (p. BR16). Retrieved from www. nytimes.com/2012/04/08/books/review/language-the-cultural-tool-by-daniel-leverett.html Moro, Andrea. (2016). Impossible languages. Cambridge, MA: MIT Press. Radford, Tim. (2012, 15 March). [Review of the book Language: The cultural tool]. The Guardian. Retrieved from www.theguardian.com/books/2012/mar/15/ language-cultural-daniel-everett-review Sullivan, Louis H. (1896). ‘The tall office building artistically considered’. Lippincott’s Magazine 57(3): 406.

14  Defining ‘Formalism’ and ‘Functionalism’ Tomlinson, Antony. (2012, September/October). [Review of the book Language: The cultural tool]. Philosophy Now 92(2). Retrieved from https://philosophynow. org/issues/92/Language_The_Cultural_Tool_by_Daniel_Everett. Williams, R. (2012, 2 May). ‘Review of Daniel Everett’s new book Language: The Cultural Tool’ [web log comment]. Retrieved from https://philosophyandpsychol ogy.wordpress.com/2012/05/02/review-of-daniel-everetts-new-book-languagethe-cultural-tool/

2 Background to the Current Debate

2.1  Is There a Starting Point? Recalling that the lists in the Appendix characterize, rather than define, formalist versus functionalist linguistics, one might ask whether this juxtaposition of traits still makes sense when we look at linguistics before the present day. That is to say, is it only in contemporary study of language that we can offset, say, formalists’ embrace of abstraction; the priority they place on theory construction; or their engineer-like attention to the inner framework of language, to functionalists’ efforts to coordinate language forms and functions; their collector-like scrutiny of particulars and specific cases; and their restraint in matters of theory? To what extent did earlier study of language associate subsets of these traits into two internally coherent approaches, dissociated externally to face each other? Or, is it even possible to discern what preceded modern formalism and functionalism in these terms? Remarkably, there has been little reflection on whether the confrontation between formalism and functionalism has much of a past. If we take the terminus a quo to be self-conscious recognition that an approach to language labeled with the term ‘formalist’ stands in opposition to another labeled ‘functionalist’, then tension between the two may only go back to the 1960s or 1970s (Graffi 2001: 389; Carnie & Mendoza-Denton 2003: 375). However, this account of the debate casts functionalism as a counterpoint to the emergence in the United States of Noam Chomsky’s distinctively formalist generative theory. That account truncates the history of both formalism and functionalism. The coinage of these terms as a pair of opposites—with their rewarding alliteration and near-matched metrical structure—may be a product of the late twentieth century, but the ideas about language that they label preceded them. A partial exception to the tendency to neglect the historical backdrop goes a little further back in time to identify an antecedent to functionalism in the 1930s work of the Prague Linguistic Circle, to be discussed below. For example, Newmeyer’s (1998; 2001) exposition of formalism versus

16  Background to the Current Debate functionalism adverts to Prague Circle linguistics. On the formalist side, however, Newmeyer grounds the debate in the emergence of generative grammar (1998: 7–9), obviating the question of whether there was any pre-generativist, but still formalist, style of language study that acted as a principled alternative to functionalism before the middle of the twentieth century. With the exception of an aside in Battistella (2000: 431), reviews of Newmeyer’s book seem to take his stance for granted, and comment on his theme of conflict between formalism and functionalism as if it had no relevant antecedents (e.g., Carstairs-McCarthy 1999; Haspelmath 2000; Morvacsik 2000; Tallerman 2000; Foolen 2002; Francis 2002). In narrowly modern terms, formalism studies language as a formal object without reference to its role in communication, while functionalism insists that communication imposes such an essential shape on language that it cannot be studied meaningfully without reference to its function. In this text, I identify formalism and functionalism in terms that are somewhat broader than those that have currency right now; by those lights, both points of view have precedents long before the 1960s or 1970s. It goes beyond the ambitions of this text to narrate the full, multi-dimensional story of the development of formalism and of functionalism, assuming that that is even achievable. Still, identifying some of the ancestors to formalism and functionalism illuminates the scope of the controversy between the two. Chapter  2 first sketches some pre-twentieth-century ideas about the nature of language that serve as the distant backdrop to the modern debate. After that, we turn to identifiably formalist linguistics from 1900 up to the late middle of the twentieth century, then likewise for functionalist linguistics.

2.2  Before the Twentieth Century It is worth noting that those few publications which swim against the tide to take a longer-term view of the history of the debate (e.g., Bates & MacWhinney 1982; Dirven & Fried 1987; Croft 1993; Givón 2013) are the work of functionalists rather than formalists. Functionalists seem more willing than formalists to inquire into their own forerunners and more open to viewing their work as part of a stream of research, as opposed to emphasizing the novelty of their ideas about language. By the light of this fact, it is striking that one of the very earliest studies of language cited as part of the general background to western language science has a distinctively formalist cast: the Sanskrit scholar Pāṇini’s Aṣṭādhyāyī, created around 500 bce in what is now Pakistan (Katre 1987; Cardona 1988). The Aṣṭādhyāyī is an arch-economical exposition of the grammar of Sanskrit, written in the form of about 4,000 pithy ordered statements, or sūtras, which distill the complex facts of Sanskrit morphophonetics and morphosyntax. Pāṇinian sūtras, designed to be memorized, communicate

Background to the Current Debate 17 the grammar of Sanskrit in a uniquely parsimonious style which pares away any intrusion of redundancy or ambiguity. Separate from the Aṣṭādhyāyī but traditionally also attributed to Pāṇini are three ancillary texts: an ordered inventory of the sounds of the language; a catalogue of verbal roots; and a list of nominal stems and other lexical items to which the sūtras refer. The goal of the Aṣṭādhyāyī is to specify the forms of the language through the operation of interacting layers of rules and metarules. The system acts on roots to produce abstract intermediate forms which are modified, step by step, into the recognizable words of Sanskrit while non-occurring forms are eliminated. Pāṇini’s grammar also accounts for certain facets of dialectal variation but without losing its focus on form or expostulating on topics such as the scope of language variation or the significance of a speaker’s choice of one speech style over another. Throughout, the Aṣṭādhyāyī models a commitment to the systematicity of grammar as a set of formal operations that are fully tractable to analysis. Indian scholarship has highly valued the depth and intricacy of Pāṇini’s Aṣṭādhyāyī and responded to it with centuries of commentary and exegesis, much of which has itself been the basis of further commentary and analysis. When Europeans discovered the text and its commentarial tradition in the late 1700s, it gave westerners access to the grammar of Sanskrit and demonstrated some of the remarkable patterned complexities of the language. Eventually, the great finesse and sophistication of Pāṇini’s work dawned on scholars from Europe (Staal 1972). Once they recognized the historical relationship of Sanskrit to Latin and Greek, analysis of Sanskrit became a cornerstone in the reconstruction of the Indo-European language family, the major preoccupation of the study of language in Europe from the late 1700s through the 1800s (Morpurgo Davies 1998). Among later-generation students of Pāṇini’s grammar was the leading American formalist Leonard Bloomfield (1887–1949), a long-time admirer of the Aṣṭādhyāyī (Bloomfield 1927, 1929). Bloomfield held in high esteem Pāṇini’s descriptivism, his concise style, and his goal of comprehensive and rigorous analysis of every form of the language. Bloomfield self-consciously emulated Pāṇini in his work on the Algonquian language Menomini and in his classic 1933 text Language (Emeneau 1988). Most importantly for our concerns here, Bloomfield’s Menomini Morphophonemics (1939) embraced Pāṇini’s adoption of abstract forms and his reduction of the complexities of a natural grammar to the output of a set of ordered operations—both hallmarks of formalism. The formalist complexion of the Aṣṭādhyāyī is enhanced by its separation of the rules governing morphophonology and morphosyntax from lists of roots and sounds, a move that seems to foreshadow the notion of modularity in the organization of a grammar. Among Pāṇini’s contemporary admirers is Noam Chomsky, who identifies the Aṣṭādhyāyī as a generative grammar (Chomsky 1965: v; Dillinger &

18  Background to the Current Debate Palácio 1997). Unlike Bloomfield, who acknowledged modeling his work on Pāṇini’s grammar, Chomsky does not claim a direct line of descent from Pāṇini to generative grammar. Rather, Chomsky views generative grammar as an independent innovation that shares some surprising common ground with ancient Indian grammatical analysis. In these ways, Pāṇini’s work prefigured some traits of modern formalism. Among other ancient-world scholars who contributed to the study of language, Plato (fourth century bce) and Aristotle (384–322 bce) are prominent, although neither wrote a full-scale grammar. Granted the breadth and complexity of both philosophers’ ideas and their pervasive impacts on western culture, it is not obvious how to map them onto either formalism or functionalism—or even whether that mapping is fruitful. Nevertheless, one can’t miss that in the writings of modern formalists the name of (‘idealist’) Plato is much more likely to surface than that of Aristotle (who famously observed and collected all manner of natural phenomena)—and vice versa among functionalists. For example, Chomsky (1986) popularized the expression ‘Plato’s problem’ with reference to a passage in Plato’s dialogue Meno in which Socrates elicits the principles of geometry from an uneducated slave boy. Chomsky’s point is that child language learners can be shown to have surprising depth of understanding of grammar which is neither taught nor demonstrated to them: like Plato’s slave boy, they know more than can be readily accounted for on the basis of their outward experience. On the opposite side, the prominent ‘west coast functionalist’ Talmy Givón (b. 1936) routinely cites Aristotle’s articulation of the core notion of functionalism in biology, namely, that the forms of living beings match their functions, so that, for example, birds that live in swampy areas have long legs and webbed feet, while those that consume nuts have short, sturdy beaks. For Givón, the best point of departure for functionalism resides in biology, a discipline into which Aristotle infused a functionalist perspective over two thousand years ago (2002: 1–7, 2013). In these ways, both Chomsky and Givón find warrants for certain facets of their views about language by looking backward to ancient Greek scholarship. In addition, Plato’s most extensive treatment of language matters, his famous dialogue Cratylus, centers on an issue that divides modern formalists from functionalists. Cratylus is a meditation on the question of whether the shapes and sounds of words are intrinsic to their meanings, or whether the association of words to meanings is merely conventional (Joseph 2000). It is framed as a conversation between Socrates and two collegial adversaries who hold opposite positions on the nature of words and meanings. One is the conventionalist Hermogenes, who asserts that words are connected to meanings arbitrarily by human habit or custom. The other is Cratylus, who argues that the relationship of words to meanings has natural force, with sounds intimately connected to meanings. Plato’s dialogue works these ideas out in the cultural and intellectual terms of ancient Greece, many of

Background to the Current Debate 19 which are understandably alien to modern language scholarship. Overlooking the centuries that separate Plato’s day from ours, there is an echo of Cratylus in modern formalists’ presupposition that words and forms are arbitrarily connected to their meanings, as opposed to functionalists’ commitment to iconicity as an important force in phonology, word-formation, and grammar. On most readings, Plato does not settle the issue in Cratylus. Contemporary debate about iconicity does not invoke Plato’s authority, but it is salient that this facet of what divides modern formalists and functionalists has such a long history. Much more could be written about the development, between the third or fourth century bce and the beginning of the twentieth century, of the ideas about language relevant to modern formalism and functionalism. One landmark might be Chomsky’s historiographical research centering on Cartesian linguistics (1966), a potential counter-example to formalist disinclination toward the history of linguistics. However, Chomsky’s concern in this problematic text is with Cartesian rationalism, which he contends shares some of the characteristics of generative theory (Thomas 2009). Whether or not Chomsky’s reading of seventeenth-century texts is persuasive, rationalism is not inherent to formalism, so that Cartesian Linguistics makes little contribution to identifying the ancestors of formalism. Moreover, Chomsky is virtually alone among generativists in his interest in historical matters. Modern generativists sometimes cite Chomsky’s historiography, but few either address the critical response to it or independently probe earlier versions of formalism. Taking up the thread again at the beginning of the 1900s, the Swiss linguist Ferdinand de Saussure (1857–1913) is conventionally identified as a turning-point between nineteenth-century historical-comparative linguistics and modern structuralist linguistics, which conceives of language as a complex relationship of systems. In this sense, Saussurean structuralism pervades all modern linguistics up to the present day. However, the central exposition of Saussure’s linguistics, the posthumously published Course in General Linguistics (1916 / 1983, the ‘Cours’) as assembled by his students, presents formidable hermeneutic challenges, so that it is not clear how his ideas have affected the debate between formalists and functionalists. Dirven and Fried (1987: xi) imply that Saussure’s influence is felt more broadly among functionalists but deny that he can be identified with either side. Joseph (2002) has studied how Bloomfield and Chomsky (two formalists already part of our narrative) each positioned themselves relative to Saussure’s linguistics. In both cases, Joseph finds a complex record of selective readings and mis-readings of the Cours and shifts in positions over time. Among functionalists, Givón (1995) stakes out a strong anti-Saussurean position. He considers three of Saussure’s trademark contributions to have strengthened formalism at the expense of functionalism: the Saussurean doctrine of arbitrariness ‘detached the linguistic sign—the visible behavior—

20  Background to the Current Debate from its invisible mental correlates’; Saussure’s distinction between langue and parole idealizes away the rich, indeterminant, and essential parts of language; and Saussure’s segregation of diachrony from synchrony similarly dismisses the complexities of variation and change which functionalists find valuable (1995: 5–7). Not all functionalists would agree with Givón, nor do all formalists accept the full scope of Saussure’s ideas. Nevertheless, those ideas have helped each group define its own commitments.

2.3  Twentieth-Century Formalist Linguistics to 1950 Dirven and Fried (1987: xi) present an intriguing graphic display summarizing their view of the separate descents of formalism and functionalism from 1900. On the functionalist side, they link Saussure by lines of influence to multiple schools of European structuralism: Geneva, Prague, London, and the Netherlands, each of which heads a list of adherents. (Note that with this, Dirven and Fried counter Givón’s reading, which puts Saussure at odds with functionalism.) The formalist side is comparatively underpopulated, with Saussure’s influence extending to a single school, that in Copenhagen. An independent line of descent of formalism runs parallel, but is unlinked to Saussure, labeled ‘Descriptivism’. Under this heading the American anthropologist Franz Boas first subsumes ‘Bloomfield etc.’. Chomsky’s generative grammar follows under Bloomfield, with ‘Cognitivism’, identified with George Lakoff and Ronald Lang­acker, conceived as a second generation below Boas. Kenneth Pike is entered off to the side, as a non-Boasian formalist. Dirven and Fried’s diagram could be criticized from several angles. Many would dispute the classification of Lakoff, Langacker, and Pike as formalists; Chomsky would reject the implication that his work descends from that of Bloomfield; and the absence of any role for Boas’s prominent student Edward Sapir is conspicuous. But putting these reservations aside, Dirven and Fried’s attribution of twentieth-century formalism to American scholars is striking, as is the priority they give to Boas. Franz Boas (1858–1942) was trained in physics and geography, then left Germany for what is now the Canadian territory of Nunavut in 1883 to research the cultural effects of extreme cold weather. On site, he was captivated by the culture and languages of the Inuit people. Despite having no formal training in linguistics, Boas eventually made a career out of studying Native American languages, especially those of the Pacific Northwest. He developed a working style that proved very influential: in direct face-to-face interactions with a speaker, Boas elicited and transcribed oral narratives, then translated them word-by-word. The output of his method was a description of the grammar of the target language arrived at by close scrutiny of the transcribed corpus without relying on the features or analytic categories of any other language.

Background to the Current Debate 21 Boas is now considered one of the founders of American anthropological linguistics. Both his methods and his focus on the study of Native American languages had lasting impact. Boasian insistence on inferring a grammar from the output of a speaker, meticulously recorded and analyzed, provided one of the building blocks of what became American structuralist linguistics. The term ‘structuralism’ is a fraught one (Hymes & Fought 1981). Even restricting its use to the context of the study of the language in the United States, it cannot be identified in any facile way with the term ‘formalism’. Nevertheless, American structuralist linguistics has a formalist orientation. Some of that character is registered in Boasian empiricism, but Boas also reflected at length on language, culture, and thought (1911), so that Hymes (1961: 90) characterizes Boas as ‘clearing the way for, but not quite occupying the ground of, his structurally-minded successors’. Among Boas’s most prominent successors was Leonard Bloomfield. Bloomfield was never a direct student of Boas, but he inherited Boasian methods and brought them together with other intellectual forces to create a distinctive and highly influential style of linguistics. One of those forces was his admiration for Pāṇini; another was his drive to establish an autonomous discipline of linguistics in the United States that was recognizably a science of language. With the prestige of science ascendant, Bloomfield assumed both institutional leadership as a founder of the Linguistic Society of America, and intellectual leadership in striving to replace philosophical speculation about language, mind, and culture with an emphasis on attested facts. In early mid-career, Bloomfield adopted behaviorism, a school of psychology in which, applied to language, the goal was to eliminate unobservable ‘mentalistic’ concepts like ‘mind’, ‘memory’, ‘thought’. He favored the terms ‘mechanistic’ for his preferred style of analysis, and ‘descriptivism’ for his approach overall. In this way, Bloomfield prioritized observable forms, and called attention to his principled reticence to approach language from the point of view of its role in communication. A short 1926 article published in Language, ‘A Set of Postulates for the Science of Language’, illustrates Bloomfield’s formalism. On the model of mathematics, he attempted to specify a set of axioms—definitions of terms and assumptions following from those definitions—on which basis a linguistics independent of anthropology, acoustics, physiology, and (in particular) psychology could be built. Bloomfield’s postulates concentrate on defining the forms of language and the characteristics of those forms, with meaning conceived of as a ‘recurrent stimulus-reaction which corresponds to a form’ (p. 155). Contrary to his reputation, Bloomfield did not reject analyses that took recourse to meaning, either the meaning of words or the meanings of constructions. Rather, he considered semantics an underdeveloped subfield, so that (in his view) the scientific basis of linguistics should start with the material forms of language, rather than what those forms communicated. He wrote: ‘linguists, confronted with the parallelism of form and meaning,

22  Background to the Current Debate choose form as the basis for classification’ (1926: 157). With this, Bloomfield implicitly rejected the functionalist maxim ‘form follows function’. Bloomfieldian formalism and Bloomfield’s drive to create a science of language had a major impact on linguistics in early twentieth-century America. A group of his students and followers that goes by the name of the ‘post-Bloomfieldians’ developed a distinctive body of formalist work that carried forward some, but not all, of his ideas. Although they inherited Bloomfield’s conception of linguistics as a science of language, they did not adopt behaviorism. The post-Bloomfieldians extended the prioritization of form over meaning beyond Bloomfield’s reticence with semantics, to develop a style of linguistic analysis that some called ‘distributionalism’. Distributionalism attempted on principled grounds to describe linguistic structure solely with reference to the co-occurrence of linguistic units. For example, a distributionalist would declare that, in English, /p/ and /b/ represent independent phonemes not because a pen is where farm animals live and Ben is a man’s name, but rather because initial /p/ is aspirated and /b/ is not aspirated, and because the sequence /mp/ appears as a syllabic coda, whereas /mb/ does not. A theoretical innovation of one of the post-Bloomfieldians demonstrates their formalist orientation. A paper by Charles Hockett (1954) explored what he called ‘two models of grammatical description’: ‘item-and-process’ versus ‘item-and-arrangement’. In the former, two linguistic units might be compared as outputs of differing processes that operate on the same root, such as suffixation of the word chat to yield chatting, versus reduplication to yield chitchat. In contrast, an ‘item-and-arrangement’ approach would compare the two units without presupposing that they are derived from the same root: in one unit, a free form (chat[t]) precedes a bound morpheme (-ing); in the other unit, two near-identical free forms (chit; chat) occur adjacent to each other. Hockett even-handedly assessed the strengths and weaknesses in both approaches, even as he found item-and-arrangement more tractable because he considered it to have been more thoroughly formalized—and to Hockett, formalization of a theoretical tool was a top priority. Another post-Bloomfieldian innovation illustrates how far some distributionalists pushed their style of formalism. ‘Immediate Constituent Analysis’ was an analytic technique they developed to display the internal structure of words and groups of words as nested, hierarchically organized constituents. In Hockett’s example (1958: 152), the string a man are (in The sons and daughters of a man are his children) would not form an immediate constituent on the basis of the observation that alone it constituted only ‘a meaningless sequence of morphemes’. In contrast, Gleason (1961: 132) distrusted reliance on ‘uncontrolled intuition’ which he felt was implied in this way of defining constituent boundaries. Gleason proposed alternative

Background to the Current Debate 23 criteria for establishing constituent boundaries, based on substitutability (a man is a constituent because it can be substituted by he, whereas there is no obvious substitute for a man are) or freedom of occurrence (the utterance a man can stand alone, but not so a man are). Gleason’s version of Immediate Constituent Analysis comes closer to the spirit of distributionalism, but both Gleason and Hockett take formalist approaches, designed to meet a formalist goal, namely the identification and organization of units and sub-units of language. Notice, for instance, how devoid these techniques of analysis are of reference to speakers’ communicative intentions. In Chapter 3 we will return to the development of the legacies of Bloomfield and the postBloomfieldians, to bring out both continuity and innovation in twenty-first century formalism.

2.4 Twentieth-Century Functionalism from the Prague Circle to Pike While the study of language evolved in the United States to meet the specifically formalist goals of early twentieth-century American science of language, in Europe functionalism flourished. Locally based scholars in Geneva, Prague, London, and the Netherlands emerged and developed their own working styles. Saussure’s Cours was a major stimulus, although each group defined its own position relative to Saussurean doctrines, and relative to the powerful model of Prague Circle functionalism. It is worthwhile keeping in mind from the start that some participants in early twentieth-century linguistics in Europe and the U.S. disrupt this general scenario. In Denmark, for example, the Linguistic Circle of Copenhagen was founded on the direct inspiration of the functionalist-oriented Prague Linguistic Circle. However, the central figure of the Copenhagen circle, Louis Hjelmslev (1899–1965), developed a grammatical theory on the model of mathematics, which he called ‘glossematics’. The goal of glossematics was to deduce from formal statements an inventory of categories for the analysis, at a rigorously high level of abstraction, of both sounds and forms in any language. Hjelmslev’s austere formalism stands out; however, he also developed a sophisticated semiotics that went beyond Saussure’s notions of signifier versus signified to analyze the relationships of expression, content, form, and substance. In this facet of his work, Hjelmslev was a major influence on the twentieth-century functionalist Michael Halliday, discussed in Chapter 4. An older compatriot of Hjelmslev, the Danish grammarian and historical linguist Otto Jespersen (1860–1943), likewise cannot easily be identified with either side of the formalist/functionalist divide. Jespersen distanced himself from Saussure and did not ally himself with any of the post-Saussurean schools, regardless of their formalist or functionalist tendencies. Instead,

24  Background to the Current Debate Jespersen produced a wealth of sensitive, innovative writings on phonetics, the history and structure of English, language teaching—which moved deftly from form to function and vice versa. Jespersen also insisted that language scholarship be rooted in the record of real language data, as used by real people, to communicate real messages. Both Hjelmslev and Jespersen contributed to the linguistics of the day from outside its general trends. Moreover, although a formalist mindset dominated mainstream American linguistics in the relevant period, there were figures (discussed below) whose work does not fit this profile. With that caveat in place, what follows is a sketch of how functionalist ideas infused language scholarship in Europe in the first half of the twentieth century (Toman 1995; Akamatsu 2001). Under most accounts, a significant event was the foundation of the Prague Linguistic Circle in 1926, which began as a series of meetings for informal discussion among three linguists, the Czech Vilém Mathesius (1882–1945) and Russian émigrés Nikolaj Trubetzkoy (1890–1938) and Roman Jakobson (1886–1982). The three shared a dissatisfaction with the Neogrammarian historical-comparativists in Leipzig, who asserted the mechanical exceptionlessness of sound laws. They also shared an attraction to Saussure’s claim that languages can be studied as coherent systems at any point in time. From the start, in the text of their ‘Ten Theses’ published in 1929 as the group’s manifesto, the Prague scholars boldly announced their functionalist orientation: ‘language is a system of means of expression adapted to a goal’ (Vachek 1967: 33). Jakobson (1965), for example, collected examples of how language form adapts to its function: Veni, vidi, vici communicates not only Caesar’s deeds, but the order in which he accomplished them; in high, higher, highest the three-stage incremental increase in word length iconically signals incremental increase in height; a shared rhyme in the words bash, smash, clash, splash, etc., represents a shared semantic feature (perhaps, ‘loss of control’). For members of the Prague Circle, units at various levels of organization of a language served the purposes of communication: communication of meaning; of the speaker’s identity or emotional state; of an effect the speaker wishes to induce in the listener. The Prague Circle enthusiastically engaged in research on many language phenomena, including literature, especially poetry; language change; language standardization and planning; language typology; language and culture. Here we will examine two of their most consequential and best-developed topics, with the clearest ties to modern functionalism: phonology and syntax. The Prague Circle separated phonology from phonetics on functional grounds: phonetics studies the substance and production of sounds, but since phonetic units do not communicate meaning, phonetics lies outside the center of linguistics proper. Phonology, on the other hand, is a core subfield of linguistics, because sounds map onto functions: unlike for the American distributionalists, for the Prague Circle, English /p/ and /b/ represent distinctive phonemes because a pen is where farm animals live and Ben

Background to the Current Debate 25 is a man’s name. The first job of the linguist is not to record a language’s inventory of sounds, but to locate what counts as a phonological opposition: the functional difference between pen and Ben signals that voicing of initial stop consonants is a significant phonological opposition in English. From the early 1900s many scholars had been working out various notions of what became the concept of the ‘phoneme’: some abstract, some material, some psychological, some as more or less discrete units. Among Prague Circle members, Jakobson’s and Trubetzkoy’s conceptions of the phoneme did not wholly converge. But as a minimum the two agreed that phonemes comprised bundles of phonic properties later called ‘distinctive features’. The notion of ‘archiphoneme’—a label for a phonological unit that neutralizes the difference between two distinctive phonemes that appear in specific positions—briefly flourished in Prague Circle writings. Archiphonemes and distinctive features signal the Prague Circle’s openness to abstract analyses and their interest in locating language universals, two traits more typical of formalists than functionalists. Nevertheless, their basic methodological commitment to locating differences in form via differences in meaning grounds the essentially functionalist character of their work. In syntax, the functionalism of the Prague Circle is very fully displayed. The group developed a ‘Functional Sentence Perspective’ (FSP) as a principled means to bring their ideas to bear on word order and other syntactic facts cross-linguistically. The FSP divides sentence constituents into ‘theme’ (old information; information previously registered in the discourse) and ‘rheme’ (new, unpredictable information; information that drives the discourse forward), and then analyzes the position of thematic versus rhematic elements. For example, the sentence ‘Shakespeare wrote Othello’ positions ‘Othello’ in the sentence-final rhematic (‘new information’) position; it is likely the answer to the question ‘What did Shakespeare write?’ In contrast, the sentence ‘Othello was written by Shakespeare’ treats ‘Othello’ as the theme (‘old information’) and ‘Shakespeare’ as the rheme; it is likely the answer to the question ‘Who wrote Othello?’ In this example, what the FSP addresses is how communicative function maps onto the syntax of English subjects and objects. The FSP goes beyond word order to examine other factors that bear on information structure, including intonation and word choice (e.g., the differing argument structures of the verbs in I like cold weather versus Cold weather suits me). A latter-day Prague scholar, Jan Firbas (1921–2000) contributed the notion of ‘communicative dynamism’, which admits that linguistic units exhibit relative degrees of communicative dynamism rather than an all-or-nothing identification as theme (less communicatively dynamic) versus rheme (highly dynamic). As an aside, if one were to apply the Prague Circle’s FSP to Louis Sullivan’s famous saying, ‘form follows function’, a buried linguistic irony emerges. In the expression itself the word ‘form’ precedes, not follows,

26  Background to the Current Debate the word ‘function’ so that the structure of this aphorism actually contradicts its meaning. Insofar as functionalists have reflected on this twist, they may simply savor its incongruity. Alternatively, they might notice that in ‘form follows function’ what counts as presupposed, thematic, material with reduced communicative dynamism—the word ‘form’—appears earlier in the expression, while the more salient, more communicatively dynamic element—‘function’—appears in final, rhematic, position. World War II interrupted the careers, and in some cases destroyed the lives, of many participants in early twentieth-century European functionalism. Some fled Europe for the United States. Among them was Roman Jakobson, who brought his brand of Prague Circle functionalism to New York and later to Harvard and MIT. Although Jakobson became an internationally known figure, Prague Circle linguistics played a more limited role in the history of functionalism in the United States, with its influence falling mostly on one subgroup, what Newmeyer (2001) calls ‘formal functionalists’ (see Chapter 4). Another scholar who emigrated to New York during the war, André Martinet (1908–1999), later returned to Paris to build up an important strand of functionalist linguistics. His work was based on a not uncritical reading of Saussure, which integrated some touches of both Praguean and even Hjelmslevian linguistics. A key principle of Martinet’s functionalism is that of ‘relevance’ (sometimes ‘communicative relevance’), by which he aims to pinpoint meaning-bearing elements of language. For example, Martinet points out that order is highly relevant at the level of phonemes: the words clay, lake, and kale comprise the same three phonemes, but instantiated in three different orders, whereas in syntax he asserts that ‘the relevancy of order is far from general: it is fairly immaterial whether I say the one I like is Paul or Paul is the one I like’ (1962: 41)—a claim some functionalists might dispute. Martinet brought his functionalism to bear on the sound patterns of language in extensive discussion of phonetics and phonetic change, intonation, pitch, tone, and syllable structure, but also addressed topics in morphosyntax, language typology, diachronic linguistics, and dialectology. The notion of ‘relevance’ shares with the Prague School term ‘communicative dynamism’ an intuitive appeal, a certain vagueness of definition—and a clear functionalist orientation. Although Martinet’s functionalism is well-established in Europe, especially in France, it had less impact on American linguistics. However, at least two other important figures in early-to-mid-twentieth-century American linguistics are recognizably functionalist. The first was one of Boas’s most prominent students, Edward Sapir (1884–1939). Sapir began his career following Boas’s footsteps, carrying out linguistic-ethnographic fieldwork in the Pacific Northwest. His work shows unusually keen powers of observation and analysis. Later, as a faculty member at the University

Background to the Current Debate 27 of Pennsylvania, Sapir worked with Tony Tillohash, a young speaker of Southern Paiute whom the Bureau of Indian Affairs had displaced from his home in Utah to attend the Carlisle Indian School in Carlisle, Pennsylvania. Working with Tillohash, Sapir developed his conception of the psychological reality of the phoneme. He found that Tillohash perceived the sounds of Southern Paiute as distinctive not with reference to their perceptible acoustic properties, but rather with reference to his feeling for their roles in the sound system of the language. That is to say, Tillohash prioritized the function of a sound over its structural properties. Sapir’s work overall was unabashedly functionalist in that he viewed language as a psychologically embedded cultural practice, which shapes speakers’ experience: Language is at one and the same time helping and retarding us in our exploration of experience, and the details of these processes of help and hindrance are deposited in the subtler meanings of different cultures. [. . .] It is this constant interplay between language and experience which removes language from the cold status of such purely and simply symbolic systems as mathematical symbolism of flag signaling. [. . .] language may not only refer to experience or even mold, interpret, and discover experience, but [. . .] it also substitutes for it. (Sapir 1933 / 1949: 11) Sapir and Bloomfield were near-contemporaries, although Sapir’s humanism contrasts strongly with Bloomfield’s behaviorism and pursuit of formalist rigor. Bloomfield had greater effect on the institutionalization of mid-century American structuralist linguistics, but Sapir inspired many functionalistoriented linguists, and continues to be cited respectfully by contemporary functionalists (e.g., Givón 1995: 1; Halliday 1999; Van Valin 2007). A generation after Bloomfield and Sapir, another American linguist, Kenneth Pike (1912–2000) worked out an original grammatical theory with a functionalist character. Pike was associated throughout his long career with the Christian evangelical organization now known as SIL International, which trains missionaries in linguistic fieldwork and Bible translation. Pike served as a missionary-linguist studying Mixtec in Mexico in the early 1930s before entering a doctoral program at the University of Michigan. He eventually became a faculty member there while continuing as an active fieldworker and trainer of missionary-linguists. Pike developed an ambitious, synthetic grammatical theory, ‘tagmemics’. The word was coined (by Bloomfield but developed by Pike) on the model of ‘phonemics’ to label the minimal grammatical element. Tagmemics analyzed how ‘fillers’ correlated with syntactic ‘slots’ as, for example, the fillers my roommate or Klara may fill the same slot in [ ] applied to law school. A sequence of tagmemes makes up a ‘syntagmeme’, and so on up to higher and higher levels

28  Background to the Current Debate of organization. For Pike, those higher and higher levels extended seamlessly from sentences to paragraphs to discourse to gesture to nonverbal behavior to all manner of cultural practices. Because for Pike language was a non-autonomous resource within general human communicative practice, the analytic categories and tools of linguistics could apply across the board: Pike (1954–1960 / 1967) famously ends with an analysis in the spirit of tagmemics of a football game. Pike’s work orbited far from the formalism of his contemporaries, in several ways. It was explicitly designed to help fieldworkers make sense of a foreign-language corpus, in their preparation for Bible translation. Pike did not participate in the distributionalists’ reluctance to use meaning as an analytic tool or to rely on speakers’ intuitions. Most of all, Pike’s integration of language with culture made tagmemics distinctively functionalist. It is notable that Dan Everett, author of Language, the Cultural Tool, first met the Pirahã people as a member of SIL International, the institution that Pike led.

2.5  Why the Background Matters The goal of Chapter 2 has been to point out some of the precursors to modern formalism and functionalism. Early twenty-first century formalist linguistics has unique features, as does early twenty-first century functionalism. Still, many of the fieldmarks that differentiate the two approaches are recognizable as far back in the history of the study of language as the eye can see. Pāṇini aimed to build a comprehensive grammar; the engineer in him produced a penetrating, concise, self-contained model of Sanskrit. In Cratylus, Plato gave to Cratylus the role of champion of the notion that words mirror the essential nature of their referents, an idea that Saussure challenged but that Jakobson would adopt and remodel in hoisting a functionalist flag over his evidence that iconicity shapes grammar, word-forms, and sound patterns. In the early twentieth-century, post-Bloomfieldian linguist-engineers dominated in the United States, while linguist-collectors amassed evidence for how the Prague Circle’s communicative dynamism, Martinet’s ‘relevance’, and Sapir’s and Pike’s different commitments to the integration of language and culture made the study of language (in their eyes) broader, richer, and more powerful. None of this earlier work precisely foreshadows how formalism and functionalism developed after the 1950s, but the existence of that work prepares us to better understand how more recent work arose, and to notice what is novel in it.

References Akamatsu, Tsutomu. (2001). ‘The development of functionalism from the Prague School to the present’. In Sylvain Auroux, E. F. K. Koerner, Hans-Josef Niederehe,

Background to the Current Debate 29 and Kees Versteegh (Eds.), History of the language sciences, vol. 2 (pp. 1768– 1789). Berlin: Walter de Gruyter. Bates, Elizabeth and Brian MacWhinney. (1982). ‘Functionalist approaches to grammar’. In Eric Wanner and Lila R. Gleitman (Eds.), Language acquisition: The state of the art (pp. 173–218). Cambridge: Cambridge University Press. Battistella, Edwin. (2000). [Review of the book Language form and language function]. Journal of Linguistics 36: 431–439. Bloomfield, Leonard. (1926). ‘A set of postulates for the science of language’. Language 2: 153–164. Bloomfield, Leonard. (1927). ‘On Some Rules of Pāṇini’. Journal of the American Oriental Society 47: 61–70. doi: 10.2307/593241. Bloomfield, Leonard. (1929). [Review of the book Konkordanz Pāṇini-Candra]. Language 5(4): 267–276. doi: 10.2307/409597. Bloomfield, Leonard. (1933). Language. New York: Henry Holt. Bloomfield, Leonard. (1939). ‘Menomini morphophonemics’. Travaux du cercle linguistique de Prague 8: 105–115. Boas, Franz. (1911). ‘Introduction’. In Handbook of American Indian Languages (pp. 1–83). Washington DC: Bureau of American Ethnology. Cardona, George. (1988) Pāṇini: His work and its tradition: Vol. 1. Background and introduction. Delhi: Motilal Banarsidass. Carnie, Andrew and Norma Mendoza-Denton. (2003). ‘Functionalism is/n’t formalism: An interactive review of Darnell et al. (1999)’. Journal of Linguistics 39: 373–389. Carstairs-McCarthy, Andrew. (1999). [Review of the book Language form and language function]. Anthropological Linguistics 41: 426–429. Chomsky, Noam. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press. Chomsky, Noam. (1966). Cartesian linguistics: A chapter in the history of rationalist thought. New York: Harper and Row. Chomsky, Noam. (1986). Knowledge of language. New York: Praeger. Croft, William. (1993). ‘Functional-typological theory in its historical and intellectual context’. Sprachtypologie und Universalienforschung (STUF) 46(1/4): 15–26. Dillinger, Mike and Adair Palácio. (1997). ‘Generative linguistics: Development and perspectives. An interview with Noam Chomsky’. Chomsky no Brasil/Chomsky in Brazil. Special edition of Revista de Documentação de Estudos em Lingüística Teórica e Aplicada 13: 159–194. Dirven, René and Vilém Fried. (1987). ‘By way of introduction’. In René Dirven and Vilém Fried (Eds.), Functionalism in linguistics (pp. ix–xvii). Amsterdam/ Philadelphia: John Benjamins. Emeneau, Murray B. (1988). ‘Bloomfield and Pāṇini’. Language 64(4): 755–760. Foolen, Ad. (2002). [Review of the book Language form and language function]. Functions of Language 9(1): 87–103. Francis, Elaine J. (2002). [Review of the book Language form and language function]. Language Sciences 24: 29–56. Givón, Talmy. (1995). Functionalism and grammar. Amsterdam/Philadelphia: John Benjamins.

30  Background to the Current Debate Givón, Talmy. (2002). Biolinguistics: The Santa Barbara lectures. Amsterdam/ Philadelphia: John Benjamins. Givón, Talmy. (2013). ‘On the intellectual roots of functionalism in linguistics’. In Shannon T. Bischoff and Carmen Jany (Eds.), Functional approaches to language (pp. 9–29). Berlin: Walter de Gruyter. Gleason, Henry Allan. (1961). An introduction to descriptive linguistics (Rev. ed.). New York: Holt, Rinehart and Winston. Graffi, Giorgio. (2001). 200 Years of syntax: A critical survey. Amsterdam/Philadelphia: John Benjamins. Halliday, Michael A. K. (1999). ‘The notion of “context” in language education’. In Mohsen Ghadessy (Ed.), Text and context in functional linguistics (pp. 1–24). Amsterdam/Philadelphia: John Benjamins. Haspelmath, Martin. (2000). ‘Why can’t we talk to each other?’ [Review of the book Language form and language function]. Lingua 110: 235–255. Hockett, Charles F. (1954). ‘Two models of grammatical description’. Word 10(2– 3): 210–234. Hockett, Charles F. (1958). A course in modern linguistics. New York: Macmillan. Hymes, Dell and John Fought. (1981). American structuralism. The Hague: Mouton. Jakobson, Roman. (1965). ‘Quest for the essence of language’. Diogenes 13(51): 21–37. Joseph, John E. (2000). Limiting the arbitrary: Linguistic naturalism and its opposites in Plato’s Cratylus and modern theories of language. Amsterdam/Philadelphia: John Benjamins. Joseph, John E. (2002). ‘Bloomfield’s and Chomsky’s readings of the Cours de linguistique générale’. In John E. Joseph (Ed.), From Whitney to Chomsky: Essays in the history of American linguistics (pp. 133–155). Amsterdam/Philadelphia: John Benjamins. Katre, Sumitra Mangesh. (1987). Aṣṭādhyāyī of Pāṇini. Austin, TX: University of Texas. Martinet, André. (1962). A functional view of language. Oxford: Clarendon Press. Morpurgo Davies, Anna. (1998). History of linguistics, volume IV: Nineteenthcentury linguistics. New York: Addison Wesley Longman. Morvacsik, Edith. (2000). [Review of the book Language form and language function]. Language 76: 168–170. Newmeyer, Frederick J. (1998). Language form and language function. Cambridge, MA: MIT Press. Newmeyer, Frederick J. (2001). ‘The Prague school and North American functionalist approaches to syntax’. Journal of Linguistics 37(1): 101–126. Pike, Kenneth L. (1954–1960). Language in relation to a unified theory of the structure of human behavior. Glendale, CA: Summer Institute of Linguistics. 2nd rev. ed., 1967. The Hague: Mouton. Sapir, Edward. (1949). ‘Language’. In David G. Mandelbaum (Ed.), Selected writings of Edward Sapir (pp. 7–32). (Original text published 1933 in Encyclopedia of the Social Sciences. Vol. 9, pp. 155–169. New York: Macmillan) Saussure, Ferdinand de. (1983). Course in general linguistics (Charles Bally and Albert Sechehaye, Eds.; Roy Harris, Trans.). London: Duckworth. (Original text published 1916)

Background to the Current Debate 31 Staal, J. F. (Ed.). (1972). Reader on the Sanskrit grammarians. Cambridge, MA: MIT Press. Tallerman, Maggie. (2000). [Review of the book Language form and language function]. Studies in Language 24: 423–439. Thomas, Margaret. (2009). [Review of the book Cartesian Linguistics (3rd ed.)]. Language and History 52: 201–204. Toman, Jindřich. (1995). The magic of a common language: Jakobson, Mathesius, Trubetzkoy, and the Prague Linguistic Circle. Cambridge, MA: MIT Press. Vachek, Josef. (1967). A Prague School reader in linguistics. Bloomington, IN: Indiana University Press. Van Valin, Robert D., Jr. (2007). ‘Some thoughts on the reasons for the lesser status of typology in the USA as opposed to Europe’. Linguistic Typology 11: 253–257.

3 Contemporary Formalist Linguistics

3.1  The Internal Structure of Modern Formalism Contemporary formalist and functionalist linguistics are both animated and contentious intellectual communities, but their internal organization differs starkly. Formalist linguistics from the 1960s to the present day is unusual in that it centers on the output of a single scholar, Noam Chomsky. Chomskyan generative theory is controversial—even among formalists—but its influence is pervasive, especially in the United States, in the sense that even those who oppose Chomsky tend to define their own work relative to his. Chomsky’s is the first name, and frequently the only name, that American non-linguists identify with the discipline. His stature is more fraught in Europe, but since the 1970s there have been proponents of Chomskyan formalism at major centers for the study of linguistics, even where they do not form a majority (Haider, Prinzhorn, & van Riemsdijk 1987). The professional organization Generative Linguists of the Old World (‘GLOW’) specifically fosters research in the generative tradition among European linguists. Founded in Amsterdam, GLOW has sponsored a newsletter and a lively annual conference continuously since 1978. Moreover, generative theory has a strong independent presence in Korea and in Japan (Fukui & Saito 1994). Granted the centripetal structure of modern formalist linguistics, it makes sense to start Chapter 3 at that center. Section 3.2 sketches out 60 years of Chomskyan formalist linguistics. Section 3.3 expands the tableau by examining modern formalist linguistic theories which depart from generativism while still remaining recognizably formalist.

3.2  What Makes Modern Formalism Formalist? Chomsky has been an industrious, creative, and fully engaged scholar for more than 60 years. The convention is to divide his output into stages, which are roughly internally coherent, but which have to be studied with

Contemporary Formalist Linguistics 33 the understanding that incremental shifts took place in his thinking at many points, so that ideas from one stage have bled into adjacent stages. We will ground discussion of what is formalist in Chomskyan formal linguistics in a conventional three-stage chronology anchored by some of his major publications: in subsection 3.2.1, early transformational grammar (Chomsky 1957, 1965); in 3.2.2, principles and parameters (Chomsky 1984, 1986); and in 3.2.3, the Minimalist Program (Chomsky 1995, 2000). My goal is not to give a full explication of Chomsky’s linguistics, since many such expositions already exist, at various levels of detail—and readers of this text may be familiar with at least some version of generative theory. Rather, my goal in section 3.2 is to narrate the development of generative theory from 1957 to the early twenty-first century with reference to how its specifically formalist character has evolved over the years. 3.2.1  Early Transformational Grammar Chomsky studied as an undergraduate at the University of Pennsylvania under American structuralist Zellig Harris (1909–1992), absorbing from him many of the traits of the linguistics of the day: conceptualization of the discipline as a science; the assumption that a grammar is a concise specification of the structure of a language; concern for methodology. According to Derwing (1980), Chomsky adopted from Bloomfield (likely via Harris) the core formalist notion that language can be analyzed an autonomous system of formal rules—autonomous from the context of its use by speakers and hearers, and autonomous from related disciplines. On the other hand, Chomsky strongly rejected Bloomfieldian behaviorism, and asserted that a grammar should be modeled on speakers’ intuitions about acceptable versus unacceptable utterances, rather than on a corpus of attested utterances. Insiders to generative grammar consider this to constitute a sharp rupture with American structuralism, using the terms ‘revolution’ or ‘paradigm shift’ to describe its emergence in a disciplinary context dominated by postBloomfieldian linguistics (Joseph 1995). However, Chomskyan generative linguistics has at least its formalist character—present from Chomsky’s earliest publications—in common with its direct disciplinary ancestors. Chomsky compiled Syntactic Structures (1957) from his notes for an introductory course in linguistics that he taught to MIT engineers and mathematicians. His goal was to disabuse them of their assumption that a grammar could be conceived of as the output of a finite-state machine. A finite-state machine is a simple computational model, often represented as a flow chart comprising boxes and arrows connected at nodes which stand for decision points. At each decision point, input from the environment selects one of a limited number of options, which leads to a predetermined subsequent state. The machine cannot ‘look back’ to earlier states or decision points, nor does

34  Contemporary Formalist Linguistics it anticipate what lies ahead. In place of this limited and mechanical model for the structure of human language, Chomsky introduced the notion of a three-component grammar, consisting of a set of phrase structure rules that generate ‘kernel sentences’; freely containing abstract elements like ‘past’ or ‘aux(iliary)’ as the economy of the analysis demands; transformational rules that operate on the output of phrase structure rules by copying, moving, and deleting elements of the kernel sentences; and morphophonemic rules that derive surface forms from the output of the transformational component, such as replacing ‘sing + past’ with ‘sang’. In Aspects of the Theory of Syntax (1965) Chomsky developed a wider context for his ideas, discussing competence/performance; linguistic universals (as his use of the expression ‘universal grammar’ emerged); deep structure/surface structure; acceptability versus grammaticality; description versus explanation; the organization of ‘the base’ (input to the transformational component) and of the lexicon. Throughout, Chomsky’s stance is resolutely formalist. His linguistic theory defines the inner structure of human language, which assumes the autonomy of syntax from semantics and discourse, the autonomy of grammar from general cognitive faculties, and the central importance of studying what people know about language (their competence) as opposed to the degraded expression of that knowledge evident in what people say (their performance). Chomsky proposed that a grammar comprises independent modules that interact with each other, so that although speakers’ syntactic knowledge comes to bear on their phonological competence, the two can be studied separately. In Aspects, Chomsky identified the primitive units of the syntactic component according to their forms, as ‘noun’ or ‘preposition’, rather than relationally as ‘subject’, ‘predicate’, or ‘object’. His argument against what he called those ‘functional notions’ was that most of what they contribute to grammatical analysis can be derived directly from formal configurational relationships (1965: 68–74), therefore to admit them into a grammar is redundant. In a footnote, Chomsky further rejected the notions of ‘topic’ and ‘comment’—terms fundamental to many functionalist analyses—on a similar basis: a topic is ‘the leftmost NP that is immediately dominated by S in the surface structure and that is, furthermore, a major [syntactic] category’ (1965: 221). Moreover, from the beginning of his career Chomsky has employed idealized, decontextualized data, both grammatical and ungrammatical, to illustrate the operation of the system (e.g., Sincerity frightens the boy; *The boy frightens sincerity). It is worthwhile reviewing one of the most memorable, and most glamorous, of Chomsky’s early-period proposals, ‘Affix-hopping’ (Lasnik 2000). All parts of it have been superseded by his subsequent theorizing, but it still stands as a remarkable museum piece of late 1950s formalist style. Affix-hopping—technically, the ‘Auxiliary Transformation’ (1957: 113)— names a transformation in English grammar that governs the distribution of

Contemporary Formalist Linguistics 35 auxiliaries and the heterogeneous forms of the verb write in sentences like those in (1a—f). (1)

a. Camila writes for the Globe b. Camila is writing for the Globe c. Camila wrote for the Globe d. Camila has written for the Globe e. Camila has been writing for the Globe f. Camila could have been writing for the Globe

Chomsky generalized pre-verbal markers for tense and aspect into a single category ‘Af[fix]’, and identified auxiliary have, be, do, modals, and lexical verbs as members of a category he called ‘v’. The operation of Affixhopping attaches any member of ‘Af’ onto the end of the next adjacent ‘v’, iteratively, from left to right. That is to say, (1d) derives from the deep structure represented in (2a), which contains two affixes: an abstract tense marker pres[ent], and en, which is part of the abstract perfect aspect marker have + en. The transformation Affix-hopping applies to (2a), yielding (2b). The relevant morphophonemic rules then apply to (2b) to produce the pronounceable forms (Camila) has written (for the Globe). (2) a. Camila pres have en write for the Globe b. Camila have.pres write.en for the Globe Affix-hopping interacts with a ‘Do Transformation’, which inserts expletive do in contexts where no ‘v’ appears to the left of an affix. In the derivation of the negative sentence Camila did not write for the Globe, (3a) shows that the affix past cannot ‘hop’ to the right, because the next adjacent word, not, is not a member of the class of ‘v’. In (3b) the Do Transformation inserts do to rescue the stranded affix. After the relevant morphophonemic rules apply, the outcome is (Camila) did not write (for the Globe), with the past tense marker realized on do rather than on write. (3) a. Camila past not write for the Globe b. Camila do.past not write for the Globe Virtually none of the particulars in these early drafts of Chomsky’s ideas has survived into subsequent versions of generative grammar. However, their bold formalism is consistent with later developments: grammar is a set of precise operations, sometimes involving abstract forms that are remote from what we perceive in normal speech, which can be studied independent of how speakers use language (i.e., how speakers build their identities, express perceptions, exchange messages). Needless to say, a functionalist’s

36  Contemporary Formalist Linguistics point of view on the phenomenon of Affix-Hopping would bring very different issues to light: the salience of the employment of do in the context of negation; the communicative significance of the fact that negation adds rather than takes material away. A functionalist point of view might also take on other challenges, such as accounting for the role of language processing granted that English tense markers precede aspect markers, while verbs follow auxiliary elements. Goldberg and Del Giudice (2005), for example, present functional motivations for the related construction of subject-auxiliary inversion (e.g., Does Camila write for the Globe?). 3.2.2  Principles and Parameters A second stage in the development of generative grammar significantly increased its scope and sophistication, and the ambition of the theory to account for what makes human languages both resemble and differ from each other. By 1980, Chomsky had survived major challenges to his work, and had attracted a cohort of colleagues and students who were assiduously extending generative concepts and tools to new languages and to new niches of familiar languages. Writing from within this dynamic, internally contentious research program, Chomsky (1984, 1986) proposed that a grammar can be construed as a set of universal principles intrinsic to a human language faculty, which include specific points of parametric variation. Learners select from limited parametric options embedded in those principles, on the evidence of their observations of the surrounding language; those options govern the differences between, say, Italian and Spanish or Italian and Cherokee. The goal was to account for ways that human grammars vary structurally, by conceiving of them as consequences of a limited inventory of parameter settings. A generative grammar capable of meeting these goals required building up an extensive theoretical apparatus fine-tuned to accommodate crosslinguistic differences. Among proposals developed in this interval were constraints on syntactic movement, including Subjacency; locality conditions; syntactic chains; binding theory; the inventory of empty categories and their properties; overt and covert case-marking; c-command and government; theta-marking; the Projection Principle; traces; X-bar theory; and the distinction between I(nternal)-language versus E(xternal)-language. (An alternative label for this stage of Chomsky’s theorizing, ‘government and binding’, refers to two key proposals.) For our purposes what is more important than either the content of these notions, or their longevity versus ephemerality, is the fact of their consistently formalist architecture. The apparatus of principles and parameters maintained its focus on the form of language, not discourse, not a speaker’s expressive choices, and not semantics that went beyond judgments of, say, the scope of negation in two utterances or whether two sentences are synonymous.

Contemporary Formalist Linguistics 37 A consistent embrace of abstraction also stands out. Early transformational grammar trafficked in terms like ‘Af’ and ‘v’, and elements like pres and en. With this Chomsky not only prioritized competence over performance, but also took a strategic move away from the observable material of language—words, sounds—to define metalinguistic terms that would meet the exigencies of his analysis. This initial step up the ladder of abstraction is commonplace in analysis of language. Still, although Chomsky took that step in the interest of accounting for specific low-level facts in a specific language (the grammar of English auxiliaries) the formalist assumptions behind Affix-hopping are worth noting. Metaphorically, from the very start Chomsky didn’t study elements perceptible to our senses, like ‘salt’ or ‘water’. Rather, he studied (his conception of ) the components of their inner identity, like ‘sodium’ and ‘oxygen’. The 1980s principles and parameters version of generative grammar extended this initiative much farther, so that the terms of analysis became increasingly removed from actual language data. For example, within their separate domains, X-bar theory climbs higher up the ladder of abstraction than does Affix-hopping: it theorizes about classes of syntactic patterns derived from the analysis of both lexical and non-lexical categories in many languages. With the advent of the principles and parameters version of generative grammar, the object of study went to a new, more general level: no longer ‘sodium’ and ‘oxygen’, but ‘electrons’ and ‘protons’. Modern formalists applaud this trajectory as a sign of scientific progress. Andrea Moro, for example, would argue that study of electrons and protons unquestionably opens more powerful insight into the nature of matter than study of sodium or oxygen, much less of salt or water. The principles and parameters model has a lot of appeal, in that it wrestles directly with questions about how languages do and do not differ from each other. Out of those struggles came a more highly articulated notion of universal grammar, and extensive reflection on learnability issues. Chomsky has always taken seriously the need to ensure that proposals about the structure of a grammar are feasible in the sense that children can determine the full grammar of any language without requiring more than (in his view) the normal fragmentary and inexplicit evidence available to learners by observation of ordinary language use. An additional virtue of the principles and parameters model is that it can be extended to make predictions—about language structure, and about cross-linguistic variation—that are open to empirical testing. Moreover, it is relatively easy to explain, so that even decades after its heyday, principles and parameters is still a touchstone to students in introductory courses on generative grammar. It is worth noting that although the next stage in the development of Chomsky’s theory vitiated much of the conceptual apparatus of principles and parameters, proposals developed in the 1980s such as binding and X-bar theory still feature in ongoing research.

38  Contemporary Formalist Linguistics 3.2.3  The Minimalist Program From the mid-1990s, generative theory turned another corner. Some observers consider minimalism a still sketchy and speculative initiative, while others (e.g., Boeckx 2006) consider it a major revolution. Chomsky emphasizes that it constitutes only a ‘program’, which is to say, a framework within which research can be carried out rather than a more-developed ‘theory’, the truth or falsity of which can be demonstrated. One goal of minimalism is to elucidate the extent to which human language is ‘perfect’ in the sense that its design is optimal granted human computational, conceptual, and physical limits. Another goal is to meet the highest possible standards of economy in the structure of a grammar—a goal Chomsky would seem to share with Pāṇini’s grammar of Sanskrit, across vast differences of time and context, and notwithstanding the fact that for Chomsky this goal extends to a grammatical model adequate for all languages. For example, minimalism proposes to reduce all structure-building operations (word-formation processes, adjunction, raising, all manner of NP or WH movement, etc.) to a single, iterative, process, ‘Merge’. Merge represents the logical endpoint of a development that began at the end of the first stage of generative grammar. At that point the number of transformational rules that built up, re-ordered, or otherwise operated on sentence constituents had ballooned almost to the point where it matched one-to-one the number of constructions in the language. Generativists recognized the intolerability of this state of affairs, so that the proliferation of transformational rules helped propel the field in another direction. At the principles and parameters stage, Chomsky aimed to economize on construction-specific rules by resolving them into the operation of a few parameters embedded within language-general principles; however, eventually the inventory of principles and their associated parameters also complexified beyond the point that could reasonably meet learnability criteria. In this third, minimalist stage of the development of generative grammar, the proposal of Merge signals an intention to define the most general possible account of structure-building operations, reducing the mechanism of the grammar to its barest necessary content. To accomplish this, minimalism licenses grammarians to freely climb as far up the ladder of abstraction as needed to achieve that goal. Motivating the operation of Merge within the constraints of economy led to the re-working of notions of constituent structure and the development of a notion of feature-checking, wherein abstract features associated with certain lexical items undergo Merge so that their features can be ‘checked’. For example—shifting first to a different analytic vocabulary—transitive verbs must appear in construction with direct objects. In minimalist terms, a verb like nominate carries the feature [uN], with ‘u’ signaling that the verb is uninterpretable until it Merges with a noun (‘N’), such as Christina. Insofar as Christina appears in the appropriate

Contemporary Formalist Linguistics 39 configurational position relative to nominate, it checks the feature [uN] to create a new syntactic unit, nominate Christina. Features that succeed in being checked cause the associated lexical item to become interpretable (that it, assigned a set of pronounceable phonetic features), so that the derivation of the superstructure can proceed, with Merge continuing to operate on pairs of lexical units as sanctioned by their assigned features. Features that remain unchecked cause a derivation to ‘crash’. Overall, the Minimalist Program adopts a style of analysis in which elementary units are attenuated by several orders of magnitude from recognizable units of language. It is salient how inward-looking its overarching goals are: to discover how closely human languages approach perfection of design, and to articulate that design in the most perspicuous possible fashion. Granted that minimalism is a program, not a theory, it is a program that exhaustively inspects itself for redundancy or extraneous mechanisms. Ultimately, its concern is with human language and with the limits on crosslinguistic variation; but in its present state minimalists’ best energies seem to be directed at theory-internal engineering. For instance, minimalists have worked hard to eliminate the multiple levels of representation that earlier versions of generative grammar took for granted, variously labeled ‘deep structure’, ‘s-structure’, ‘phonetic form’, etc. Chomsky (1993) proposed that a grammar need support only a single level of representation. The majority of Chomsky’s argument is theory-internal: that is, he argues for the superiority of a radically ‘flat’ derivational system on the basis of its suitability to the emerging demands of minimalism. He does, however, supplement the argument by citing its consequences for certain classic facts to which generations of generativists have attended, such as constraints on whmovement, or anaphora in picture-noun constructions. Still, improving the design of the theory itself drives Chomsky’s proposal forward, more than concern with any specific empirical fact. Metaphorically speaking, far from dealing with ‘salt’ or ‘sodium’ or even ‘electrons’, minimalism turns inward on itself to try to resolve whether quantum-scale objects can be best represented as waves or as particles. In linguistics, as in physics, those debates are not carried out in the absence of reference to material reality. But in both cases, material reality is attenuated from the terms of the debate. On the other hand, minimalism sometimes pivots to face issues outside the typical purview of formalism. In fact, at least some of its rhetoric resembles that of functionalism. This startling observation emerges in a development that intersects with generative theory, namely ‘biolinguistics’. In the words of Cedric Boeckx, one of its chief proponents, biolinguistics is ‘a branch of cognitive science that seeks to uncover the biological underpinnings of the human capacity [to acquire languages]’ (Boeckx 2013: 316). Biolinguistics is centered directly on minimalism, with which its leaders are identified, but still maintains an independent identity. It is aggressively interdisciplinary: much of the data and conceptual vocabulary of biolinguistics derive

40  Contemporary Formalist Linguistics from genetics, primatology, immunology, neurology, evolutionary developmental biology, and computer science. Biolinguistics situates Chomsky’s linguistics within this company. But it also sometimes appears to open a door facing functionalism. In a review of biolinguistics, Chomsky wrote approvingly that to conceive of the language faculty as a biological system requires attending to three factors: (1) genetic factors, ‘apparently near uniform for the species’, that is to say, universal grammar; (2) experience, which introduces variation in language use; and (3) principles not specific to the language faculty, including computational capacities (2005: 9–10). Factors (2) and (3) have long been pursued by functionalists, who conceive of formalists as having deliberately and on principle neglected experience and non-language-specific faculties. Both biolinguistics and the Minimalist Program are controversial and rapidly developing. Whether they represent a real rapprochement between functionalism and formalism is a topic to which we return in Chapter 6. 3.2.4 The Common Formalist Character of Generative Grammar Before proceeding, it may be valuable to summarize the character of generative formalism. The principled separation of form from function—in the autonomy of syntax from semantics or discourse, of grammar from general cognition, and of competence from performance—stands out. We have also seen abundant evidence that generative grammar is unfazed by abstraction. Moreover, a consistent factor in generative theory is its prioritization of the internal over whatever is conceived as external: deep structure over surface structure; competence over performance; I-language over E-language. In the Minimalist Program, Chomsky (2007) suggested that rather than trying to determine how much of universal grammar must be part of the language faculty, a better approach would be to determine how little of UG will suffice. He calls this move ‘approaching UG from below’, a coinage which keeps alive generative grammar’s option for the internal over the external. In addition, generative theory foregrounds universals over particulars, and articulates an array of ideas about how a grammar can be organized, sometimes attending more to those ideas than to linguistic data itself. It is controversial whether the new field of biolinguistics, which looks to incorporate the latest version of generative theory into a larger disciplinary landscape, disrupts the otherwise formalist identity of generativism.

3.3  Diversity Within Contemporary Formalism Generative grammar—at all its stages of development—is the foremost modern exemplar of formalism in linguistics. Its formalism runs deep, and in that sense, it provides an excellent foil for the contrast between formalism

Contemporary Formalist Linguistics 41 and functionalism. However, there also exist non-generativist, and countergenerativist, varieties of formalist linguistics. None are as well-known or dynamic as Chomskyan linguistics, but they, too, belong in the picture. A number of competing non-generativist formal grammatical theories emerged in the late 1970s into the 1980s as attempts to address what were perceived as shortcomings in generative grammar. As a class, they accept Chomsky’s formalism, and aim to create mathematically precise and computationally tractable models of language. They tend to be ‘flatter’ than 1980s generative grammar, which at that time assumed multiple levels of representation (deep/D-structure, phonetic form, etc.), and to shift the work of the transformational component of generative grammar into the lexicon. Generalized Phrase Structure Grammar (GPSG; Gazdar et al. 1985), out of which later grew Head-Driven Phrase Structure Grammar (Pollard & Sag 1994) has probably had the greatest impact. In some ways GPSG seems optimistic that the structure of a language can be represented with a lower burden of theoretical abstraction—no transformations; only two X-bar levels, and only for the major categories of nouns, adjectives, verbs, and prepositions; no abstract categories like ‘infl’. GPSG analyzes syntactic categories into assemblies of features with associated values. Features project up and down trees, to guarantee that, for example, the plurality in the noun children matches the plurality in its specifier, these, in the expression these children (subject to certain restrictions: a Head Features Convention ensures that only some features spread across trees). Each word’s entry in the lexicon includes complex information about its features and their relevant values. An additional layer of Feature Co-occurrence Restrictions specifies certain dependency relationships, such as the requirement in English that complementizer for only appear with the non-finite form of the verb (e.g., For us to be late would be rude). The elimination of transformational rules is a particularly significant innovation. GPSG resolves phrase structure rules into statements of immediate dominance and linear precedence, then adopts ‘metarules’ that do some of the work of transformations in that they map statements of dominance or precedence onto other statements of dominance or precedence. For example, a metarule in the lexicon applies to any verb that selects a direct object, replacing generative theory’s Passive transformation; it licenses an additional construction in which that verb appears with passive morphology, and precedence and dominance relations are redistributed. Also significant is GPSG’s account of the long-distance dependencies that had inspired Chomsky to develop transformations. GPSG admits null elements into trees marked with the feature ‘slash’ (‘/’), specified for the identity of the relevant null argument, so that ‘S/NP’ marks a sentence with a null NP. A Foot Feature Convention allows ‘/’ to spread upward until it is satisfied on encountering a feature-matched constituent, so that in Soda I’m sure he never drinks, the slash feature percolates upwards from NP/NP in the direct object position of drink until it is satisfied by the topicalized NP soda.

42  Contemporary Formalist Linguistics An important issue of relevance to the formalism of GPSG is the status of semantics. In early generative grammar, the autonomy of form from meaning— of syntax from semantics—was presupposed, but its nature unsettled: if syntax were truly autonomous from semantics, in what sense could a Passive transformation operate on The Globe hired Camila to generate Camila was hired by the Globe, since the passive and active options are related precisely by meaning? GPSG, on the other hand, assumes that formal rules map onto semantic translations: it builds rich meaning constructs into syntax (Gazdar et al. 1985: 182–244). It is worth noting, however, that GPSG semantics excludes pragmatics, rejecting the functionalist conception of a continuum between the two. Head-driven Phrase Structure Grammar (HPSG) grew out of and has overtaken GPSG. It is explicitly counter-generative in that it criticizes generative grammar as imprecise and inadequately formalized, and aims to create a grammatical model accessible to natural language processing initiatives. Lexical entries in HPSG comprise complex ‘attribute value matrices’ associated with fully inflected word forms. The attribute value matrices list elaborate information about the morphology, argument structure, and combinatorial privileges of words, represented as features with associated values (e.g., for the word children, the feature ‘number’ is associated with the value ‘plural’). The goal is to shift into the lexicon even more of the grammar, dispensing with both transformations and the use of trees as a formalization of phrase structure. HPSG has been implemented in several language processing projects, with their outcomes feeding back to increase the accuracy and coverage of HPSG itself. A second non-generative formalist linguistic theory that arose alongside GPSG has a different flavor. Lexical-Functional Grammar (LFG) probably peaked in the early 1990s, but still holds the attention of a minority following. The classic source text is Kaplan and Bresnan (1982), and now Bresnan (2001). In part, it aimed to replace Chomsky’s 1970s work with a more psychologically plausible theory, a theory which could be shown through psycholinguistic experimentation to better represent the structure of competence. Like GPSG (and HPSG), LFG eschews transformations, but it retains a good deal of the apparatus of generative grammar, including X-bar theory and the binding principles. It assumes that lexical entries provide richly detailed information about the properties of words, which are entered in their fully inflected forms (such as eaten or multi-dimensional or wiring). But LFG differs from both generative grammar and GPSG in that it identifies as primitives the notions of ‘subject’, ‘object’, ‘oblique goal’, etc. The term ‘functional’ in the name of the theory does not refer to functionalism as a counterpoint to formalism; rather it refers to the mathematical sense that a function of some set x relates x to some other domain y: when the function ‘subject’ is applied to the sentence Sissi left early, it yields Sissi.

Contemporary Formalist Linguistics 43 LFG posits two levels of syntactic description: ‘c(onstituent)-structures’ and ‘f(unctional)-structures’. C-structures provide phrase-structural and configurational information; f-structures encode surface grammatical relations, such as ‘subject’ or ‘adjunct’, and serve as input to a semantic component of the grammar. C-structures display how languages organize the information contained in f-structures: some languages allow null copulas with adjectival predicates, like He busy, while some require an overt copula, like He is busy; the c-structures of the two languages would record that difference. However, in LFG, c-structures aren’t structural skeletons which constrain the order and relationships of lexical items. Instead, c-structures come together with f-structures, which already contain structural information without specifying order or constituency relationships. F-structures connect features, or attributes, to values: an f-structure connects the feature ‘pred(icate)’ to the value ‘love’, with the lexical entry for ‘love’ specifying that (among other properties) it takes a subject and a direct object. The nodes of LFG trees carry notations indicating each terminal unit’s f-structural properties, some of which percolate up to the f-structures of non-terminal nodes. For example, the plural feature on children percolates up to the f-structure of the mother NP node, ensuring that the plurality of children will be accessible. Well-formedness conditions ensure that every function has a value, and there are neither too many nor too few functions than the relevant lexical entries call for. With this basic plan for a grammar in place, LFG has been developed to address perennial topics of interest to syntacticians, such as anaphora, long-distance dependencies, conjunction, and so forth. Like GPSG and HPSG, LFG has been adopted by computer scientists as a feasible basis for machine translation and natural language processing initiatives. Compared to generative grammar, none of these additional twentiethcentury formalisms has the rich philosophical contextualization that Chomsky has provided for his own work. Defining language universals did not have a prominent role in their development, and only LFG has been attentive to learnability issues. GPSG/HPSG and LFG have focused on constructing economical and maximally watertight systems for accounting for the structural properties of language, with one eye on the value of training computers to model human linguistic abilities. Although all three aimed to replace some of Chomsky’s formalist proposals, it was not the formalism of transformations or of multi-stratal architecture that GPSG/HPSG and LFG rejected. In fact, they abundantly display a formalist orientation in the goals they pursue, in their prioritization of theory over data, in their modular organization, and in their inventive initiatives to ensure that all the ends meet, as they assume they must, in the representation of words, their features, and their formal properties—in the absence of interaction between forms and semantic, pragmatic, or discourse factors. In all this, the engineering-like nature of their approach stands out.

44  Contemporary Formalist Linguistics

References Boeckx, Cedric. (2006). Linguistic minimalism: Origins, concepts, methods, and aims. Oxford: Oxford University Press. Boeckx, Cedric. (2013). ‘Biolinguistics: Fact, fiction, and forecast’. Biolinguistics 7: 316–328. Bresnan, Joan. (2001). Lexical-functional syntax. Malden, MA: Blackwell. Chomsky, Noam. (1957). Syntactic structures. The Hague: Mouton. Chomsky, Noam. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press. Chomsky, Noam. (1984). Lectures on government and binding. Dordrecht: Foris. Chomsky, Noam. (1986). Knowledge of language: Its nature, origin, and use. New York: Praeger. Chomsky, Noam. (1993). ‘A minimalist program for linguistic theory’. In Kenneth Locke Hale and Samuel Jay Keyser (Eds.), The view from building 20: Essays in linguistics in honor of Sylvain Bromberg (pp. 1–52). Cambridge, MA: MIT Press. Chomsky, Noam. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, Noam. (2000). New horizons in the study of language and mind. Cambridge: Cambridge University Press. Chomsky, Noam. (2005). ‘Three factors in language design’. Linguistic Inquiry 36(1): 1–22. Chomsky, Noam. (2007). ‘Approaching UG from below’. In Uli Sauerland and Hans-Martin Gärtner (Eds.), Interfaces + recursion = language? (pp. 1–30). Berlin: Mouton de Gruyter. Derwing, Bruce. (1980). ‘Against autonomous linguistics’. In Thomas A. Perry (Ed.), Evidence and argumentation in linguistics (pp. 163–189). Berlin: De Gruyter. Fukui, Naoki and Mamoru Saito. (1994). ‘Noam Chomsky’s impact on linguistics in Japan’. In Carlos P. Otero (Ed.), Noam Chomsky: Critical Assessments. Part I: Linguistics. Tome II (pp. 735–739). London: Routledge. Gazdar, Gerald, Ewan Klein, Geoffrey K. Pullum, and Ivan Sag. (1985). Generalized phrase structure grammar. Cambridge, MA: Harvard University Press. Goldberg, Adele E. and Alex Del Giudice. (2005). ‘Subject-auxiliary inversion: A natural category’. The Linguistic Review 22: 411–428. Haider, Hubert, Martin Prinzhorn, and Henk van Riemsdijk. (1987). ‘Syntaxreport – generative Grammatik in Europa. Report on generative grammar in Europe’. Folia Linguistica 21(2–4): 471–488. Joseph, John E. (1995). ‘The structure of linguistic revolutions’. Historiographia Linguistica 23: 379–399. Kaplan, Ronald M. and Joan Bresnan. (1982). ‘Lexical-functional grammar: A formal system for grammatical representation’. In Joan Bresnan (Ed.), The mental representation of grammatical relations (pp. 173–281). Cambridge, MA: MIT Press. Lasnik, Howard. (2000). Syntactic structures revisited: Contemporary lectures on classic transformational theory. Cambridge, MA: MIT Press. Pollard, Carl and Ivan Sag. (1994). Head-driven phrase structure grammar. Chicago: University of Chicago Press.

4 Contemporary Functionalist Linguistics

4.1  The Internal Structure of Modern Functionalism Chapter 4 surveys varieties of modern functionalism, which does not have the centripetal organization of formalist linguistics. Some scholars have independently developed functionalist approaches; others have drawn on each other’s work, or reacted against it. Graffi (2001: 389) adverts to ‘the galaxy of functionalism’, an expression that captures the image of a crowded array of competing functionalist proposals and counter-proposals, spread out over a panoramic space, within which no single star indisputably outshines the others. Chapter 4 explores this wide disciplinary space. Section 4.2 presents three case studies of contemporary functionalist linguistics, as a sample of what lies within the functionalist horizon. Section 4.3 goes on to review several commentators’ attempts to organize the galaxy of modern functionalism into multiple internally consistent solar systems, in which like-minded functionalists orbit around shared ideas or positions. Section 4.4 concludes by returning to differences between formalists and functionalists, offering an analysis of graphic style practices between the two, which, arguably, communicates something about their differing orientations.

4.2  The Scope of Modern Functionalism Although there is variety within modern formalist linguistics, including the multiple versions of generative theory, functionalist linguistics is more heterogeneous as a class. Van Valin (1990: 171) attributes to Elizabeth Bates the playful remark that ‘functionalism is like Protestantism: it is a group of warring sects which agree only on the rejection of the authority of the Pope’. Bates probably overstates the case. Nevertheless, it makes sense that while engineers’ plans may converge—blueprints for a garage, a library, and a summer cottage share recognizable features—collectors differ widely from each other in what, how, why they collect the objects they

46  Contemporary Functionalist Linguistics collect, and in how they organize and evaluate their collections. Moreover, among functionalists whose work we discuss below, the lack of a dominating intellectual leader or even a dominating geographical center adds to their diversity. In Europe, functionalists have continued to acknowledge the influence of the Prague Circle (e.g., Halliday, Dik). In America, as Newmeyer (2001) has pointed out, linguists such as George Lakoff developed recognizably functionalist views as part of a late 1960s rebellion against Chomskyan formalism called ‘generative semantics’. Other functionalists (Langacker; Givón; Goldberg) neither participated in generative semantics, nor acknowledge descent from the Prague Circle (Fortis 2015). In this section I  will bring forward three significant modern function­ alist theories, as a sample of the kaleidoscopic content of the total group: Halliday’s systemic functional linguistics; Langacker’s cognitive linguistics; and the functionalist work of Givón. These three samples by no means exhaust the scope of functionalism. They only represent three especially lively modern versions of it, strategically selected from different neighborhoods within the galaxy of ideas. To expand on this initial presentation, one might turn next to the functional grammar Dutch linguist Simon Dik (1940–1995) and his followers, or to Robert Van Valin, Jr.’s (b. 1952) Role and Reference Grammar. Both scholars come up in passing in section 4.3. Butler (2003) provides more extensive treatment. Note that what I label with the cover term ‘functionalism’ some commentators separate into ‘functional linguistics’ versus ‘cognitive linguistics’ (e.g., Siewierska 2013: 485). Functional linguistics focuses on how the pressures of communication impose a shape on language; cognitive linguistics on how humans categorize, store, access, and represent meanings and the effect of those processes on the form of language. Other commentators do not make this distinction, since many ‘functional linguists’ (in the broad sense of the term) view both communication and cognition as essential factors. 4.2.1  Systemic Functional Linguistics Michael A. K. Halliday (1925–2018) was a British-born scholar who spent most of his professional life affiliated with the University of Sydney. He developed an internationally influential linguistic theory which has had particular impact in the study of language in education, child language development, language and social class, textual analysis, and stylistics. Halliday’s approach is thoroughly functionalist and social-semiotic in the sense that he views language as formed by its role in communicating meaning within a social context: ‘The particular form taken by a grammatical system of language is closely related to the social and personal needs the language is required to serve’ (Halliday 1970: 142). His rejection of the autonomy of grammar from socially embedded communication is absolute; he described language as ‘the creature and creator of human society’ (2002: 6–7).

Contemporary Functionalist Linguistics 47 Halliday worked out an intricate network of terms to label the parts of his theory, which extends far beyond the formalist’s central concerns with phonology, morphology, lexis, and syntax. Among other conceptual tools, systemic functional linguistics identifies three ‘meta-functional constructs’, namely, the ideational (the capacity of language to inform); the interpersonal (the linguistic expression of human relations); and the textual (the internal organization of a text, written or spoken). Halliday associated the three meta-functional constructs with three features of what he called the ‘context of situation’. The ideational meta-function is allied with the ‘field’ of discourse, such as the choices a speaker makes to use a transitive rather than an intransitive construction to communicate a particular meaning. The interpersonal meta-function Halliday associated with the ‘tenor’ of discourse, which opens the analysis to how speakers encode social distance, solidarity, or authority in language. The textual meta-function aligns with the ‘mode’ of discourse, its internal organization as signaled by such features as ellipsis, repetition, or the use of anaphoric terms. With these analytic tools in hand, Halliday studied many facets of language, always under the assumption that what he calls the ‘lexicogrammar’ of a particular language both emerges out of the interpersonal and material experiences of its speakers and provides those speakers with a means with which to construe the world. The scope of Halliday’s work is wide: he has done close analyses of literary and non-literary texts, samples of spontaneous conversation, newspaper headlines, and advertising slogans. The non-formal character of systemic functional grammar is particularly clear in Halliday’s insistence that speakers make choices within a highly flexible and open-ended system. That is to say, his goal is not to identify what is grammatical versus ungrammatical but rather to understand why certain speakers, in certain contexts, with certain intentions, encode messages the way they do. This approach doesn’t distill data for generalizations, nor does it propose the existence of patterns in language and then evaluate their validity. Butler (2003) notes that Halliday’s epistemology is inconsistent with that of contemporary formalist linguistics, and hence so are his methods: ‘the methodology of setting up explicit hypotheses which are then rigorously tested is not one which finds much favour in the majority of the systemic literature’ (2003, Part 1: 202). Rather, Halliday collects and examines language data in the context of its production to discover ways that context, meaning, and linguistic structure shape each other. A topic of special attention in systemic functional linguistics is textual coherence. Halliday and Hasan (1976) examine many samples of natural language in English—written and spoken—derived from popular and academic texts, literature (including poetry), transcribed interviews, and informal conversations. Their materials also include many short, evidently invented, passages. Halliday and Hasan examine the techniques speakers and writers use to build coherence, which, in their view, are features of a text that depend on

48  Contemporary Functionalist Linguistics the presence of some other feature for their full interpretation. A particular concern is with what gives a text coherence across sentences. Halliday and Hasan identify multiple cohesion devices that ensure the parts of the text fit together: pronouns and demonstratives; ellipsis; conjunctions and adverbials; replacement of one element by another (she left and so did I; what the committee aimed to achieve . . . their goal); and all manner of lexical chains: repetition, (near) synonyms, superordinate terms. From a formalist perspective, those devices exist on many levels of grammatical organization. They frequently cross sentence boundaries and therefore lie outside the formalist’s basic unit of analysis. From a systemic functionalist point of view, the interest lies in observing the match between a writer’s/speaker’s choices to his or her inferred communicative intentions and in discovering what kinds of cohesion devices, represented by what lexicogrammatical items, lead readers to identify a text as tightly versus loosely coherent. 4.2.2  Langacker’s Cognitive Grammar Ronald Langacker (b. 1942) is an American linguist, and a central player in the emergence since the early 1980s of cognitive grammar (an expression used both generically, and to label the specific theories of certain scholars, among them Langacker). Like Halliday’s systemic functional grammar, cognitive grammar now has an international following. It is unquestionably anti-formalist in its rejection of the autonomy of syntax from semantics and pragmatics, and of grammar within general human cognitive processes. The relevant cognitive processes, in Langacker’s view, include association, automatization, categorization, and schematization (‘extracting the commonality inherent in multiple experiences’ [2008: 17]). Cognitive grammar offers alternatives to generativist practices and proposals at virtually every point. It insists that, rather than being the product of a discrete faculty, language is integral to human cognition and is shaped by general faculties of perception, memory, and categorization. It objects globally to categorical judgments such as grammatical/ungrammatical; content/ function word; synchrony/diachrony; derivation/inflection—on the grounds that in all cases members of each pair differ in matters of degree, and that intermediate cases can be found that challenge simple dichotomies. Not surprisingly, Langacker’s cognitive grammar also finds fault in the methodology of generative theory, including the focus on economy, which values rules over lists, in what he perceives as a premature push to formalize, generalize, and predict. Against these objections, cognitive grammar presents a very different style of study of language. Langacker (1987: 11–55) starts with the Saussurean doctrine that the signs of a language associate semantic structures and phonological representations, with the lexicon, morphology and syntax forming a continuum of symbolic structures. The emphasis on the symbolic

Contemporary Functionalist Linguistics 49 nature of language ensures that meaning and form are bound up together: Langacker (2008: 21) gives the example of sharp < sharpen < sharpener < pencil sharpener < electric pencil sharpener to illustrate how increasing complexity in form mirrors increasing complexity in meaning. Cognitive grammar is ‘usage-based’, not focused on establishing (un)grammaticality, but rather sensitive to degrees of routinization of particular expressions. The term accepts that speakers learn, categorize, and retain thousands of specific compositions of symbolic units—the inventory of a language’s constructions—which vary freely across languages. Constructions are on a continuum from conventionally routinized form/meaning relationships (for example, for speakers of English the expression brothers and sisters is thoroughly entrenched) to novel symbolic assemblies (such as the expression paper and roses). The different usage status of brothers and sisters versus paper and roses means only that the second expression signals some speaker’s less-commonly instantiated knowledge, which has at least the potential to be categorized, stored, and then automatized, if it were to achieve sufficient frequency. Cognitive grammar asserts that figurative language, including metaphors and conventional expressions, are not special cases to be treated as extra-grammatical exceptions. They are typical constructions, with no clear boundary separating figurative from literal speech: to Langacker, literal speech is only speech in a context where familiarity has eroded its essentially figurative nature. In cognitive grammar, the lexicon, morphology and grammar lie on a continuum, comprising ‘a structured inventory of conventional units’ (1987: 57). Speakers assemble smaller symbolic units within larger ones, so that even very complex sentences are made up of overlapping conventional expressions. The array of conventional units within the inventory opens to speakers many choices among ways to code a message. For example, in 1970s formalist analyses, take the trash out is derived from take out the trash via syntactic movement, which displaces the particle out while otherwise maintaining the meaning relationships of the expression. From the perspective of cognitive grammar, both conventional units are available, with take the trash out representing a different image and different viewpoint from take out the trash. In Langacker’s writings the focus is on cognition, that is, on the mind’s capacities to perceive, generalize, categorize, etc., the symbolic structures of language. However, he considers cognitive grammar a species of functional grammar, which he understands as insisting on the foundational, grounding, nature of social interaction in language (2008: 7–8). The concerns of cognitive grammar, such as speakers’ abilities to access stored material, recognize differences, combine smaller elements into larger structures, and so forth underlie their capacity to use language in social interaction. In this sense Langacker treats his work as both fully cognitivist and fully functionalist.

50  Contemporary Functionalist Linguistics 4.2.3  Talmy Givón As a third exemplar of modern functionalism, I take the work of Thomas (‘Talmy’) Givón (b. 1936). He was born and educated in what is now Israel, earned a PhD from UCLA, and spent most of his career at the University of Oregon. Givón calls his work simply ‘functionalism’, although others subsume it within the informal label of ‘west coast functionalism’ (Noonan 1999: 11). Although he has a reputation for taking strong anti-formalist stances, much of his writing seems directed at discouraging extremism on both sides (Givón 1995: 1–23). Both cognitive and social-communicative factors are important in Givón’s analyses. A good example is his treatment of negation. Givón (1995: 42–43) provides evidence for the markedness of negation over affirmation (negation, not affirmation, adds material; a text count shows that negation is less frequent than affirmation; negation is more complex cognitively). In an extended treatment, Givón (2001, Vol. I) first points out the inadequacy of conflating negation in natural language with negation in logic, since language communicates many subtler grades of negation than simply ‘pqr’ versus ‘not pqr’. He then proceeds to examine many aspects of the pragmatics of negation, and how the grammar of negation depends on our expectations about social norms: Givón marks the sentence There once was a man who didn’t look like a frog ‘pragmatically odd’ (p. 371) because it disrupts our presupposition that people generally do not look like frogs. He goes on to point out that because direct negative statements assert that the presumed beliefs of the hearer are mistaken, speakers’ use of them is highly sensitive to the social roles of the interlocutors. As an illustration, Givón cites a passage from a 1985 novel depicting life in small-town America, in which two interlocutors elaborately negotiate differences of opinion about a person’s appearance, while going to great lengths to avoid explicit use of negation that contradicts each other’s statements (pp. 377–378). Givón’s point is that this text represents communicative practice within a ‘society of intimates’, in which the social context shapes language use. Mindful of functionalists’ recognition of the flexibility of language, Givón follows up with examples of how, paradoxically, negation (combined with irrealis operators) can soften a message delivered to a presumed social superior, as in Wouldn’t it be better if . . . ? (p. 378). The title of one of his early books, On Understanding Grammar (1979), is revealing in that it signals Givón’s concern that functionalist analyses explain why the facts of grammar are what they are. To Givón, ‘explanation’ and ‘understanding’ necessarily involve factors such as discourse pragmatics, processing, cognition, perception, and speakers’ world-views (1979: 3–4). An example illustrates the point. Givón (2001, Vol. I: 281) adverts in passing to an adjacency principle that he considers a reflex of iconicity in syntax: ‘Items that are closer functionally or cognitively tend to be placed closer in the linguistic code’. Exceptions to that principle are not

Contemporary Functionalist Linguistics 51 unexpected, and arise from conflicts between the pressures to meet other principles. In Givón’s example, the same adjacency principle is at work differentially in (1a) and (1b). (1) a. A guy who spoke no English came by yesterday, and. . . b. A guy came by yesterday who spoke no English, and. . . In (1a) the (near) adjacency of guy and the relative clause predicate spoke no English are satisfied; in (1b) the adjacency of guy and the main clause predicate came by yesterday are satisfied. The two applications of the principle cannot be met at the same time, but each sentence satisfies some minimal requirement for coordination of closeness in function to closeness in syntax. Compare (2a) and (2b): (2) a. She went to a restaurant she owned with a friend b. ? She went to a restaurant with a friend she owned In (2a) a restaurant is adjacent to the relative clause she owned, at the expense of the adjacency of the predicate went to a restaurant to its adjunct with a friend: another instance of conflict in the application of the adjacency principle that is resolved to meet a minimal standard. But in (2b), the inverse resolution is unavailable: the adjacency principle applies to interpret she owned as a modifier of a friend, setting up a ‘semantic incompatibility’ that blocks a felicitous reading of the sentence. Givón’s point here is that the adjacency principle, ‘like many general principles interacting in a complex system . . . is seldom absolute’ (p. 281): grammatical patterns are more than the output of mechanically applied rules, mechanically distilled from formal facts.

4.3  Taxonomies of Modern Functionalist Linguistics The three case studies described above depict a range of modern functionalist approaches, and contrast functionalist and formalist analyses. Recalling that present-day functionalism is more diverse than formalism, this subsection communicates how different commentators have tried to organize the present-day ‘sects’, warring or not, of functionalism. To begin with a scholar who has extensively studied formalism versus functionalism from a formalist perspective, Frederick Newmeyer (1989: 10) starts by excluding a group he labels ‘formal functionalists’. This group comprises linguists who, in Newmeyer’s judgment, don’t strictly qualify as ‘functionalists’ because they accept core formalist doctrines such as the autonomy of syntax. They also are more likely to acknowledge the Prague Circle as an inspiration to their work (Newmeyer

52  Contemporary Functionalist Linguistics 2001). Susumu Kuno (b. 1933) is a central member; Ellen Prince (1944–2010) was also an active contributor. Kuno’s work argues that generative grammar needs to be augmented by a functional module that would rectify points where he feels that generativists have forced unnatural formal analyses in the place of more convincing functional solutions. For example, Kuno (1987: 183–192) gave a functionalist analysis of the intractable properties of reflexives embedded in ‘picture noun phrases’, a construction with which a generation of formalists struggled. He resolved the difference in acceptability between *John gave Mary an accurate description of herself and John gave Mary a renewed faith in herself into the operation of a host of factors such as the awareness of the trigger NP (here, Mary) that the picture noun is about herself; the extent of overt or covert agency of the trigger NP; the humanness of the trigger NP; etc. Similarly, what formalists attributed to a structural asymmetry on quantifier scope interpretation, Kuno (1991) reassigned to a functionalist principle, namely that a hierarchy based on semantic features like subject/ non-subject, human/non-human, speaker or hearer/third person, etc. constrains the scope of quantified expressions relative to each other. In these ways formal functionalists like Kuno graft functionalist concepts onto the body of formalist grammar, a move which leads Newmeyer to exclude them from the functionalist camp. Setting aside ‘formal functionalists’, Newmeyer defines three subgroups of functionalists proper: external, integrative, and extreme. External functionalists are those who accept what Newmeyer calls ‘the idea of a synchronic semiotic system’ (p. 14; emphasis on the original), which is to say, who accept that languages comprise formal elements (‘noun’; ‘adverbial’) that are linked to semantic/pragmatic elements. Into this group Newmeyer would place Halliday’s systemic functional grammar, Langacker’s cognitive grammar, and the work of cognitive linguist George Lakoff, best recognized for his writings on how metaphors and figurative language shape thought and political life (Lakoff & Johnson 2008). In addition, Newmeyer identifies as ‘external’ other influential branches of formalism, including Role and Reference Grammar (Foley & Van Valin 1984); Simon Dik’s functional grammar (1978); and the Competition Model of Bates and MacWhinney (1989). The traits Newmeyer attributes to this big-tent group are familiar: meaning is based on speakers’ construal, not truth conditions; continuity between semantics and pragmatics; prototypicality rather than categorical distinctions; grammatical constructions pair meanings and forms and have real cognitive status; language is designed to communicate meaning (pp. 15–16). As a second subgroup, Newmeyer identifies what he calls ‘integrative functionalism’. He cites only a single figure, Paul Hopper (1987), who

Contemporary Functionalist Linguistics 53 adopts an ‘emergent’ view of grammar. In this view, grammar is not an abstract, mentally represented set of rules that speakers have access to, but rather a complex adaptive system that emerges on a provisional and context-dependent basis in the course of communication to meet speakers’ needs. In Hopper’s words, structure ‘comes out of discourse and is shaped by discourse as much as it shapes discourse in an on-going process’ (1987: 142). Hopper recognizes that the concept of emergent grammar steeply challenges some of the basic premises of both formalists and functionalists, insofar as their work attempts either to pair forms and functions or to dispute that such pairing is possible. Newmeyer does not dwell on integrative functionalism, but comments that some functionalists not allied with one of the brand-name theories may express a commitment to it, while carrying out their work under the practices of external functionalism (1989: 17). Newmeyer’s third subgroup, ‘extreme functionalism’, is similarly underpopulated, with only one attested member, a ‘Columbia School’ associated with the American linguist William Diver (1921–1995), of Columbia University. The Columbia School considers language to be composed of individual paired signs and signals. It denies the existence of an independent syntactic component, and accounts for grammatical patterns as the properties of lexical and morphological signs acted upon by discourse and processing. It resists cross-linguistic generalization in favor of puzzling out low-level phenomena in specific languages: arrive at night versus arrive in the morning. The Columbia School is uncompromisingly functionalist, to the point where it remains encapsulated from the rest of the formalist/ functionalist debate. In short, Newmeyer’s taxonomy excludes formal functionalists as too amenable to formalism to count as functionalists; places the majority of other scholars in the single heterogeneous category of external functionalists; and sets aside two additional small subgroups of integrationist and extreme functionalists. Unsurprisingly, the panorama Newmeyer offers does not satisfy all parties. In a 2013 essay, Polish-born linguist Anna Siewierska (1955–2011) organizes the landscape differently. She first distinguishes functional linguistics (which emphasizes the communicative basis of language) from cognitive linguistics (which emphasizes how language reflects human conceptualization). Both major groups branch internally. First among functionalists, Siewierska mentions the Prague Circle and Martinet as forerunners. Among modern exponents of functionalism, she positions Newmeyer’s formal functionalists on one edge. On the other edge she includes Givón and Hopper as members of a ‘west coast’ group (an epithet presumably recognizing Givón’s career at the University of Oregon, although Hopper’s affiliations have varied). Between these two extremes are what Siewierska considers the most well-developed types of functionalism, which she calls ‘structuralist-functionalist’ grammars: Role and

54  Contemporary Functionalist Linguistics Reference Grammar and Dik’s functional grammar (which she distinguishes as both having strong language-typological commitments); and Halliday’s systemic functional grammar. At a finer level of resolution, Siewierska names and depicts local schools that have descended from each of the three branches of structuralist-functionalist grammar. She also points out that although many members of her functionalist macro-group aspire to incorporate into their work models of psychological adequacy and linguistic processing, in practice this has not been a strength. Facing Siewierska’s functional linguistics are the less-diversified group of cognitive linguistics, which subsumes Langacker’s work, that of George Lakoff, Adele Goldberg’s (1995) Cognitive Construction Grammar, and William Croft’s (2001) Radical Construction Grammar. Cognitive linguists ‘consider constructions, conceived as symbolic units, rather than rules, to be the primary objects of linguistic description’ (Siewierska 2013: 495). Constructions include individual words, productive morphology, idioms, and (what in formalist models would be called) syntactic structures like passives or relative clauses. Goldberg’s Cognitive Construction Grammar has been instrumental in developing the usage-based model and in amassing psycholinguistic evidence that supports its claim that language development is driven by learners’ acquisition of thousands of specific form-function pairings to create a network of constructions. Speakers innovate by filling open slots in constructions in novel ways: the novelty paper and roses might be based on paper and pencil (or perhaps guns and roses). Siewierska’s cognitivists converge on assuming that speakers organize constructions into networks, but they diverge in the extent to which they replace the familiar vocabulary of linguistic analysis by conceptually based categories. In Croft’s Radical Construction Grammar the identity of what are conventionally called the parts of speech are worked out within a three-by-three ‘conceptual space’ with rows labeled ‘objects’, ‘properties’, ‘actions’, and columns headed by ‘reference’, ‘modification’, and ‘predication’ (Croft 2001: 63–107). Siewierska notes that none of these three subtypes of cognitive grammars has, in her opinion, tied cognitive linguistics very closely to present-day cognitive psychology, so there remains too informal a relationship between the analysis of language and topics like perception, schemas, or prototypes. However, she credits Croft’s work as having made more progress in bringing cognitive linguistics to bear on typological issues. A third taxonomy of functionalists was compiled by Johanna Nichols (b. 1945), who is sympathetic to functionalism but best known for her work in language typology. Nichols (1984) opens with a fine-grained analysis of the multiple senses in which discussion of functionalism uses the word ‘function’: ‘function’ in the sense of what speakers do with language (request; insult; query); ‘function’ of language in the sense of the context of the speech event; etc. She divides functionalism—more precisely, ‘functionalist statements’ (p. 102) into three types: conservative, moderate, and extreme.

Contemporary Functionalist Linguistics 55 Conservative functionalism, like the work of Kuno, can be made amenable to formalism. Moderate functionalism aims to replace formalist analyses of language phenomena to meet the diverse standards of functionalism: how speakers’ meanings, human cognitive faculties, or the demands of communication shape language. Extreme functionalism, according to Nichols, can go so far as to claim that ‘structure is only coded function’ (p. 103). Nichols’s assignment of linguists to her three types reveals a few surprises: She tentatively identifies Simon Dik as a conservative functionalist, while Givón is a moderate who sometimes makes extreme functionalist pronouncements. In addition, Nichols adds a few more figures. Michael Silverstein (b. 1945) re-defines grammatical case as the interaction between a noun’s lexical content, its agent/patient status, and seemingly distant phenomena such as clause-linkage and switch-reference. Jeffrey Heath (b. 1949), whose work Nichols places on the border between moderate and extreme functionalism, found that inflectional morphology in the Australian language Ngandi suffices to identify the roles of nouns, obviating the formal notion of the clause. Dwight Bolinger (1907–1992), who by dates and by training belonged to the heyday of American structuralism, left a legacy of keen observation of semantic differences, and argued in a distinctively functionalist tone against the arbitrariness of word/meaning relationships and in support of gradient, not categorical, differences at many levels of linguistic organization. To end this overview of modern functionalism, it is useful to return to our first example, Everett’s Language, the Cultural Tool. Where might this text be placed in the array of functional linguistics? Recall that Everett identified his work as compatible with functionalism, but distinct from it in his insistence that culture shapes language—or, more precisely, his claim is that culture shapes language via cognition. For example, Everett asserts that because the Pirahã ‘don’t have a money-based economy and do not otherwise need numbers’ (2012: 260), therefore they have only a rudimentary sense of numeracy. As a result, although the language has words to communicate meanings like ‘small quantity’, it wholly lacks ordinal numbers. Similarly, the absence of an elaborated temporal and aspectual system in Pirahã grammar reflects speakers’ here-and-now orientation to time. Unlike Langacker or his fellow cognitive-constructivists, Everett is essentially a field linguist who does not analyze how perception, categorization, processing, etc., shape language. Rather, he draws on the resources of cognitivefunctional linguistics to make sense of his data.

4.4 Graphic Style in Modern Formalism and Functionalism We conclude Chapter 4 by calling attention to the distinctive graphic styles of modern formalists versus functionalists. Authors vary in their practices,

56  Contemporary Functionalist Linguistics but it is common for linguists to embed in running texts lists, formulas, diagrams, or tabular displays, to help readers visualize ideas or directly confront data. Tables and images can be highly effective in summing up complex relationships. They also can backfire: Lehmann (2000: 126) points out how tables, sketches, and formulas may both richly communicate linguistic data, and distort those same data. Modern formalism and functionalism exhibit strikingly different graphic styles. Formalists aim to create a model of the grammar of a language; generativists (and non-generative formalists) abstract away from the properties of individual languages to model the inner framework of human language. By convention, generativists have represented the organization of grammar in flow chart-like diagrams with boxes and arrows (e.g., Chomsky 1986: 68; Boeckx 2006: 72–82). In all their iterations, these diagrams affirm the separation of syntactic, semantic, and phonological components, position them variously relative to each other, and propose interactions among them. They also communicate that utterances undergo step-wise development, with each module contributing in turn—at least metaphorically, if not as an assertion of psycholinguistic fact. The Minimalist Program version of the diagram flattens, but does not do away with, the representation of how syntax interacts with phonetic and logical forms (Chomsky 1995: 22). Taken as a group, generativists’ flow charts materialize their assumptions about how to organize a grammar in the most economical and satisfying manner. Generative as well as lexical-functional formalists also employ syntactic trees to elucidate configurational relationships. In early transformational grammar, tree diagrams illustrated structural relationships among words: Chomsky (1957: 27), for example, provided a syntactic tree indicating the constituents of the sentence [[The man] [hit [the ball]]]. With the advent of X-bar theory, trees hosted increasingly abstract features, such as S’, pro, INFL, t, or AgrsP, augmented with subscripted indices or lines indicating paths of syntactic movement. In Chomsky’s work from the 1980s onward, lexical items like hit or the ball appeared less frequently on the terminal nodes of trees in favor of (for example) Agro or tsubj, representing a shift away from specific language data toward the general features of grammar. Additionally, formalists take seriously the value of a notational system with a high level of clarity and precision—both to ensure theoretical integrity, and to make the theory accessible to scholars in other disciplines. Proponents of both LFG and GPSG/HPSG self-consciously formalize their theories with abbreviatory conventions and mathematical and logical notation that computational studies of language processing can exploit. The early chapters of Gazdar et al. (1985) each end with a section formalizing the preceding content into a tissue of definitions and postulates. The pages

Contemporary Functionalist Linguistics 57 of Bresnan (2001) bristle with equations, formal definitions of terms, and diagrams. Whether or not computer scientists make use of this material, once language data is distilled into logical or graphic form, those forms take on their own life: the exposition concentrates on tree structures or formulas, working out the most perspicuous and powerful generalizations about, say, grammatical case, on the basis of its formal representation. In this way, graphic style plays a role in analysis. Interestingly, although Chomsky freely uses elementary mathematical devices such as sets, variables, and logical symbols, he considers full formalization a lesser goal: ‘the serious problem is to learn more, not to formalize what is known’ (1990: 147). Functionalists’ graphic style also provides readers with revealing visual clues to their conception of language. While formalists represent linguistic generalizations in pictorial models or formulas, a common feature of functionalist texts is the extensive citation of language data. Halliday and Hasan (1976), for example, include at least one quoted passage on virtually every page, with frequent citation of dialogue. In addition, they embed tables or lists of data in text: lists of ‘The forms of one, and related items’ (p. 106); classes of conjunctions (pp. 242–243); types of lexical cohesion (p. 288). Unlike the diagrams and formulas typical of formalist literature, Halliday and Hasan’s lists seem designed to illustrate rather than define. Their point is to provide examples of, say, varieties of conjunctions, not to specify the limits of what counts as a conjunction. In Givón (2001), data takes up more space than its analysis. There are dense blocks of narrative text; lists of independent examples under labels or headings; paradigms of forms arranged to induce comparison; and abundant non-English materials with conventional three-tiered glosses. Givón also uses tree diagrams (always with lexical items as terminal nodes), and tables which array data, features of data, or grammatical categories in rows and columns (Vol. I: 376), or which indicate the frequencies of observations of specific traits (Vol. I: 423). What these practices communicate is that, to Givón, data is neither a scaffolding over which he builds his argument, nor is it merely illustrative. Rather, the collected linguistic material is the central content of his exposition. The surrounding text merely comments on it. A final example of functionalist graphic style derives from Langacker (2008). In an unusual move, Langacker addresses the reader directly about the informal diagrams and sketches that populate his writing, representing both grammatical constructions and semantic representations (pp. 9–12). He defends them as heuristic in nature, rejects attempts at ‘mathematically respectable formalization’, and acknowledges the difficulty of representing meaning pictorially at the very fine level of resolution that satisfies cognitive linguists (p. 10). Langacker’s (1987: 295–296) comparison of the expressions corn kernel versus kernel of corn is telling. He notes that formalists simply derive corn kernel from kernel of corn, whereas for cognitive

58  Contemporary Functionalist Linguistics linguists these are different constructions with different semantics: relative to corn kernel, kernel of corn focuses on the ‘notion of a relationship between an integrated whole and a restricted subpart’ (p. 296). Langacker provides a series of three sketches representing how, in corn kernel, kernel designates a specific grain within the mass of corn to produce the composite expression. Kernel of corn is more complex, requiring five sketches: of signals a stative relationship with respect to the mass of ‘corn’; kernel then picks a single grain out ‘of corn’ to yield kernel of corn. Langacker’s sketches ground the fine distinction he makes between the two expressions, and emphasize the functionalist theme of how difference in form reflects difference in meaning. Moreover, the prolixity of these graphics in Langacker’s work, and his attention to the specifics of form and meaning in each instance, contrasts with formalists’ pursuit of underlying generalizations. Whether or not a picture is worth a thousand words, both formalists and functionalists exploit pictures effectively to augment their words.

References Bates, Elizabeth and Brian MacWhinney. (1989). ‘Functionalism and the competition model’. In Elizabeth Bates and Brian MacWhinney (Eds.), The cross-linguistic study of language processing (pp. 3–73). Cambridge: Cambridge University Press. Boeckx, Cedric. (2006). Linguistic minimalism: Origins, concepts, methods, and aims. Oxford: Oxford University Press. Bresnan, Joan. (2001). Lexical-functional syntax. Malden, MA: Blackwell. Butler, Christopher S. (2003). Structure and function: A guide to three major structuralfunctional theories. Part 1: Approaches to the simplex clause; Part 2: From clause to discourse and beyond. Amsterdam/Philadelphia: John Benjamins. Chomsky, Noam. (1957). Syntactic structures. The Hague: Mouton. Chomsky, Noam. (1986). Knowledge of language: It nature, origin, and use. New York: Praeger. Chomsky, Noam. (1990). ‘On formalization and formal linguistics’. Natural Language and Linguistic Theory 8(1): 143–147. Chomsky, Noam. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Croft, William. (2001). Radical construction grammar: Syntactic theory in typological perspective. Oxford: Oxford University Press. Dik, Simon. (1978). Functional grammar. Amsterdam: North-Holland. Everett, Daniel L. (2012). Language: The cultural tool. New York: Random House. Foley, William A. and Robert D. Van Valin, Jr. (1984). Functional syntax and universal grammar. Cambridge: Cambridge University Press. Fortis, Jean-Michel. (2015). ‘Generative grammar and cognitive linguistics: On the history of a theoretical split in American linguistics’. In Viviane Arigne and Christiane Rocq-Migette (Eds.), Metalinguistic discourses (pp. 53–88). Newcastle upon Tyne: Cambridge Scholars. Gazdar, Gerald, Ewan Klein, Geoffrey K. Pullum, and Ivan Sag. (1985). Generalized phrase structure grammar. Cambridge, MA: Harvard University Press.

Contemporary Functionalist Linguistics 59 Givón, Talmy. (1979). On understanding grammar. New York: Academic Press. Givón, Talmy. (1995). Functionalism and grammar. Amsterdam/Philadelphia: John Benjamins. Givón, Talmy. (2001). Syntax: An introduction, vols. I–II. Amsterdam/Philadelphia: John Benjamins. Goldberg, Adele E. (1995). Constructions: A construction grammar approach to argument structure. Chicago: University of Chicago Press. Graffi, Giorgio. (2001). 200 Years of syntax: A critical survey. Amsterdam/ Philadelphia: John Benjamins. Halliday, M. A. K. (1970). ‘Language structure and language function’. In John Lyons (Ed.), New horizons in linguistics (pp. 140–165). Harmondsworth: Penguin. Halliday, M. A. K. (2002). ‘Introduction: A personal perspective’. In Jonathan Webster (Ed.), The collected works of M. A. K. Halliday. Vol. 1: On grammar (pp. 1–16). London: Continuum. Halliday, M. A. K. and Ruqaiya Hasan. (1976). Cohesion in English. London: Longman. Hopper, Paul. (1987). ‘Emergent grammar’. Proceedings of the Annual Meeting of the Berkeley Linguistics Society 13: 139–157. Kuno, Susumu. (1987). Functional syntax: Anaphora, discourse, and empathy. Chicago: University of Chicago Press. Kuno, Susumu. (1991). ‘Remarks on quantifier scope’. In Heizo Nakajima (Ed.), Current English linguistics in Japan (pp. 261–287). Berlin: Mouton de Gruyter. Lakoff, George and Mark Johnson. (2008). Metaphors we live by. Chicago: University of Chicago Press. Langacker, Ronald W. (1987). Foundations of cognitive grammar. Vol. I: Theoretical prerequisites. Stanford, CA: Stanford University Press. Langacker, Ronald W. (2008). Cognitive grammar: A basic introduction. Oxford: Oxford University Press. Lehmann, Winfred P. (2000). ‘The development of adequate formalism in linguistics’. In David G. Lockwood, Peter H. Fries, and James E. Copeland (Eds.), Functional approaches to language, culture and cognition (pp. 125–145). Amsterdam/ Philadelphia: John Benjamins. Newmeyer, Frederick J. (1989). Language form and language function. Cambridge, MA: MIT Press. Newmeyer, Frederick J. (2001). ‘The Prague school and North American functionalist approaches to syntax’. Journal of Linguistics 37(1): 101–126. Nichols, Johanna. (1984). ‘Functional theories of grammar’. Annual Review of Anthropology 13: 97–117. Noonan, Michael. (1999). ‘Non-structuralist syntax’. In David G. Lockwood, Peter H. Fries, and James E. Copeland (Eds.), Functional approaches to language, culture and cognition (pp. 11–31). Amsterdam/Philadelphia: John Benjamins. Siewierska, Anna. (2013). ‘Functional and cognitive grammars’. In Keith Allan (Ed.), The Oxford handbook of the history of linguistics (pp. 485–502). Oxford: Oxford University Press. Van Valin, Robert D., Jr. (1990). [Review of the book Functional syntax: Anaphora, discourse, and empathy]. Studies in Language 14(1): 169–219.

5 Formalism and Functionalism in Action

5.1  Applying Formalism and Functionalism Chapter 5 brings formalism and functionalism, in parallel, to bear on three topics of perennial interest to linguists. The goal is to observe each approach in action, that is to say, to examine how formalists versus functionalists differentially address specific language issues. We start small, by comparing formalist versus functionalist treatments of two syntactic phenomena: word order and transitivity. A second focus for comparison is the two sides’ conceptualizations of a subfield of linguistics, the study of child language learning, and the assumptions formalists versus functionalists make about how language comes to children. The final comparison thinks big, juxtaposing formalist versus functionalist expositions of a long-standing, macroscopic issue, namely, speculation about the origin of human language. Chapters 2 through 4 have emphasized the range of what counts as formalist versus functionalist linguistics. In Chapter 5, the focus is selective rather than exhaustive, singling out for comparison how certain instantiations of formalism versus functionalism deal with specific language issues, rather than attempting to survey all relevant work. The hope is that this strategy will provide sufficiently representative contrasts to illustrate how the two approaches can differ, overlap, compete, or complement each other. The last section of Chapter 5 returns to this point.

5.2  In the Analysis of Syntactic Phenomena 5.2.1  Word Order Realistically, word order can hardly be thought of as a ‘small’ issue, since on some definitions it comprises the heart and soul of syntax, a major preoccupation of linguists across time and space. But for the purpose of this comparison, we will conceive of word order narrowly, as a language’s patterned sequencing of major syntactic or relational categories: subject, object, verb; noun, modifier; adposition, object; etc. In this limited sense, how have formalists and functionalists accounted for regularities of word order?

Formalism and Functionalism in Action 61 Among generative formalists, the question has been addressed multiple times in the evolution of the theory. In early transformational grammar, phrase structure rules recorded word order relationships, with the output of those rules subject to re-ordering by transformations. In other words, the order of a noun and its modifiers, or of a verb relative to its subject, were taken to be arbitrary, observable, language-specific, facts. Options were limited: nouns either precede or follow their modifiers (and likewise for verbs and objects, etc.), and phrase structure rules did not link the order of nouns and adjectives to (say) the order of verbs and direct objects. The business of the grammarian was to distinguish underlying word order sequences from surface permutations, so as to build the most accurate, most perspicuous model of a speaker’s competence. In the 1980s, generative grammar replaced language-particular phrase structure rules in pursuit of the more ambitious objective of accounting for word order phenomena across syntactic categories as well as across languages. Recall that generative theory at that time posited a human language faculty which specified a set of principles of universal grammar, some of which included a limited number of parametric options. Learners selected the appropriate parametric option by exposure to the language they are learning. Under these new assumptions, and accepting that universal grammar comprised multiple interacting modules, generativists re-worked how to represent word order facts. They proposed a single, maximally broad, structural configuration governing word order for all syntactic classes, as part of the module ‘X-bar theory’ (Haegeman 1991: 94–96). According to X-bar theory, the maximal projection of any category X dominates the specifier of X and a phrasal exponent of X, as in (1); that phrasal exponent of X dominates in turn the head X and YP, the maximal projection of the complement of X, as in (2). The position of adjuncts could be accounted for by adding another level that incorporates the phrasal exponent of X with ZP, an adjunct, as in (3). (1) XP → Spec; Xˈ (2) Xˈ → X; YP (3) Xˈ → Xˈ; ZP To account for cross-linguistic variation, the use of semicolons in (1) through (3) signals that the schema makes no claim about the order of elements. That is, (2) sanctions both the English (S)VO word order in (4) and the Japanese (S)OV word order in (5). (4) Ken ate the sandwich (5) Ken wa sandoichi o tabeta topic sandwich acc ate Ken ‘Ken ate the sandwich’

62  Formalism and Functionalism in Action The presumption is that, on hearing only a few samples of VO order in the surrounding language, learners of English can quickly and decisively select a version of (2) in which heads precede their complements—and that learners of Japanese can select the opposite option with equal ease on encountering samples of OV order. 1980s generative grammar went further to float ideas about how (1–3) could be adapted to account for inconsistency in the kinds of complements syntactic classes can appear with. For example, the transitive verb eat can take a noun complement (sandwich), but nouns themselves do not take noun complements. An additional module of universal grammar, case theory, interacted with X-bar theory to sustain the generality of (1–3). Case theory specifies that all nouns must be assigned case, which is parameterized to be either overtly marked, as in Japanese (note the accusative marker o in (5)), or abstract as in English, where case marking is not explicit in surface structure. Although transitive verbs like eat assign case to their complements, nouns themselves cannot assign case. Therefore, a rogue noun appearing in the position of complement of a noun would violate the requirement that every noun be assigned case. In this way, interaction of the X-bar theory schema in (1–3) with case theory ensures that, across different languages and different syntactic classes, complements will either appear or not, as appropriate. Kayne (1994) sought to extend the spirit of this approach by speculating that all languages are underlyingly SVO, attributing Japanese-style SOV word order to movement of the object leftward across the verb—a move designed to increase the elegance of X-bar theory by removing the burden of parameterizing the order of O and V. Of course, assuming that SVO is a default order comes at a certain cost, presumably borne elsewhere in the grammar, insofar as leftward movement would have to be thoroughly motivated and its consequences worked out in detail. From the point of view of 1980s formalism, the principles and parameters approach to word order phenomena satisfied many desiderata, in that it captured generalizations about multiple ordering facts, at multiple levels of organization, arguably for all human languages. Moreover, it did so while respecting the poverty of the stimulus, that is, the requirement that a language be learnable under the normal conditions in which (according to generative theory) children are exposed only to random and incomplete evidence for the structure of the target language. Setting aside further developments brought by the Minimalist Program, this account of the grammar of word order is representative of generative formalism in its drive to define the inner structure of human language with maximum breadth and economy. Functionalist accounts of word order have a different complexion, and make themselves responsible for a different domain of language facts. I survey here three separate discussions by contemporary functionalists. First, an article by the Czech linguist Eva Hajičová (1994) reviews and updates

Formalism and Functionalism in Action 63 Prague Circle research on word order and information flow, which maintains the notions of theme/rheme (or, sometimes with a slight shift in meaning, ‘topic/focus’) and degrees of communicative dynamism as the critical factors. The functionalist basis of this work is obvious, in that it derives word order facts directly from the roles words play in the transmission of a speaker’s intention to a hearer. Its basic claim is that the theme, or topic of a sentence—elements that are most given, most known, most established in the discourse—‘[functions] as a certain point of departure of the utterance’ (Hajičová 1994: 247). The theme precedes sentence elements that push the communication forward, variously identified as the rheme, focus, or comment. Hajičová adverts to research that has variously extended the Prague Circle analysis of theme/rheme to word order within smaller units (between main and subordinate clauses, and among the constituents of NPs and VPs) as well as research that has brought that analysis to bear on word order phenomena within larger units, namely connected texts. Hajičová acknowledges a frequently expressed critique of Prague Circle research on word order, that the notions of communicative dynamism and of theme/rheme are poorly defined or even circular. The gist of the problem is that rhematic material is identified as material that is more communicatively dynamic and therefore appears late in word order, while whatever material appears late is defined as more communicatively dynamic and therefore rhematic. Hajičová also concedes that in some languages have rigid formal principles of word order which override functionalist factors. Nevertheless, her article indicates that the distinctively functionalist research on word order that developed in Prague in the 1930s remains relevant. Two additional initiatives illustrate diverse functionalist research on word order phenomena. A contribution by Givón (1988) proposes a principle of ‘communicative task urgency’, originating in his objection that the Prague Circle account rests on ill-defined terms. He set out to derive a means of measuring the predictability or continuity of topics by counting their recurrence in a text. Givón also analyzed how recurrences of topics are morphophonologically encoded, and discovered that the less predictable or continuous a topic is, the more phonologically explicit it is, on a cline between a null pro-form (for a very predictable topic) to an unstressed pronoun, to a contrastive pronoun, all the way to a restrictively modified definite noun (for an unpredictable topic). He then inspected texts in many languages, with both flexible and variously rigid word orders, for correlations between pre- versus post-verbal subjects and objects and the extent of explicitness of expression. Givón found that information that was less predictable was pre-posed and more explicitly encoded. His conclusion was that, in languages where word order is not defined grammatically, nouns which are more communicatively urgent (more important, less predictable) are pre-posed. This finding runs against the grain of the Prague Circle’s

64  Formalism and Functionalism in Action principle that theme precedes rheme. But by analyzing how speakers’ cognition and discourse pragmatics are at work behind syntactic patterns, Givón offers another demonstrably functionalist account of word order phenomena. The issue continues to attract attention: although not couched in formalist versus functionalist terms, Arnold et al. (2000) investigate the role of processing and speech planning on word order variation in a study based on corpus analysis and psycholinguistic experimentation. A final example of functionalist research on word order is Payne (1999). Payne comments on both functionalist and formalist approaches, identifying two fundamental discrepancies between them. A first discrepancy lies in the data to be explained: formalists have carried out fine-grained studies of, for example, the word order privileges of full lexical nouns versus bound or free pronouns, whereas functionalists have been more concerned with word order and the communicative forces that act on speakers, and their motivations. Second, Payne identifies a difference between formalists’ versus functionalists’ notions of what counts as explanation for a linguistic fact: formalists, she asserts, are satisfied that a fact has been ‘explained’ when it is shown to be consistent with a theory-internal principle; functionalists ‘explain’ language facts by articulating connections between form and function. Against this background, she rejects formalist accounts of word order phenomena based on constituency and hierarchical relationships, branding Kayne’s (1994) proposal that all languages are underlyingly SVO as especially reductionistic (p. 146). Instead, she offers a list of discourse and cognitive factors that she feels can explain word order facts: sentence parsing; speakers’ and hearers’ management of attention; the effects of building and storing mental representations. Payne’s conclusion is that ‘the formalist and functionalist traditions are not, in my view, complementary in any crosslinguistic way’, even though she allows that certain accounts of language phenomena try to bring the two together (p. 157). 5.2.2 Transitivity In research on word order phenomena, formalists and functionalists recognize the same challenges—accounting for cross-linguistic patterns in the distribution of subjects, verbs, and objects, nouns and modifiers, adpositions and objects, etc.—but approach them with their own tools and assumptions. A different logical relationship exists in the case of transitivity, in which an essentially unified formalist understanding of the phenomenon contrasts strongly with at least one influential functionalist understanding of it. Formalist accounts of transitivity take for granted that it is a property inherent to lexical items, most particularly to verbs. The verb divulge, for example, is inherently transitive; that is to say, it must co-occur with what is traditionally called a direct object, as in divulge your password. Vanish is inherently intransitive: The stain vanished, and not *The detergent vanished

Formalism and Functionalism in Action 65 the stain. Give is ditransitive, hosting both direct and indirect objects, as in They gave their inheritance to Doctors Without Borders. In some analyses, prepositions can also carry features for transitivity, so that for is transitive (for our team) while in may be either transitive (in the garage) or intransitive (fade in)—although other analyses identify in in fade in as an adverb or as a particle. This informal exposition of transitivity, and the status of transitivity as an attribute of specific syntactic classes, is inscribed in the various idioms of formalism. The post-Bloomfieldian Charles Hockett (1958: 204) attributed (in)transitivity to predicates, or clauses, which constitute a verb plus its ­dependents—but it is still the core verb that is the seat of transitivity. Chomsky’s Syntactic Structures (1957: 111; see Lasnik 2000: 125–126) seemingly ignored transitivity, in that he established a phrase structure rule that expanded VP into a verb and an NP without acknowledging that intransitive verbs like vanish do not appear with direct objects. As generative theory developed, Chomsky returned to the issue in Aspects of the theory of syntax to propose that lexical entries for verbs carry features marking them as [+ transitive] versus [– transitive] (Chomsky 1965: 93–99). 1980s generative grammar refined the insight by positing a ‘Theta Criterion’ as one of the modules of universal grammar: verbs are intrinsically associated with an inventory of thematic, or theta, roles such as ‘patient’ or ‘goal’ that they must assign to arguments within a particular configuration: divulge assigns a patient theta role; vanish does not. An unassigned theta role, or an argument to which no theta role is assigned, violates the Theta Criterion (Chomsky 1986: 93–95). The Minimalist Program operates with a different grammatical-theoretical tableau, but still carries over from earlier versions of generative grammar the notion that specific features inhere in specific lexical items—with respect to transitivity, now expressed as case features that a verb must assign—and that those features control lexical items’ configurational privileges (Chomsky 1995: 225–235). Non-generative formalists express essentially the same conception of transitivity, in their own terms. In Lexical-Functional Grammar, a verb’s feature matrix specifies its argument structure, including (in)transitivity, which is then represented in c-structure (Bresnan 2001: 100–101). In Generalized Phrase Structure Grammar, lexical entries for verbs contain (along with phonological forms, syntactic categorical features, annotations of any irregular morphology, etc.) a feature ‘subcat’ and its numerical attribute. That number identifies the particular configuration of arguments the verb participates in (Gazdar et al. 1985: 31–35). What all these formalist accounts hold in common is the understanding that transitivity is grammatically consequential; that it resides in the properties of the lexical items (verbs, and in some accounts, prepositions); and that those lexical items impose particular patterns on the surrounding syntax. A famous 1980 article by two linguists identified with the ‘west coast functionalist’ group, Paul Hopper and Sandra Thompson, takes a different

66  Formalism and Functionalism in Action approach to transitivity. Their core insight is that it is a global property of clauses, which should be analyzed componentially by examining the contributions of the verb, its agent, its object, and other sentence elements. Clauses are not simply [+ transitive] versus [– transitive], but rather located on a continuum of greater and lesser degrees of transitivity. Hopper and Thompson identify 10 factors that bear on transitivity, paraphrased in (6) through (15); in parentheses, the value associated with high transitivity is listed first, with the low-transitivity value listed second. (6) Number of participants (high: 2 or more, agent and object; low: 1 participant) (7) Kinesis of verb (high: action; low: non-action) (8) Aspectual content of the clause (high: telic; low: atelic) (9) Punctuality of the clause (high: punctual; low: non-punctual) (10) Agent’s volitionality in acting on object (high: volitional; low: non-volitional) (11) Affirmation expressed in the clause (high: affirmative; low: negative) (12) Realis/irrealis encoding of events (high: realis; low: irrealis) (13) Agency of the agent (high: potent agency; low: low agency) (14) Affectedness of the object (high: object totally affected; low: object unaffected) (15) Individuation of the object (high: object noun is proper, human/animate, concrete, singular, count, referential/definite; low: object noun is common, inanimate, abstract, plural, mass, non-referential) In this way, Hopper and Thompson fracture the notion of transitivity as a categorical feature of verbs, and reconceive it as built up incrementally according to the properties of many clausal elements. They provide some telling examples: the sentences Jerry likes beer and Jerry knocked Sam down share high-transitivity properties for their number of participants (two), affirmation, and (ir)realis. But unlike Jerry likes beer, the sentence Jerry knocked Sam down also has high scores for kinesis, agency, punctuality, affectedness, and individuation of the object. Therefore, overall, Jerry knocked Sam down is more transitive than Jerry likes beer. Hopper and Thompson go on to point out that many languages, including Spanish, encode sentences equivalent to Jerry likes beer as intransitive clauses with beer treated as the subject and the drinker demoted to an oblique case, as if the total ensemble doesn’t meet the threshold for full transitivity: gusta la cerveza (16) Me 1.p.s.dat pleases the beer ‘I like beer’ (Hopper & Thompson 1980: 254)

Formalism and Functionalism in Action 67 Having established this counter-cultural notion of transitivity, Hopper and Thompson further propose that, for all languages, when two features for any of the factors in (6) through (15) must appear in a clause, they will be concordant in their values for transitivity. For example, in Indonesian (as in English), the equivalent of the verb sell can be immediately followed by either a post-verbal object (sell something to someone) or by a benefactive or dative NP (sell someone something). In the latter case, the Indonesian verb obligatorily takes the suffix -kan (informally: ‘sellkan someone something’), which marks increased transitivity because the human object someone scores higher for individuation than does the object that is sold, something. On this basis, Hopper and Thompson predict correctly that in other constructions suffixation by -kan increases transitivity: it renders adjectives and intransitive verbs causative, and it signals that the action of a verb totally affects its object. Most of Hopper and Thompson’s paper is taken up analyzing specific cases, in almost 40 languages, where high transitivity features co-occur with other high transitivity features (and likewise for low transitivity). They then introduce what they consider to be the key to understanding transitivity, namely, that it is an instrument which speakers use to mark grounding in discourse. Foregrounded information, which carries the main thrust of a discourse, is high in transitivity; backgrounded information, which ‘merely assists, amplifies, comments’ (p.  280), is low in transitivity. Hopper and Thompson test this claim by analyzing three short narratives written in English. First, they distinguished foregrounded and backgrounded material in each of the texts. Next, they counted the proportion of high versus low markers of transitivity in foregrounded versus backgrounded material, walking through (6) through (15). Their results are positive: foregrounded clauses are more likely than backgrounded clauses to contain two or more participants; they are more likely to depict actions; they have high values for telicity and punctuality; etc. Conflating scores across all 10 factors, 78% of foregrounded clauses evinced high transitivity, versus 39% of backgrounded clauses (p. 288). Hopper and Thompson thus extend the analysis of transitivity beyond the formal features of verbs to a global property of clauses, then, in an explicitly functionalist move, analyze transitivity as a device by which speakers mark the differential status they assign to information in discourse.

5.3 In Conceptualization and Study of Child Language Learning In reviewing Moro’s and Everett’s books in Chapter 1, we saw that formalists and functionalists sometimes study different language phenomena. In

68  Formalism and Functionalism in Action this chapter we have seen that they sometimes examine the same language phenomena but perceive them differently. In addition, formalists and functionalists sometimes approach the same language phenomena with different questions in mind. A case in point is how linguist-engineers versus linguistcollectors study how language comes to children. To start by glancing back at early twentieth-century formalism, Bloomfield famously declared that what he called a child’s acquisition of the ‘habits of speech’ constituted ‘the greatest intellectual feat any of us is ever required to perform’ (1933: 29). As a committed behaviorist, he resolutely grounded that feat in experience: children imitate what they hear in their environment; adults provide models for children and correct them, shaping their emerging habits; child speech gradually increases in range and accuracy. To this scenario Bloomfield’s student Hockett (1958) added an apparently functionalist touch: he proposed that there is a pre-linguistic stage in which a child’s earliest communicative signals are iconic rather than arbitrary, in the sense that there is some ‘geometrical resemblance between signal and meaning’ (p. 354). But Hockett specified that true language does not begin until the advent of imitation, and only later, when the capacity to analogize arises, has the child ‘begun to participate in a genuine, if still highly idiosyncratic, language’ (p. 357). In a review originally published in 1941, Leopold (1941 / 1971: 10–11) noted that relative to psychologists and educators, linguists came late to research on child language. He further remarked that linguists linked the study of child language to historical linguistics, citing Bloomfield’s assertion that the development of language in the child represents a compressed version of the slow change of language over time. There is little evidence, however, that mid-century American linguists considered child language data relevant to theoretical linguistics. This is not true of their direct disciplinary successors, generative formalists, nor is it true of modern functionalists: both present-day formalists and functionalists take child language data as a key to understanding the nature of language. But how formalists versus functionalists construe child language and the processes involved in acquisition brings out sharp contrasts between the two approaches. In Chomsky’s earliest writings, there is not much discussion of language learning. Through the 1960s, he occasionally remarked in passing that input to learners is more degenerate or limited than had been recognized. During the next decade, generativists began to argue that mere exposure to adults’ speech in the environment is insufficient to account for all of what children demonstrably know about that language. This amounts to a ‘logical problem of language acquisition’ (Baker & McCarthy 1981) insofar as one assumes, against the grain of generativism, that children’s observation of adult speech is what drives language learning. From around 1980 (Thomas 2002), Chomsky began to articular the notion of the poverty of

Formalism and Functionalism in Action 69 the stimulus as a premise in the argument that input to language learners is incommensurate with the rich and detailed knowledge that they end up exhibiting. By problematizing what comes to children from the outside, generativists sanctioned efforts to define what children bring to language acquisition from inside their competence. Since competence cannot be directly observed, only inferred via performance, generativists developed observational and experimental techniques for probing what children know and don’t know about language, how they come to know it, and how that knowledge changes over time. Hyams (1999) illustrates generativist research on language acquisition in a study of the ‘telegraphic’ speech of one- and two-year-olds, that is, the stage when young children produce utterances that often—but not always— omit functional categories such as auxiliaries, inflectional markers, or pronouns (as opposed to lexical categories such as nouns and verbs). Analyzing a corpus assembled from samples of child learners of several European languages, she focuses on the absence of determiners and of finite inflections on verbs, as in the English example (17), compared to the adult utterance The man drives the truck, and the absence of subject pronouns and presence of default infinitival inflection on verbs as in (18) in languages like German, which explicitly mark infinitival forms of verbs (Hyams 1999: 389–390). (17) Man drive truck (18) Matratze schlafen mattress sleep.inf ‘(He) sleeps on the mattress’ Hyams argues on two grounds against the conclusion that children at this stage lack grammatical knowledge of functional categories. First, she cites research indicating that, in the terms of 1980s generative theory, very early learners can accurately set other parameters tied to functional categories, such as verb raising, ‘V2’ or verb-second phenomena, or the null subject parameter. Second, she highlights that learners omit only certain finite inflections, subject pronouns, and determiners, rather than doing so across the board. On this basis Hyams rules out the possibility that young children haven’t yet learned how to inflect verbs or select the appropriate forms of pronouns. They do not lack an underlying grammar of functional categories and their distribution; the problem is to determine what controls the differential display of that facet of their competence. To account for ‘telegraphic’ speech without conceding her position on young children’s underlying competence, Hyams scrutinizes the corpus more closely (relying on research she co-authored). She notices that verbs not marked for finiteness tend to co-occur with null subjects, as in (18). In fact, 83% of verbs not marked for finiteness appear with null subjects, as

70  Formalism and Functionalism in Action opposed to only 23% of verbs with finite inflections (Hyams 1999: 396). Hyams then extends the notion of finiteness to both verbs and nouns, predicting that in early child language non-finite verbs will appear with nonfinite nouns, and likewise for finite verbs and finite nouns. Non-finite verbs are those which lack inflectional features, or verbs that are explicitly marked as infinitival, as in (18). She defines non-finite nouns as null subjects and bare nouns without determiners, and finite nouns as those that appear with determiners or plural markers. Hyams excludes proper names and overt pronouns. Testing these predicted correlations against various sub-sets of the data in her corpus, Hyams finds only very rare counterevidence. There remains the question of why early child learners pass through a stage in which they correlate finiteness across verbal and nominal domains, in the absence of models for this grammatical feature in the adult grammar. Hyams points to generativist research that concluded that the heads of functional categories, including finite markers on verbs and determiners on nouns, are pronominal in nature and therefore must have their reference fixed. The adult grammar identifies the reference of functional heads grammatically (through binding relationships, on the model of how the reflexive himself is bound to John in John fooled himself ). In child grammar, Hyams proposes that learners permit the reference of functional heads to be identified either grammatically, or pragmatically (on the model of John fooled him, where the referent of him lies outside the sentence). What children have to learn as they abandon ‘telegraphic’ speech is that functional heads must be identified grammatically, not pragmatically, because the presence of a grammatical option blocks the possibility of identifying a pronominal pragmatically. Hyams’s research is salient to our purposes because she directly counters alternative functionalist analyses of her data. She reports a study disconfirming a functionalist proposal that early learners use finite versus nonfinite verbs to encode different meanings (desideratives with non-finites; ongoing activity with finites). She further comments that that proposal, even if it were empirically valid, would not account for the correlation of nonfinite verbs with non-finite nouns, and of finites with finites. She rejects another functionalist proposal, namely that child learners conserve processing resources by omitting predictable or recoverable material, on the grounds that her data show an opposite, all-or-nothing association: more elaborated, finite subjects co-occur with more elaborated, finite, verbs whereas less elaborated, non-finite subjects co-occur with less elaborated, non-finite verbs. Hyams’s analysis is demonstratively formalist in the classic generative style: from the start, it presupposes the competence of child learners; it surveys data from multiple languages and assumes a common inner structure across all of them; it readily proposes an abstract feature that extends across verbs and

Formalism and Functionalism in Action 71 nouns; it seeks a theory-internal explanation for the gap between adult and child grammar; it assumes a modular grammar which separates syntactic and pragmatic competence. Moreover, throughout her exposition, Hyams remains committed to the high relevance of child language data to linguistic theory. Modern functionalists are also committed to the relevance of child language data, so that how language comes to children is a topic of extensive discussion and research within many streams of functionalism. The influential work of Elizabeth Bates and Brian MacWhinney, which Nichols (1984: 113–114) characterizes as a type of extreme functionalism, serves as an example. In one of their most-cited papers, Bates and MacWhinney (1987) presented their Competition Model as a direct challenge to parameter-setting accounts of language acquisition, on several grounds. First, they characterize the generativist model as predicting sudden, irreversible, changes in child grammar once a parameter is set, whereas Bates and MacWhinney see diversity in the data and gradual shifts over time, during which interval children’s grammars appear to cycle among competing hypotheses. Second, the generativist model predicts that children follow a uniform path in the acquisition of language, whereas Bates and MacWhinney perceive individual variability in learning processes, as children exploit their diverse strengths as learners. Bates and MacWhinney depict the Competition Model as a ‘data-driven, connectionist model. . . [which] . . . allows statistical properties in the input to play a major role in determining order of acquisition as well as the final state’ (1987: 158). With this, the Competition Model announces its functionalist-friendly consciousness of variation in multiple domains: in the input to learners, in their paths of acquisition, and in the language children produce. In fact, the Competition Model insists that variation is essential, given that compared to the infinite range of functions and meanings that language communicates, languages have limited formal resources for signaling those messages: words, morphology, syntax, intonation. Meanings therefore compete with each other to be mapped onto forms. A favorite example of Bates and MacWhinney is how some languages mark subjects morphologically, others through word order, while many others provide multiple markers for subjecthood, such as position, case, agreement, definiteness, etc. Where a language provides multiple cues, other meanings compete to map onto those same markers, so that identifying a subject is less a once-and-for-all decision than an exercise in recognizing what counts as the best fit among several candidates. The task for learners is to figure out, with respect to the target language, which markers of subjecthood are most reliable—that is, learners need to weigh the ‘cue validity’ of position, case, agreement, and so forth. They also need to weigh ‘cue strength’, which is a function of both cue validity and frequency. Through the process of interpreting cue strength

72  Formalism and Functionalism in Action and frequency in the input, learners converge on the formal properties of the target language. Bates and MacWhinney cite a study showing that for two-year-old learners of English, word order is the first cue they rely on in interpreting potentially ambiguous sentences, whereas two-year-old learners of Italian first rely on animacy (1987: 172–173). But for neither group does the process resemble a sudden, deterministic setting of a parameter. Rather, learners gradually zero in on the adult grammar. The Competition Model conceives of child language learning in functionalist terms at every step. It is of interest that Hyams (1999: 407–408) ends with a direct challenge to Bates and MacWhinney. She concedes that their Competition Model can explain how input may lead children to retreat from overgeneralization, for example if, in Hyams’s own data, children were to extend overt determiners from subject position of finite verbs to subject position of non-finite verbs. But what an input-driven model cannot explain, according to Hyams, is why child learners so rarely overgeneralize. To Hyams, children’s access to formal grammatical principles alone explains the absence of overgeneralization. Bates and MacWhinney do not take up this specific challenge issued by Hyams. However, in commentary on Hyams’s work on null subjects in child grammars cross-linguistically, they cite experimental evidence disputing the expected correlations across null subject-related phenomena: Englishlearning children’s grammars don’t exhibit the relevant formal grammatical features of the null-subject parameter (1989: 28). Bates and MacWhinney counter-propose that children adopt null-subject phenomena as their awareness of the statistical properties in the input advances. In this exchange, neither formalists nor functionalists satisfy the critique each makes of the other. Still, unlike early twentieth-century formalists, both groups invest child language data with high theoretical value.

5.4  In Debate About the Origin of Human Language A third case study of formalism versus functionalism in action turns away from the nitty-gritty of transitivity or telegraphic speech to a dramatic, unwieldy, interdisciplinary question at the nexus of linguistics, evolutionary biology, neurophysiology, primatology, and cognitive science: how did human language first arise, and how has it evolved? The question has intrigued laypeople and scholars for centuries and, since the 1990s, has had a prominent role in debate between formalists and functionalists. It is particularly difficult to resolve because of the great time depth involved, the sparsity of relevant material evidence, and perhaps (as some have remarked) linguists’ general lack of training in the abutting natural sciences. Chomsky’s work, provocative as ever, is a major stimulus behind the recent upsurge in publications on the origin of language. In texts and

Formalism and Functionalism in Action 73 interviews from the 1970s, he has evinced skepticism about the relevance of the default neo-Darwinian theory of evolution to the origin and development of language. In brief, neo-Darwinism assumes that natural selection routinely operates on the inheritable variation that random mutation introduces into species. Birds’ beaks are an often-cited example (Wang et al. 2017). As the class of what became birds developed from reptiles, the keratinous material that comprises birds’ beaks gradually emerged through natural selection, inhibiting the establishment of teeth. Individuals who benefited from having beak-like characteristics succeeded in meeting their need for food, survived to reproduce, and as a result their characteristics became entrenched in the community. Over very long spans of time, birds whose diets included seeds encased in hard shells went on to evolve short, strong beaks, while birds who fed on nectar evolved long, curved, probing beaks. The total picture of the origin and evolution of biological traits is of course far more complex, but its basis in neo-Darwinian evolution and natural selection is the consensus view. According to Chomsky, how bird beaks came to be and how language came to be have little in common. Although he steadfastly identifies language as a biologically (not culturally) based phenomenon, he suggests that it arose in human cognition essentially intact as an abrupt, adventitious outcropping. He speculates that language is a kind of side effect of other changes in the course of human evolution, most likely a sudden enlargement and complexification of the brain that comprised a ‘Great Leap Forward’ in human history (Chomsky 2005: 11). With this opinion, Chomsky adopts a ‘saltationist’ position, which is to say, he accepts that species can undergo large, precipitous, mutational change in as little as a single generation. (Saltation was a popular pre-Darwinian concept largely abandoned in modern evolutionary theory, although support for it exists in limited cases.) Moreover, Chomsky does not accept that language has adapted and complexified since its emergence through the shaping force of natural selection. It is not irrelevant that through the 1980s, the prominent American evolutionary theorist Stephen Jay Gould (1941–2002) developed a parallel anti-gradualist, anti-natural selective argument about the origin and development of language (Gould 2002). Gould also championed the notion of ‘punctuated equilibrium’, the claim that evolutionary change proceeds in bursts followed by periods of stasis, rather than through slow, steady change (Eldredge & Gould 1972). Gould’s contribution was to modulate and augment the received neo-Darwinian theory of evolution, not to repudiate it; Chomsky’s citations of Gould capitalize on the fact that, 200 years after Darwin’s birth, not all questions about evolution have been settled. Chomsky’s treatment of language as exceptional follows directly from his formalist commitments. Among those commitments is his drive to position linguistics as a branch of biology (earlier, psychology [Chomsky

74  Formalism and Functionalism in Action 1968: 1, among other places]), which should investigate language—that is, ‘I[nternal]-language’—using the tools and assumptions of the natural sciences, including that it is rooted in a dedicated faculty which is part of human genetic endowment, on a par with vision or digestion. Moreover, the very exceptionality of language, from Chomsky’s point of view, supports a foundational principle of generative formalism: although the general cognitive abilities of humans—memory, inference-making, attention, rational thought—might have been built up through natural selection, the content of the language faculty is ‘so eccentric, so opaque, so peculiarly linguistic’ (Bates, Thal, & Marchman 1991: 35) that it is impossible to view it as a reflex of cognition. Lightfoot (2000: 245) enlarges on the point: principles of universal grammar such as the heterogeneous conditions that block syntactic extraction from a subject noun phrase, are ‘dysfunctional . . . byproduct[s] of something else and not the result of adaptive change favoring survival’. The autonomy of the language faculty from general cognition suggests to Chomsky that it spun off from a ‘Great Leap Forward’, rather than having been honed by evolution. His anti-natural selective conception of the origin of language is furthermore consistent with the formalist assumption that language is essentially a vehicle for thought and reasoning, which has only incidentally been drafted into playing a role in information transfer. Chomsky’s views on language evolution are not only deeply formalist, but also distinctively generativist in their formalism. Hauser, Chomsky, and Fitch (2002) introduce an important distinction between ‘the language faculty in the broad sense’ (FLB) and ‘the language faculty in the narrow sense’ (FLN). The distinction seems to culminate Chomsky’s several efforts to pare down to the precise essence of his object of inquiry: deep structure, not surface structure; competence, not performance; I-language, not E-language; FLN, not FLB. The FLB subsumes the FLN, combining with it two additional systems which humans share with non-human species: a sensory-motor system (the sound patterns of a language alongside the means for externalizing them through speech—or, for signed languages, through movement in space), and a less well-defined conceptual-intentional system (a repository of speakers’ intended meanings). The sensory-motor and conceptual-intentional systems are external to the FLN, but the sites of interface with it. Chomsky proposes that the FLN is unique to humans and unique to language itself; it is ‘an abstract computational system alone, independent of the other systems with which it interacts’ and which comprises, perhaps as its only content, the capacity for recursion (Hauser, Chomsky, & Fitch 2002: 1571). By ‘recursion’ Chomsky refers to the property of ‘discrete infinity’, which is to say, the formally unlimited capacity of an operation to operate on itself: a clause can be embedded within a clause; a reduplicated form can itself be reduplicated. The proposal to reduce all forms of syntactic movement to the operation of Merge is a step in the direction of radically

Formalism and Functionalism in Action 75 reducing the computational mechanism. In a perfectly designed system, the contents of universal grammar would be nil or virtually nil, and the full complexity of the grammar of natural languages would (somehow) fall out from the interaction of the two external interfaces. In an essay clarifying the definitions of FLB/FLN, Fitch, Hauser, and Chomsky (2005: 180) depict the FLB as ‘language’ in the sense that biologists and psychologists have conventionally employed the term, roughly ‘the communication system used by humans’. The FLN refers to linguists’ conception of language in the sense of the ‘abstract core of computational operations’. Chomsky’s focus on the FLN is consistent with the thrust of the Minimalist Program, with which much of his work on the origin and evolution of language has been carried out. It also has the hallmarks of formalism: it materializes autonomy and modularity; it narrows the scope of inquiry, directing attention specifically to formal matters and deflecting into the FLB most of what functionalists value; it tries to define the innermost structure of human language. Functionalists have actively responded to Chomsky’s writings on the origin and evolution of language. So have certain high-profile formalists, and not uncritically (Pinker & Bloom 1990; Newmeyer 1991; Jackendoff & Pinker 2005). Both groups converge on certain themes in their critiques. Among them is reluctance to follow Chomsky in abandoning the relevance of neo-Darwinian evolution to language. Both functionalists like Bates, Thal, and Marchman (1991) and Givón (2002), and formalists like Pinker and Bloom (1990) make strong cases for the relevance of Darwin. Anderson (2013) is more equivocal, in a discussion of whether the human language faculty satisfies three Darwinian criteria that must be met for a trait to be a product of natural selection. The first two criteria are whether the language faculty is variable (yes, as indicated by the existence of language disorders), and whether that variability is heritable (yes, on the evidence of twin studies and familial language disorders). The third criterion is harder to assess: does the language faculty confer adaptive advantages? No one disputes that having an intact language is advantageous to humans compared with not having a language. But can that advantage be affirmed over the whole course of its emergence, which (outside of Chomsky’s saltationist views) would extend over a very long interval? The issue is sometimes stated as analogous to the problem of establishing the adaptive value of having 5% or 10% of an eye, granted that natural selection is a protracted process. Anderson, for one, considers the matter undecided, but still concludes that neo-Darwinism offers the best account of the origin of language. Another prominent theme in critique of Chomsky’s views arises from the evolutionary timeframe, a vexed problem since it involves diverse kinds of speculation and evidence, all disputed. Berwick and Chomsky (2016: 148–149) argue that language probably emerged for the first time between the appearance of anatomically modern humans 200,000 years ago, and the

76  Formalism and Functionalism in Action emergence of evidence for behaviorally modern humans 60,000 years ago. But in the absence of a saltationist, ‘Great Leap Forward’ origin for language, that is far too short an interval for a single mutation to propagate through an entire community. Nor does it account for the evidence, even if it is disputed, that Homo neanderthalensis (who diverged from Homo sapiens 500,000 years ago) had cultural practices advanced enough to warrant language (Dediu & Levinson 2014). If language emerged gradually via natural selection, 500,000  years might be sufficient. Other estimates go back to 600,000 years, or even much farther (Diller & Cann 2013). These are some of the bases on which functionalists, and some formalists, have objected to Chomsky’s specifically formalist account of the origin of language. In addition, some functionalists have reflected independently on the topic, from diverse points of view. In the early 1990s, two separate target articles (Pinker & Bloom 1990; Newmeyer 1991) on the topic of the origin of language were published alongside multiple peer commentaries. Among authors of those commentaries were several who spoke up from functionalist points of view. These include Bates and MacWhinney, Hopper, and Thompson—to cite only members of the dramatis personae in Chapter 4. Talmy Givón has given sustained attention to the puzzle of the origin of language, always speaking from a position that identifies language as ‘a profoundly adaptive biological phenomenon . . . impossible outside an evolutionary framework’ (Givón & Malle 2002: vii; see also Givón 2002). Interestingly, few functionalists assigned to Newmeyer’s middle-ofthe-road ‘external’ subgroup, Siewierska’s ‘structuralist-functionalists’, or Nichols’s ‘moderates’ (e.g., Van Valin; Halliday; Dik) have written extensively about the origin of language. In fact, the bulk of what could be called ‘functionalist-friendly’ scholarship on the origin of language has been produced not by linguists, but by psychologists and neuroscientists who adopt positions coherent with functionalism. One prominent contributor is psychologist Michael Tomasello (2014), who argues that language developed as humans extended the intentional pointing gestures of apes first to pantomime, then to arbitrary vocal conventions, in the course of specifically human cooperative communication. In Tomasello’s view, humans did so not by exploiting a language faculty, but by applying their general cognitive skills, intentionality, and capacity to imitate and analogize. He proposes a functionally organized developmental path along which human language moved from a pointing- and gesturebased grammar of requesting, to a grammar of informing, which mixes gesture with vocalization, to a grammar of sharing, in which vocal language dominates (2014: 292–295). Tomasello cites approvingly the usageoriented and construction-based functionalism of Goldberg, Langacker, and Croft: if not only words but grammatical constructions are conventionalized, arguably streamlining their transmission, then ‘naturally meaningful,

Formalism and Functionalism in Action 77 action-based gestures . . . could have at some point turned into full-blown communicative conventions, culturally-created and learned’ (p. 243). In other words, language came to humans neither through a saltationist breakthrough, nor in the way that beaks came to birds. Rather, starting with crude gestures, humans worked out more sophisticated and more arbitrary ways to communicate cooperatively, while the messages they communicated advanced in complexity. Neuroscientist Michael A. Arbib (2012, 2015) has also directly challenged Chomsky’s formalist account of language evolution. Arbib capitalized on research into ‘mirror neurons’ in the brain, first discovered in monkeys, which fire not only (for example) when an animal grasps something manually, but also when it observes another animal executing the same gesture. Arbib and his colleagues located mirror neurons for grasping in human brains in or near Broca’s area, the site in the frontal cortex conventionally associated with speech production. This is significant because, to Arbib, mirror neurons provide a mechanism by which gesture and the imitation of gesture could evolve into a signed human language, and eventually into a spoken language, making the leap from motor control of gestures to speech production. He conceives of the earliest protolanguages as holophrastic, representing ‘chunks’ that might comprise actors plus actions, or objects plus states or attributes. Early humans then decomposed holophrastic units to yield words. Eventually, they re-aggregated words into constructions, so that (like Tomasello) Arbib finds the functionalist construction grammars of Goldberg or Croft ‘more hospitable’ (Arbib 2015: 605) as an account of grammatical patterning than formalists’ autonomous modules. Danish cognitive scientist Morten H. Christiansen offers a different functionalist-friendly account of the origin of language. A 2013 article (expanded in Christiansen & Chater 2016) used computer modeling and corpus analysis to show how language processing might exploit multiple cues probabilistically, similar to the operation of vision and sensorimotor control. Tasked with identifying parts of speech, analysis of phonological cues such as syllable length, the number of initial consonants, and final devoicing resulted in around 60% correct identification of nouns, and 70% for verbs: ‘nouns tend to sound like other nouns and verbs like other verbs’ (Christiansen & Chater 2016: 16). Christiansen reports experimental studies of child learners and lexical decision tasks with adults that produced results consistent with those from computer modeling: evidently people, too, are faster and more accurate at making syntactic judgments of phonologically typical words. In real speech recognition, the domain of cues is larger than in experimental work, because hearers recruit distributional, prosodic, semantic, and pragmatic information to identify the most likely candidate. According to Christiansen, these findings bear on the origin of language because they undermine generativism’s notion of a dedicated language faculty. What they show instead is that human

78  Formalism and Functionalism in Action cognition constrained the evolution of language, and continues to shape language to make it learnable and accessible. Christiansen calls this ‘cultural evolution’: a process by which ‘socio-pragmatic considerations, the nature of our thought processes, and perceptuo-motor factors, as well as cognitive limitations on learning, memory, and processing. . . [are] amplified and embedded in language’ (2013: 43). To borrow the words of Bates, Thal, and Marchman (1991: 35), language is a ‘new machine built out of old parts’, an identifiably functionalist conception.

5.5 Gaps, Overlap, and Combat Between Formalism and Functionalism In this chapter we have viewed formalism and functionalism in media res, as each wrestled with complex language data. It is worthwhile taking a step back to survey each of these paired case studies for the evidence they give about the nature of formalism versus functionalism. In its treatment of word order phenomena, generative formalism grants a central role to theoretical issues: the focus is on building a theory of language with reference to language facts, rather than on the facts themselves. In commenting on the transition between the principles and parameters model and the Minimalist program, Chomsky (1995: 7) wrote that ‘the primary [task] is to show that the apparent richness and diversity of linguistic phenomena is illusory and epiphenomenal, the result of [the operation of] fixed principles under slightly varying conditions’. To formalists, ‘word order’ in the sense addressed in Chapter 5 is epiphenomenal: it’s a secondary consequence of the interaction of X-bar and case theories (in the terms of one stage in the development of generative theory). To functionalists, however, word order is a primary marker of how speakers organize discourse, whether according to the Prague Circle’s communicative dynamism or according to Givón’s principle of urgency. Therefore, there is a significant gap between formalism and functionalism in their treatment of word order phenomena. The two approaches don’t pursue the same goals, and, as Payne (1999: 157) remarks, they aren’t obviously complementary. A different logical relation holds in the case of transitivity. We saw that diverse species of formalism attribute (in)transitivity to a lexical feature inherent in verbs (or adpositions). Hopper and Thompson’s (1980) functionalist treatment locates transitivity more broadly. They subsume the classic, discrete, signal of transitivity (the number of arguments a verb selects) within a scenario of incremental assignment of transitivity according to the semantic properties of the verb, its subject, and its dependents. Ultimately, Hopper and Thompson conceive of transitivity as an instrument for foregrounding versus backgrounding information in discourse. Here the formalist/functionalist contrast is striking. However, rather than being in a disjoint relationship, the two overlap in that both attend to the argument structure

Formalism and Functionalism in Action 79 of the verb. Hopper and Thompson’s functionalism starts with the same observation that formalism makes, but builds beyond it. In contrast, the formalist and functionalist accounts of first language acquisition that we examined in Chapter 5 simply oppose each other. Hyams’s formalist initiative prioritizes competence, seeks out cross-linguistically valid principles that govern the form of child language, embraces abstraction and a modular organization of grammar—all moves that functionalists reject. In particular, Bates and MacWhinney’s Competition Model rejects the premise that child language development proceeds through deterministic setting of parameters within universal grammar. They replace it with the notion that learners weigh multiple, sometimes contradictory, cues as they sort through the evidence for the structure of a language, following an often circuitous path to the adult grammar. This difference signals the fundamental incompatibility of Hyams’s versus Bates and MacWhinney’s conceptions of first language acquisition. One of these views will not necessarily emerge as triumphant at the expense of the other; in fact, in the years since the publication of the two articles, both sides have revisited some of their claims and assumptions. But it is unclear how they could both ‘win’. Our final case study, debate about the origin of language, results in a similar opposition. Chomsky’s anti-gradualist, anti-selective position is bold and coherent with his overall linguistic theory. It is in direct competition with the better-developed, default, neo-Darwinian evolutionary theory, which even some formalists find persuasive. ‘Functionalist-friendly’ research has proposed alternatives to Chomsky’s account of the origin of language that are consistent with neo-Darwinian evolution. Again, it is not clear that one of these stances on the origin of language will necessarily succeed. But they cannot both do so. More likely, the descendants of one will displace the descendants of the other. In this way, we see that formalist and functionalist accounts can have various relations relative to each other. The particular positions of the two that come to light in Chapter 5 are not, of course, inevitable. That is to say, the incommensurability of formalist with functionalist accounts of word order isn’t inherent to applications of the two approaches to syntax; it is a product of the specific illustrations discussed here. Other examples would turn up overlap, or flat-out combat, in formalist versus functionalist research on syntax. Nevertheless, it is important to bring to light examples, even arbitrarily chosen ones, of the range of positions the two can hold relative to each other.

References Anderson, Stephen R. (2013). ‘What is special about the human language faculty and how did it get that way?’ In Rudolf Botha and Martin Everaert (Eds.), The

80  Formalism and Functionalism in Action evolutionary emergence of language (pp. 18–41). Oxford: Oxford University Press. Arbib, Michael A. (2012). How the brain got language. Oxford: Oxford University Press. Arbib, Michael A. (2015). ‘Language evolution: An emergentist perspective’. In Brian MacWhinney and William O’Grady (Eds.), The handbook of language emergence (pp. 600–623). Chichester, West Sussex: John Wiley and Sons. Arnold, Jennifer E., Anthony Losongco, Thomas Wasow, and Ryan Ginstrom. (2000). ‘Heaviness vs. newness: The effects of structural complexity and discourse status on constituent ordering’. Language 76: 28–55. Baker, C. L. and John J. McCarthy (Eds.). (1981). The logical problem of language acquisition. Cambridge, MA: MIT Press. Bates, Elizabeth and Brian MacWhinney. (1987). ‘Competition, variation, and language learning’. In Brian MacWhinney (Ed.), Mechanisms of language acquisition (pp. 157–193). Hillsdale, NJ: Lawrence Erlbaum Associates. Bates, Elizabeth and Brian MacWhinney. (1989). ‘Functionalism and the Competition Model’. In Brian MacWhinney and Elizabeth Bates (Eds.), The crosslinguistic study of sentence processing (pp. 3–73). Cambridge: Cambridge University Press. Bates, Elizabeth, Donna Thal, and Virginia Marchman. (1991). ‘Symbols and syntax: A Darwinian approach to language development’. In Norman A. Krasnegor, Duane M. Rumbaugh, Richard L. Schiefelbusch, and Michael StuddertKennedy (Eds.), Biological and behavioral determinants of language development (pp. 29–65). Hillsdale, NJ: Lawrence Erlbaum Associates. Berwick, Robert C. and Noam Chomsky. (2016). Why only us. Cambridge, MA: MIT Press. Bloomfield, Leonard. (1933). Language. New York: Henry Holt. Bresnan, Joan. (2001). Lexical-functional syntax. Malden, MA: Blackwell. Chomsky, Noam. (1957). Syntactic structures. The Hague: Mouton. Chomsky, Noam. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press. Chomsky, Noam. (1968). Language and mind (Enlarged ed.). San Diego: Harcourt Brace Javanovich. Chomsky, Noam. (1980). Rules and representations. New York: Columbia University Press. Chomsky, Noam. (1986). Knowledge of language: Its nature, origin, and use. New York: Praeger. Chomsky, Noam. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, Noam. (2005). ‘Three factors in language design’. Linguistic Inquiry 36(1): 1–22. Christiansen, Morten H. (2013). ‘Language has evolved to depend on multiple-cue integration’. In Rudolf Botha and Martin Everaert (Eds.), The evolutionary emergence of language (pp. 42–61). Oxford: Oxford University Press. Christiansen, Morten H. and Nick Chater. (2016). Creating language: Integrating evolution, acquisition, and processing. Cambridge, MA: MIT Press. Dediu, Dan and Stephen C. Levinson. (2014). ‘The time frame of the emergence of modern language and its implications’. In Daniel Dor, Chris Knight, and Jerome

Formalism and Functionalism in Action 81 Lewis (Eds.), The social origins of language (pp. 184–195). Oxford: Oxford University Press. Diller, Karl C. and Rebecca L. Cann. (2013). ‘Genetics, evolution, and the innateness of language’. In Rudolf Botha and Martin Everaert (Eds.), The evolutionary emergence of language (pp. 244–258). Oxford: Oxford University Press. Eldredge, Niles and Stephen Jay Gould. (1972). ‘Punctuated equilibria: An alternative to phyletic gradualism’. In Thomas J. M. Schopf (Ed.), Models in paleobiology (pp. 82–115). San Francisco: Freeman, Cooper and Company. Fitch, W. Tecumseh, Marc D. Hauser, and Noam Chomsky. (2005). ‘The evolution of the language faculty: Clarifications and implications’. Cognition 97: 179–210. Gazdar, Gerald, Ewan Klein, Geoffrey K. Pullum, and Ivan Sag. (1985). Generalized phrase structure grammar. Cambridge, MA: Harvard University Press. Givón, Talmy. (1988). ‘The pragmatics of word-order: Predictability, importance and attention’. In Michael Hammond, Edith A. Moravcsik, and Jessica R. Wirth (Eds.), Studies in syntactic typology (pp. 243–284). Amsterdam/Philadelphia: John Benjamins. Givón, Talmy. (2002). Bio-linguistics: The Santa Barbara lectures. Amsterdam/ Philadelphia: John Benjamins. Givón, Talmy and Bertram F. Malle (Eds.). (2002). The evolution of language out of pre-language. Amsterdam/Philadelphia: John Benjamins. Gould, Stephen Jay. (2002). The structure of evolutionary theory. Cambridge, MA: Harvard University Press. Haegeman, Liliane. (1991). Introduction to government and binding theory. Oxford: Blackwell. Hajičová, Eva. (1994). ‘Topic/focus and related research’. In Philip A. Luelsdorff (Ed.), The Prague School of structural and functional linguistics: A short introduction (pp. 245–275). Amsterdam/Philadelphia: John Benjamins. Hauser, Marc D., Noam Chomsky, and W. Tecumseh Fitch. (2002). ‘The faculty of language: What is it, who has it, and how did it evolve?’ Science 298: 1569–1579. Hockett, Charles F. (1958). A course in modern linguistics. New York: Macmillan. Hopper, Paul J. and Sandra A. Thompson. (1980). ‘Transitivity in grammar and discourse’. Language 56(2): 252–299. Hyams, Nina. (1999). ‘Underspecification and modularity in early syntax: A formalist perspective on language acquisition’. In Edith Moravcsik, Michael Darnell, Frederick Newmeyer, Michael Noonan, and Kathleen Wheatley (Eds.), Functionalism and formalism in linguistics, vol. I (pp. 387–413). Amsterdam/Philadelphia: John Benjamins. Jackendoff, Ray and Steven Pinker. (2005). ‘The nature of the language faculty and its implications for evolution of language (Reply to Hauser, Fitch, and Chomsky)’. Cognition 97: 211–225. Kayne, Richard S. (1994). The antisymmetry of syntax. Cambridge, MA: MIT Press. Lasnik, Howard. (2000). Syntactic structures revisited: Contemporary lectures on classic transformational theory. Cambridge, MA: MIT Press. Leopold, Werner F. (1971). ‘The study of child language and infant bilingualism’. In Aaron Bar-Adon and Werner F. Leopold (Eds.), Child language: A book of readings (pp. 1–13). Englewood Cliffs, NJ: Prentice-Hal. (Original published in 1941)

82  Formalism and Functionalism in Action Lightfoot, David. (2000). ‘The spandrels of the linguistic genotype’. In Chris Knight, Michael Studdert-Kennedy, and James R. Hurford (Eds.), The evolutionary emergence of language: Social functions and the origins of linguistic form (pp. 231–247). Cambridge: Cambridge University Press. Newmeyer, Frederick J. (1991). ‘Functional explanations in linguistics and the origins of language’. Language and Communication 11: 3–28. Nichols, Johanna. (1984). ‘Functional theories of grammar’. Annual Review of Anthropology 13: 97–117. Payne, Doris. (1999). ‘What counts as explanation? A functionalist’s account of word order’. In Edith Moravcsik, Michael Darnell, Frederick Newmeyer, Michael Noonan, and Kathleen Wheatley (Eds.), Functionalism and formalism in linguistics, vol. I (pp. 137–165). Amsterdam/Philadelphia: John Benjamins. Pinker, Steven and Paul Bloom. (1990). ‘Natural language and natural selection’. Behavioral and Brain Sciences 13: 707–727. Thomas, Margaret. (2002). ‘Development of the concept of “the poverty of the stimulus” ’. The Linguistic Review 19: 51–71. Tomasello, Michael. (2014). Origins of human communication. Cambridge, MA: MIT Press. Wang, Shuo, Josef Stiegler, Ping Wu, Cheng-Ming Chuong, Dongyu Hu, Amy Balanoff, Yachun Zhou, and Xing Xu. (2017). ‘Heterochronic truncation of odontogenesis in theropod dinosaurs provides insight into the macroevolution of avian beaks’. PNAS 114: 10930–10935.

6 Formalism and Functionalism Juxtaposed

6.1  Formalism and Functionalism in Silhouette This chapter juxtaposes formalism and functionalism, bringing the two together in silhouette. That is to say, the focus is less on their internal composition than on what each looks like when offset against the other. We take three perspectives in succession: first, we observe formalism and functionalism juxtaposed face to face, as mutually-critical conversational partners: how does each assess the other? Second, we compare the two as though oriented side by side, probing formalists’ versus functionalists’ views on two contentious matters important to them both: the matter of what counts as an explanation in linguistics, and the matter of the (non)complementarity of formalism and functionalism. Finally, we assemble the two approaches together as though back to back, a juxtaposition that reveals their asymmetrical positions within the field of linguistics.

6.2  Face to Face A first approach is to observe what linguist-engineers say to and about linguist-collectors, and vice versa. The corpus is not large. In fact, it is a motif in literature on formalism and functionalism to lament a dearth of direct, sustained, exchanges (Croft 1995: 490; Carstairs-McCarthy 1999: 426; Tallerman 1998: 423). For the most part, each approach discounts the other, so that scholars tend to speak, write, and circulate their ideas within their own segregated communities. Haspelmath (2000) brings the point to a head in the plaintive title of a review article, ‘Why Can’t We Talk to Each Other?’ 6.2.1  In Conversation This subsection examines imagined dialogues in which the two sides speak directly to each other, in two unusual texts that whimsically play out linguistic theory through interpersonal dynamics. Newmeyer (1998: 3–5) opens his exposition of formalism versus functionalism with a fictitious conversation

84  Formalism and Functionalism Juxtaposed between two new PhDs killing time as they wait to be interviewed for the same academic job. ‘Sandy Forman’, fresh out of MIT, speaks for formalism; ‘Chris Funk’, a graduate of the University of California at Santa Barbara, represents functionalism. The conceit allows Newmeyer to parade before readers the chief grounds for conflict, including whether or not communication is the primary function of language; the purported lack of functional motivation for structure dependence or gender marking; the difficulty of establishing non-circular criteria for what counts as ‘form follow[ing] function’; and the problem of accounting for language variation in the face of universal grammar. Newmeyer’s rhetoric probably intentionally polarizes Funk and Forman to heighten readers’ interest in the text that follows. Still, both sides use oppositional diction, and refuse to concede value of the other party’s positions. The self-attributions of Funk the functionalist include having common sense, being a reasonable person, and being committed to explaining grammatical structure, while—from Funk’s point of view—Forman’s pursuits are ‘trivial’, ‘mechanical’, ‘hopeless and ridiculous’: they are ‘artifacts’ that involve merely ‘pushing symbols around’. Forman the formalist lays claim to ideas that ‘make sense’, capture ‘profound generalizations’, and display the hallmarks of scientific progress. By Forman’s lights, Funk has adopted hazy or fuzzily speculative positions that ‘[don’t stand] the test of time’, about which Forman is skeptical or flat-out ‘not impressed’. Funk accuses Forman of caricaturizing functionalism; Forman criticizes Funk for falling into a logical trap. Each attributes to the other a ‘head-in-the-sand’ attitude and loyalty to a linguistic theory that lacks consistency and stability. Even granting to Newmeyer a certain license for dramatic overstatement, the exchange between Forman and Funk leaves the impression that face-toface conversation between the two is doomed to sterility. In fact, that may be some real linguists’ experience: that nothing can grow out of confrontation between formalism and functionalism, either because one or both sides harbor beliefs that the other won’t tolerate or because both are complacently encapsulated within their own views. Newmeyer returns to Forman and Funk at the end of his book (pp. 365–369), expressing the hope that they both would have found something in his text that led them to value the work of the other. Whether that hope is realized or not, Newmeyer’s sketch defines one posture that formalists and functionalists might take, a posture of minimal rapport. A second imaginary dialogue between the two parties models a different kind of give-and-take, and a different outcome. Carnie and Mendoza-Denton (2003) published a review of the two-volume proceedings of a major 1996 University of Wisconsin–Milwaukee conference on formalism and functionalism (Darnell et al. 1999). The review takes the form of the transcript of a fictionalized coffee-shop war of words between the two authors. Carnie plays himself, ‘a junior formal syntactician’, as does Mendoza-Denton, ‘a junior

Formalism and Functionalism Juxtaposed 85 variationist sociophonetician’ (p. 373). Although the main business of the text is to assess the books under review while giving voice to both formalist and functionalist readings of them, there is an arc to the conversation. It begins with playful name-calling, in which Carnie labels Mendoza-Denton with the same epithet that Newmeyer’s Forman directs at Funk—‘fuzzy’— while Mendoza-Denton echoes Funk’s term for Forman (‘symbol pusher’) in a swipe at Carnie’s devotion to ‘contentless formalisms’. But the two parties move on to discover some signs of compatibility, on the evidence from the conference papers that in phonology, at least, both formalist and functionalist evidence is salient. From there, Carnie and Mendoza-Denton proceed, circuitously, to grudging acceptance of shared assessments of some of the papers and of shared stances (that syntax can be said to, at least, interact with discourse; that synchronic variation may initiate diachronic change; that both functional pressures and formal structural properties have implications for typology). In the end they resolve that, after all, formalists and functionalists may not be so far apart: ‘it seems like a lot of the “debate” about formalism and functionalism revolves around people characterizing each other’s positions in extreme terms, setting up unrealistic expectations’ (p. 387). Carnie and Mendoza-Denton’s review makes a tidy narrative—a conversion story of sorts—that locates formalists versus functionalists differently relative to each other, compared to the stalemate in which Newmeyer’s Forman and Funk are locked. Neither tableau seems unrealistic. There likely have been actual conversations between differently oriented linguists who, if they have the patience to walk through evidence bearing on formalism versus functionalism together, converge in their views more than they had anticipated without abandoning the perspectives they started from. Carnie and Mendoza-Denton’s sketch models this benign outcome. There are, however, other imaginable positions that formalists and functionalists could take relative to each other: positions from which—in the popular folkloristic terms—‘symbol pushers’ unilaterally criticize ‘fuzzies’ or vice versa; from which one side asserts ownership of the other’s achievements; or from which the two come to view their goals as incompatible and their differences too extensive to be reconciled. 6.2.2  How Formal Is Functionalism? Continuing with our agenda to eavesdrop on exchanges between the two parties (but shifting from imaginary conversations to published literature), in this subsection we listen to how formalists have talked back to functionalists. A matter of particular interest is formalists’ claims that functionalists have converged toward formalism or appropriated quasi-formalist claims and assumptions.

86  Formalism and Functionalism Juxtaposed We know that formalism and functionalism differ in many ways: in their conception of the fundamental nature of language with respect to communication and cognition; in their essential theoretical constructs; in their methodological practices and treatment of data; in how they organize a grammar; in their relationships with other disciplines; in their typical working habits. With these dimensions of difference in mind, and moreover recalling the heterogeneity of modern functionalism, one might expect formalist critiques of functionalism to be both many and diverse. In fact, formalists engage in less analysis of the faults of functionalism than one might expect, and what critique they do offer is surprisingly convergent. An exception to this generalization is Frederick Newmeyer, a scholar who has written extensively about formalism and functionalism (1991, 1992, 1998, 1999, 2003, 2005a, 2005c). Although Newmeyer is a formalist by training and commitment, he happily adopts the characterization ‘functionalist Chomskyan’ (2005a: 232). He has looked critically and in depth into the conceptual basis and empirical research on both sides. His conclusion is that there is little principled incompatibility between them. On the way to that conclusion, a theme in his writing is that functionalism is more consistent with formalism than it may appear to be. That is to say, Newmeyer is quick to identify formalist-like traits within functionalism, or at least traits not threatening to formalism, and quick to undercut facets of functionalism that challenge formalism. An example of this comes to light in his treatment of a core principle of formalism, autonomy. Newmeyer (1998: 23–94) places autonomy at the center of his formalist’s-eye-view of the debate. He begins his exposition by defining three partially independent hypotheses about autonomy to which many formalists adhere: (1) that syntax is autonomous from semantics and discourse; (2) that competence is autonomous from performance, and from ‘the social, cognitive, and communicative factors contributing to use [of competence]’ (p. 24); and (3) that grammar is autonomous from general cognition. Functionalists do not necessarily accept any of these notions of autonomy. But from Newmeyer’s point of view, functionalist research that questions (1) through (3), ultimately fails to debunk the autonomy hypotheses. For example, with respect to (1), Goldberg’s Cognitive Construction Grammar identifies the central use of the English ditransitive construction ‘V NP NP’ (Sally baked her sister a cake; Goldberg 1995: 141–151) as involving transfer of an object by a volitional agent to a willing recipient; speakers then extend the construction to sanction its metaphorical use, as in The music lent the party a festive air. Newmeyer, however, denies that the association which Goldberg asserts between syntax and meaning threatens the autonomy of syntax: ‘simply showing that there is an intimate relationship

Formalism and Functionalism Juxtaposed 87 between form and meaning or even that metaphors have grammatical consequences is not sufficient to defeat [(1)]’ (1998: 45). Likewise, Newmeyer denies that Hopper’s notion of emergent grammar undermines (2), the autonomy of competence and performance. Emergent grammar claims that there is no ‘competence’ in the sense that language has no independent existence outside of specific discourse acts. Instead, it is made of prefabricated parts which speakers put together improvisationally. Newmeyer (pp. 59–64) objects that it is unclear how children could learn a language comprised of bits and pieces encountered incidentally—or how to account for uniformity in the course of child language acquisition. With respect to (3), Newmeyer turns up functionalists who, on his reading, either support the autonomy of grammar from general cognition (Givón); are agnostic (Langacker); or reject it while asserting what amounts only to an apparent counter-position, that study of language should be informed by cognitive psychology and neuroscience (Lakoff). Therefore, according to Newmeyer, functionalists either do not succeed in repudiating the bedrock formalist assumptions (1) through (3), or do not try to repudiate them. In a 1992 article, he took a similar approach on the topic of iconicity, with which many functionalists oppose formalist assumptions about arbitrariness. Newmeyer makes three arguments: that iconicity is irrelevant to generative grammar; that (in any case) iconicity poses no challenge to generative grammar; and that iconicity is already presupposed in some versions of generative grammar (1992: 756). He examines data and arguments put forth by most of the functionalist pantheon mentioned in Chapter 4, but reaches a conclusion about iconicity that parallels his conclusion about autonomy, namely, that functionalists fall not all that far from formalists. Not all readers agree, however. In commentary on Newmeyer, Hopper objected to what he sees as Newmeyer’s assimilation of functionalism into formalism: he characterized Newmeyer as ‘tr[ying] to interpret “functionalism” in terms of the formalist paradigm’ (1991: 46). Hopper might have the same reaction to an essay by Anderson in the proceedings of the University of Wisconsin–Milwaukee conference. Anderson’s assignment was to comment on ‘what formalists can learn from functionalists in syntax’ (1999: 112). His answer is, essentially, ‘not much’, since he finds little in functionalism that disrupts his formalist outlook. Most tellingly, what exceptions Anderson admits are instances where functionalism conforms to formalism: ‘some functionalist work is in fact close enough to that of your canonical formalist to make dialogue and reciprocal interaction fairly straightforward’ (p.  118). Note that Anderson’s syntax specifies that it is the approach of functionalism to formalism that facilitates dialogue—not

88  Formalism and Functionalism Juxtaposed mutual rapprochement, and not a movement of formalism in the direction of functionalism. Anderson cites as an example Kuno’s research on empathy, research which (as we saw in Chapter 4) is usually classified on the far edge of functionalism abutting formalism. That is, Anderson comes closest to investing value in functionalism when it speaks to him in formalist terms. In assessing functionalism, the question he asks is: ‘How formalist is functionalism?’ At the risk of overstating the point, one might notice that there are other options. Croft (1995), like Newmeyer, discusses the issue of autonomy. He defines a number of positions linguists have taken with respect to the autonomy of syntax, of grammar, and of language, noting that some seem to mix formalist and functionalist stances. He cites an analysis of Babungo, spoken in Cameroon, where noun class agreement phenomena are partially syntactic, and (in an apparent challenge to the autonomy of syntax) partially assigned by semantic or discourse factors. Croft asserts that it is possible to ‘smuggle’ semantics into syntax by replicating the relevant semantic features in the syntactic component (pp. 501–503). What is salient in Croft’s presentation of this problem, and its proposed solution, is that he conceives of it as part of a ‘near-continuum of [formalist] and functionalist theories’ (p. 526) rather than as an assimilation of one to the other, or an attempt to interpret one in the terms of the other. It remains to be determined what kind of account provides more insight into the nature of language: a mixed account that admits ‘smuggling’ across the border between syntax and semantics; a formalist one that prohibits ‘smuggling’ on principle; or a functionalist one that denies that there is such a border. Anderson (1999: 127–129) comments on Croft’s Babungo data, remarking that he finds in it no challenge to a strictly formalist account. In his analysis, ‘smuggling’ need not take place: general principles of agreement which also operate between modifiers and heads, or between verbs and their arguments, govern the syntactic components of Babungo agreement. Whatever residue of agreement is controlled by semantics and discourse is ‘not at all a matter of sentence syntax’ (p. 128) and would be accounted for elsewhere in the grammar. With that, Anderson implicitly rejects Croft’s notion of a continuum between formalism and functionalism. By extracting for separate treatment just those facets of the Babungo data that are amenable to formal analysis while setting aside facets that depend on semantics or discourse, he supplants Croft’s ‘mixed’ functional proposal with a purely formalist analysis. Another criticism that has been directed at functionalists has a different status. Many, including Anderson, have complained that functionalist texts are marred by inexplicit, vague, intuitive, or obscure language in the place of formalized terms and concepts (Labov 1987: 313;

Formalism and Functionalism Juxtaposed 89 Anderson 1999: 11, 115, 120; Newmeyer 2001: 114–116)—leading, fairly or not, to the epithet ‘fuzzy’. Anderson does not consider inadequate formalization intrinsic to functionalism, but simply a too-common practice among those who take that approach. Formalists, of course, value their trademark style of explicit definitions and axioms, which, in their eyes, renders their ideas clear and unambiguous. Insofar as adequately formalizing a claim means simply expressing it with clarity and precision, that ideal should transcend differences between the two approaches. The tricky part lies in adjudicating what counts as adequately precise language. As Hengeveld (1999) points out, some varieties of functionalism, such as Foley and Van Valin’s Role and Reference Grammar (1984) and Dik’s functional grammar (1978, 1989), formalize their proposals in their own idiosyncratic, intrinsically functionalist, idioms which still aspire to the kind of explicitness that formalists value. For other functionalists (Halliday, Givón) formalization is not a priority: for them, clarity and precision are achieved by other means. For still other functionalists, the critique misfires: ‘functional research has been severely held back by inappropriate or premature demands for rigor, abstractness, generality, and so on, stated as absolute, a priori criteria of science’ (Beaugrade 1994: 176). 6.2.3  How Functionalist Is Formalism? In short, much of formalist critique seeks to reconcile functionalism in the direction of formalism. Now we hand the microphone over to functionalists: how have they talked back to formalists? The overall tenor of functionalist critique of formalism is, unsurprisingly, more heterogeneous than the converse body of critique. It is also more profuse. In general, functionalists seem more self-conscious about tension between their ideas and those of formalists than vice versa (a point to which we return in section 6.4). Some functionalists comment in depth on formalism and on the gap between it and their own work. Other functionalists are more restrained in their objections to formalist policies and practices. Halliday belongs to this second group: he tended to put forth his own ideas rather than argue against views he rejected. But a critique is present nonetheless. For example, in discussing the labeling of grammatical categories, Halliday emphasized in the following passage that systemic functional linguistics does not start with an inventory of ready-made labels or concepts (like ‘theme’ or ‘NP’). . . . we do not use the name to impose artificial rigour on a language. Linguistic phenomena tend to be indeterminate, with lots of ambiguities, blends and ‘borderline cases’. The categories of analysis take this

90  Formalism and Functionalism Juxtaposed into account, allowing us to treat it not as something exceptional or dysfunctional, but as a natural and positive feature of an evolving system. (Halliday 1992 / 2003: 201) Halliday tacitly characterized formalist labels as ‘impos[ing] artificial rigour’, in opposition to the ‘natural and positive’ practices of systemic functional linguistics, which are emergent, flexible, and prototype-centered. In the same text and through similar rhetorical moves, Halliday dismissed formalist pursuit of a discipline based on the natural sciences; the sources and treatment of data typical of formalism; formalist notions of ‘explanation’; formalist reification of ‘structure’ over ‘system’ (pp. 200–212). Langacker, on the other hand, tends to be more straightforward in his criticism of formal grammar. About his own cognitive grammar, he wrote that ‘Some linguists view it with disdain, as it challenges fundamental dogmas and requires alternative modes of thought and analysis’ (2008: 5). On Halliday’s topic of the labeling of grammatical units, he continues, ‘Received wisdom—repeated in every linguistic textbook—holds that notions like noun and subject are purely grammatical categories not susceptible to any general semantic characterization’, whereas cognitive grammar ‘denies this by claiming that all valid grammatical constructs are symbolic, hence reducible to form-meaning pairings’ (pp. 5–7). In the same passage, Langacker links the matter to the formalist conception of the autonomy of syntax, to which he confronts cognitive grammar’s conception of ‘a gradation consisting solely in assemblies of symbolic structures’. He goes on to counter ‘the distorting lens of contemporary linguistic theory’; to mention what he perceives as a fallacy leading some formalists to exaggerate the scope of functionalists’ claims; and correct the confused understanding that positing overlap across the lexicon, syntax, and morphology necessarily means that cognitive grammarians do not distinguish among them. In these ways, Langacker directly engages with formalism, laying explicit boundaries down between cognitive grammar and opposing approaches. Givón sometimes takes a more combative tone in his critique of formalism, for example, in his 1979 analysis of ‘the threefold methodological delusion’ of generative grammar (p. 8); ‘the gutting of the database’ by which formalists impoverish linguistics (p. 22); and the ‘unprecedented’ synthesis of ‘theoretical vacuity’ and ‘empirical irresponsibility’ that comprises Chomsky’s proposals (p. 44). However, in other writings—without retreating from his repudiation of formalism—Givón (1995) argues that both functionalists and formalists fall into black-and-white thinking that warrants correction. He details the sources and consequences of numerous extreme positions on both sides, such as the ‘naïve functionalism’ of ‘[b]ecause structure is not 100% arbitrary, it must be 100% iconic’

Formalism and Functionalism Juxtaposed 91 juxtaposed against equally pernicious ‘naïve Platonic reductionism’ of ‘[e] ither all form-function pairings are non-arbitrary, or else functionalism is empirically vacuous’ (pp. 10–11). Against this kaleidoscopic corpus of criticism of the many, by the many, from many directions, it is reasonable to ask whether functionalists, like formalists, see the other party as approaching (or encroaching) on their own positions. The assimilatory stance which formalists sometimes adopt (discussed in subsection 6.2.2) also shows up in functionalist critiques of formalism. An example is Bates and MacWhinney’s (1990) response to Pinker and Bloom’s (1990) article questioning Chomsky’s views on the origin of language. Bates and MacWhinney advert to their idea that there are four levels of increasingly deeper commitment to the ideals of functionalism, starting with a Level 1 which (merely) accepts that communication plays a role in shaping language form at some stage of its historical development. They attribute Level 1 functionalism to Pinker and Bloom on the evidence that, contra Chomsky, Pinker and Bloom argue that natural selection found something to favor in the early emergence of human language, then capitalized on it. Bates and MacWhinney also find evidence elsewhere in Pinker’s and Bloom’s work of covert Level 3 functionalism, which asserts that children exploit form/function links in acquisition. Their evidence resides in Pinker’s (1984) and Bloom’s (1990) research showing that children deduce the syntactic properties of words from their meanings via what Pinker called ‘semantic bootstrapping’. On this basis, Bates and MacWhinney perceive that Pinker and Bloom are moving away from doctrinaire formalism. On a triumphant note, they title their commentary ‘Welcome to Functionalism’. In their response, Pinker and Bloom deflect the assimilatory welcome, edging away from Bates and MacWhinney. They dryly acknowledge that they do believe that semantics exists and that children make use of semantics in acquiring syntax, but protest that ‘it’s not clear who isn’t a functionalist in this sense’ (1990: 740)—including Chomsky. At issue in this exchange seems to be the foundational, unsettled matter of what the autonomy of syntax from semantics means to formalists, to functionalists, and within each side’s understanding of the other side’s position. There is also a stream of literature in which functionalists take assimilation to the maximum, to declare victory over formalists. Beaugrade (1994) confidently announces that ‘the tide is turning’ toward functionalism and that ‘I see no way the shift can be postponed if the science of language is to regain some solid footing’ (p. 1950). He asserts that ‘pure formalism runs aground on its own austere principles and is trapped in irresolvable dilemmas’ (p. 177), clearing the way for a full-voiced functionalism that replaces, instead of merely augments, formalism. Geeraerts (2010) announces that cognitive grammar has passed through

92  Formalism and Functionalism Juxtaposed pioneering and building phases into a stage of consolidation. He provides bibliographical data over a 20-year interval showing an abrupt increase in references to ‘cognitive grammar / linguistics’, with a correlative decline in references to ‘generative grammar / linguistics’. Enfield (2013) asserts that ‘[u]niversalist, rationalist approaches have recently gone from dominant to outdated’, marking the abandonment of ‘desiccated and socioculturally impoverished versions of linguistics’ (pp. 155). Enfield cites a host of publications as witness. A less sweeping, but provocative assimilatory claim argues that Chomsky himself has adopted functionalist ideas. Golumbia picks up on traits of the Minimalist Program, which in his eyes mark it as ‘hard to distinguish from functionalist theories’ (p. 40). As mentioned in Chapter 5, the Minimalist Program raises the profile of the two external interfaces that interact with the formal computational mechanisms of the language faculty ‘in the narrow sense’: the conceptual-intentional system and the sensorymotor system. Golumbia emphasizes that Chomsky’s renewed attention to the coordination of meaning with sound and his radical hollowing out of universal grammar—formerly, the hallmark of generative theory, and a focus for conflict between formalists and functionalists—brings minimalism closer to functionalism.

6.3  Side by Side A next step is to juxtapose formalism and functionalism laterally, in the sense of examining issues that bear equally on both approaches, and that may help us see where each stands relative to the other. Among issues that could be explored fruitfully in this way, two are conspicuous. The first is the fraught matter of explanation: how do formalists versus functionalists explain why the facts of language are what they are? What counts as an ‘explanation’ under each of these two approaches? The second issue is the question of complementarity: can formalism and functionalism be brought together in some way, so that each complements the other to provide a fuller account of the nature of language? Or do they compete for a unique role within a limited conceptual space? 6.3.1 Explanation The capacity to explain why the shape of language is what it is, is crucial to both formalists and functionalists. The two approaches rival each other, side by side, in laying claim to the prestige of being able to explain language facts. They also stand side by side in viewing their counterparts’ explanations skeptically. The issue is sufficiently developed that a matched pair of

Formalism and Functionalism Juxtaposed 93 terms have emerged: formalists pursue ‘internal’ explanations, and functionalists ‘external’ explanations. Generativists hold a proprietary relationship with the notion of explanation going back to Chomsky (1965: 30–37). In distancing himself from his formalist predecessors, Chomsky articulated three ascending levels of adequacy that a grammar must achieve. A grammar that meets the criterion of ‘observational adequacy’, accurately labels language data; ‘descriptive adequacy’, specifies the rules that generate those data; and ‘explanatory adequacy’, gives principled reasons for valuing one possible grammar over another, and makes predictions that go beyond the observed data. In pursuing the ideal of explanatory adequacy, Chomsky reacted against an austere streak in post-Bloomfieldian linguistics which abjured any departure from description to try to explain language facts: the post-Bloomfieldian Martin Joos (1907–1978) famously remarked that ‘Children want explanations, and there is a child in each of us; descriptivism makes a virtue of not pampering that child’ (1958: 96). Generativists explicitly reject that posture by aiming to explain the facts of language within terms of their own theory. In fact, prioritizing explanation over description is often identified as a distinguishing feature of generative theory. As an example of what it means to explain language facts from a formalist perspective within the terms of generative theory, recall Hyams’s (1999) account of children’s telegraphic speech. Her data showed that 83% of verbs that child learners left unmarked for finiteness appeared with null subjects, compared to 23% of verbs with finite inflections. Moving from observation to description, Hyams generalized the notion of (non-)finiteness from verbs to nouns, drawing on generative theory’s disposition to discover symmetry across syntactic classes. She proposed that young children generate non-finite verbs (verbs lacking inflectional features) in the context of non-finite nouns (which include null subjects, and nouns lacking determiners or plural markers). From here Hyams went on to seek an explanation for why finiteness correlates across verbal and nominal domains in child language, granted that this is not a property of the adult grammar. She recruited a proposal from then-current generative grammar that identified the heads of functional categories as pronominal in nature and therefore underspecified. She invoked a principle that blocks the construal of pronominals pragmatically when an anaphoric option is available. Adults acquire this principle as they gain expertise in language use, as it is not part of universal grammar. Until children acquire it, they allow pragmatic construal of pronominals in telegraphic speech. It is worthwhile reflecting on the structure of Hyams’s argument as an example of what it means to explain an observation about child language

94  Formalism and Functionalism Juxtaposed within generative theory. Notice that her explanation is built within a dense network of theory-internal proposals: about cross-categorical symmetry; about the pronominal character of the heads of functional categories; about the interaction of grammatical and pragmatic principles, with the latter driven by environmental learning and the former by universal grammar. One might add that Hyams’s starting point is also part of the explanation, in that ab initio she is disposed to expect children’s performance with functional categories to under-represent their actual competence. Hyams’s explanation for telegraphic speech is internal in the sense that it rests on an elaborate tissue of interlocking theory-internal proposals. Insofar as these proposals successfully hang together in the explanation of multiple linguistic facts, so far the whole apparatus of generative theory is strengthened. It should also be noted that internal explanation in generative theory often invokes the innateness of universal grammar (Hoekstra & Kooji 1988). In the case at hand, Hyams assumes that child learners bring intact to the context of learning elaborate, abstract knowledge about functional categories and their properties, knowledge that cannot be attributed to general cognitive skills. All that they need to learn from environmental input is the pragmatic principle that anaphoric identification of a pronominal preempts its identification through discourse. The readiness of formalists to locate elaborate, abstract knowledge of language in an innate language faculty is, of course, a point of maximum contention with functionalists. For many functionalists, to say that ‘XYZ is innate’ does not and cannot serve as an explanation for the existence of XYZ; it merely articulates an unsupported claim. In the words of Givón, ‘without reference to function and evolution no explanation of the structural properties of an organism is possible’ (1979: 22; emphasis in the original). Givón’s objection to the role that generativists assign to innateness (which many have developed: Derwing 1973: 63–77; Sampson 2005; Fischer 2007: 71–74) appears in an extended rebuke of their conception of what counts as an explanation for linguistic facts (1979: 3–22). The gist of Givón’s complaint is that formalism rules out a priori what seem to him to be the obvious, compelling sources of explanation: meaning, discourse pragmatics, processing constraints, human cognitive structure, diachronic change, and so forth. All of these are touchstones of functionalism’s conception of how to explain language. In their place, according to Givón, formalists create an artificial model of language, dub it a ‘theory’, and then invest that theory with explanatory power. Pivoting now to focus on functionalist conceptions of explanation, functionalists—like their formalist counterparts—also claim special insight into explanation. But the grounds on which they do so differ. Insofar as functionalists interpret the motto ‘form follows function’ to mean that function

Formalism and Functionalism Juxtaposed 95 imposes a shape on language forms, then the relevance of exactly those forces that formalists reject come into play: meaning, discourse pragmatics, processing constraints, human cognitive structure, diachronic change, etc. (Notice that the conventional characterization of these forces as ‘external’, privileges a formalist view of them, in the sense that they are external to the core language faculty. From a functionalist point of view, meaning, discourse, processing, etc., are ‘internal’, not ‘external’, factors.) Speaking for functionalism, Payne (1999: 142–143) provides a memorable analogy. Holding up a (randomly chosen, unremarkable) leaf she asks, ‘Why is this leaf flat?’ A formalist might respond that the leaf is flat because more of its cells are arranged in one dimension than in the other. According to Payne, this ‘explanation’ misfires because it merely re-describes what it means to be ‘flat’. Another formalist may respond that the leaf is flat because of its genes; this corresponds to the invocation of innateness. A functionalist, on the other hand, might respond that the leaf is flat because flatness is functional: flatness enhances the leaf’s exposure to sunlight, which therefore increases photosynthesis. To Payne, only this third response is truly explanatory: leaves are flat because their form follows their function of converting light into chemical energy. Newmeyer (2005c: 174–175) digs deeper into form and function in distinguishing two views of their relationship in the context of explanation. Some functionalists seek evidence directly connecting specific functions to specific forms; others connect functions indirectly to forms through constraints on acquisition and language change. The first option Newmeyer calls ‘atomistic functionalism’, the second ‘holistic functionalism’. Although Newmeyer concedes a place for ‘holistic functionalism’ in which functional pressures act indirectly on form, he rejects atomistic functionalism, which is prevalent in functionalist literature. He gives as an example Dik’s (1989: 215) analysis of English ditransitive sentences, which asserts that form connects directly to function in ditransitive structures. In The man gave the book to the boy, form reflects the functional principle of iconicity, since the order of major elements in the sentence represents the actual transaction originating in the agent (man), who transmits the object (book) into the possession of the recipient (boy). Moreover, according to Dik, form also reflects function in the alternative sentence, The man gave the boy the book, in the guise of a principle that prioritizes the animate, human noun boy over an inanimate book. In this way, atomistic functionalism purports to link every element in grammar point-by-point to a functional motivation. Newmeyer objects that in atomistic functionalism the scope of potential motivations is large, and indeterminant: iconicity, prototypicality, markedness, prominence, topic maintenance, focus, economy, parsing ease, memory constraints, and many other factors have been nominated. Functionalists may be at ease with such

96  Formalism and Functionalism Juxtaposed an open-ended inventory of explanatory instruments. In the language of Payne’s metaphor, some leaves are flatter (or larger, differently shaped, differently colored) than others because behind the full scope of forms lies a large inventory of explanatory functions, including maximizing photosynthesis; shade; rainfall; humidity; length of the growing season, etc. For functionalists, this is all grist for the mill. But many formalist critics agree with Lass (1980: 70), who remarked (in a discussion of functional explanation for language change) on the ‘virtual invulnerability of functional arguments’ in that ‘[t]he trouble with arguments like this is that you can’t lose’. As a final example of what emerges in comparing formalist and functionalist conceptions of explanation side by side, some functionalists have elaborated on generative grammar’s notion of explanatory adequacy. Initiated by Dik (1989), Butler (2003, Part 2: 485–489) exploded Chomsky’s idea of three levels of adequacy with the goal of explaining ‘why the language we describe is as it is’ under a new, expanded, definition of explanatory adequacy. To Butler, a grammar that explains language must meet criteria of discoursal adequacy (taking into account authentic, dynamic data produced by speakers in particular relations to each other); sociocultural adequacy (with reference to the relevant social and cultural environment, and the immediate environment of the preceding discourse); psychological/cognitive adequacy (taking on processing; inferencing; the mental representation of communicative intentions); and acquisitional adequacy (grammars must show how children acquire not only the core grammar and lexicon, but how they arrive at the norms for discourse in the language). With this enlarged list of desiderata, Butler mapped out an ambitious, maximally external, functionalist notion of what counts as an explanation. It contrasts with a formalist notion of explanation such as that assumed by Hyams, which looks inward to assess the theoryinternal coherence and consistency of a claim for why language is what it is. 6.3.2 (Non)complementarity Surveying the mutual criticism of formalism and functionalism, and working through their contrasting notions of what counts as an explanation, sharpens our sense of how the two parties differ. In this subsection we continue to juxtapose the two side by side to ask what kinds of relationships might hold between them: if not harmony, at least complementarity—or not. As it turns out, there is no consensus about how to position formalism and functionalism relative to each other. But it is important to understand the range of positions that participants in the debate take. Among attempts to articulate a complementary relationship between formalism and functionalism, one position acknowledges the different domains of their concerns while incorporating the formalist conception

Formalism and Functionalism Juxtaposed 97 of modularity. This is the stance of ‘formal’ or ‘conservative’ functionalists, who assign functional factors to a separate component of the grammar. Kuno makes the point clear: given a linguistic process that is governed by both syntactic and, say, discourse factors, the syntactic aspect will be formulated in the syntactic component, while discourse factors will be described in the discourse component of the grammar. . . . There need not be any disagreement between the two. (Kuno 1987: 1) In fact, this position seems to be tacitly assumed by some formalists tout court. Recall that Hyams (1999) invokes a pragmatic principle which interacts with children’s grammatical competence to account for telegraphic speech. Hyams’s concern is with grammatical competence, and she presupposes the autonomy of syntax from pragmatics. Therefore she can incorporate that pragmatic principle into her argument, adopting Kuno’s no-disagreement stance. Kuno and Takami (1993: 165) urge linguists to be on the lookout for points like this where syntactic factors interact with discourse or pragmatics. They criticize generativists for proposing purely formal accounts of phenomena that they feel can be better accounted for by the interaction of syntactic and functional constraints. Among those phenomena, Kuno and Takami include Heavy NP Shift, reflexives in picture nouns, and sentences with multiple WH words. Formal functionalists extend formalist methodology and practices to data analysis: they assume the sentence is the basic unit of analysis; freely idealize language data; rely on native speaker intuitions; downplay inter-speaker variation. They are undisturbed by the epistemological gaps between formalism and functionalism, and remain serenely confident that the two approaches, though distinct, can work together to make the best account of linguistic patterns. A different stance on the complementarity of formalism and functionalism appears in the work of Newmeyer. He staunchly defends the compatibility of the two on the grounds that grammar can be both autonomous and externally motivated (1998: 365–369, 1999: 469, 2005b: 277–278). Earlier in this chapter, we reviewed his claims that functionalism is more consistent with formalism than it may at first appear. In addition to this perception, Newmeyer sees the same developments in the Minimalist Program and biolinguistics that Golumbia (2008) remarked on as evidence of incipient compatibility between formalism and functionalism. Among that evidence, Newmeyer includes erosion of a commitment to the autonomy of syntax—which he laments (2005b:

98  Formalism and Functionalism Juxtaposed 278–279)—and Chomsky’s (2005) recognition of the ‘third factor’, that is, the importance of principles not specific to language that bear on language design. It is important to recognize that the purported convergence of recent generative grammar with functionalism does not resemble Kuno’s deliberate revision of formalism to accommodate functional factors. It is also not due to the success of an assimilatory initiative on the part of functionalists. Nor does it derive from self-conscious rapprochement of generativists in the direction of functionalists. Rather, it is an independent development within generative theory, a development that, arguably, coincides with views that functionalists also hold. In this scenario, the Minimalist Program approaches functionalism because, unprovoked by functionalism, generative grammar has discovered the salience to linguistic theory of discourse, language processing, and so forth. There are also participants in the debate who do not see formalism and functionalism as complementary. In particular, Newmeyer’s alignment of the two is controversial. Recall that Hopper objected to what he perceived as unwarranted assimilation of functionalism into formalism. Likewise, Foolen (2002: 99) wrote that for at least some functionalists, ‘only a functionalist approach from the beginning to the end can lead to a proper description and explanation’. He went on to quote Langacker’s insistence that functional considerations are necessarily foundational, not subsidiary, to the analysis of language. D’Alessandro and van Oostendorp (2017: 3) object that integration of the two into ‘a mosaical view’ cannot be achieved at present because the two adhere to different definitions of the object of study and rely on different data sources. Butler spells out a strong anti-complementarian stance in a response to Newmeyer’s (2005b) review of Butler (2003): formalism and functionalism are simply incompatible, because ‘there is no room in Chomsky’s theory, now as in the past, for the view that usage of language is itself a determining factor in the synchronic grammar of a language as well as in the process of language change’ (Butler 2006: 205). To illustrate what he sees as the essential incompatibility of formalism and functionalism, Butler drills down on the relation of syntax and semantics. Many, including Newmeyer, have pointed out that throughout the history of a theory dominated by attention to syntax, generativists have in fact assigned various supporting roles for semantics: in Logical Form; in the specification of theta-roles; in the conceptual-intentional system. But to Butler, ‘[i]n a functionalist account, the claim is not that we can give a semantic interpretation to syntactic phenomena, but that semantics is the driving force, and syntax reflects, at least in part, the meanings that it is there to convey’ (p. 208). On this and other bases, Butler denies that formalism and functionalism are converging, or that there are grounds for anticipating convergence.

Formalism and Functionalism Juxtaposed 99 Participants in the debate between formalism and functionalism have staked out at least these three positions relative to each other. What is surprising is that although the discussion spans contrasting, and strongly held, points of view, exchanges between the two are still relatively sparse. The best sources are texts that expressly contrast the two approaches (e.g., Darnell et al. 1999), and target articles in scholarly journals offset by accompanying peer commentary (e.g., Pinker & Bloom 1990; Newmeyer 1991, 2003 [with commentary surrounding Newmeyer 2005a]; Evans & Levinson 2009 [extending to Rooryck et al. 2010]). Target articles with peer commentary are designed to provoke discordant opinions, which may exaggerate the expression of conflict between formalists and functionalists. Conflict is good and strengthens the discipline, but it also has a polarizing effect, muting points of view that lie between the extremes, and thus may paradoxically increase insularity on both sides.

6.4  Back to Back There remains an additional juxtaposition of formalism and functionalism. Aligning the two back-to-back calls attention to an important imbalance, namely, that they are not arrayed symmetrically in disciplinary space. Unlike the imaginary conversational partners Forman and Funk, or the interlocutors in Carnie and Mendoza-Denton (2003), real participants in real debates about formalism and functionalism do not hold congruent positions relative to each other. At least in North America and East Asia, formalism is entrenched as the default to which functionalism reacts. In Europe, functionalism has a higher profile, but it still seems to adopt a defensive posture relative to formalism. As evidence for this asymmetry, one notices that remarks about the lack of communication across the two parties are largely made by functionalists (both North American and European), not by formalists. It was a functionalist who asked ‘Why Can’t We Talk to Each Other?’ (Haspelmath 2000), and another who titled an article ‘What Formalists Seem Not to Understand About Functionalists’ (Nuyts 1986). One can hardly imagine formalists posing Haspelmath’s question, or specifying what functionalists do not understand about formalists. Furthermore, functionalists seem to take the presence of formalism for granted in ways that formalists rarely reciprocate. For example, Geeraerts (2010) announces the ascendance of functionalism, but throughout his article keeps measuring that success against the status of generative theory. Moreover, it seems significant that no companion text to Newmeyer (1998) has appeared, in which an avowed functionalist would introduce and explain the debate from a sympathetic but pro-functionalist point of view. These observations support a sense that, contra Battistella (2000: 431), the tableau is not a simple one in which formalists ‘push’ and functionalists

100  Formalism and Functionalism Juxtaposed ‘pull’. Rather, formalists occupy a dominant position in the sociology of the field, and do not look over their shoulders anxiously at functionalists. In contrast, functionalists are on guard against their counterparts, expecting to have to argue for the legitimacy of their own views. Comparison of the books by Moro (2016) and by Everett (2012) with which this text began provides an illustration. Moro the formalist brushes away concerns that have absorbed the careers of functionalists in an offhand remark: the capacity of language to transmit meaning and to subserve communication is, to Moro, merely ‘trivial’ (2016: 2). Everett the functionalist, on the other hand, raises, explains, and responds to a large array of formalist claims, among them the autonomy (or not) of grammar, cognition, and culture; brain specialization and language; the status of recursion and evidence for it in Pirahã; etc. It is worth adding that, even if this impression of the present-day disposition of formalism relative to functionalism is accurate, it has more sociological than theoretical significance. Formalist versus functionalist tendencies have been in tension for a long time in linguistics; what rises to the surface at any one moment isn’t a judgment on its overall adequacy as an account of the nature of language. Other factors affect the bearing of formalism versus functionalism. For example, in the current century, the alliance of formalism with the natural sciences boosts its public profile. Generative formalism has also benefitted from the sustained leadership of a single perseverant, charismatic figure, even acknowledging that not all formalists do align themselves with Chomsky. (A paradox lies here, in that Chomsky has consistently asserted that his ideas represent the views of a tiny minority of linguists [2004: 67–69]—an opinion few observers share.) As opposed to these public-relations advantages enjoyed by formalists, functionalism is very disparate, with no dominant figure, school, or even geographical locale. Functionalism is also averse on principled grounds to the imposition of uniformity. Many of its adherents seem most at home in the relatively private act of collecting, handling, and assessing the actual data of language: one thinks of Halliday’s close examination of writers’ options for building cohesion in a text, Goldberg’s analyses of the many constructions that make up a language, or Langacker’s sketches representing his differential breakdown of kernel of corn versus corn kernel. In contrast, many formalists make claims about language that display counter-intuitive audacity and a willingness to take risks that adds glamour to the public face of the linguist-engineer: the insight that looked beyond the surface of English auxiliaries to recognize abstract patterns of remarkable regularity; Hyams’s readiness to impute to pre-schoolers elaborate sensitivity to functional categories; Chomsky’s saltationist stance. Juxtaposing the two approaches from diverse angles brings this matter to light, as well

Formalism and Functionalism Juxtaposed 101 as bringing to light more substantive differences between formalism and functionalism.

References Anderson, Stephen R. (1999). ‘A formalist’s reading of some functionalist work in syntax’. In Edith Moravcsik, Michael Darnell, Frederick Newmeyer, Michael Noonan, and Kathleen Wheatley (Eds.), Functionalism and formalism in linguistics, vol. I (pp. 111–135). Amsterdam/Philadelphia: John Benjamins. Bates, Elizabeth and Brian MacWhinney. (1990). ‘Welcome to functionalism’ [Commentary on Pinker & Bloom 1990]. Behavioral and Brain Sciences 13: 727–728. Battistella, Edwin. (2000). [Review of the book Language form and language function]. Journal of Linguistics 36: 431–439. Beaugrande, Robert de. (1994). ‘Function and form in language theory and research: The tide is turning’. Functions of Language 1(2): 163–200. Bloom, Paul. (1990). ‘Syntactic distinctions in child language’. Journal of Child Language 17(2): 343–355. Butler, Christopher S. (2003). Structure and function: A guide to three major structuralfunctional theories. Part 1: Approaches to the simplex clause. Part 2: From clause to discourse and beyond. Amsterdam/Philadelphia: John Benjamins. Butler, Christopher S. (2006). ‘On formalism and functionalism: A reply to Newmeyer’. Functions of Language 13(2): 197–227. Carnie, Andrew and Norma Mendoza-Denton. (2003). ‘Functionalism is/n’t formalism: An interactive review of Darnell et al. (1999)’. Journal of Linguistics 39: 373–389. Carstairs-McCarthy, Andrew. (1999). [Review of the book Language form and language function]. Anthropological Linguistics 41: 426–429. Chomsky, Noam. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press. Chomsky, Noam. (2004). The generative enterprise revisited: Discussions with Riny Huybregts, Henk van Riemsdijk, Naoki Fukui, and Mihoko Zushi. Berlin: Walter de Gruyter. Chomsky, Noam. (2005). ‘Three factors in language design’. Linguistic Inquiry 36: 1–22. Croft, William. (1995). ‘Autonomy and functionalist linguistics’. Language 71: 490–532. D’Alessandro, Roberta and Marc van Oostendorp. (2017). ‘On the diversity of linguistic data and the integration of the language sciences’. Frontiers in Psychology 8: Article 2002. https://doi.org/10.3389/fpsyg.2017.02002 Darnell, Michael, Edith Moravcsik, Frederick Newmeyer, Michael Noonan, and Kathleen Wheatley (Eds.). (1999). Functionalism and formalism in linguistics: Volume I: General papers. Vol. II: Case studies. Amsterdam/Philadelphia: John Benjamins. Derwing, Bruce L. (1973). Transformational grammar as a theory of language acquisition. Cambridge: Cambridge University Press. Dik, Simon C. (1978). Functional grammar. Amsterdam: North-Holland.

102  Formalism and Functionalism Juxtaposed Dik, Simon C. (1989). The theory of functional grammar. Part 1: The structure of the clause. Dordrecht: Foris. Enfield, N. J. (2013). ‘What I’m reading: Language, culture, and mind: Trends and standards in the latest pendulum swing’ [Review of the book Language: The cultural tool]. Journal of the Royal Anthropological Institute 19: 155–169. Evans, Nicholas and Stephen C. Levinson. (2009). ‘The myth of language universals: Language diversity and its importance for cognitive science’. Behavioral and Brain Sciences 32(5): 429–448. Everett, Daniel L. (2012). Language: The cultural tool. New York: Random House. Fischer, Olga. (2007). Morphosyntactic change: Functional and formal perspectives. Oxford: Oxford University Press. Foley, William A. and Robert D. Van Valin Jr. (1984). Functional syntax and universal grammar. Cambridge: Cambridge University Press. Foolen, Ad. (2002). [Review of the book Language form and language function]. Functions of Language 9(1): 87–103. Geeraerts, Dirk. (2010). ‘Recontextualizing grammar: Underlying trends in thirty years of cognitive linguistics’. In Elzbieta Tabakowska, Michel Choinski, and Lukasz Wiraszka (Eds.), Cognitive linguistics in action: From theory to application and back (pp. 71–102). Berlin: Walter de Gruyter. Givón, Talmy. (1979). On understanding grammar. New York: Academic Press. Givón, Talmy. (1995). Functionalism and grammar. Amsterdam/Philadelphia: John Benjamins. Goldberg, Adele E. (1995). Constructions: A construction grammar approach to argument structure. Chicago: University of Chicago Press. Golumbia, David. (2008). ‘Minimalism is functionalism’. Language Sciences 32: 28–42. Halliday, M. A. K. (2003). ‘Systemic grammar and the concept of a “science of language” ’. In Jonathan Webster (Ed.), Collected works of M. A. K. Halliday. Vol. 3. On language and linguistics (pp. 199–212). London: Continuum. (Original work published 1992) Haspelmath, Martin. (2000). ‘Why can’t we talk to each other?’ [Review of the book Language form and language function]. Lingua 110: 235–255. Hauser, Marc D., Noam Chomsky, and W. Tecumseh Fitch. (2002). ‘The faculty of language: What is it, who has it, and how did it evolve?’ Science 298: 1569–1579. Hengeveld, Kees. (1999). ‘Formalizing functionally’. In Edith Moravcsik, Michael Darnell, Frederick Newmeyer, Michael Noonan, and Kathleen Wheatley (Eds.), Functionalism and formalism in linguistics, vol. II (pp. 93–105). Amsterdam/ Philadelphia: John Benjamins. Hoekstra, Teun and Jan G. Kooji. (1988). ‘The innateness hypothesis’. In John A. Hawkins (Ed.), Explaining language universals (pp. 31–55). Oxford: Basil Blackwell. Hopper, Paul J. (1991). ‘Functional explanations in linguistics and the origins of language’ [Commentary on Newmeyer 1991]. Language and Communication 11(1/2): 47. Hyams, Nina. (1999). ‘Underspecification and modularity in early syntax: A  formalist perspective on language acquisition’. In Edith Moravcsik, Michael Darnell, Frederick Newmeyer, Michael Noonan, and Kathleen Wheatley (Eds.),

Formalism and Functionalism Juxtaposed 103 Functionalism and formalism in linguistics, vol. I (pp. 387–413). Amsterdam/ Philadelphia: John Benjamins. Joos, Martin. (1958). Readings in linguistics: The development of descriptive linguistics in America since 1925 (2nd ed.). New York: American Council of Learned Societies. Kuno, Susumu. (1987). Functional syntax: Anaphora, discourse and empathy. Chicago: University of Chicago Press. Kuno, Susumu and Ken-Ichi Takami. (1993). Grammar and discourse principles: Functional syntax and GB theory. Chicago: University of Chicago Press. Labov, William. (1987). ‘The overestimation of functionalism’. In René Dirven and Vilém Fried (Eds.), Functionalism in linguistics (pp. 311–332). Amsterdam/ Philadelphia: John Benjamins. Langacker, Ronald W. (2008). Cognitive grammar: A basic introduction. Oxford: Oxford University Press. Lass, Roger. (1980). On explaining language change. Cambridge: Cambridge University Press. Moro, Andrea. (2016). Impossible languages. Cambridge, MA: MIT Press. Newmeyer, Frederick J. (1991). ‘Functional explanation in linguistics and the origins of language’. Language and Communication 11(1/2): 3–28. Newmeyer, Frederick J. (1992). ‘Iconicity and generative grammar’. Language 68: 756–796. Newmeyer, Frederick J. (1998). Language form and language function. Cambridge, MA: MIT Press. Newmeyer, Frederick J. (1999). ‘Some remarks on the formalist – functionalist controversy in linguistics’. In Edith Moravcsik, Michael Darnell, Frederick Newmeyer, Michael Noonan, and Kathleen Wheatley (Eds.), Functionalism and formalism in linguistics, vol. I (pp. 469–486). Amsterdam/Philadelphia: John Benjamins. Newmeyer, Frederick J. (2001). ‘The Prague school and North American functionalist approaches to syntax’. Journal of Linguistics 37(1): 101–126. Newmeyer, Frederick J. (2003). ‘Grammar is grammar and usage is usage’. Language 79: 682–707. Newmeyer, Frederick J. (2005a). ‘A reply to critiques of “Grammar is grammar and usage is usage” ’. Language 81: 229–236. Newmeyer, Frederick J. (2005b). [Review of the book Structure and function: A guide to three major structural-functional theories]. Functions of Language 12(2): 275–283. Newmeyer, Frederick J. (2005c). Possible and probable languages. Oxford: Oxford University Press. Nuyts, Jan. (1986). ‘What formalists seem not to understand about functionalists’. Belgian Journal of Linguistics 1(1): 225–237. Payne, Doris. (1999). ‘What counts as explanation? A functionalist’s account of word order’. In Edith Moravcsik, Michael Darnell, Frederick Newmeyer, Michael Noonan, and Kathleen Wheatley (Eds.), Functionalism and formalism in linguistics, vol. I (pp. 137–165). Amsterdam/Philadelphia: John Benjamins. Pinker, Steven. (1984). Language learnability and language development. Cambridge, MA: Harvard University Press.

104  Formalism and Functionalism Juxtaposed Pinker, Steven and Paul Bloom. (1990). ‘Natural language and natural selection’. Behavioral and Brain Sciences 13: 707–727. Rooryck, J., N. V. Smith, A. Liptak, and D. Blakemore (Eds.). (2010). Special issue on Evans and Levinson’s ‘The myth of language universals’. [Special issue]. Lingua 120(5). Sampson, Geoffrey. (2005). The ‘language instinct’ debate. London: Continuum. Tallerman, Maggie. (1998). [Review of the book Language form and language function]. Studies in Language 24: 423–439.

7 Conclusion

7.1  Engineers and Collectors, Revisited In this text, we first encountered formalist and functionalist linguistics in two books written for a non-specialist readership. We probed the history of linguistics for earlier attestations of the general orientation of the two approaches, then surveyed varieties of mid-twentieth-century formalism and functionalism up to the present day. Next, we sampled analyses by linguist-engineers versus linguist-collectors on three topics: word order and transitivity in syntax; child language learning; and ideas about the origin of language—analyses which are sometimes coordinate, sometimes in conflict, and sometimes simply discrepant. Finally, we listened to formalists and functionalists talk to each other and talk about each other, and considered their differential status in disciplinary culture. Throughout, the goal has not been to determine the superiority of one side over the other, but rather to understand what makes each distinctive, and what each brings to modern study of language. This concluding chapter offers two final perspectives on formalism and functionalism. One is inspired by a work of art that captures something essential about formalism; we then turn that image inside-out to imagine a correlative pictorial essence of functionalism. The second compares a matched set of self-representations, by which formalists and functionalists present themselves to the public.

7.2  Portrait of a Linguistic Mindset An image included in Boeckx’s (2006: 96–98) exposition of the Minimalist Program is arresting. Boeckx reproduces three multi-part sets of lithographs created in late 1945 into 1946 by Pablo Picasso, each of which portrays the figure of a single powerful, standing bull viewed from the side and facing to the right. According to Wye (2010: 71–77), Picasso first generated a fully developed image that represented the bull in good anatomic detail, with

106  Conclusion shadow and shading that gave the figure a plausible three-dimensionality. He then created a second version of the bull that withheld some of that detail, and superimposed on the image a network of lines tracing out what Picasso saw as meridians of tension across the animal’s body, diagonally from groin to neck, and arching from the lower back across the flank to the front legs. The third version discarded more of the surface features of the bull’s body and limbs to present an increasingly schematic two-dimensional image, with the meridians made more prominent by a sparsity of background detail. The fourth and final version comprised a highly abstract distillation of the shape of the bull: the network of superimposed lines has faded so that what remains is only a sketchy outline of the tail, legs, horns, body, and—Picasso being Picasso—the animal’s genitals. He iterated these steps three or four times, so the full collection of Picasso’s lithographs captures multiple views of the bull at multiple levels of resolution. Boeckx sees these lithographs as communicating Picasso’s search for the essence of a bull, which he likens to the trajectory of Chomskyan formalism, culminating in the Minimalist Program’s search to reduce the whole of a grammar to its sparest architectural core. That was Pāṇini’s aim as well, and a goal that Bloomfield’s ‘postulates’ (1926) approached in a different intellectual-cultural milieu. Moreover, formalism often takes itself as its own object. Pāṇini’s Aṣṭādhyāyī attracted a large commentarial literature; modern formalist graphic style sublimates language data to attend to its own analysis of underlying human linguistic competence. (Recall Fischer’s characterization of formalism: ‘the system of grammar [is] more important as an object of study than the actual language data’; 2007: 54; emphasis in the original.) About the creation of the bull lithographs, Wye (2010: 15) quotes Picasso as having remarked that ‘I’ve reached the moment . . . when the movement of my thought interests me more than the thought itself’. Boeckx’s (2006) sole concern is formalism, so he does not suggest a functionalist counterpart to Picasso’s lithographs. But one can imagine how a functionalist artist working under the motto of ‘form follows function’ might attempt to represent the essence of a bull. Such an artist might create a multi-layered, multimedia mosaic that collected diverse evidence for the conventional roles of the animal in human culture: close-up images of Spanish bullfighting, registering the views of both participants and opponents; evidence of the worship of sacred bulls in ancient Minoan civilization; data about the commercial value of bulls in agriculture and animal husbandry; a video of Pamplona’s ‘running of the bulls’; Paleolithic cave drawings of bulls; Arturo Di Modica’s 1989 bronze statue ‘Charging bull’ as a symbol of Wall Street’s entrepreneurial spirit; bulls in comparison to their near-relatives water buffalo, bison, domesticated cattle; and so forth. Just as functionalist linguists insist on analyzing non-idealized language as produced by real speakers in real speech contexts, functionalist artists would feature genuine, particularized bulls carrying out conventional activities in their natural milieu. Functionalist artists would, of course, vary in

Conclusion 107 their tactics. Some might organize representations of bulls culture by culture, or distinguish artistic versus scientific representations. Others might extend their work to incorporate the olfactory and kinesthetic traits of bulls: manure; the odor of bovine sweat; the humid, grassy, blast of a bull’s breath; the swishing of its tail, its snorting and charging. All this would be brought to bear on how form follows function, for example, how competition among bulls for reproductive opportunities privileges larger body mass; how the high cultural value of a bull makes it the natural centerpiece in Paleolithic art; how human projection of virility to bulls makes their slaughter in bullfighting a paradoxical communitybuilding activity. Picasso’s ‘internalist’ representation of bulls looks more and more deeply inward, removing successive layers of surface matter to get at the gist of what a bull is. In contrast, our imagined functionalist artist’s ‘externalist’, culturally and physically contextualized representation collects and interprets evidence linking what a bull is to what a bull does. It is not clear that it would make sense to coordinate these two representations. However, they share a common ambition: to understand the nature of what makes a bull, a bull. Formalist and functionalist linguists both subscribe to that goal with respect to language. The linguist-engineer builds—or distills—a model of the underlying framework of the object of interest, while the linguist-collector assembles samples of that object, and makes sense of those samples with respect to the total collection.

7.3 Formalism and Functionalism Face the World Online One final perspective on formalism and functionalism turns to how the two represent themselves to the public. Newmeyer (1998) introduces his fictional protagonists ‘Sandy Forman’ and ‘Chris Funk’ as graduates of MIT and of the University of California at Santa Barbara, respectively—institutions that Newmeyer doubtless chose for their reputations as polar opposites in orientation. With new insight into the debate between formalism and functionalism, we can analyze how these two prominent American doctoral programs construct their divergent public faces. We will draw as evidence the material posted on their departmental websites assuming that, granted the medium’s flexibility and ease of design, texts disseminated on an institutional website have been groomed to accurately depict the group’s self-image. At the top of the homepage of the MIT Department of Linguistics and Philosophy, a brief institutional self-introduction sets the scene: Our research aims to discover the rules and representations underlying the structure of particular languages and what they reveal about the general principles that determine the form and development of language in the individual and the species. (http://linguistics.mit.edu; accessed 3 August 2018)

108  Conclusion Every word of this passage carries strategic weight. The verb discover suggests an empirical approach. The expression rules and representations, echoing the title of Chomsky (1980), presupposes both that grammar is rule-governed, and that its analysis must consider multiple levels of representation, such as those which (over the history of generative theory) have been variously labeled ‘deep structure’, ‘Logical Form’, or ‘the external interfaces’. Both those presuppositions are strengthened by the specification that rules and representations ‘underl[ie] the structure of particular languages’, an expression that admits the existence of abstract levels of language structure beneath an overlying surface. The text goes on to balance ‘particular languages’ against ‘general principles that determine the form . . . of language’. The existence of those general principles is taken for granted—as signaled by the definite determiner the in construction with initial-reference use of the phrase general principles—and their first, defining, role is to determine the ‘form’ of language. (Note also the earlier use of the with initial-reference rules and representations: the writer presupposes the existence, and obvious relevance to the reader, of ‘rules and representations’ as well as ‘general principles’.) The reference to ‘development of language in the individual’ registers a role for the study of language acquisition, an issue of long-sustained importance in the generative version of formalism. In the last line, and the species may be a shoutout to the recent debate about the origins of language in which generative grammarians have played a leading role. Perhaps most telling of all, notice that this passage asserts straightaway a central formalist position: that it is ‘rules and representations’ that determine the form of language. Behind this seems to lie the commitment that it is neither communication, nor cognition, nor culture that imposes a form on language. In short, this passage posted as if over the doorway into the online presence of linguistics at MIT displays many formalist themes and motifs, and in particular those of generative formalism. The remaining content of the website is consistent with that first impression. In a text reflecting on the department’s history, linguistics at MIT is depicted as ‘a leading center for research on formal models of human-language phonology, morphology and syntax’. Use of the terms ‘formal’ and ‘model’ is salient, as is the absence of reference to domains of human language other than phonology, morphology, and syntax. The next paragraph expands in a similar vein: ‘A distinctive feature of the Linguistics Program at MIT has been its insistence on explicit theories of language formalized as grammatical rules and constraints’. In a search of over 8,000 words posted on the website, the word ‘function’ (or its morphological derivatives) never appears in the sense that contrasts with ‘formal’.

Conclusion 109 Of course, neither Newmeyer’s fictional caricature of a new PhD nor a single-sentence summary of something as multifarious as a graduate program does justice to the complexity of the real-life phenomena they stand for. But imagining that ‘Sandy Forman’ actually had been educated at MIT, the views attributed to Forman in conversation with his counterpart Funk align well with those articulated, and implied, in the institution’s website. Forman’s interlocutor, ‘Chris Funk’, is a graduate of the Linguistics Department of the University of California at Santa Barbara—located geographically on the opposite continental coast, and by reputation intellectually at odds with MIT. The UCSB website’s closest analog to MIT’s one-sentence institutional self-introduction does not appear on the homepage (www.lin guistics.ucsb.edu/home), but rather in a text linked to the homepage under ‘Santa Barbara Sneak Peek’: The UCSB Department of Linguistics studies the ways that languages and language varieties around the world are used in everyday life. We look to social interaction, social and cultural change over time, and cognitive and biological processes to understand why languages work the way they do. . . . We therefore have a strong commitment to using linguistics to advance social justice as well as scientific knowledge. (www.linguistics.ucsb.edu/news/announcement/620; accessed 4 August 2018) The hallmarks of functionalism are conspicuous. The department’s object of study is not ‘language’ in the abstract but plural ‘languages’ worldwide; the text further raises the ante to include differential context-dependent manifestations of languages as well (‘language varieties’). The target is how languages are ‘used in everyday life’, obviously invoking functionalist concern with usage. Moreover, the text registers the investment functionalism makes in explanation by listing the top reasons why (from a functionalist perspective) ‘languages work the way they do’: ‘social interaction, social and cultural change . . . and cognitive and biological factors’. Lest any unclarity remain about the orientation of the department to formalism versus functionalism, the passage goes on to mention two long-term goals: first, ‘to advance social justice’, and secondly—in syntactically subordinate position—‘[to advance] scientific knowledge’. All of the information about linguistics and about graduate study posted on the UCSB website is coherent with the functionalist content of this introductory passage. For example, the value placed on explanation comes up repeatedly: the program ‘focuses on the discovery of general, theoretically significant explanations’ (‘Linguistics—Graduate Program’); ‘The . . . faculty share a commitment to asking why languages are as they are. This fundamental question drives the pursuit of functional explanation’ (‘Research’). The word ‘function(al)’ appears 13 times in a 4,800-word corpus of the text

110  Conclusion posted on the website; the word ‘form’ (in its relevant sense), three times. Arguably, there is also a shadow of the defensiveness that sometimes characterizes functionalist rhetoric. A passage on the homepage declares that: As more and more researchers across all fields of linguistics are seeking well-motivated explanations and firmly grounded empirical evidence for claims about the nature of language, UCSB’s longstanding leadership in this enterprise puts the department at the cutting edge of linguistic scholarship, developing ideas and methods that are critical for moving the field of linguistics into a new era. (www.linguistics.ucsb.edu/home; accessed 4 August 2018) The tone that surfaces here—somewhere between defensive and aspirational— may only communicate context-appropriate self-promotion. But it is salient that, in contrast, the MIT department frames the warrant for its own value point-blank as a long-established fact: ‘the Linguistics Program at MIT rapidly acquired an international reputation as a leading center for research’ (http://linguistics.mit.edu/graduate/). A final touch nails in place the orientation of the UCSB department: in the upper right-hand corner of the website’s homepage, and repeated as the featured graphic on the department’s Facebook page, is a square box in which, in bold upper-case letters and without commentary, appear the words ‘ form follows function ’. In photographs of students and faculty, the same famous aphorism appears inscribed on what seem to be department-issued royal-blue hats. As we know, the fictional encounter between Newmeyer’s Forman and Funk did little to advance their mutual understanding, or even mutual toleration. Some observers believe the two orientations can complement each other; others disagree. But observers, as well as participants who identify as either linguist-engineers or linguist-collectors, can all profit from better insight into what is at stake in the debate between formalism and functionalism.

References Bloomfield, Leonard. (1926). ‘A set of postulates for the science of language’. Language 2: 153–164. Boeckx, Cedric. (2006). Linguistic minimalism: Origins, concepts, methods, and aims. Oxford: Oxford University Press. Chomsky, Noam. (1980). Rules and representations. New York: Columbia University Press. Newmeyer, Frederick J. (1998). Language form and language function. Cambridge, MA: MIT Press. Wye, Deborah and Pablo Picasso. (2010). A Picasso portfolio: Prints from the Museum of Modern Art. New York: Museum of Modern Art.

Appendix Characteristics of Formalism Versus Functionalism

Formalism Profile, in Short Linguist-engineers define inner structure of human language ‘Austere, diagrammatic, ruleconscious’ Defaults to ‘internalist’ explanations Linguistic competence is central Fieldmarks Embraces abstraction Analyses presuppose, and build, theoretical constructs Searches actively for generalizations Values rigorous formalization of findings States generalizations in absolute terms May admit interaction between formal and functional constraints Pursues narrow range of grammatical facts ‘Grammaticality’ is a key construct Language is essentially a vehicle for thought or reasoning, which incidentally plays a role in transfer of information Focus on ideal speaker/hearer; postpones or disregards individual variation

Functionalism Linguist-collectors assemble ‘cabinets of curiosity’ ‘Loose, approximate, contingent’ ‘Externalist’ explanations Attends to linguistic performance Deals with particulars, idiosyncrasies Analyses presuppose connection of language, communication, cognition Can be reticent in pursuit of generalizations Values fidelity in representing diverse data States generalizations as statistical probabilities Functional forces are a foundational, direct, influence on the shape of language Pursues totality of language ‘Prototypicality’ and ‘markedness’ are key constructs Language most essentially performs social-interactive functions Explores individual, group, and community-wide differences (Continued)

(Continued) Formalism

Functionalism

Assumes homogenous speech community Distant goal: specify precise contents of human nature

Sensitive to issues of power and solidarity Distant goal: subserve social progress

Methodological Practices; Treatment of Data Deductive methods Inductive methods Open to bold speculation Deep analysis of particular cases Idealized, decontextualized data Data retain features of natural speech/ text in context Grammaticality judgments; speakers’ Case studies; corpora (spoken and intuition and introspection textual) Defining constituency is foundational Analysis highlights dependency to analysis relationships Little engagement with written Addresses speech/writing differences language Assumptions About the Organization of Grammar Assumes autonomy and modularity in All parts of language are integral to organization of grammar each other Syntax is an autonomous mental Syntax is a tool for encoding meaning, faculty comprising an operational shaped by speakers’ intended code messages Sentence is the typical unit of analysis Highlights discourse Semantics based on truth conditions Semantics based on speaker construal Assumes ‘encapsulated’ modularity Semantics/pragmatics form a continuum Arbitrariness of the lexicon Champions iconicity, metaphor Grammatical relations are key units: Parts of speech are key units: noun; subject; object; topic; theme; verb; auxiliary; CP; INFL . . . focus . . . Parts of speech resolvable into features ‘Fuzzy’ distinctions among parts of speech Categories are sharply delineated, Categorization of elements is subject to defined by necessary and sufficient gradation: prototypical vs. extended conditions cases

Index

abstraction: in analysis of child language 70, 79, 94; in formalism in general 5, 7, 15, 41; functionalist avoidance of 53, 89, 94; increase over history of generativism 37–38, 56; in generative grammar 9, 34 – 5, 40, 74 – 5, 100, 108; in pre-modern formalism 17, 23, 25; see also data, linguistic acquisition of language: according to Bloomfield 68; according to Chomsky 18, 37, 68 – 9; and explanation 93–4; formalist accounts 9, 67– 71, 79, 94; functionalist accounts 71 – 2, 79, 91, 96; see also child language adjacency principle 50 – 1 Affix hopping 34 – 6, 37 Akamatsu, Tsutomu 24 American structuralism see structuralism Anderson, Stephen R. 75, 87 – 9 anthropology 10–11, 20 – 1 Arbib, Michael A. 77 Aristotle 18 Arnold, Jennifer E. 64 Aṣṭādhyāyī 16–17, 106 autonomy: of competence from performance 34, 86 – 7; critique from functionalist perspective 46, 48, 88, 90; of grammar from cognition 33 – 4, 74, 86 – 7; of grammar from discourse 34, 40; Newmeyer’s threepart analysis of 86 – 8; of syntax from semantics 34, 40, 42, 51, 88, 91, 97 Babungo 88 Baker, C.L. 68 Balanoff, Amy 73 Banawá 10

Bates, Elizabeth 16, 45, 74–6, 78: critique of formalism 91; functionalist account of language acquisition 52, 71 – 2, 79 Battistella, Edwin 16, 99–100 Beaugrade, Robert de 89, 91 behaviorism 21, 22–27, 33, 68 Berwick, Robert 75 binding 36–37, 42, 70 biolinguistics 39 – 40, 97–8 Blakemore, D. 99 Bloom, Paul 75 – 6, 91, 99 Bloomfield, Leonard: 19 – 20; 21 – 2, 23, 27; 33; language learning 68; Pāṇini 17 – 18; ‘Postulates’ 21; 106; Saussure 19; see also post-Bloomfieldian linguistics Boas, Franz 20 – 1, 26 Boeckx, Cedric 38 – 9, 56, 105 – 6 Bolinger, Dwight 55 Bresnan, Joan 42, 56–57, 65; see also Lexical-Functional Grammar Broca’s area 11 – 12, 77 Butler, Christopher S. 46, 96, 98 Cann, Rebecca L 76 Cardona, George 16 Carnie, Andrew 2, 15, 84–5, 99 Carstairs-McCarthy, Andrew 16, 83 Cartesian Linguistics 19 case theory 62 Chater, Nick 77 child language: functional versus grammatical categories in 69 – 70, 93–4, 100; relevance to linguistic theory 68, 71; telegraphic speech 70, 72, 93 – 4, 97; see also acquisition of language

114 Index Chomsky, Noam: as generative theorist 8, 20, 78; as prominent formalist 8, 15, 32–3, 40, 106, 108; covert functionalism in 92, 98; critique of 75 – 7, 90, 91, 98; graphic style 56 – 7; history of generative grammar 17 – 19, 20; language acquisition 9, 18, 68 – 9; leadership as generativist 32, 100; levels of explanatory adequacy 93, 96; Minimalism 38 – 40, 65, 74 – 5, 78; non-generative formalism 40 – 3, 46; origin of language 72 – 6, 79; principles and parameters 36 – 7; transformational grammar 33 – 6, 65; see also generative grammar; see also poverty of the stimulus Christiansen, Morten H. 77 – 8 Chuong, Cheng-Ming 73 cognition: and explanation 50, 96; role in evolution of language 77 – 8 Cognitive Construction Grammar 54, 76 – 7, 86, 100 cognitive grammar (Langacker) 48 – 9, 52, 54, 90 – 2 cognitive linguistics 46, 53–4 collecting see functionalism Columbia School 53 communicative dynamism 25 – 6, 28, 63, 78 communicative task urgency 63, 78; see also theme/rheme competence/performance: 37, 61, 74, 79, 106; autonomy of 40, 86 – 7; in child language acquisition 68–69, 93–94, 97; Chomsky’s definition of 34; in Lexical-Functional Grammar 42 Competition Model 52, 71 – 2, 79 conceptual-intentional system 74, 92, 98 construction: in functionalism 49, 52, 57 – 8, 67; see also Cognitive Construction Grammar Cratylus 18 – 19, 28 Croft, William 16, 54, 76, 77, 83; autonomy of syntax and semantics 88 Curnow, Timothy J. 2, 5 D’Alessandro, Roberta 98 Darnell, Michael 84, 99 data, linguistic: formalist versus functionalist treatment 4, 6–7, 64, 68, 70–2, 97–8; formalist deprioritization

6, 37, 39–40, 43, 106; formalist idealization 9, 34, 93, 97; functionalist fidelity to 6, 11, 24, 47, 71, 96, 100; graphic representation 56 – 7; see also abstraction Dediu, Dan 76 deep structure/surface structure 34–35, 39 – 40, 74, 108 Derwing, Bruce 33, 94 dialogue between formalism and functionalism 83, 99; Carnie and Mendoza-Denton 84 – 5, 99; Newmeyer 83 – 4, 99, 107, 109 Dik, Simon 46, 55, 76, 95–96; functional grammar 52, 54, 89 Diller, Karl C. 76 Dillinger, Mike 17–18 Dirven, René 16, 19–20 discourse 28; and explanation 50, 96; and word order phenomena 78; autonomy of grammar from 36, 40, 43, 86, 88; grounding in 67, 78 ; integration of grammar and 47, 53, 63 – 4; old/new information in 25, 63–4 distributionalism 22 – 24, 28 ditransitivity 86, 95 Diver, William 53 E-language 40, 74 economy: formalist prioritization 16 – 17, 38 – 9, 62; functionalist resistance 48 Eldredge, Niles 73 Emeneau, Murray B. 17 emergent grammar 53, 87, 90 Enfield, N. J. 10, 92 engineering see formalism Evans, Nicholas 99 Everett, Daniel L. 10 – 13, 28; as functionalist 55, 100; language and brain 11–12; language as cultural tool 10 – 11, 67 evolution of species, theories of: neo-Darwinian 72 – 3, 75, 79; punctuated equilibrium 73; saltationism 73, 75 – 7, 100 explanation: Butler’s expanded levels of adequacy of 96; Chomsky’s levels of adequacy of 93, 96; formalist critique of functionalist 72, 84, 95 – 6; formalist versus functionalist 64, 92 – 6; functionalist critique

Index  115 of formalist 90, 94 – 5; in Givón’s functionalism 50 – 1; internal versus external 6, 70–71, 93 – 4, 96; in post-Bloomfieldian linguistics 93; in transformational grammar 34 feature checking 38 – 9 features, distinctive 25 features, lexical 52, 65–7; Generalized Phrase Structure Grammar 41 – 2, 65; Head Drive Phrase Structure Grammar 42; Lexical-Functional Grammar 43, 65 figurative language 49, 52 finite-state grammar 33 Fischer, Olga 2, 6, 94, 106 Fitch, W. Tecumseh 74 – 5 Foley, William A. 52, 89 Foolen, Ad 16, 98 formal functionalism 26, 51 – 53, 97 formalism and functionalism: in architecture 5 – 6; in dialogue 83 – 5; definitions of 5 – 7; employment of terms 15 – 16; explanation in 92 – 6; formalist critique of functionalism 83 – 9; functionalist critique of formalism 83–5, 89 – 92; mutual segregation 83; (non) complementarity between 96 – 9; in phonology 2, 85; relation to each other 78 – 9; relative status in disciplinary culture 99 – 101 formalism: abstract forms and operations in 7, 9; compared to engineering 6 – 8, 15, 28, 43, 45, 100, 107; diversity within 12 – 13, 40 – 3; early twentieth-century 20 – 3; explanation in 92–6; in Europe vs. North America 32; premodern versions 16 – 20; purported convergence toward functionalism 91 – 2, 97–8; see also abstraction; see also economy formalization of linguistic theory 22, 42, 56 – 7, 108; Chomsky on 57; functionalism and 48, 57, 88–9 ‘form follows function’ 5, 21, 25 – 6; as functionalist motto 13, 84, 94–5, 106 – 7, 110 Fortis, Jean-Michel 46 Fought, John 21 Francis, Elaine J. 16

Fried, Vilém 16, 19–20 Fukui, Naoki 32 functional grammar see Dik, Simon functional linguistics versus cognitive linguistics 46, 49, 53 Functional Sentence Perspective 25 functionalism: compared to collecting 7 – 8, 10, 15, 28, 45 – 6, 68, 100, 107; diversity within 12 – 13, 45 – 6, 51 – 5; early twentieth-century 23 – 8; explanation in 94 – 6, 109 – 10; in Europe versus North America 20, 23–24, 26, 46, 99; purported convergence toward formalism 85 – 9; sense of history 16; taxonomy of varieties 51 – 5 Gazdar, Gerald 41 – 2, 56, 65; see Generalized Phrase Structure Grammar Geeraerts, Dirk 91 – 2, 99 Generalized Phrase Structure Grammar (GPSG) 41 – 3, 56, 65 generative grammar 9, 12, 18–19; international profile of 32; stages in development of 16, 32 – 3, see also transformational grammar; see also principles and parameters; see also Minimalist Program Generative Linguists of the Old World 32 generative semantics 46 German 69 gesture 28; role in origin of language 76 – 7 Ginstrom, Ryan 64 Giudice, Alex Del 36 Givón, Talmy: critique of formalism 90, 94 ; as functionalist 46, 50 – 1, 53, 55, 63 – 4, 87, 89; graphic style 55, 57; on history of functionalism 16, 18, 27; on origin of language 75 – 6; Saussure 19 – 20 Gleason, Henry Allan 22 – 3 Goldberg, Adele E. 36, 46, 54, 76 – 7, 86, 100 Golumbia, David 92, 97 Gould, Stephen Jay 73 Graffi, Giorgio 15, 45 grammar, modular organization of 17, 33 – 4, 43, 71, 75 grammatical roles 25, 34, 42–3, 60–1; in child language 69–72

116 Index grammaticalization 2 graphic style: in formalism 55 – 7, 106; in functionalism 57 – 8 Haegeman, Liliane 61 – 2 Haider, Hubert 32 Hajičová, Eva 62 – 3 Halliday, Michael A. K. 23, 27, 46 – 8, 57, 76; critique of formalism 89 – 90; textual cohesion 48 , 100; see also systemic functional linguistics Harris, Zellig S. 33 Hasan, Ruqaiya 47–48, 57 Haspelmath, Martin 2, 16, 83, 99 Hauser, Marc D. 74 – 5 Head-Driven Phrase Structure Grammar (HPSG) 42 – 3, 56 Heath, Jeffrey 55 Hengeveld, Kees 89 Hjelmslev, Louis 23–24, 26 Hockett, Charles 22 – 3, 65, 68 Hoekstra, Teun 9, 94 Hopper, Paul J. 52 – 3, 65 – 7, 76, 78 – 9; on non-complementarity of formalism and functionalism 87, 98 Hu, Dongyu 73 Hyams, Nina 69 – 72, 79, 93–94, 96–97, 100 Hymes, Dell 21 I-language 40, 74 iconicity 19, 28, 50, 87, 95 – 6 Immediate Constituent Analysis 22 – 3 Indo-European languages 17 Indonesian 67 interface, linguistic 74 – 5, 92, 108 item-and-process/-arrangement 22 Jackendoff, Ray 75 Jakobson, Roman 24 – 6, 28 Jespersen, Otto 23 – 4 Johnson, Mark 52 Joos, Martin 93 Joseph, John E. 18–19, 33 Kaplan, Ronald M. 42 Katre, Sumitra Mangesh 16 Kayne, Richard S. 62, 64 kernel sentence 34 Klein, Ewan 41 – 2, 56–57, 65; see Generalized Phrase Structure Grammar

Kooji, Jan G. 94 Kuno, Susumu 52, 55, 88, 97–8 Lakoff, George 20, 46, 52, 54, 87 Langacker, Ronald 20, 46, 48 – 9, 52, 54 – 5, 76, 87; critique of formalism 90, 98; graphic style 57 – 8, 100; see also cognitive grammar language and brain 11 – 12, 73, 77 language and cognition 10 – 11, 40, 55, 64; in cognitive grammar 48 – 9; in functionalism 46; in Givón 50 language and communication 9, 24, 46, 50 – 1, 55, 62 – 3, 76; formalist view of 16, 21, 75 language and culture 10 – 12, 24, 27–8, 55; and explanation 96 language faculty, human 9, 11, 36, 48, 61, 77; and natural selection 75–6; in broad/ narrow sense 92; comprised of three factors 40; functionalist skepticism about 76, 92, 95; see also universal grammar language, ‘impossible’ 8 – 10, 11 – 12 Lasnik, Howard 34 , 65 Lass, Roger 96 learnability 37, 38, 43, 87; of word order in generative grammar 61 – 2, 72; see also poverty of the stimulus Lehmann, Winfred P. 56 Levinson, Stephen C. 76, 99 Lexical-Functional Grammar (LFG) 42 – 3, 56, 65 lexicogrammar 47 Lightfoot, David 74 Linguistic Society of America 21 linguistics and natural sciences 9, 21–2, 39–40, 72, 74, 91, 100 Liptak, A. 99 Losongco, Anthony 64 MacWhinney, Brian 16, 52, 71 – 2, 76, 79, 91 Malle, Bertram F. 76 Marchman, Virginia 74–75, 78 Martinet, André 26, 28, 53 McCarthy, John J. 68 McWhorter, John 10. meaning and form 18–19, 24, 88; in child language 71, 91; in formalism 9, 21 – 2; in functionalism 25 – 6, 28, 46 – 9, 52, 55, 86–7; in Generalized

Index  117 Phrase Structure Grammar 42; see also iconicity; see also autonomy Mendoza-Denton, Norma 2, 15, 84 – 85, 99 Merge 38 – 9, 74–5; see also Minimalist Program Minimalist Program 33–38, 65, 75, 78; 105 – 6; formalist nature of 38 – 40, 97–8; and functionalism 39 – 40, 92, 97–8; graphic style 56 mirror neurons 77 missionary linguistics 11, 27, 28 MIT 26, 33, 84, 107 – 9, 110 Moro, Andrea: as formalist 8 – 10, 12–13, 37, 100; goal of linguistics 8; language and brain 11 – 12; notion of ‘impossible language’ 8 – 9 Morpurgo Davies, A. 17 Morvacsik, Edith 16 Native American languages 20–21, 27 natural language processing 42–43, 56 Neogrammarians 24 Newmeyer, Frederick J. 2, 99, analysis of autonomy in formalism 86 – 8; complementarity of formalism and functionalism 86, 97 – 8; explanation in formalism and functionalism 95 – 6; on history of formalism and functionalism 15 – 16, 46; imagined dialogue between formalism and functionalism 83 – 5, 99, 107, 109–10; on origin of language 75–76; Prague Circle 15–16, 26; taxonomy of functionalists 26, 51 – 3, 76 Nichols, Johanna 54 – 5, 71, 76 Noonan, Michael 50 null subjects 69 – 70, 72, 93 Nuyts, Jan 99 Oostendorp, Marc van 98 origin and evolution of human language: formalist accounts 72 – 6, 79, 108; functionalist accounts 76 – 79; timeframe for 75 – 6; gesture and 76 – 7 Palácio, Adair 17–18 Pāṇini 16 – 18, 28, 38, 106 Payne, Doris 64, 78, 95 – 6 phonology 2, 19, 24 – 5, 85

phrase structure rules 34, 41–2, 61 Picasso, Pablo 105 – 7 Pike, Kenneth 20, 27 – 8 Pinker, Steven 75 – 6, 91, 99 Pirahã 10 – 11, 28, 55, 100 Plato 18 – 19, 28, 91 Plato’s problem 18 Pollard, Carl 41; see Head-Driven Phrase Structure Grammar post-Bloomfieldian linguistics 22 – 3, 28, 33, 65, 93 poverty of the stimulus 9, 37, 62, 68–69; see also learnability pragmatics 42, 48, 50, 52, 63–4, 77–8, 95; in child language 70 – 1, 94, 97 Prague Linguistic Circle 15 – 16, 20, 23 – 6, 28, 46, 51–2, 53; phonology 24 – 5; syntax 25, 62 – 4, 78; see also theme and rheme principles and parameters 36 – 8, 78; and word order in 62 ; in language acquisition 69 – 70, 79 Prinzhorn, Martin 32 processing, language 56; in accounts of origin of language 77 – 8; as factor in functionalist analyses 12, 36, 50, 53 – 4, 64, 94 – 6; rejected as factor in formalist analyses 70; see also natural language processing Pullum, Geoffrey K. 41 – 2, 56–7, 65; see Generalized Phrase Structure Grammar Radford, Tim 10 Radical Construction Grammar 54 rationalism 19, 92 recursion 9, 11–12, 74, 100 relevance, communicative 26, 28 Riemsdijk, Henk van 32 Role and Reference Grammar 46, 52, 54, 89 Rooryck, J. 99 ‘rules and representations’ 107–8 Sag, Ivan 41 – 2, 56, 65; see Generalized Phrase Structure Grammar; see Head-Driven Phrase Structure Grammar Saito, Mamoru 32 saltationism see evolution of species

118 Index Sampson, Geoffrey 94 Sanskrit 16 – 17, 38 Sapir, Edward 10, 20, 26 – 7, 28 Saussure, Ferdinand de 19 – 20, 23–24, 26, 28 sensory-motor system 74, 92 Siewierska, Anna 46, 53 – 4, 76 SIL International 27, 28 Silverstein, Michael 55 Smith, N.V. 99 Southern Paiute 27 Spanish 66 Staal, J. F. 17 Stiegler, Josef 73 structuralism 19 – 21; American structuralism 21, 27, 33, 55 Sullivan, Louis 5–6, 25–6; see also ‘form follows function’ syntactic movement 36, 38–9, 49, 56, 62, 74; see also Merge; see also transformational rules systemic functional linguistics 46 – 8, 52, 54, 89 – 90; meta-functional constructs 47 ; functionalist methods 47

transitivity: in formalism 62, 64 – 5, 78; in functionalism 47 , 65– 7, 78, 86, 95 tree, syntactic 41–43, 56, 57 Trubetzkoy, Nikolai 24 – 5

tagmemics 27 – 8 Takami, Ken-Ichi 97 Tallerman, Maggie 16, 83 Thal, Donna 74–75, 78 theme and rheme 25, 63 – 4 Theta Criterion 65 theta roles 36, 65, 98 Thomas, Margaret 19, 68 – 9 Thompson, Sandra A. 65 – 7, 76, 78 – 9 Tillohash, Tony 27 Toman, Jindřich 24 Tomasello, Michael 76 – 7 Tomlinson, Antony 10 transformational grammar 33 – 6; word order in 60 – 1 transformational rules 34 – 6, 38, 41 see also syntactic movement

Wang, Shuo 73 Wasow, Thomas 64 websites of departments of linguistics 107 – 10 Wheatley, Kathleen 84, 99 Williams, R. 10 word order: in generative formalism 60 – 2, 78–79; in functionalism 25, 62 – 4, 71 – 2, 78–79 World War II 26 Wu, Ping 73 Wye, Deborah 105 – 6

universal grammar: and learnability 37, 43, 61, 79, 93–4; in early transformational grammar 34; in Minimalism 40, 74–75, 92; Prague Circle universals 25; in principles and parameters 36–37, 61–62, 65 University of California at Santa Barbara 83 – 4, 107, 109 – 10 University of Wisconsin-Milwaukee conference (1996) 84, 87 usage-based grammar 49, 54, 76 Vachek, Josef 24 Van Valin, Robert D., Jr. 27, 45–46, 52, 76, 89; see also Role and Reference Grammar variation, linguistic 17, 84–85; formalist perspective on 8, 20, 36 – 7, 39 – 40, 61, 97; functionalist perspective on 20, 64, 71, 109

X-bar theory 36 – 7, 41 – 2, 56, 61 – 2, 78 Xu, Xing 73 Zhou, Yachun 73