The Handbook of Discourse Analysis

By Deborah Tannen, Heidi E. Hamilton and Deborah Schiffrin

The second edition of the highly successful Handbook of Discourse Analysis has been expanded and thoroughly updated to reflect the very latest research to have developed since the original publication, including new theoretical paradigms and  discourse-analytic models, in an authoritative two-volume set.

• Twenty new chapters highlight emerging trends and the latest areas of research
• Contributions reflect the range, depth, and richness of current research in the field
• Chapters are written by internationally-recognized leaders in their respective fields, constituting a Who’s Who of Discourse Analysis
• A vital resource for scholars and students in discourse studies as well as for researchers in related fields who seek authoritative overviews of discourse analytic issues, theories, and methods

• Language: English
• Publisher: Wiley
• Release date: Apr 28, 2015
• ISBN: 9781118584187

Deborah Tannen

Deborah Frances Tannen is an author and professor of linguistics at Georgetown University in Washington, D.C.

Book preview

Preface to the Second Edition

DEBORAH TANNEN AND HEIDI E. HAMILTON

The success of the first edition of The Handbook of Discourse Analysis has been gratifying, and sets the bar high for this second edition. Our goal for this edition, as it was for the first, is (1) to provide a vital resource for scholars and students in discourse studies as well as for researchers in related fields who seek authoritative overviews of discourse analytic issues, theories, and methods; (2) to serve the needs of students and scholars in professional and academic domains such as education, law, medicine, business, government, and media who may consult the Handbook as they consider how fine-grained examinations of discourse can illuminate central problems in their fields; and (3) to constitute an essential addition to personal, academic, and professional libraries around the world, as new collaborators join the area of discourse studies.

During the nearly 15 years since the publication of the first edition, new research has been conducted in all areas covered by the original 41 chapters. New theoretical frameworks have taken on importance even as existing ones have been expanded and enriched both by young scholars who have risen to the forefront of the field and by established researchers who have built on their own prior advances. Moreover, new types of discourse have appeared with the invention and adoption of new technologies. To capture and reflect these developments, we invited 20 new chapters for the second edition. In order to accommodate them, 19 chapters from the first edition were of necessity replaced. We regret their loss, as all made significant contributions to the field, and we hope and expect that readers will continue to consult them in the first edition.

Of the 22 chapters remaining from the first edition, 19 have been updated and one is an entirely different chapter by the same author (Emanuel Schegloff). The remaining two are unchanged because the nature of their contents is not affected by the passage of time: John Gumperz (who, sadly, passed away in 2013) provided a personal perspective on his founding of the field of Interactional Sociolinguistics, while Robin Lakoff illustrated how a single communicative act, apology, can be best understood by the application of multiple approaches. We are gratified that, in addition to adding the work of scholars who have come to prominence since the publication of the first edition and while retaining the voices of many who helped establish the field of discourse analysis, we have also been able to add chapters by leading scholars who were missing from the first edition. We have sought to maintain the international character of perspectives represented, as reflected by the fact that the contributors hail from 11 countries.

Given that nearly half the chapters in the current edition are new and that almost all of the rest are significantly revised, it was clear that the organization needed to be reconceptualized. The new organization progresses from a focus on the linguistic analysis of discourse (Part I) to increasingly broad perspectives on the world outside language: the range of academic approaches and methodologies (Part II); the individual, society, and culture (Part III); and the real-world contexts that are in part created by discourse as they are sites for its use (Part IV). We have slightly revised and significantly shortened the original introduction, retaining those sections that remain relevant and excising those that no longer apply.

It is our hope that this new edition of The Handbook of Discourse Analysis will not only reflect the range, depth, and richness of current research in the field but also inspire new, illuminating work by providing students, scholars, and practitioners with state-of-the-art discussions of key aspects of this now well established but still burgeoning field. We look forward to continuing to engage in vibrant scholarly conversations as researchers in a broad range of disciplines explore the complexity of discourse and the numerous ways in which its analysis enhances understanding of human communication and its role in tackling key problems confronting our world and the people who live in it.

We would like to express our gratitude to those who helped in a multitude of ways. First, our sincere thanks go to the contributors for their huge investments of time and creativity. We know that all have many demands on their time, and we are grateful to them for choosing to devote such a full measure of it to this project. We ourselves learned much from each chapter, and we know readers will as well.

With equal fervor, we express our deep gratitude to Gwynne Mapes for her unwavering, proactive, and perspicacious efforts on behalf of the Handbook. Gwynne's dedicated oversight, organizational genius, and consummate communication gifts shepherded the chapters through the twisting byways from submission to publication. We cannot imagine having brought this volume to fruition without her.

We are grateful, as well, to the students, staff, and our faculty colleagues in Georgetown University's Department of Linguistics. The entire department, and in particular our students and colleagues in the sociolinguistics concentration, inspire and educate us daily, as they create the intellectually stimulating and interpersonally supportive environment that grounds and nurtures all our work.

Deborah would like, in addition, to express her gratitude to the Center for Advanced Study in the Behavioral Sciences at Stanford University, where she was a fellow during the academic year 2012–13, for providing an otherworldly beautiful and academically inspiring environment in which to edit chapters as they arrived.

In closing, we express our enduring gratitude, admiration, and affection for our treasured colleague and dear friend Deborah Schiffrin. Her vision, dedication, and hard work were pervasive at every stage of the first edition of this Handbook, and in the conceptualization of this second edition. Although health challenges precluded her participation in the execution of this edition, it nonetheless benefits from her significant influence throughout. We felt her spirit beside us always, as we will going forward. We see this volume as a part of her legacy, a testament to the enormous role she played in the establishment and development of the field of discourse analysis at Georgetown University and within the field of linguistics.

Introduction to the First Edition

DEBORAH SCHIFFRIN, DEBORAH TANNEN, AND HEIDI E. HAMILTON

What Is Discourse Analysis?

Research in the rapidly growing and evolving field of discourse analysis flows from numerous academic disciplines that are very different from one another. Included, of course, are the disciplines in which models for understanding, and methods for analyzing, discourse first developed, such as linguistics and anthropology. But also included are disciplines that have applied, and extended, such models and methods to problems within their own academic domains, such as communication, cognitive psychology, social psychology, philosophy, literary criticism, and artificial intelligence.

Given this disciplinary diversity, it is no surprise that the terms discourse and discourse analysis have different meanings to scholars in different fields. For many, particularly linguists, discourse has generally been defined as anything beyond the sentence. For others (e.g., Fasold 1990: 65), the study of discourse is the study of language use. These definitions have in common a focus on specific instances or spates of language. But critical theorists and those influenced by them can speak, for example, of discourses of power and discourses of racism, where the term discourses not only becomes a count noun but further refers to a broad conglomeration of linguistic and non-linguistic social practices and ideological assumptions that together construct or reinforce power or racism. So abundant are definitions of discourse that many linguistics books on the subject now open with a survey of definitions. In their collection of classic papers in discourse analysis, for example, Jaworski and Coupland (1999: 1–3) include 10 definitions from a wide range of sources. They all, however, fall into the three main categories noted above: (1) anything beyond the sentence, (2) language use, and (3) a broader range of social practice that includes non-linguistic and non-specific instances of language.

Issues associated with definitions are by no means unique to discourse and discourse analysis. In his two-volume reference book on semantics, for example, Lyons (1977) illustrates 10 different uses of the word mean, and thus an equal number of possible domains for the field of semantics. In his introductory chapter on pragmatics, Levinson (1983) discusses 12 definitions of the field of pragmatics (including some that could easily cover either discourse analysis or sociolinguistics). Since semantics, pragmatics, and discourse all concern language, communication, meaning, and context, it is perhaps not surprising that these three subfields of linguistics are those whose definitions seem to be most variable.

Rather than seeking to establish a single definition, the variety of papers in this Handbook reflects the broad array of definitions of – and approaches to – discourse analysis. The different understandings of discourse represented reflect the rising popularity and ever-expanding range of the field. Our own intellectual/academic histories – all in linguistics – reveal some of the different paths that have led us to an interest in discourse. Since each of our paths is different, we here speak in our own voices – in the order in which we arrived at Georgetown University.

Deborah Tannen

When I decided to pursue a PhD in linguistics, I held a BA and MA in English literature and had for several years been teaching remedial writing and freshman composition at Lehman College, the City University of New York. Restless to do something new, I attended the 1973 Linguistic Institute sponsored by the Linguistic Society of America at the University of Michigan. That summer I fell in love with linguistics, unaware that language in context, the theme of that Institute, did not typify the field. Inspired by A. L. Becker's introductory course and by Robin Lakoff's course on politeness theory and communicative strategies, as well as by Emanuel Schegloff's public lecture on the closings of telephone conversations, I headed for the doctoral program at the University of California, Berkeley, where Robin Lakoff was on the faculty. There I discovered, in addition to Lakoff, Professors Charles Fillmore (then interested in frame semantics), Wallace Chafe (then interested in theories of frames and scripts as well as the comparison of speaking and writing), and John Gumperz (then developing the field that later became known as Interactional Sociolinguistics). Not for a moment did I think I was doing anything but linguistics. The word discourse was neither a category with which I identified nor a term I regularly heard. There were no journals with the word discourse in their titles. The only journal that specialized in language in context was Language in Society, which had a strongly anthropological orientation. I vividly recall the sense of excitement and possibility I felt when a fellow graduate student mentioned, as we stood in the halls outside the linguistics department, that another journal was about to be launched: Discourse Processes, edited by psychologist Roy Freedle.

When I joined the faculty of the sociolinguistics program at Georgetown University in 1979, I redefined myself as a sociolinguist. That year I submitted an abstract to the annual LSA meeting and checked the box sociolinguistics to aid the committee in placing my paper on the program. But, when I delivered the paper, I found myself odd man out as the lone presenter analyzing transcripts of conversation among a panel of Labovians displaying charts and graphs of phonological variation. I promptly redefined what I was doing as discourse analysis – the name I also gave to courses I developed at Georgetown. When invited to organize a Georgetown University Round Table on Languages and Linguistics in 1981, I titled the meeting (and the book that resulted) Analyzing Discourse: Text and Talk and invited as speakers linguists, anthropologists, and psychologists, all of whom were examining language in context. During these early years, a number of journals appeared that reflected and contributed to the development of the field: Text, the first of several journals founded and edited by Teun van Dijk in Amsterdam (it was later renamed Text & Talk, under the editorship of Srikant Sarangi), and Journal of Pragmatics, co-edited by Jacob Mey and Hartmut Haberland in Denmark. As the years passed, many other journals were added – too many to name them all, but including Discourse & Society, Discourse Studies, Journal of Linguistic Anthropology, Journal of Sociolinguistics, Multilingua, Narrative Inquiry, Pragmatics, and Research on Language and Social Interaction. In recent years, the list has expanded to include Critical Discourse Studies, Discourse & Communication, Gender and Language, Journal of Language Aggression and Conflict, Journal of Language and Politics, and Pragmatics and Society. The proliferation of journals in itself testifies to the upsurge of interest in discourse analysis, and its many incarnations.

The changes I have seen since I first began defining myself as a discourse analyst reflect the tremendous growth in this area. Work in discourse analysis is now so diverse that discourse is almost a synonym for language – coming full circle to where I saw such work at the start.

Deborah Schiffrin

I discovered linguistics and discourse analysis in a very roundabout way. In my senior year of college at Temple University, I read Erving Goffman's Presentation of Self in Everyday Life during a course in sociological theory (the last requirement of my major). I was so excited by his work that I went on to read everything else he had written and then decided to continue studying face-to-face interaction in a PhD program in sociology at Temple. There my studies included an eclectic blend of sociological and social theory, semiotics (which included initial forays into structural and transformational linguistics), statistics, and urban studies. While still at Temple, I wrote an article on the semiotics of the handshake, which I boldly sent to Goffman. What followed was an invitation to a personal meeting and then his permission to audit a course with him. (The course prerequisite was to read all his work before the first class!) When my advisor at Temple decided to leave for another position, I had already decided to try to work with Goffman. Ironically, it was Goffman himself who first turned my thoughts toward a PhD in linguistics: during our first meeting, he proclaimed his belief that linguistics could add rigor and respectability to the analysis of face-to-face interaction.

Once I was enrolled in the PhD program in linguistics at the University of Pennsylvania, I quickly learned that, although linguists knew that understanding social interaction was important, the study of social interaction itself had a somewhat peripheral role in the linguistics curriculum. What I found instead was Labov's sociolinguistics: an energizing mix of fieldwork, urban ethnography, variation analysis, and narrative analysis. I gladly immersed myself in the life and work of the faculty and students in the sociolinguistics community: we interviewed people, measured vowels, coded narratives, and wondered (and worried) about how to measure different styles. Although many of my teachers published articles about discourse (Bill Labov on narrative and ritual insults; Ellen Prince on syntax, presupposition, and information status; Gillian Sankoff on grammaticalization in Tok Pisin), there was little sense of collective interest or of a community of discourse analysts.

As it became time for me to write my dissertation, I decided that I wanted to use what I had learned as a linguist to study social interaction. I remember my sense of confusion, though, when I tried to use what I had learned about the systematicity of language, as well as to follow the advice of both Labov and Goffman. Labov presented me with one mission: solve an old problem with a new method. But Goffman presented me with another: describe something that had not yet been described. After spending some time trying to apply these directives to the study of everyday arguments, I ended up focusing on discourse markers.

When I joined the faculty of Georgetown in 1982, I was immersed in the study of discourse, even though I was hired as a sociolinguist who could teach pragmatics and speech acts. Discourse analysis gradually filtered into those courses, as did face-to-face interaction, variation analysis, fieldwork, and even my old friend sociological theory. These various interests further jelled when I organized a Georgetown University Round Table on Languages and Linguistics in 1984, with the title Meaning, Form and Use in Context: Linguistic Applications. Thanks to the interest in discourse created by Deborah Tannen and the receptiveness of my sociolinguistics colleagues Roger Shuy and Ralph Fasold, I found – and continue to find – a community of faculty and students eager to pursue a collection of interests similar to my own under the rubric of discourse analysis.

Heidi E. Hamilton

My motivation to study discourse came from my real-life experiences with what Gumperz has called crosstalk. After receiving my bachelor's degree in German language and literature and cross-cultural studies, I worked in the field of international education for four years. Day after day I witnessed misunderstandings related to (what I would later learn were called) contextualization cues, framing, and complementary schismogenesis. I decided it was time to search for a graduate program to study the linguistic underpinnings of these misunderstandings. After culling through numerous graduate catalogs, I discovered that the courses that I had identified as the ones that seemed most intriguing and relevant led to a degree in linguistics at Georgetown University with a concentration on sociolinguistics. So off I went.

I was fortunate to begin my studies in 1981. The Georgetown University Round Table focusing on discourse had just been organized by Deborah Tannen. The entire department – students and faculty alike – was infused with a sense of excitement and open-ended possibility regarding the future of discourse studies. It was within this context that I worked as Deborah's research assistant and took her eye-opening courses on the analysis of conversation. In my second year of graduate study Deborah Schiffrin arrived at Georgetown as a new assistant professor, bringing with her a deep understanding of sociology and an approach to the analysis of discourse that was greatly influenced by Labov's work on variation. We graduate students were in the enviable position of working with two of the most innovative young discourse scholars at the time – a situation that became even more apparent to us a couple of years later.

In the summer of 1985, Georgetown University hosted 600 students and faculty who came from around the world to participate in the LSA Linguistic Institute organized by Deborah Tannen. Through the whirlwind of courses, lectures, and discussions, the interactional sociolinguistic approach to discourse analysis that we had been steeped in for several years was taking shape and gaining in prominence. Those of us educated at Georgetown kept hearing how very lucky we were to have the opportunity to study this kind of linguistics year-round. In retrospect, these comments seem to have foreshadowed the movement of the study of discourse from the fringes to a more mainstream position within linguistics.

Though my initial interest in crosstalk within international contexts never diminished (I came close to writing my dissertation on directness in German conversational style while living in Berlin for several years), I ended up shifting gears to another type of problematic talk – that of Alzheimer's disease. Little did I know that, with that choice of dissertation topic, I was jumping headfirst into a paradigmatic maelstrom. Being trained as an interactional discourse analyst, I was attempting to study a population that was firmly entrenched in the territory of neuro- and psycholinguistics. Time after time I found myself having to justify (to linguists and to gerontologists and neurologists alike) my attempt to marry the odd couple of Interactional Sociolinguistics and Alzheimer's disease. In the process, I learned quite a bit about how to talk across disciplinary boundaries, an enterprise that can be both frustrating and invigorating.

In 1990, when I joined the Georgetown Linguistics Department faculty, the program in discourse analysis was already very well established. Graduate students were entering our program better prepared than ever before and were ready to take their study of discourse to a new level. The field was mature enough to be expanded to include the study of exceptional discourse, which in turn can illuminate the often invisible workings of more ordinary, everyday discourse.

Purpose of the Handbook

Our own experiences in the field have led us to the conviction that the vastness and diversity of discourse analysis is a strength rather than a weakness. Far from its being a liability to be lamented because of the lack of a single coherent theory, we find the theoretical and methodological diversity of discourse analysis to be an asset. We thus envision this handbook as fostering the cooperative use – by linguists and others interested in empirically grounded studies of language – of the many theoretical and analytical resources currently proliferating in the study of discourse.

This collection of articles suggests that the future cooperation that we hope will emerge will respect the many differences that distinguish the approaches reflected here. There are differences in the types of data drawn upon, ranging from political speeches to everyday conversation to literary texts. There are also differences in the types of context considered, including community, institutional, and ideological contexts. Finally, there is a varied range of theoretical paradigms, such as Interactional Sociolinguistics, Conversation Analysis, Critical Discourse Analysis, and Systemic Functional Linguistics; and of methodology, including interpretive, statistical, and formal methods. As a result, the articles collected here suggest a foundational paradigm for discourse analysis that should be broad enough to support a wide range of assumptions, approaches, methods, analyses, and even definitions of discourse.

We hope that the range of chapters, and connections across them, will enhance the ability of discourse analysts to deal with a variety of problems and phenomena in ways that are not only internally coherent but also enriched by multiple resonances with one another. We also hope that the wide scope of chapters will reinforce the synergy between theory and data analysis that is reflected in the pervasive understanding of discourse analysis as the examination of actual (not hypothetical) text and/or talk. Although the authors have pursued a range of formats within the general topic assigned to them, we have encouraged them – in keeping with the term discourse analysis as well as the strong empirical bent that has characterized the field – to illustrate and substantiate general points by drawing upon concrete analyses of real discourse data. This springs from our conviction that theory and data are inseparable and mutually enriching: theoretical insights are needed to move the analysis of discourse beyond instance-specific insights, at the same time that analysis must be grounded in actual instances of language in order to provide both realistic constraints and empirical bases for theory-building. Though we have not asked contributors to address the need for – or even the desirability of – a single discourse theory, what contributors chose to include and emphasize, the themes and problems they address from the perspective of their specific areas, and the analyses and findings that they report all reveal the richness that must be respected and encompassed in discourse theories.

We hope, finally, that the breadth of articles collected here will provide a comprehensive view of the central issues in contemporary discourse analysis that is both accessible to students and informative to scholars. To this end, we have included articles by leading scholars in the field that provide an overview of their previous work, as well as chapters that survey the history of an area and summarize recent developments. In other articles, firmly established domains are assessed in order to link past approaches and findings with future challenges. In still others, authors develop relatively new fields of inquiry. Thus, we hope that the Handbook will serve not only as an authoritative guide to the major developments of discourse analysis but also as a significant contribution to current research.

Conclusion

We return, in conclusion, to the question, What is discourse? Years ago, Charles Fillmore captured the essence of discourse by presenting the following two sentences, each of which appeared as a sign at a swimming pool. One sign said, Please use the toilets, not the pool. The other said, Pool for members only. Read separately, each sign is reasonable enough. But, when the two sentences are read as if they were part of a single discourse, the second sentence forces a reinterpretation of the first that provokes laughter. Fillmore's example captures what we might call the gift of discourse: new meanings are created through the relationships between sentences. But the example also illustrates what we might call the challenge of discourse: Since more than one meaning can be created, how do we decide which meaning is intended, is justifiable, and/or makes the most sense?

We hope, through this Handbook, to offer a comprehensive sense of the scope and possibilities of discourse analysis, like the gift of multiple meanings. We know that some will see areas we have omitted or pathways we could have walked down that, due to the usual vagaries of human fallibility, we either did not pursue or were not able to realize. These omissions, though regrettable, are inevitable given the challenge of discourse: the directions in which its meanings may fan out are limitless. We have tried, at least, to provide a starting point from which the major highways emanate.

REFERENCES

Fasold, Ralph. 1990. Sociolinguistics of Language. Oxford: Blackwell.

Jaworski, Adam and Nikolas Coupland. 1999. The Discourse Reader. London and New York: Routledge.

Levinson, Stephen. 1983. Pragmatics. Cambridge: Cambridge University Press.

Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press.

I Linguistic Analysis of Discourse

1 Discourse and Grammar

MARIANNE MITHUN

0 Introduction

Language has traditionally been understood as a hierarchical system of systems: phonology, morphology, syntax, etc. A tenet of much of linguistic theory, particularly the American Structuralist and Generative approaches that arose during the twentieth century, was that intellectual rigor depended on a strict separation of these levels as autonomous, self-contained domains. For practical reasons, work began at the smaller, more concrete levels. Phonology was the study of the patterning of sounds; morphology how morphemes are combined to form words; syntax how words are combined to form sentences. Within mainstream theory in America, the focus had not yet moved to discourse, presumably the study of how sentences are combined to form texts, that is, structure beyond the sentence.

But running alongside this mainstream trajectory throughout most of the century was an interest in discourse in other circles. Members of the European Structuralist Prague School, founded in 1929, articulated their influential theory of Functional Sentence Perspective (Firbas 1966, 1992). Other scholars in North America integrated discourse into their work on language structure early on, among them Pike (1945, 1964a, 1964b, 1967, 1983), Bolinger (1964, 1968, 1972, 1982, 1989), Grimes (1971, 1975, 1978, 1982a, 1982b), Longacre (1977, 1978, 1982, 1983, 1984, 1985, 2003), Longacre and Shin (2012), and Halliday (1967–8, 1973, 1975, 2002; also active in Britain and Australia). References cited here represent only a small sample of the work of these productive scholars. All looked at language as an integrated communicative phenomenon.

As described by Tannen (Schiffrin, Tannen, and Hamilton 2001: 2–3), the last quarter of the twentieth century saw a blossoming of the status of the field of discourse analysis. Symposia devoted to discourse analysis began to spring up, first at Georgetown University and then elsewhere, as did journals such as Discourse & Society, Discourse Studies, Journal of Linguistic Anthropology, Journal of Pragmatics, Journal of Sociolinguistics, Multilingua, Narrative Inquiry, Pragmatics, Research on Language and Social Interaction, and Text. In certain quarters, work on grammar began to include consideration of the discourse context and the cognitive factors behind discourse structure. Among the important figures leading this were Chafe (1976, 1980, 1987, 1994) and Givón (1979, 1983, 1990; Givón and Gernsbacher 1994). (Both of these authors have continued to produce pioneering work.) All discourse analysis work shares a focus on extended bodies of speech in its communicative context. It is generally strongly empirically based. But it is not a monolithic endeavor characterized by a single set of questions, a single focus of inquiry, a single methodology, or a single theory. The variety of interests and approaches that characterize the field is richly exemplified in this volume.

For those interested in language structure, it is now generally recognized that discourse is more than an autonomous level beyond the sentence. Grammar provides speakers with tools for packaging information. And how information is packaged depends on the larger discourse context, the flow of thought through time, the communicative and social goals of the speaker, the presumed knowledge state of the audience, and more. Many of the grammatical choices speakers make at all levels – morphology, simple clause structure, and complex sentence structure – can be detected and understood only with respect to the discourse situation. At the same time, a full understanding of the discourse structures of a language depends on the recognition of the grammatical devices that signal them. Discourse structure is indicated by markers at all levels. It is more than the simple manipulation of sentences.

The relationship between discourse and grammar goes deeper. Recurring patterns of expression play a major role in the development of grammatical structures through time. What speakers choose to say the most often in the course of their daily interactions can become crystallized in grammar. In some cultures, for example, acceptable patterns of speech include specification of the source of information. With use, an expression such as ‘they say’ can become routinized, processed as a single unit. Over time, the expression may lose its internal compositionality and erode phonologically, until it is just a particle, a clitic, or an affix. It may even become obligatory. As Ariel puts it, discourse depends on grammar, which in turn depends on discourse (2009: 5).

A central aspect of the study of grammar is discovering what features all languages share and the ways they can differ. But, as long as our vision stops at the sentence, we will miss too much. The study of speech in its full discourse contexts can reveal cross-linguistic differences at all levels that may not be obvious when grammatical analyses focus on one level of structure at a time, each in isolation from the others. This chapter illustrates the kinds of intimate relations that hold between discourse and grammar in a language that is typologically quite different from more familiar major world languages. This is Mohawk, an Iroquoian language of northeastern North America, spoken primarily in Quebec, Ontario, and New York State. Much of the essence of the language could go unnoticed without examination of spontaneous, interactive speech in its discourse context.

1 The Basic Sentence

Pedagogical grammars are often rich in sentences constructed in isolation. Those in (1) all occurred in materials designed for teaching Mohawk. (Spelling has been regularized.1)

All of the words here are morphologically correct. It is unlikely that any of these sentences was ever uttered spontaneously by a speaker, however. If grammatical descriptions of the language were based on such sentences alone, the essence of the language would be severely distorted. The organization of information here, both the packaging of ideas into words and the combination of words into clauses and sentences, is decidedly un-Mohawk. Other grammatical features that are pervasive in normal speech are simply absent. The following sections will show how even seemingly straightforward grammatical structures cannot be understood fully without a consideration of their uses in their discourse contexts. All of the Mohawk material cited from this point on is drawn from a conversation at Kahnawà:ke, Quebec, involving a lively group of speakers ranging in age from their mid-twenties through their mid-seventies.2 All examples are from first-language speakers over the age of 60. They are arranged such that each line represents a separate intonation unit or prosodic phrase. In some cases, the larger context is provided just in free translation, but the original was in Mohawk.

2 The Word

One obvious difference between Mohawk and English is the packaging of information into words. Often a single Mohawk word corresponds to multiple English words. A typical example is in (2).3

But speakers have choices. Direction or location, for example, can be indicated by a verbal prefix, such as the cislocative te- ‘hither, here, there’ in (2), but also by a separate word, such as kén: ‘here’ or thó ‘there.’ Referents are identified by pronominal prefixes, such as sewa- ‘you all’ in (2), but also with separate words, such as í:se' ‘you.’ Objects can be identified inside words, such as ‘lunch’ here, or by separate nouns, such as atennà:tshera' ‘lunch, groceries.’ As will be seen throughout this chapter, such choices are not random.

2.1 Lexical categories

A basic unit of grammatical structure is the word class. Languages differ in the ways information is distributed among kinds of words. Mohawk contains just three lexical categories, clearly distinguished by their internal morphological structure: particles, nouns, and verbs.

Particles have no internal structure and serve a variety of functions, such as ì:'i ‘I,’ ne ‘the aforementioned,’ kí:ken ‘this,’ and the interrogative ken in the examples in (1) above.

Morphological nouns contain a gender or possessive prefix, a noun stem, and a noun suffix. The neuter prefix is ka-, o-, or zero, as in atókwa' ‘spoon’ in (1). A possessive prefix rao- ‘his’ can be seen in rao-wennókwas ‘his radio.’ The most common noun suffix -a' appears at the end of atokw-a'‘spoon.’ Nouns generally function syntactically as arguments, as would be expected.

Morphological verbs contain minimally a pronominal prefix and a verb stem. The prefix identifies the core arguments, one for intransitives and two for transitives. The verb wa'kkontsherárho' ‘I painted’ in (1a), for example, contains the first-person-singular prefix k-, the stem -kontsherarho- ‘paint,’ and the perfective suffix -'. Verbs may also contain numerous other elements. They can function syntactically as predicates, as in (1a) ‘I painted the chair,’ but they can also serve as full clauses. The word wa'kkontsherárho' is a complete grammatical sentence in itself: ‘I painted it.’

Morphological verbs can also serve other syntactic functions. They can be used as referring expressions with no change in form, such as kawennókwas ‘radio,’ and function syntactically as arguments.

Many morphological verbs, such as ‘radio,’ have been lexicalized as referring expressions. If a Mohawk speaker were asked about the meaning of kawennókwas, the first answer would probably be ‘radio.’ Lexicalization is a matter of degree: some verbs are normally used as nominals, others as both arguments and predicates, and still others only as predicates. Some other examples of verbal arguments from this conversation are in (4).

Mohawk contains no adjective category. Properties expressed with adjectives in other languages are conveyed in Mohawk with verbs.

Morphological verbs can also function as adverbials, like sewatié:ren's ‘sometimes’ in (6).

Due in part to their ability to function syntactically like the clauses, predicates, arguments, adjectives, and adverbs of other languages, morphological verbs are extremely frequent in Mohawk speech. When Wallace Chafe counted the proportion of nouns to verbs in a corpus of English conversation, he found a noun-to-verb ratio of about 1:1. A count over a similar Mohawk corpus yielded a proportion of 1:17.

The difference is not confined to morphological category. It appears in syntactic predicate-to-argument ratios as well. Patterns of idiomaticity vary interestingly across languages: what might normally be expressed in a predicate in one language might be expressed more often in an argument in another. During the conversation examined here, one speaker rehashed the morning's activities for a latecomer. What she later translated into English as ‘we did work’ was expressed with just the Mohawk predicate ‘we worked.’ What she translated as ‘a lot of’ was expressed in the Mohawk predicate ‘it was much.’

This conversation was full of similar differences. What was translated as an English possessed noun phrase ‘your habit’ in (8) was packaged in Mohawk in a predicate ‘how you are habituated.’

A sentence translated ‘I am waiting to have some soup’ contained no noun ‘soup’ in Mohawk. The idea of soup was conveyed by a predicate based on the verb stem -atshori ‘slurp.’

The sentence ‘he’ll still be a young man' contained no noun ‘young man.’ The idea was expressed in a predicate based on the verb root -nekenhter- ‘be good looking, be a young man.’

Mohawk speakers often use predicates for the idiomatic expression of ideas that English speakers convey with arguments. The difference is striking, but it emerges most clearly in unscripted speech in context.

2.2 Incorporation

Mohawk speech is characterized by a higher proportion of predicates for another reason. It contains a robust noun-incorporation construction, a kind of noun–verb compounding that yields a complex verb stem. Incorporated nouns are somewhat rarer in isolated constructed sentences than in spontaneous speech, though they do occur in lexicalized expressions such as ka-wenn-ókwas ‘it-word-scatters’ = ‘radio.’ Mohawk verbal counterparts to attributive adjectives in other languages often contain incorporated nouns.

Some adverbial notions are expressed with incorporating verbs.

But the full nature of noun incorporation cannot be appreciated in isolated sentences. Some of the motivations behind speakers' choices between independent and incorporated nouns can be seen by tracing the use of the noun root -wenn- ‘word, language’ through a stretch of the current conversation. The remark in (7) above, ‘This morning we did a lot of work,’ was addressed to a man who had just arrived. It was followed by (13). This first mention of the language to the newcomer was accomplished with an independent noun: onkwawén:na' ‘our language.’

The group lamented the difficulty of speaking Mohawk without reverting to English. The new arrival said (in Mohawk), ‘My older brother's like that. When we get together and talk, he starts speaking English to me. And he's my older brother.’ In the next sentence, ‘You would think he would push the language,’ the noun -wenn- ‘language’ was incorporated. The language was already the established topic of conversation, so a separate word was not necessary to focus special attention on it.

Incorporation is used as a rhetorical device for controlling the flow of information. One speaker could have said simply, ‘You’ll add to the story.' Instead, he developed his point in two intonation units, two clauses, the second, with an incorporated noun, an elaboration of the first.

Incorporated nouns do not bear a specific grammatical relation in the clause. They simply evoke a kind of entity, much like the non-head in English noun–noun compounds.

There is more to noun incorporation in Mohawk than online management of attention. Speakers do not necessarily produce language morpheme by morpheme as they speak. Frequently recurring chunks of language become routinized over time. As noted earlier, many verbs containing incorporated nouns have become lexicalized, stored as unitary expressions for single concepts. Lexicalization is a significant factor in noun incorporation: in speech, in many cases, both those like ‘radio’ and those like the alternation between incorporated and independent ‘language,’ incorporation is not an online process of word formation but rather a choice between existing alternatives.

Lexicalization can extend beyond the boundaries of the word, a fact that also affects the frequency of incorporation. A speaker remarked:

The language was already under discussion, so the incorporation of -wenn- could be attributed to its information status. But there was another factor. The construction consisting of the particle tsi plus a verb containing the partitive prefix (here ni-), an incorporated noun, and the verb root -o'ten ‘be a kind of’ is well established in the language. It is the way one talks about what something is like.

Frequency of use is an important aspect of incorporation. Some verb roots can appear with or without incorporated nouns. But some never incorporate, some rarely incorporate, some often incorporate, some usually incorporate, and some always incorporate. Some verbs that always incorporate denote relative properties, such as -iio ‘be good.’

Some verbs that always or usually incorporate contribute little independent information of their own, such as -ien ‘lie,’ which often serves simply to indicate the presence or absence of a referent.

Noun stems show a similar range of frequency of incorporation. Some are never incorporated, some rarely, some often, and some always. Many of those that incorporate more frequently have more general, even abstract meanings, such as -'nikonhr- ‘mind’ in verbs pertaining to mental phenomena, -ia't- ‘body’ in verbs pertaining to physical properties of animate beings, and -rihw- in verbs pertaining to abstract matters.

Without discourse, our understanding of noun incorporation would be superficial at best. Noun incorporation allows speakers to package familiar unitary concepts in single, lexicalized words, and also to carry established referents within verbs in ongoing speech without drawing special attention to them.

3 The Clause

In most models of syntax, a basic clause is assumed to consist of a predicate and one or more arguments. As seen in the previous section, the two may be packaged in a single word in Mohawk, a verb, such as Tesewatenna'tsherénhawe' ‘You all brought your lunches.’ Arguments can also be identified by additional words, as in Aonsetewatshèn:ri' nonkwawén:na' ‘We could find our words,’ with nonkwawén:na' ‘our words.’

3.1 Arguments

As in other languages, arguments in Mohawk may be identified by a simple pronoun or noun, or a more elaborate construction. The isolated sentences in (1) seen above show argument structures similar to those of English. A look at discourse shows a quite different story.

3.1.1 Pronominal arguments

In addition to the pronominal prefixes in verbs, Mohawk contains independent pronouns.

But these pronouns are actually rare in speech. Such patterns have sometimes been referred to as ‘pro-drop’: the pronoun is assumed to be present to begin with, then dropped under certain circumstances, as when reference is otherwise clear.

In the conversation discussed here, there are 195 first-person references, of which 12 are independent pronouns; there are 128 references to second persons, of which eight are independent pronouns. Given the numbers, the hypothesis that independent pronouns are dropped when reference is clear would be difficult to defend. All Mohawk verbs contain obligatory pronominal prefixes identifying their core arguments, so reference is always clear, even when an independent pronoun is used. The clause in the third line of (19), for example, contains both the independent pronoun í: and the pronominal prefix k- in the verb.

The pronominal prefixes actually make more distinctions than the independent pronouns. There are, for example, distinct prefixes for first-person-singular agent, first-person-inclusive-dual agent, first-person-exclusive-dual agent, first-person-inclusive-plural agent, first-person-exclusive agent, first-person-singular patient, first-person-dual patient, first-person-plural patient, first-person-singular inalienable possessor, first-person-dual inalienable possessor, first-person-plural alienable possessor, first-person-singular alienable possessor, first-person-dual alienable possessor, and first-person-plural alienable possessor. All of these categories are expressed with the same independent pronoun: ì:'i, often shortened to i:.

The independent pronouns have special discourse functions. One is to mark a shift in topic, as in (19) above: ‘You could tell a story, then I could continue …’ Another is to mark focus, information that the speaker deems especially important. Speaker A below was making fun of the dialect spoken in a neighboring community. Speaker B, who was born there, protested.

This focus construction is often characterized by distinctive intonation as well. The focused element is pronounced with extra-high pitch, visible in the bump in the pitch trace in Figure 1.1.

Figure 1.1 Focused pronoun ‘me.’

Independent pronouns are also often used to highlight a focus of contrast.

3.1.2 The Determiner Phrase

In most current syntactic theory, arguments are analyzed as clausal constituents, Determiner Phrases, with an internal structure of their own. The prototypical Determiner Phrase consists of a determiner (article or demonstrative), optional adjectives, and a noun: a nice book, this lovely house. As noted, Mohawk has no adjective category. It does, however, contain both an article and demonstratives.

If we look at the isolated sentences in (1) earlier, the Mohawk article ne seems comparable to English ‘the.’

The absence of Mohawk ne corresponds to an indefinite article in the English translation.

As described in detail by Chafe (1976, 1994), the English definite article indicates identifiability: it signals that the speaker assumes the hearer can identify the referent. Identifiability can come from various factors: uniqueness (Don't look at the sun [there is only one sun]), common knowledge (I've already fed the dog [we have just one dog and we both know who it is]), prior mention (I bought a coat and matching scarf. The coat …), or association with something identifiable (I bought a coat. The sleeves …). Judging from (1b) and (1d) above, Mohawk ne seems to mark general identifiability.

But, in more extensive bodies of speech, Mohawk ne sometimes appears in contexts where English the would not.

And it is sometimes absent from contexts where English the is used.

An accurate understanding of ne emerges only from discourse. Speaker A below brought up a word she had heard used for ‘thousand,’ iohsóhserote'. (The entire conversation was in Mohawk.)

Mohawk ne does not mark general identifiability but rather previous mention within the discourse. The first time the word iohsóhserote' was mentioned, there was no ne. The second time it was preceded by ne: ‘Is it an old word, ne iohsóhserote’?' The ne next appears with a possessed noun: ‘ne our language,’ pronounced nonkwawén:na'. At this point the language was already under discussion. The ne appears again in the following line, this time before ‘our words,’ also a central topic of the ongoing discussion. Two lines later, it appears before ‘word’ (ne owén:na'), again a referent established a few lines before. Finally, in the last line, it occurs before a complement clause: ‘our renting it.’ This clause, the argument of the matrix iotié:ren ‘it is surprising,’ is functioning as a nominal, identifying a previously introduced idea. The Mohawk ne is thus better rendered as ‘the aforementioned.’ It often appears to function like the English definite article the, because previous mention is a common way of establishing definiteness. But the actual meaning ‘the aforementioned’ can only be seen in discourse.

This refined understanding now allows us to make sense of the two sentences seen earlier. The sentence ‘All of us hold ne our language’ in (22) occurred in the midst of a discussion about the Mohawk language. When the speaker remarked in (23), ‘I thought maybe she was in (the) hospital,’ this was the first mention of the hospital, so there was no ne, even though there is only one hospital in this community. Sentences constructed in isolation, even by skilled native speakers, often do not reflect the functions of markers whose meanings depend on a larger discourse context.

It is generally assumed that a fundamental element of the Determiner Phrase cross-linguistically is the demonstrative. Judging from the isolated sentences in (1) earlier, the Mohawk kí:ken and thí:ken seem comparable to English ‘this/these’ and ‘that/those.’

At first glance, spontaneous speech appears to reflect a similar structure.

The prosody reveals a different structure. The group had been discussing dialect differences between communities. In Kahnawà:ke, where this conversation took place, the cluster /ts/ is pronounced [dz] before a vowel: [odzì:dza?] ‘flower.’ In another community, Ahkwesáhsne, it is pronounced [dʒ]: [odʒì:dʒa?]. The utterance in (25) actually consisted of two prosodic sentences. The first ended with thí: ‘that’ and a full terminal fall. It was separated from the next by a response from a listener. The second sentence began with a high-pitch reset on the stressed syllable. (The pitch appears extra high because of the tone, characterized by an extra-high rise followed by a steep plunge.)

A pitch trace can be seen in Figure 1.2. The sequence thí: otsì:tsia' does not constitute a single constituent.

Figure 1.2 Demonstrative thí: ‘that.’

Larger stretches of discourse show that demonstratives are rarely elements of a Determiner Phrase constituent in Mohawk. Demonstratives do occur on occasion before nouns, but they are usually referring expressions on their own. Furthermore, they more often serve functions beyond what would be predicted from the expected Determiner Phrase structure. As seen earlier, a Mohawk verb can constitute a complete sentence on its own. Additional elaboration is possible with lexical arguments. But, as pointed out by Chafe (1987, 1994), speakers are careful not to introduce too many major new ideas at once. One strategy for conveying one new idea at a time is to begin with a predicate (perhaps with particles) followed by a demonstrative. The demonstrative serves as a place holder, signaling that further specification is to follow. In (27), as throughout, each intonation unit is presented on a separate line. The proximal demonstrative is kí:ken ‘this,’ often shortened to kí:.

The segmentation of ideas into phrases can be seen in the sound wave and pitch trace in Figure 1.3.

Figure 1.3 Demonstratives as place holders.

The Determiner Phrase, considered a fundamental element of syntactic structure in most current models of syntax, might appear to be a language universal on the basis of sentences constructed in isolation. It is well known, of course, that languages vary in the order of elements within their Determiner Phrases; in some languages, for example, determiners and other material precede the noun, as in English, while in others they follow, as in Japanese. Examination of spontaneous speech in context, however, indicates that there can be deeper differences involving the relationships between these constituents.

3.2 Core and oblique

Mohawk speakers differ not only in how they distribute information over words within clauses but also in how they distribute ideas over clauses within sentences and beyond. Such patterns and the reasons behind them are not always obvious from isolated sentences.

As mentioned, a basic notion in syntax is that the clause consists of a predicate, one or two (or three) core arguments, and any number of obliques (adjuncts). In English, obliques are usually marked with prepositions.

In some languages, obliques are marked with case endings. Mohawk has neither adpositions nor case endings. Core arguments are identified by a pronominal prefix in the verb, but the roles of lexical nominals are simply inferred. In (29), the location is identified by the word Kahnawà:ke. This is a placename, but there is nothing in the sentence to indicate its syntactic role. The same form would be used if the speaker were saying ‘Kahnawà:ke is a beautiful place’ or ‘We were discussing Kahnawà:ke.’

But this language differs in a subtle way from canonical expectations. Mohawk clauses are not stacked with arguments. Ideas expressed in obliques in other languages are often expressed in other ways in this language. One is with noun incorporation, like the boat and the island in (30). Their semantic roles are often inferred from the verb, such as ‘encircle.’

In (31) the instrument ‘elbow’ is incorporated into the applicative verb ‘hit-with.’

But often the additional referent is introduced in a separate clause, like the location ‘baskets’ in (32) and the companion ‘my grandmother’ in (33).

On their own, these examples do not appear unusual. But monoclausal alternatives like ‘Gold was handed to them in baskets’ are rare in spontaneous discourse. When asked directly for a translation of the English ‘She fried her eggs with butter,’ a Mohawk speaker provided the bi-clausal construction in (34).

In (35) the time was introduced in one sentence and the language in another.

Rather than presenting the idea ‘The learned people call it polysynthetic’ in a single sentence, the speaker packaged it in two sentences, three intonation units:

As can be seen in the pitch trace (Figure 1.4), each sentence begins with a pitch reset.