Parts of the empirical research reported in this book were supported by research funds from the Arts and Humanities Research Board/Council, the Nuffield Foundation, and IELTS.
For permission to reproduce material I am grateful to various colleagues with whom I have co-researched and/or co-written: Janet Cowper, Stephen Cox, Tess Fitzpatrick, Kazuhiko Namba, and John Staczek. Thanks also to Mary Goebel Noguchi, Dawn French, Sue Hunter, Sarah Lloyd, Jon Plowman, and various publishers of material already in the public domain for helping me obtain permission to reproduce it. For support in the development of the idea for the book, and the preparation of the manuscript, my sincere thanks to Natasha Forrest, Julia Sallabank, Cristina Whitecross, Maggie Shade, and Henry Widdowson.
Finally, for putting up with me, even when it meant sitting indoors instead of walking outdoors, and for unstinting encouragement, support, and inspiration, huge thanks to Mike Wallace.
Copyright statement
Every effort has been made to trace and contact copyright holders. Any apparent infringement of copyright is unintentional, and the publisher will be pleased to rectify any errors or ommissions at the earliest opportunity.
PART ONE
Determining boundaries
1
Introduction
Boundaries
When we express ourselves using language, we encode our ideas and attitudes by combining units of form that are associated with meanings. A dominant view in linguistics has been that lexical units are small, and the combinatory rules quite complex. However, there is increasing evidence that some of the units are larger and contain within them tacit complexity that would otherwise be under the remit of the combinatory rules. These larger units still combine with others, but the overall amount of grammatical activity entailed in expressing a complete message is reduced, and some of it is potentially different in kind, compared with that entailed in expressing the same message using smaller units. Since the smaller units also exist, this account simply proposes a greater flexibility in how language is expressed and understood than was previously assumed. Very few, if any, of the explanatory advantages of the ‘small unit’ view are lost, while a number of powerful opportunities arise for explaining patterns observed in language, particularly the complete or relative absence of certain configurations that the small unit view predicts to be possible.
‘Formulaic language’1 is a term used by many researchers to refer to the large units of processing – that is, lexical units that are more than one word long. Although it will be demonstrated later that only certain theories can sustain a fundamental distinction between large and small lexical units, for now it will be adequate, and indeed clearer, to adhere to this preliminary characterization. Research into formulaic language has proliferated in recent years, particularly since computer technology has made it possible to search effectively for examples of recurring multiword strings in large corpora – though opinions differ as to the precise relationship between frequency and formulaicity.
Yet there remain mysteries at the heart of the phenomenon: What exactly is formulaic language? Why do we have it? How does it arise? How can it help with practical matters like learning another language or communicating when one has a language disorder? In my 2002 book, Formulaic Language and the Lexicon (Wray 2002b), I offered possible answers to these and other questions, by locating formulaic language within the larger frame of how we direct our limited language processing capacities towards formulating the messages needed to achieve certain important goals of social interaction. The key proposals developed in that book are recapped in Chapter 2 of this one, as the springboard for new issues that form the focus of the remaining chapters.
This book aims to discover more about the nature and extent of formulaic language by examining a broad range evidence and theory ‘at the boundaries’. Furthermore, in the light of the claim that formulaic language is a linguistic solution to a non-linguistic problem – namely our need to promote and protect ourselves in relation to others (Wray 2002b: 101) – it explores ways in which a greater understanding of the scope of formulaic language might inform our engagement with practical challenges in relation to communication.
The unfolding book addresses inherent tensions regarding the identification, definition, and theoretical modelling of formulaicity. Although the Needs Only Analysis model that underpins this book is not truly falsifiable in the Popperian sense, the attempt is made, nevertheless, to examine the extent of its explanatory power by testing the boundaries of its scope. It is at the boundaries that the real test cases are to be found, and grappling with them provides both valuable insights into the robustness of one’s existing beliefs, and ground rules for evaluating the place of formulaic language in common and unusual linguistic situations.
At the boundaries
How does one engage at the boundaries? The answer is that one must take risks about what one examines. If one does not, one is not really at the boundaries at all. One must anticipate that at least some of the material one examines is possibly not formulaic – the point is to explore where the borderline is best placed. As in my previous publications, particularly Wray (2002b), my approach in this book is to err on the side of including too much in the first instance, rather than too little, on the assumption that it is better to examine and discard something than to overlook it.
Thus, in this book, the notion of formulaicity is extended to a range of linguistic and non-linguistic phenomena that may, for some, seem too extreme to really be counted as formulaic. They are included because they fall within, or very close to, the boundaries of the theoretical model arising from my 2002b account. The material explored here therefore encompasses not only idioms and common multiword strings with a specific communicative purpose, but also deliberately memorized phrases, extensive oral narratives, performance scripts, and even visual signalling systems. They have in common that a communicative act is effected using one or more internally complex, predetermined units of form. As the various data-based discussions throughout the book will reveal, lengthy memorized texts are evidently internalized as series of smaller chunks, at the boundaries of which variations may occur (see particularly Chapter 20). We might argue the toss, in such cases, about what the formula is: the smaller chunk or the larger script. And that is precisely the point of the exercise – to use a range of evidence and reasoning to come closer to a robust account of the essence of formulaicity.
Formulaic language is investigated here by looking at what happens when language operates beyond its normal scope, where there are unusual constraints on communication that make formulaic language more evident, or where language users choose, or are forced, to favour previously assembled output over something more spontaneous. In many different ways, we are at the boundaries – of language behaviour, of communicative potential, and of linguistic theory – and aim to see what happens when you squeeze a phenomenon until (as we say formulaically) the pips squeak.
Five questions about formulaic language
The impetus of the book is the exploration of five key questions arising from published claims about formulaic language. Each question has one or more dedicated chapters in Part Four of the book, and answers are explored with reference to the theory and data presented in the first three parts.
The first question under consideration is: Do we use formulaic language by default? If we do, it means that we attempt first to work with pre-existing multiword forms, and only move to smaller ones if it becomes necessary – very much as Sinclair (1991) proposed with his ‘idiom principle’ and ‘open choice’ principle. To what extent are our assumptions about building up utterances using words and rules just a product of our cultural conditioning? Could Biggs’ (1998) claim that the Maori language cannot be learnt other than by viewing the phrase as the base component be indicative of alternative ways of viewing what is ‘normal’? The question is addressed in Chapter 16, and Chapter