The Prosody of Formulaic Sequences

A Corpus and Discourse Approach

Phoebe Lin

• Publisher: Bloomsbury Academic
• Year: 2018
• ISBN:
  9781441100856
• Edition: 1
• Topic: Languages & Linguistics
• Subtopic: Syntax in Linguistics
• Index: Languages & Linguistics

To apply the same approaches to analysing spoken and written formulaic language is problematic; to do so masks the fact that the contextual meaning of spoken formulaic language is encoded, to a large extent, in its prosody. In The Prosody of Formulaic Sequences, Phoebe Lin offers a new perspective on formulaic language, arguing that while past research often treats formulaic language as a lexical phenomenon, the phonological aspect of it is a more fundamental facet. This book draws its conclusions from three original, empirical studies of spoken formulaic language, assessing intonation unit boundaries as well as features such as tempo and stress placement. Across all studies, Lin considers questions of methodology and conceptual framework. The corpus-based descriptions of prosody outlined in this book not only deepen our understanding of the nature of formulaic language but have important implications for English Language Teaching and automatic speech synthesis.

1

Introduction

Formulaic language, which is ‘any sequence of two or more words that are perceived to be more constrained than usual in their co-occurrence’ (Hudson and Wiktorsson, 2009, p. 81), is ubiquitous in everyday language. In the past two decades, many corpus-based studies have been conducted to reveal the use of formulaic language in spontaneous speech (e.g. Altenberg, 1998; Altenberg and Eeg-Olofsson, 1990; Biber, 2006; Biber and Barbieria, 2007; De Cock, 1998, 2000, 2007; Simpson, 2004). A very popular approach has been to examine highly recurrent, contiguous sequences of word forms extracted from a spoken corpus using automatic extraction tools like WordSmith (Scott, 2012). This approach has revealed many important findings; however, treating spoken and written data in the same way may mask the fact that meaning in spoken language is, to a great extent, encoded in speech prosody. Words only hold part of the overall contextual meaning of an utterance. To fully understand meaning in context, one must also consider the prosody of the utterance. The English cliché ‘It’s not what you say, but how you say it’ captures the essence of this argument. Nonetheless, many spoken corpora (with the notable exceptions of the Spoken English Corpus and the London-Lund Corpus of Spoken English) still do not provide prosodic information either through prosodic transcriptions or through audio streams synchronized with the transcripts. This explains, in part, why researchers in the past often treated spoken formulaic language and written formulaic language in the same fashion.

If it is found that there is a particular way in which formulaic language should be spoken, the impact of such a discovery could be tremendous since it concerns not only the teaching and learning of formulaic language, but also formulaic language research methodology. In relation to language teaching, researchers including Pawley and Syder (1983), Wray (2002) and Wood (2012) have long argued that if learners want to achieve native-like fluency and proficiency in English as a second or foreign language they need to have a similarly sized store of formulaic language as native speakers. A very important point that has been overlooked so far is that every entry of formulaic language in the mental lexicon must contain the phonological form of the formulaic language in addition to its orthographic, structural and semantic information (see also Lin, 2012, 2018). Therefore, to be able to use formulaic language effectively, English language learners must also master the prosody of formulaic language. If there is a particular prosody with which formulaic language is said, English language learners should be aware of it and teachers should likewise consider teaching it. English prosody, however, has not, until now, been a priority in the English language teaching (ELT) syllabus. This reflects a lack of awareness of the importance of prosody in conveying (and contradicting) meaning. The work of phonologists including Bolinger (1989), Crystal (1969, 2003), Halliday (1967) and Wennerstrom (1994, 1998, 2001, 2006) has shown that prosody is a complex system that reflects emotion, contextual meaning, grammatical structure and cohesion. Aijmer (1996), for example, discussed the distribution of nuclear tones in the multi-word formula thank you and how tone choice makes a difference to the emotional weight of the phrase (see also Wells, 2006, Section 2.19). One can imagine learners learning to use thank you but without an understanding of the tones required to add their implied meaning to the formula, thus impeding their ability to communicate the meaning fully. Likewise, Cowie (1988, p. 134) used the example of do you know as in the utterance Do you know, he was still in bed! to illustrate the argument that intonation is an indispensable part of knowledge in the successful use of formulaic language:

In order to use this formula successfully, the speaker requires knowledge of invariant form, syntactic position (initial rather than final) and intonation (fall-rise on know).

In the field of formulaic language research, investigations into the prosody of formulaic language are also likely to have a profound impact on research methodology. Nowadays corpus-based automatic extraction is the most commonly used identification method of formulaic language. This identification method makes it possible for researchers to look at patterns of formulaic language use in a language corpus on a large scale. However, the key problem, as mentioned earlier, is that it treats spoken and written formulaic language in the same fashion, due to the neglect of phonological evidence (see also Section 2.3.1). In fact, many researchers (e.g. Aijmer, 1996; Altenberg and Eeg-Olofsson, 1990; Baker and McCarthy, 1988; Bloom, 1973; Hickey, 1993; Peters, 1977, 1983; Plunkett, 1990; Weinert, 1995; Wray, 2002, 2004; Wray and Namba, 2003) have suggested that formulaic language can be identified by its phonological and prosodic features. These prosodic features that are particularly associated with formulaic language will be reviewed in Chapter 3. The key point to be noted here however is that, so far, no empirical study has been conducted that has comprehensively investigated the hypothesis that formulaic language can be identified by its unique prosodic features (see Section 3.3). If phonological and prosodic cues are to assist in the identification of formulaic language in adult speech, an investigation into the prosodic features of formulaic language is a prerequisite. It is hoped that this will be achieved through study one and study two.

1.1 Is there a prosody of formulaic language?

The previous section presented the potential impacts of the discovery that formulaic language has a distinctive prosody on ELT and formulaic language methodology. The key question being asked here is whether there is a prosody specific to formulaic language that is different from the prosody of spoken English. Phonologists including Ashby (2006) and Wells (2006) have provided an affirmative answer to this question.

Ashby (2006) gives the example of to have eyes in the back of one’s head to illustrate how the prosody of idioms differs from the prosodic rules governing general spoken language. When this idiom is used in context as an utterance to describe a person who is very vigilant about everything around him or her (see (1) below), it must be delivered with stress or pitch prominence on the word head as in (1a). However, general prosodic rules should predict (1b) or (1c) instead because, given that everybody has eyes in the front of their head, the ‘surprising’ and ‘new’ information in the utterance that should be highlighted prosodically is eyes or back. However, it is clear that neither (1b) nor (1c) is appropriate because they convey to the hearer that a woman ‘literally’ has eyes in the back of her head.

(1) She has eyes in the back of her head.

(1a) She has eyes in the back of her HEAD.

*(1b) She has EYES in the back of her head.

*(1c) She has eyes in the BACK of her head.

In the book English Intonation: An Introduction, Wells (2006) comprehensively presents the rules governing the prosody of spoken English. In presenting the prosodic rules, he notes some exceptional cases which cannot be explained by the overarching rules. Because no logical explanation can be found for these cases, he declares these to be ‘idiomatic cases’, as the two quotes below clearly show:

Some instances of a speaker accenting repeated words do not seem to have a logical explanation, and must be regarded as idiomatic. (Wells, 2006, p. 179)

and

Rather than seek a logical explanation for this tonicity, perhaps we should regard such cases as merely idiomatic. (Wells, 2006, p. 181)

Examples of idiomatic cases include accenting the ‘empty word’ some in what Wells calls ‘idiomatic expressions’ such as for SOME reason, in SOME cases and SOME days (see Wells, 2006, p. 150), and the accenting of the verb to be in what he calls ‘intonational idioms’ such as the trouble IS, the thing IS, the difficulty IS and the snag IS (see Wells, 2006, p. 146). Other examples of idiomatic cases provided by Wells (2006) are summarized in Table 3.4.

It is clear that from the phonologists’ (i.e. Ashby, 2006; Wells, 2006) perspective, a prosody of formulaic language exists that cannot be explained by general English prosodic rules. This is true at least in terms of stress placement. However, going beyond the work of Ashby (2006) and Wells (2006), it can be seen that the distinctive prosody of formulaic language in fact also extends to the assignment of intonation breaks or pauses. Utterance (2), below, is a concordance line taken from the British National Corpus. Any proficient speaker of English will recognize that there are only two options as to where to assign a break or a pause within the utterance, as in (2a) and (2b). In both cases, the ‘legitimate’ locations for the breaks and pauses are right at the boundaries of formulaic sequences, that is the thing is, organic produce and isn’t cheap, but not elsewhere. In fact, the break after the thing is appears to be obligatory as (2c), even if it exists, would be extremely rare. Hence, the break after the thing is is interesting because, if syntactic rules govern the prosody of formulaic language just as they govern the prosody of general spoken English, we would expect a break between the subject and the predicate, (in (2d) – The thing is the subject and is organic produce isn’t cheap is the predicate).

(2) The thing is organic produce isn’t cheap.

(2a) The thing is, organic produce isn’t cheap.

(2b) The thing is, organic produce, isn’t cheap.

*(2c) The thing is organic produce, isn’t cheap.

*(2d) The thing, is organic produce isn’t cheap.

Based on the accounts of the phonologists, it is clear that there is no shortage of examples in which no logical explanation can be found for the prosodic features observed other than to regard such cases as ‘idiomatic’. However, attention must be drawn to the fact that the discussions provided by phonologists have often been based on anecdotal and introspective data. Instead of putting the prosody of formulaic language at the centre of study, phonologists have a tendency to leave idiomaticity at the periphery and use the concept only as an explanation for ‘exceptional cases’ which general English prosodic rules cannot explain. From the perspective of formulaic language researchers, the phonologists’ approach is insufficient. First, formulaic language research requires an objective definition or a system by which the formulaicity of word sequences can be validated. However, the phonologists’ identification of formulaic language may appear ad hoc. Secondly, phonologists’ introspection cannot replace the empirical observations of the prosodic features of formulaic language based on real, spontaneous speech data. Thirdly, the introspective approach makes it difficult to establish the scale of the phenomenon of prosodic fixedness.

On this basis, there is a great need for empirical investigations into the prosodic features of formulaic language which adopt a systematic approach not only to the prosodic description but also to the identification of formulaic language. This need is fulfilled in the three empirical studies presented here.

1.2 The three empirical studies

This book reports three empirical studies into the prosody of formulaic language. Before introducing the three studies, it is necessary to point out that these three studies represent the journey of how my consideration over time of the overarching question can we identify formulaic language based on prosodic cues? has changed. To address this overarching research question, the original plan was to examine empirically whether formulaic language can be identified based on tracking intonation unit boundaries. This was what study one of the book accomplished. However, in examining the literature and the results of study one, it became obvious to me that the idea that formula ic language could be identified by tracking intonation unit boundaries alone might be too simplistic and one-dimensional in two respects. First, it is possible that formulaic language could be identified on the basis of other prosodic cues such as stress and rhythmic patterns in addition to intonation. Secondly, the division of spontaneous speech into intonation units is affected by a series of globally and locally managed contextual factors such as stylistic effects, cognitive resources and semantic focus; formulaicity is only one of the variables in the formula. With the aim of expanding the scope of the exploration of the overarching research question, study two was designed as an extension to study one. The aim of this study was to determine more precisely if formulaic language demonstrates any unique patterns in terms of stress placement and rhythm that may ultimately lead to the identification of formulaic language in spontaneous speech.

In considering the use of prosodic cues to identify formulaic language in studies one and two, I realized that my exploration of the overarching research question had been particularly influenced by the unsaid assumption in the literature about how the phonological method would work. This untold assumption is that the tracking of a finite and unique set of prosodic cues should reveal the extent to which a word sequence is formulaic. This assumption influenced the design of studies one and two. However, this is only one way of interpreting the significance of the phonological method in formulaic language research. An alternative interpretation of the phonological method exists, however, which has been inspired by the recent trend towards multimodality in corpus linguistics. In thinking about the practicalities of data collection for this book, I discovered that in the past the native speaker judgement method of identifying formulaic language was conducted on textual transcripts without the provision of accompanying audio recordings. Omitting the audio component in the process of native speaker judgement is problematic because the native speaker judges have no access to meaning that is encoded in prosody. Furthermore, if formulaic language can be identified by tracking a finite and unique set of prosodic cues, it is necessary for the native speaker judges to be able to listen to the prosody of formulaic language in the identification process. That is why this alternative interpretation of the phonological method adopts a multimodal approach in the process of native speaker judgement, under which native speaker judges are allowed to listen to the original recordings of the spoken texts alongside the transcripts in which they are required to identify formulaic language. This alternative phonological method was what study three aimed to test.

All in all, the three empirical studies were designed to explore, from different angles, the question of whether and how formulaic language can be identified using prosodic cues. Furthermore, studies one to three represent how my consideration of the overarching research question developed over time. The starting point, in study one, is the simple belief that formulaic language can be identified by tracking intonation units; then study two expands the exploration by considering the stress and rhythmic patterns of formulaic language; finally, study three reflects the realization that an alternative interpretation of phonological method can be adopted, in which native speaker judges are allowed to listen to the audio recordings of the transcript in the process of formulaic language identification. This alternative interpretation of the phonological method, however, should not be considered as a replacement of the original interpretation which is reflected in studies one and two. They are both necessary.

While it is essential to discuss the overarching question that links together the three empirical studies, each of the empirical studies addressed very specific research questions about the use of prosodic cues in the identification of formulaic language. To be precise, they were designed to answer research questions one, two and three respectively.

1. To what extent do formulaic sequences align with intonation unit boundaries

a) in the fluent, spontaneous speech of native speakers and proficient English-as-a-Foreign-Language (EFL) learners;

b) when the formulaic sequences are identified by automatic extraction or by collective native speaker judgement; and

c) when the formulaic sequences are classified according to how confident the native speaker judges are of their formulaicity judgements?

2. Does formulaic language demonstrate any unique patterns in terms of tempo/rhythm and stress placement?

3. Will listening to the audio recordings in the process of native speaker judgement increase t...

Table of contents

• Cover
• Half-Title
• Series
• Title
• Contents
• List of Illustrations
• Acknowledgements
• 1 Introduction
• 2 Formulaic Language: An Overview
• 3 Can We Identify Formulaic Language Based on Prosodic Cues?
• 4 Study One: Do Formulaic Sequences Align with Intonation Units?
• 5 Study Two: A Comprehensive Profile of the Intonation, Stress and Rhythm of Formulaic Language
• 6 Study Three: A Multimodal Approach to the Identification of Formulaic Language by Native Speaker Judgement
• 7 Conclusions: The Prosody of Formulaic Language
• Appendix 1: Information on the Pilot Study
• Appendix 2: The Instructions Used in Various Sessions of the Pilot Study
• Appendix 3: Screenshots of the Interactive PowerPoint Interface with Integrated Audio Playback
• Appendix 4: The Task Booklet for the Native Speaker Judgement Process
• Appendix 5: Instructions for Prosodic Transcribers
• Appendix 6: Prosodic Transcription of Text W
• Appendix 7: Formulaic Sequences Assigned the Maximum Confidence Score in Text W
• Appendix 8: The Duration of the Task for each Native Speaker Judge
• Notes
• Bibliography
• Author Index
• Subject Index
• Copyright