Download the full version and explore a variety of ebooks

or textbooks at https://ebookmass.com

Discourse Markers and Beyond: Descriptive and

Critical Perspectives on Discourse-Pragmatic
Devices across Genres and Languages 1st ed. 2020
Edition Péter B. Furkó

_____ Tap the link below to start your download _____

https://ebookmass.com/product/discourse-markers-and-beyond-
descriptive-and-critical-perspectives-on-discourse-
pragmatic-devices-across-genres-and-languages-1st-
ed-2020-edition-peter-b-furko/

Find ebooks or textbooks at ebookmass.com today!

Here are some recommended products for you. Click the link to
download, or explore more at ebookmass.com

A Contrastive View of Discourse Markers: Discourse Markers

of Saying in English and French 1st ed. 2020 Edition Laure
Lansari
https://ebookmass.com/product/a-contrastive-view-of-discourse-markers-
discourse-markers-of-saying-in-english-and-french-1st-ed-2020-edition-
laure-lansari/

Discourse Markers: An Enunciative Approach 1st Edition

Graham Ranger

https://ebookmass.com/product/discourse-markers-an-enunciative-
approach-1st-edition-graham-ranger/

Language, Vernacular Discourse and Nationalisms 1st ed.

Edition Finex Ndhlovu

https://ebookmass.com/product/language-vernacular-discourse-and-
nationalisms-1st-ed-edition-finex-ndhlovu/

Ethical Discourse in Finance: Interdisciplinary and

Diverse Perspectives Marizah Minhat

https://ebookmass.com/product/ethical-discourse-in-finance-
interdisciplinary-and-diverse-perspectives-marizah-minhat/
Therapy as Discourse: Practice and Research 1st ed.
Edition Olga Smoliak

https://ebookmass.com/product/therapy-as-discourse-practice-and-
research-1st-ed-edition-olga-smoliak/

Foreign Languages in Advertising: Linguistic and Marketing

Perspectives 1st ed. 2020 Edition Jos Hornikx

https://ebookmass.com/product/foreign-languages-in-advertising-
linguistic-and-marketing-perspectives-1st-ed-2020-edition-jos-hornikx/

Expressions of War in Australia and the Pacific: Language,

Trauma, Memory, and Official Discourse 1st ed. 2020
Edition Amanda Laugesen
https://ebookmass.com/product/expressions-of-war-in-australia-and-the-
pacific-language-trauma-memory-and-official-discourse-1st-
ed-2020-edition-amanda-laugesen/

The Politics of Climate Change Metaphors in the U.S.

Discourse: Conceptual Metaphor Theory and Analysis from an
Ecolinguistics and Critical Discourse Analysis Perspective
Othman Khalid Al-Shboul
https://ebookmass.com/product/the-politics-of-climate-change-
metaphors-in-the-u-s-discourse-conceptual-metaphor-theory-and-
analysis-from-an-ecolinguistics-and-critical-discourse-analysis-
perspective-othman-khalid-al-shboul/

Environmental Organizations and Reasoned Discourse Richard

M. Robinson

https://ebookmass.com/product/environmental-organizations-and-
reasoned-discourse-richard-m-robinson/
POSTDISCIPLINARY STUDIES IN DISCOURSE
SERIES EDITOR: JOHANNES ANGERMULLER

Discourse Markers
and Beyond
Descriptive and Critical Perspectives
on Discourse-Pragmatic Devices across
Genres and Languages

Péter B. Furkó
Postdisciplinary Studies in Discourse

Series Editor
Johannes Angermuller
Centre for Applied Linguistics
University of Warwick
Coventry, UK
Postdisciplinary Studies in Discourse engages in the exchange between
discourse theory and analysis while putting emphasis on the intellectual
challenges in discourse research. Moving beyond disciplinary divisions
in today’s social sciences, the contributions deal with critical issues at
the intersections between language and society.
Edited by Johannes Angermuller together with members of
DiscourseNet, the series welcomes high-quality manuscripts in dis-
course research from all disciplinary and geographical backgrounds.
DiscourseNet is an international and interdisciplinary network of
researchers which is open to discourse analysts and theorists from all
backgrounds.

Editorial Board
Cristina Arancibia
Aurora Fragonara
Péter Furkó
Tian Hailong
Jens Maesse
Eduardo Chávez Herrera
Michael Kranert
Jan Krasni
María Laura Pardo
Yannik Porsché
Kaushalya Perera
Luciana Radut-Gaghi
Marco Antonio Ruiz
Jan Zienkowski

More information about this series at

http://www.palgrave.com/gp/series/14534
 Péter B. Furkó

Discourse Markers
and Beyond
Descriptive and Critical Perspectives
on Discourse-Pragmatic Devices across
Genres and Languages
Péter B. Furkó
Department of English Linguistics
Károli Gáspár University of the Reformed
Church in Hungary
Budapest, Hungary

Postdisciplinary Studies in Discourse

ISBN 978-3-030-37762-5 ISBN 978-3-030-37763-2 (eBook)
https://doi.org/10.1007/978-3-030-37763-2

© The Editor(s) (if applicable) and The Author(s) 2020

This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by
similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with
regard to jurisdictional claims in published maps and institutional affiliations.

Cover illustration: RooM the Agency/Alamy Stock Photo

This Palgrave Macmillan imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
 Contents

1 Preliminary Issues: Category Membership, Methodology,

Alternative Perspectives on Discourse Markers 1
1 Theoretical and Empirical Approaches to Discourse Markers 1
2 Characteristics of Discourse Markers, Criteria for
Discourse Marker Status 6
2.1 Non-propositionality and Optionality 6
2.2 Context-Dependence 8
2.3 Multifunctionality 8
2.4 Weak Clause Association and Variable Scope 9
2.5 Procedural Meaning/Non-compositionality 12
2.6 High Frequency, Orality, Stigmatization 13
3 Automatic Semantic Annotation: Testing Its Methods
and Precision 15
4 Corpus and Methodology 16
5 Findings 18
6 Conclusions, Utility of USAS as a Heuristic Tool 27
7 Alternative Perspectives on Discourse-Pragmatic
Devices: Outline of the Volume 28
References 32

v
vi      Contents

Part I Discourse Markers Across Genres

2 Discourse Markers in Natural Conversations,

Scripted Conversations and Political Interviews:
Core and Peripheral Uses 39
1 Introduction 39
1.1 Types of Reports: Direct Reports, Indirect
Reports and Voicing 40
1.2 Research Questions 42
2 Research Material 43
3 Automated Semantic Annotation and Keyness of
Reporting Verbs and Expressions 46
4 The Use of Discourse Markers in Reporting Based
on the Findings of Manual Annotation 50
5 Prototypical and Peripheral Uses of Discourse
Markers in Reporting Across Four Genres 52
5.1 Core and Peripheral Examples from
the MPI Sub-corpus 52
5.2 Core and Peripheral Examples from Scripted
Discourse 56
5.3 Core and Peripheral Examples from Celebrity
Interviews and Natural Conversations 57
6 Summary, Conclusions 59
References 61

3 Discourse Markers from a Critical Perspective:

Some Theoretical Issues 65
1 Introduction 65
1.1 Discourse Marker Research and Its Relevance
to (Critical) Discourse Theory 66
1.2 Perspectives on the Manipulative Potential
of Discourse Markers 67
2 Previous Micro-Analyses of Manipulative Intent
in Political Discourse 69
 Contents     vii

3 The Political News Interview as a Genre 71

4 Data and Methodology 72
5 Discourse Markers Marking Manipulative Intent
in Political Interviews 74
5.1 Evidential Markers: Suppression and Backgrounding 74
5.2 General Extenders: Playing Down the Importance
of Alternative Viewpoints 77
5.3 The Role of Quotation Markers
in Decontextualizing and Recontextualizing
Texts, Legitimizing Opinions and Polarizing
the Audience 79
5.4 Other Manifestations of Manipulation:
Conversationalization and the Exploitation
of Ambiguity 81
6 Conclusions, Directions for Further Research 83
References 85

4 Discourse Markers from a Critical Perspective:

A Case Study of Discourse Markers in Parliamentary
Speeches 91
1 Introduction: Populism and Populist Discursive
Strategies 91
2 Contextual Background 93
3 Research Questions, Corpus and Methodology 95
4 Characteristics of Parliamentary Speeches 96
5 Propositional Lexical Items and Indexicals Used
as Manifestations of Populist Discursive Strategies 98
6 Discourse Markers and Modal Adverbs Used as
Manifestations of Populist Discursive Strategies 102
7 Conclusions, Directions for Further Research 113
References 114
viii      Contents

Part II Discourse Markers Across Languages

5 The Use of Discourse Markers in Business English

Textbooks: Issues in L2 Communicative Competence
and Learners’ Input 119
1 Introduction 119
2 Components of Communicative Competence 120
3 The Role of Discourse Markers in Shaping Learners’
Communicative Competence 122
4 Characteristics of Discourse Markers as Sources
of Learners’ Difficulties 125
5 Mapping the Functional Spectrum of Discourse
Markers in a Corpus of Business English Textbooks 126
5.1 Research Process 126
5.2 The Functional Spectrum of Well in Naturally
Occurring Discourse 127
5.3 The Functional Spectrum of Of Course in
Naturally Occurring Discourse 129
5.4 The Functional Spectrum of Well and Of Course
in a Corpus of Business English Textbooks 130
6 Conclusion 136
References 138

6 Discourse Markers in Scripted Discourse I: Issues

of (Under)Specification in the Translation
of Reformulation Markers 141
1 Introduction 141
1.1 Contrastive Approaches to Discourse Markers 142
2 Preliminary Considerations 143
2.1 The Translation of Discourse Markers—Theory
and Practice 143
2.2 From Scripted Discourse to the Language of
Subtitles 146
3 Previous Accounts of Reformulation 148
3.1 Definitions and Lists of Reformulation Markers 148
 Contents     ix

3.2 The Functional Spectrum of I Mean 151

3.3 The Functional Spectrum of Actually 152
4 The Study 153
4.1 Data and Methodology 153
4.2 Findings 155
4.3 Discussion, Translation Strategies
and Implications 157
5 Conclusions, Limitations and Directions
for Further Research 159
References 160

7 Discourse Markers in Scripted Discourse II:

The Representation and Translation of Irish English
Stereotypes in Contemporary Cinematography 165
1 Introduction 165
1.1 The Concepts of “Stereotype” and “Stereotyping” 166
1.2 “Stereotype” and “Stereotyping” in Interactional
Sociolinguistics and Sociopragmatics 168
2 Examples of Discourse-Pragmatic Stereotyping 169
2.1 Request Strategies 169
2.2 Compliment Responses (CRs) 171
2.3 Responses to Thanks (RTs) 173
3 Stereotypes Incurred by the Use of Discourse Markers 174
4 Quantitative Perspectives on IrE Stereotypes 179
4.1 Keyness Analysis and Automated Semantic
Tagging of IrE Scripted Discourse
with Reference to AmE Scripted Dialogues 179
4.2 Keyness Analysis and Automated Semantic
Tagging of IrE Scripted Discourse
with Reference to BrE Scripted Dialogues 181
5 Conclusions, Fuzzy Boundaries Between
Pragmalinguistic and Sociopragmatic Features
as Well as Sociocultural Norms 183
References 184
x      Contents

8 Discourse Markers and Their Translation in Literary

Discourse: A Case Study of Discourse-Pragmatic
Devices in The Hobbit 187
1 Introduction 187
1.1 Literary Pragmatics—Perspectives and Approaches 188
1.2 Perspectives on Tolkien’s Linguistic Beliefs
and Style 189
2 Metapragmatic Awareness and Metacommunicative
Reflections in Tolkien’s Novels 190
3 The Functional Spectrum of Discourse Markers
in The Hobbit 195
3.1 The Use of Evidential Markers 195
3.2 Reformulation Markers—Self-Correction
and Mock Technicality 198
3.3 “… the Gold and the Dragon, and All
That”—General Extenders in The Hobbit 200
4 Quantitative Perspectives on Tolkien’s Literary
Style and Authentication Strategies 202
5 Some Issues Regarding the Translation
of Authentication Strategies 206
5.1 Strategies and Creativity in the Translation
of Discourse Markers in The Hobbit 208
5.2 The Translation of Speech Community
Creating Devices in the LOR Trilogy 212
6 Conclusions, Tolkien’s Philosophy of Language
Reconsidered 214
References 215

9 The Use of Discourse Markers in Bible Translations 219

1 Introduction, the Contribution of Discourse Marker
Research to the Study of New Testament Sentence
Conjunctions 219
1.1 Categorization and Typologies 220
1.2 Optionality and Stigmatization 222
 Contents     xi

2 Greek Sentence Conjunctions in John and Galatians

and Their Corresponding Discourse Markers
in Various Bible Translations 225
2.1 The Use of καί in Narrative and Expository Bible
Texts 226
2.2 The Functional Spectrum of δέ 228
2.3 The Functional Spectrum of γάρ 231
2.4 ἀλλά as a Global Marker of Contrast 234
3 Quantitative Perspectives on the Formal and Functional
Equivalence of New Testament Discourse Markers 235
4 Conclusions, Directions for Further Research 241
References 242

Concluding Remarks 247

Software and Online Tools 251

Data Availability 253

Appendix—List of Abbreviations 255

References 257

Index 283
 List of Figures

Chapter 2
Fig. 1 Concordance plots of Q2.1 and Q2.2 tags across
the four sub-corpora (Source Adopted from Furkó et al.
[2019: 253]) 48
Fig. 2 Presence and absence of discourse markers in reporting
across the four corpora (Source Adopted from Furkó et al.
[2019: 253]) 51

Chapter 3
Fig. 1 Types of reporting across four genres: scripted discourse
(SD), mediatized political interviews (MPI), celebrity
interviews (CI) and natural conversation (NC)
(Source Adopted from Furkó et al. [2019: 263]) 81

Chapter 8
Fig. 1 Translations of well in two Hungarian editions of The Hobbit 210
Fig. 2 Translations of of course in two Hungarian editions
of The Hobbit 210

xiii
xiv      List of Figures

Chapter 9
Fig. 1 Specification and underspecification in 150 verses of Galatians 237
Fig. 2 Specification and underspecification in 150 verses of John 237
 List of Tables

Chapter 1
Table 1 Alternative terms for English discourse-pragmatic devices 3
Table 2 Individual authors’ focus on categorial properties discourse
markers display 5
Table 3 Summary of discourse marker and non-discourse
marker-related semantic tags assigned to the most frequent
discourse marker types in the MPI and CI sub-corpora 19
Table 4 Inter-annotator agreement between automated and manual
tagging of discourse marker/non-discourse marker tokens 21

Chapter 2
Table 1 Semantic fields in USAS 46
Table 2 Normalized frequencies of the USAS categories
relevant to reporting 47

xv
xvi      List of Tables

Chapter 4
Table 1 Frequency and keyness of potentially populist uses
of lexical items and suffixes in speeches given by governing
and opposition parties in the period between 8 May 2018
and 18 September 2019 99
Table 2 Frequency and keyness of potentially populist uses
of lexical items and suffixes in speeches given
by governing and opposition parties surrounding
the immigration quota referendum of 2016 100
Table 3 Frequency and keyness of potentially manipulative
discourse markers in speeches given by members
of governing and opposition parties in the period
of 8 May 2018 and 18 September 2019 104
Table 4 Frequency and keyness of potentially manipulative
discourse markers in speeches given by members
of governing and opposition parties 105

Chapter 5
Table 1 The use and functional spectrum of well and of course
in BE textbooks adopted from Furkó and Mónos
(2013: 142–143) 132
Table 2 The use and functional spectrum of well and of course
in BE textbooks analysed in this chapter 134

Chapter 6
Table 1 RMs listed by author and year of publication 150

Chapter 7
Table 1 Keyness analysis of lexical items in the IEC and SD
sorted by test corpus and keyness 180
Table 2 Keyness analysis of lexical items in the IEC and BSD
sorted by test corpus and keyness 182
 List of Tables     xvii

Chapter 8
Table 1 Lexical items associated with authentication and
the pragmatic mode, sorted by test corpus and keyness 204
Table 2 Formal indicators of the use of the pragmatic mode
in the THC and the CSLC 205
Table 3 Frequency of USAS tags associated with textual,
interactional and subjectivity markers in the THC
and the CSLC sorted by log-likelihood 207
Table 4 Hungarian discourse markers associated with
spontaneous conversations in two Hungarian editions
of The Hobbit 211
Table 5 Pragmatic routines and their translation based
on Bayona (2003: 81) 213

Chapter 9
Table 1 Translation Equivalents of kαί (TEs in each row
are listed in order of frequency) 227
Table 2 Translation equivalents of δε (TT discourse markers
in each row are listed in order of frequency) 232
Table 3 Frequency of specification and underspecification
underspecification strategies in the various Bible
translations 236
Table 4 The frequency of and, but and for in the KJV and the ASV 238
Table 5 Frequency of USAS tags associated with textual,
interactional and subjectivity markers across the KJV
and the ASV of the New Testament sorted by log-likelihood 239
 1
Preliminary Issues: Category Membership,
Methodology, Alternative Perspectives
on Discourse Markers

1 Theoretical and Empirical Approaches

to Discourse Markers
The present volume is informed by research in a sub-field of pragmatics,
discourse marker research, often considered a “growth industry” (Fraser
1999: 931) and, at the same time, a “testing ground” (Bordería 2008:
1354) for pragmatic theories.
Discourse markers comprise a functional class of linguistic items that
do not typically change the propositional meaning of an utterance but
are essential for the organization and structuring of discourse, for mark-
ing the speaker’s attitudes to the proposition being expressed as well as for
facilitating processes of pragmatic inferences. A variety of approaches and
definitions have been offered, each one informed by a particular theo-
retical framework (Conversation Analysis, Interactional Sociolinguistics,
Rhetorical Structure Theory, Relevance Theory, etc., for an overview cf.
Fischer 2006; Furkó 2007; Vaskó 2016; Dér 2017).
Despite the rapidly growing body of discourse marker research, experts
in the field observe over and over again that there are still a number of
fundamental questions that need to be answered (cf. e.g. Schourup 1999;
Fraser 1999; Dér 2010, 2017; Heine 2013; Vaskó 2016). Some of the
© The Author(s) 2020 1
P. B. Furkó, Discourse Markers and Beyond, Postdisciplinary Studies
in Discourse, https://doi.org/10.1007/978-3-030-37763-2_1
2 P. B. Furkó

issues include the lack of generally accepted terminology and classifica-

tions and uncertainty regarding essential formal, semantic and pragmatic
characteristics.
The resulting terminological turmoil is illustrated in Table 1, which
provides a summary of some of the terms in classical studies and more
recent analyses referring to the respective discourse uses of well, of course,
oh, etc.:
The issue of terminology, alternative terms and overlapping cate-
gories such as pragmatic markers, connectives, contextualization cues and
enunciative markers will be the focus of Chapter 3 of the volume, while a
number of monographs (cf. Furkó 2007; Beeching 2016; Dér 2017) also
provide detailed discussions of the theoretical assumptions behind the
use of competing terms. Furkó (2007), in particular, takes the pragmatic
marker-discourse marker dichotomy as its major theme. Dér (2017: 9)
observes that the term discourse marker appears to be the most inclusive
and frequently used in the English literature, while its Hungarian mirror
translation “diskurzusjelölő” also appears to be the most widespread in
the respective academic community (Dér 2017: 10). Accordingly, I will
be using this as an umbrella term in the present introduction and make
the distinction between textual (prototypically discourse) functions and
(inter)subjective (prototypically pragmatic) functions prominent in sub-
sequent chapters where relevant.
The present volume will provide a series of empirical case studies on
the use of discourse markers across genres and languages as well as critical
analyses of the manipulative potential of discourse markers. Therefore,
the present introduction will focus on the second problem area men-
tioned above, i.e. the issue of categorization and category membership
as well as the methods by which one can identify lexical items that are
discourse markers and distinguish them from non-discourse marker uses
of the source categories.
Describing the characteristics of the functional class of discourse mark-
ers and developing criteria for deciding for every given instance whether
or not it is a discourse marker have been major preoccupations in recent
discourse marker research. Authors usually provide exhaustive lists of the
formal, functional and stylistic features that are associated with discourse
Table 1 Alternative terms for English discourse-pragmatic devices
well of course oh ah now I mean but you know
Schourup (1985) DP – DP DP DP DP – DP
Schiffrin (1987) DM – DM – DM DM DM DM
Erman (1987) PE – – – – PE – PE
Fraser (1990) DM/pause DM int/pause marker pause marker DM ‘90-PM DM PM
and (1999) marker* ‘99-DM
James (1974) int – int int – – – –
Wierzbicka – – int int – – – –
(1991)
Hirschberg and cue – – – cue – cue –
Litman (1993)
Stenström DM/IS* – IS – DM/IS* IS – DM/IS*
(1994)
Holmes (1995) PP PP – – – PP – PP
Kroon (1995) DP – – – – DP DP DP
Nikula (1996) PFM PFM – – – PFM – PFM
Fuller (2003) DM DM DM – – DM – DM
Beeching (2016) PM booster – – – PM – PM
Crible (2017) DM – DM int DM DM DM DM
Legend
DP—discourse particle
DM—discourse marker
PE—pragmatic expression
PM—pragmatic marker
int—interjection
cue—cue word/cue phrase
1 Preliminary Issues: Category Membership, Methodology …

IS—interactional signal
PP—pragmatic particle
PFM—pragmatic force modifier
*—categorized according to the position/slot they take in the utterance
3
4 P. B. Furkó

markers as a functional class (cf. e.g. Schourup 1999; Fraser 1999; Beech-
ing 2016; Brinton 2017), but empirical studies rely on different subsets
of such criterial features when identifying particular instances of dis-
course markers in a given corpus (for a detailed discussion, cf. Crible
2017). Naturally, this makes it difficult to compare the results of empir-
ical research even if similar datasets are involved. Table 2 illustrates this
problem.
An even more challenging task is to develop annotation software that
can automatically identify discourse markers in oral discourse and fil-
ter out non-discourse marker tokens of lexical items that are frequently
used as discourse marker types (e.g. adverbial uses of well or now, prepo-
sitional uses of like, etc.). Moreover, to date, few attempts have been
made to use automated means of identification involving semantic crite-
ria and semantic fields, since one of the very criterial features of discourse
markers is their semantic underspecification (cf. Crible et al. 2019), which
is a result of the diachronic process of semantic bleaching (cf. Brinton
2017: 31).
Accordingly, the present introduction will explore the utility of
using an automated semantic tagging software, USAS as a pre-
annotation tool for the identification of oral discourse markers, includ-
ing (inter)subjective as well as textual markers. After an overview of the
formal and functional features that can be used for manual annotation,
and after comparing the results of manual and automatic annotation of
selected discourse markers, I will argue that despite the semantic under-
specification of most discourse markers, automatic semantic annotation
(ASA) can be an effective tool for the disambiguation between discourse
marker and non-discourse marker uses with regard to certain items, but
needs to be complemented by extensive manual error correction and
filtering.
Table 2 Individual authors’ focus on categorial properties discourse markers display
seq. context oral synt. proced poly-func. attitude scope non-prop. inv.
Schiffrin (1987) x x x x (x)
Fraser (1990, 1999) x x x x x
Redeker (1990, x x (x)
1991)
Stenström (1994) x (x)
Kroon (1995) x x x
Knott and Sanders x
(1998)
Andersen (1998) x x x
Hansen (1998) x (x) (x) x x
Risselada and x x
Spooren (1998)
Romaine and Lange x (x) (x)
(1998)
Blakemore (1987, x
2002)
González (2004) x x x
Crible (2017) x x x
Legend
seq.—sequentiality-coherence-connectivity
context—context-dependence—context-coordination
oral.—orality
synt.—syntactic criteria (diversity, non-integration)
proced.—procedural meaning
poly-funct.—poly-functionality
1 Preliminary Issues: Category Membership, Methodology …

attitude—marking attitudes
scope—variable scope, functional scope
non-prop.—non-propositional content
inv.—invariable form
5
6 P. B. Furkó

2 Characteristics of Discourse Markers,

Criteria for Discourse Marker Status
2.1 Non-propositionality and Optionality

Many scholars (cf. Schourup 1999) consider non-propositionality (non-

truth-conditionality) as a sine qua non for discourse marker status, yet
others include propositional items such as then and after that. While it is
generally agreed that certain discourse markers (e.g. well, however, etc.)
contribute nothing to the truth-conditions of the proposition expressed
by an utterance, the non-truth-conditionality of others (frankly, I think)
have generated a great deal of controversy (cf. Infantidou-Trouki 1992;
Brinton 2017: 127ff ).
Blakemore (1987: 106) argues that a distinction has to be made
between truth-conditional and non-truth-conditional meaning, on the
one hand, and conceptual vs. procedural meaning, on the other. Thus,
many of the controversies stem from the fact that certain scholars confuse
the two distinctions and use them interchangeably. Schourup (1999), for
example, uses the compositionality test to argue in favour of the truth-
conditionality of in addition:

(1a) Owens is a respected drama critic. I tell you in addition that she
has written …
(1b) Owens is a respected drama critic. In addition, she has written …

While in addition is indeed truth-conditional, the above test would pre-

dict that frankly is also truth-conditional. However, as Blakemore (2002)
would argue, discourse marker uses of frankly are non-truth-conditional,
but conceptual. It is, therefore, important to point out that the com-
positionality test will be a useful tool in deciding whether individual
discourse markers have conceptual or procedural meaning, the truth-
functionality of discourse markers is tested more efficiently in terms of
whether they retain their original meaning when embedded in if-clauses
or under the scope of factive connectives such as because:
 1 Preliminary Issues: Category Membership, Methodology … 7

(2a) Allegedly / Obviously / Frankly, the cook has poisoned the soup.
(2b) If the cook has allegedly / ?obviously / *frankly poisoned the soup,
we can eat the meal without worrying.
(2c) We shouldn’t eat the soup, because the cook has
allegedly/?obviously/*frankly poisoned it.

The uncertainty with regard to whether or not obviously retains its orig-
inal meaning in (2c) suggests to many that the truth-functionality–non-
truth-functionality distinction should be viewed as a continuum, rather
than a dichotomy, which is consistent with the finding in grammatical-
ization theory that due to the diachronic grammaticalization processes
that are synchronically manifested in the use of discourse markers, there
is a fuzzy boundary between uses that are non-truth-conditional and
(omissible) and those that are not (for a detailed discussion, cf. Andersen
2001; Blakemore 2002; Dér 2017).
Optionality as a distinguishing feature is in many respects derivative of
the previously discussed criterion of non-propositionality, and discourse
markers are considered optional from the perspective of sentence mean-
ing because their absence does not change the conditions under which
the sentence is true.
There are, however, two further senses in which discourse markers are
claimed to be optional. Firstly, they may be seen as syntactically optional
in the sense that removal of a discourse marker does not alter the gram-
maticality of its host sentence. Secondly, they are optional in the sense
that if a discourse marker is omitted, the relationship it signals is still
available to the hearer, though no longer explicitly cued (cf. Schourup
1999: 231).
The above statement does not entail that discourse markers are useless;
rather, it reflects the view according to which discourse markers guide
the hearer towards a particular interpretation of the connection between
a sequence of utterances and at the same time rule out unintended inter-
pretations.
8 P. B. Furkó

2.2 Context-Dependence

Discourse markers’ extreme context-dependence is frequently identified

with their inherent indexicality. Aijmer, for example, considers indexi-
cality as the most important property of discourse markers, a property
whereby discourse markers are linked to attitudes, evaluation, types of
speakers and other features of the communicative situation (cf. Aijmer
2002: 5). In this respect, discourse markers can be compared to deic-
tics, i.e. another borderline phenomenon can be observed if we look at
some of the definitions of deictic expressions, which often overlap with
those of discourse markers. Both categories are usually defined in terms
of context-dependence, i.e. in terms of having meaning only by virtue
of an indexical connection to some aspect of the speech event (cf. e.g.
Sidnell 1998). Levinson (2004), in fact, considers discourse markers as
discourse deictics, other subgroups including spatial, temporal and social
deictics.
Similarities between indexicals and discourse markers are also recog-
nized by proponents of Relevance Theory. Carston, for example, notes
that the two seemingly disparate phenomena are brought together by
the fact that both encode a procedure rather than a concept, and both
play a role in guiding the hearer in the pragmatic inferential phase of
understanding an utterance (Carston 1998: 24). The difference between
the two sets of phenomena, according to Carston, is that indexicals con-
strain the inferential construction of explicatures and discourse markers
(discourse connectives in RT terms) constrain the derivation of implica-
tures (in other words, intended contextual assumptions and contextual
effects).

2.3 Multifunctionality

In addition to playing a role in pragmatic inferencing, individual dis-

course markers are also associated with a plethora of functions includ-
ing hedging and politeness functions. What is more, they can also be
salient in conversational exchanges as openers, turn-taking devices, hes-
itational devices, backchannels, markers of topic shift and of receipt of
 1 Preliminary Issues: Category Membership, Methodology … 9

information, and so on (cf. e.g. Beeching 2016: 4ff ). Discourse mark-

ers are inherently multifunctional and ambiguous, since there is a lot
of interpersonal and discourse burden on their signalling capacity. Dis-
course markers signal interpersonal and discourse functions simultane-
ously; thus, they are ambiguous between the two levels, and on the other
hand, they are vague with regard to signalling particular relations on a
given level as well (ibid.).
The multifunctionality of discourse markers also brings up the ques-
tion of whether different uses of a given marker are to be considered
incidental and unrelated (maximalist approach) or motivated and related
(minimalist approach) and whether there is an invariant “core mean-
ing” of discourse markers that is context-independent and preserves some
component of the lexeme’s original semantic meaning. Since the focus of
the present introduction is on differentiating between discourse marker
and non-discourse marker uses of a given item, further discussion will
not ensue on the multifunctionality of discourse marker uses. The the-
oretical issue will be picked up in Sect. 1.1 of Chapter 9 with regard to
connectives, often considered the most underspecified subgroup of dis-
course markers (cf. e.g. Crible 2018: 211). Multifunctionality will also be
an important consideration from an empirical perspective when develop-
ing coding schemes for manual annotation, as we will see in Sect. 4 of
Chapter 2 as well as most of the chapters in the volume.

2.4 Weak Clause Association and Variable Scope

It is frequently observed in the literature that discourse markers usually

occur either outside the syntactic structure or loosely attached to it (cf.
e.g. Crible 2017: 111). Quirk et al. classify many linguistic items that are
elsewhere included among discourse markers as conjuncts (e.g. nonethe-
less) which are considered to be clause elements but to have a detached
role relative to other, more closely interrelated clause elements such as
subject, complement and object:
10 P. B. Furkó

Conjuncts are more like disjuncts than adjuncts in having a relatively

detached and ‘superordinate’ role as compared with other clause elements
(Quirk et al. 1985: 631).

In addition, some of the items that Quirk et al. refer to as “disjuncts”

(e.g. obviously, sentence initial surprisingly and frankly) also display a
whole range of properties associated with the functional class of discourse
markers.
It is important to note that the property of weak clause association is
relative to elements external to the discourse marker’s lexical form, since
several discourse markers clearly have their own internal syntactic struc-
ture (e.g. on the other hand ) and others (e.g. y’know, I mean) constitute
clauses from a syntactic point of view despite the fact that they are no
longer considered to be compositional, but procedural (cf. e.g. Furkó
2014).
Weak clause association is frequently discussed in relation to phono-
logical independence: discourse markers often constitute independent
tone units or are set off from the main clause by “comma intonation”
(cf. Hansen 1997: 156).
Adding weak clause association and a corresponding lack of intonational
integration to our list of criteria could also be justified from the perspec-
tive of grammaticalization theory. An important clause of the definition
of grammaticalization states that it takes place in special morpho-syntactic
environments. In the case of discourse markers, this environment can
be associated with sentence-initial position, and hence, many scholars
regard quasi-initiality as yet another distinguishing feature of discourse
markers (cf. e.g. Schourup 1999). However, Dér (2017: 15) questions
the validity of these features on grounds of empirical evidence and distin-
guishes between different degrees of initiality. Moreover, once discourse
markers enter an advanced stage of grammaticalization, they become syn-
tactically independent and can appear at various parts of the sentence,
with an accompanying “comma intonation”, resulting in fuzzy bound-
aries and borderline cases in terms of syntactic dependence /indepen-
dence of individual tokens.
Discourse markers’ position in an utterance can also influence their
scope, which is variable, as is illustrated by (3a) and (3b):
 1 Preliminary Issues: Category Membership, Methodology … 11

(3a) Interviewer: I know how close you are to your mom. How old is
she?
Interviewee: Well, she probably doesn’t want me to say…
(3b) You’re not going to have quality if you can’t sleep and you itch
and you bitch and you weep and you cry and you bloat and you can’t
remember anything and you don’t have a, well, sex drive. (examples
taken from Furkó 2014)

As the examples above show, the size of the linguistic unit well can take
in its scope ranges from a whole sentence to a single word. Waltereit
(2006) observes that this variability is a remarkable property, but it is
not an exclusive feature of discourse markers, since conjunctions as a
word-class (and even some individual conjunctions as a lexical item) can
also have variable scope, giving the following sentences as examples:

(4a) Ed and Doris loved each other.

(4b) Ed worked at the barber’s, and Doris worked in a department
store.

In (4a), and has scope over two NPs, and in (4b), it has scope over two
clauses. However, the difference between and used as a conjunction and
its discourse marker use lies in the fact that the scope of the conjunction
and can always be determined in grammatical terms. It could be defined
as ranging over two constituents of the same type adjacent to and, which,
in turn, make up a constituent of again the same type. The scope of dis-
course markers, in contrast, cannot be determined in grammatical terms,
as is clear from (5) below:

(5) My husband got a notice t’go into the service

and we moved it up.
And my father died the week … after we got married.
And I just felt, that move was meant to be. (Schiffrin 1987: 53,
emphasis in the original)

Schiffrin (1987) concludes that and has “freedom of scope”, rather

than “variable scope”, since “we can no more use and to identify the
12 P. B. Furkó

interactional unit that is being continued than we can use and to identify
the idea that is being coordinated” (Schiffrin 1987:150).
Traugott (1995) relates the feature of variable scope to grammaticaliza-
tion and argues that in addition to nominal clines (nominal adposition >
case) and verbal clines (main verb > tense, aspect, mood marker), which
are “staples of grammaticalization theory”, a further cline: Clause inter-
nal Adverbial > Sentence Adverbial > Discourse Particle should be added
to the inventory (Traugott 1995: 1). According to Traugott, this cline
involves increased syntactic freedom and scope.
Brinton (2017: 24) further refines Traugott’s (1995) clines and adds
scope within the proposition > scope over the proposition > scope over dis-
course as a separate cline in the evolution from propositional to textual
and interpersonal meaning.

2.5 Procedural Meaning/Non-compositionality

Although most scholars treat non-compositionality as a property of dis-

course markers per se (for a detailed account, cf. Brinton 2017), Blake-
more (2002) associates discourse markers with procedural meaning and
uses non-compositionality as a test to decide whether individual items
are conceptual or procedural.
Blakemore also claims that if discourse markers are synonymous with
their non-discourse marker counterparts, they encode conceptual mean-
ing. Thus, seriously and in other words in (6a) and (7a) encode a concept
parallel to (6b) and (7b), respectively. On the other hand, well (as in 8a)
encodes a procedure, since it is not synonymous with well in (8b):

(6a) Seriously, you will have to leave.

(6b)He looked at me very seriously.
(7a)In other words, you’re banned.
(7b)She asked me to try and put it in other words.
(8a)A: What time should we leave?
B: Well, the train leaves at 11.23.
(8b)You haven’t ironed this very well.
 1 Preliminary Issues: Category Membership, Methodology … 13

A second test Blakemore uses is to see if a given item can combine

with linguistic items encoding conceptual meaning to produce complex
expressions.
As far as the question of synonymity is concerned, it is important to
note that the fact that, on the basis of native intuitions, no correspon-
dence can be found between the adverbial well and its discourse marker
counterpart, does not mean that such a relationship is absent (cf. e.g.
Furkó 2013). Native intuitions, naturally, disregard diachronic aspects of
individual lexical and grammatical items, and it is exactly these aspects
that account for the fuzziness of the category of discourse markers and
the resulting borderline cases.

2.6 High Frequency, Orality, Stigmatization

Last but not least, some of the stylistic features core members of the
functional class of discourse markers display need to be considered.
While semantic-functional properties are more important in determin-
ing class membership than formal and stylistic ones, stylistic criteria can
also be helpful in determining discourse marker status and differentiating
between discourse marker and non-discourse marker tokens.
It is important to note that high frequency of use is the backbone of
various processes of grammaticalization as well as pragmaticalization (cf.
e.g. Furkó 2014; Dér 2017). In other words, the more frequently an item
is used, the more likely it is that its formal-functional properties are going
to change, and once it has entered the process of grammaticalization, the
faster it is going to go through the substages of that process.
A number of studies on discourse markers observe that the frequency
of discourse markers can be primarily observed in speech (e.g. Beeching
2016); what is more, one of the most salient features of oral style is the
use of items such as well, right, ok and you know. For example, in their
classical study, Brown and Yule (1983: 17) label well, erm, I think, you
know, if you see what I mean, I mean, of course “prefabricated fillers”, when
drawing up a list of contrasting characteristics of spoken and written lan-
guage. They also point out that these items’ overuse is often stigmatized
by prescriptivists (ibid.).
14 P. B. Furkó

However, it is easy to illustrate the meaningfulness and distinctive (as

opposed to random) use of even the two most used discourse markers,
you know and I mean. As Fox Tree and Schrock (2002: 731) illustrate, it
matters where you know or I mean appears in an utterance and they are
not interchangeable:

(9a) Original: me and the Edinburgh girl got together after dinner
late in the evening and decided they’d really got us along to make
it look right, you know they had after all had candidates from other
universities.
Alternative: me and the Edinburgh girl got together after dinner you
know late in the evening and decided they’d really got us along to make
it look right, they had after all had candidates from other universities.
(9b) Original: but I don’t think it’s feasible. I mean I know this is the
first time I’ve done it, and I’m not in a main line paper, but I’m sure
it’ll take me all my time to do it in three weeks.
Alternative: but I don’t think it’s feasible. I know I mean this is the
first time I’ve done it, and I’m not in a main line paper, but I’m sure
it’ll take me all my time to do it in three weeks. (example taken from
Fox Tree and Schrock 2002: 731)

In (9a) Original, you know comments on what is meant by “look right”,

whereas in (9a) Alternative it comments on what “after dinner” means (in
other words, they differ in what they take within their scope, see Sect. 2.5
above). In (9b) Original, I mean comments on why the speaker says “I
don’t think it’s feasible”, without overwriting the statement, but in (9b)
Alternative, I mean comments on “I know”, retrospectively treating it as
a false start.
Moreover, as both manual and automatic annotation will illustrate,
there is no principled basis on which one could exclude from the func-
tional class of discourse markers connectives such as however, after all,
consequently and a range of other items characteristic of formal style,
some of which (e.g. besides, however, moreover ) are in fact included in
Brown and Yule’s (ibid.) list of the characteristics of written language.
The issue of overuse and stigmatization will be taken up in Sect. 3 of
Chapter 7.
 1 Preliminary Issues: Category Membership, Methodology … 15

3 Automatic Semantic Annotation:

Testing Its Methods and Precision
After the above discussion of the categorial features, discourse mark-
ers are usually associated with I cannot but agree with Crible (2017),
who argues that “any categorical definition is only useful insofar as it is
endorsed by an empirical model of identification and annotation” (2017:
99). She also provides an overview of most of the above features and
comes to the conclusion that they can be reduced to three criteria that are
helpful when identifying discourse marker types (1) and setting apart dis-
course marker tokens from non-discourse marker tokens (2&3): (1) mul-
tifunctionality, (2) syntactic (non-)integration and (3) functional scope,
respectively (Crible 2017: 105).
As for the issue of automatic annotation, there are a variety of comput-
erized semantic tagging (CST) systems, including artificial intelligence-
based, knowledge-based, corpus-based and semantic taxonomy-based
systems (for an overview, cf. e.g. Prentice 2010). By way of introduc-
ing the methodology used in the following chapters, the present analysis
draws on the results gained from the UCREL Semantic Analysis System
(USAS), which has the major advantage of combining these approaches.
Furthermore, USAS groups lexical items in terms of a taxonomy of
semantic fields and assigns semantic categories to all words, including
grammatical and other procedural (non-propositional) items, which is
relevant for the present dissertation in view of the fact that the lexical
items under scrutiny are highly procedural and semantically bleached (cf.
Sect. 2 above).
USAS system uses an automatic coding scheme of 21 semantic fields
(for details, cf. Table 1 in this Chapter), subdivided into 232 sub-
categories. For reasons of brevity, only the tags that have been asso-
ciated with the discourse marker types under analysis will be dis-
cussed in the present section, and the complete coding scheme can be
found at http://ucrel.lancs.ac.uk/usas/. USAS uses disambiguation meth-
ods including part-of-speech tagging, general likelihood ranking, multi-
word-expression extraction, domain of discourse identification and con-
textual rules (for a detailed discussion, cf. Rayson et al. 2004). Previous
16 P. B. Furkó

evaluations of the accuracy of the system reported a precision value of

91% (ibid.), i.e. a 9% margin of error applying to lexical items across
the board (including propositional and non-propositional items).
The research questions in the introductory study were as follows:

1. Are the disambiguation methods USAS uses sufficient for filtering out
non-discourse marker tokens of the most frequent discourse marker
types?
2. Does the margin of error reported to apply in general apply to the
identification of discourse markers as well?
3. Are individual discourse markers identified/tagged with a similar mar-
gin of error?
4. If individual discourse markers are tagged with varying precisions by
USAS, what formal-functional properties of the relevant discourse
markers might explain the differences?

4 Corpus and Methodology

In the course of the research, two sub-corpora of the same size (100,000
words each) were used:

• a corpus of the official transcripts of 39 confrontational type of medi-

atized political interviews (henceforth MPI sub-corpus) downsampled
from BBC’s Hard Talk and Newsnight (available at http://bbc.co.uk),
the transcripts comprise a total of 99,225 words ± 2%, allowing for
technical/transcript-specific information such as the indication of par-
ticipants’ names;
• a corpus of the official transcripts 50 celebrity interviews (henceforth
CI sub-corpus) downsampled from CNN’s Larry King Live (available
at http://www.cnn.com), each interview lasts approximately 50 min-
utes, the total word count (100,436 ± 2%) of the CNN sub-corpus
thus makes it comparable to the BBC sub-corpus.
 1 Preliminary Issues: Category Membership, Methodology … 17

To increase comparability, the downsampling technique (cf.

Khosravinik 2010) involved selecting the same time periods, similar
topics and themes of the interviews. Further genre-specific details about
the two sub-corpora will be provided in this Chapter of the volume. In
the present section, the results of automatic tagging will be compared
to findings based on manual annotation with respect to the D-values as
well as the functional distribution of individual lexical items (see Sect. 5
below).
The research process has been as follows: in order to identify and com-
pare the USAS tags of oral discourse markers in the two sub-corpora, the
semantic tags assigned to frequent discourse markers (e.g. I mean, you
know, in other words, so, well ) were considered, and then these semantic
tags were used to identify further types and tokens relevant to discourse
marking. It was found that 95.1% of the instances of discourse markers
trawled from the two sub-corpora through this method are either tagged
with Z4, described in the USAS manual as the “discourse bin” (including
items such as oh, I mean, you know, basically, obviously, right, yeah, yes) or
with A5.x, described as “evaluative terms depicting quality” (including
discourse markers such as well, OK, okay, good, right, alright ). The fre-
quency of the relevant tags across the two sub-corpora was compared,
as well as the ratio between discourse marker-relevant tags (i.e. Z4 and
A5.x) and non-discourse marker-relevant tags (e.g. B2, I1.1, T1.3, etc.,
see below for details).
In the second stage, a representative sample of 400 tokens in the MPI
sub-corpus was manually annotated using a numeric code of 1 for dis-
course marker and 2 for non-discourse marker tokens with a view to
comparing the results of automatic and manual tagging. When deciding
if an individual token is a discourse marker or not, Crible’s (2017) cri-
terial features (2017: 99, see also Sect. 2) were applied by a single expert
annotator. The tokens that were selected for the sample were weighted for
their frequency in the corpus, while discourse marker and non-discourse
marker tokens were included in equal proportions. For example, the 429
tokens of well comprise 19.6% of all automatically tagged items; thus,
78 tokens (39 A5.1-tagged and 39 non-A5.1 tagged by USAS) were
included in the sample.
18 P. B. Furkó

5 Findings
Table 3 summarizes the raw frequency of the relevant lexical items’ dis-
course marker and non-discourse marker-related USAS tags. Since both
sub-corpora were compiled in a way that they are of the same size of
100,000 words, the raw frequencies can also be compared as normalized
frequencies.
As a first step, the ratio of discourse marker and non-discourse marker
tokens of individual items was compared with the results of previous
research in the course of which discourse markers in the same sub-
corpora were manually annotated (cf. Furkó and Abuczki 2014 and
Sect. 3 in this Chapter). In order to gauge the categorial multifunction-
ality of discourse markers, the measure of D-function ratio or D-value
(a term proposed by Stenström 1990) was used. An individual item’s
D-value is calculated as a quotient of the number of tokens that ful-
fil discourse-pragmatic functions and the total number of occurrences
in a given corpus. The D-value of oh, for example, is 1 (100%) in the
London-Lund Corpus, since it is used exclusively as a discourse marker,
whereas well showed a D-value of 0.86 (86%), as 14% of its tokens serve
non-discourse marker (adverbial, nominal, etc.) functions (ibid.).
If we calculate the D-values of individual discourse markers based on
the above values and compare them to the findings of previous research,
we see that the results of automatic annotation and manual annotation
converge to a great extent. Mean, for example, has a D-value of 0.808 in
the MPI corpus based on automatic annotation (calculated as the num-
ber of Z4 tags divided by all tokens of mean, i.e. 141), while manual
annotation yielded a D-value of 0.797 (cf. Furkó and Abuczki 2014:
50). Similarly, manual annotation yielded a D-value of 0.82 for well in
the MPI corpus (Furkó and Abuczki 2014: 54), while Table 1 yields a
D-value of 0.839 for this lexical item (360 Z4 tags divided by the total
number of tokens, i.e. 429).
The table also correctly predicts that most of the lexical items under
scrutiny have higher D-values in the CI sub-corpus than in the MPI
sub-corpus, which is explained by the fact that there is a higher degree
of conversationalization in celebrity interviews, i.e. they are more similar
to spontaneous, informal, face-to-face conversations (cf. this chapter and
Table 3 Summary of discourse marker and non-discourse marker-related semantic tags assigned to the most frequent
discourse marker types in the MPI and CI sub-corpora
frequency of frequency of frequency of frequency of
DM-related tag in the DM-related tag in the non-DM-related tag non-DM-related tag
lexical item MPI CI in the MPI in the CI
well (429) 360xA5.1 312xA5.1 14xI1.1, 55xN5 1xA7, 2xB2, 24xN5
sort (38) 14xZ4 25xZ4 21xA4.1, 3xA1.1.1 10xA4.1
now (299) 4xZ4 1xZ4 288xT1.1.2, 7xZ5 229xT1.1.2, 6xZ5
(you) know (346) 205xZ4 455xZ4 140xX2.2, 1xZ6 307xX2.2
like (97) 6xZ4 17xZ4 51xZ5, 40xE2+ 238xZ5, 139xE2+
(I) mean (141) 114xZ4 201xZ4 27xQ1.1 30xQ1.1, 5xS2.2.2
(in other) words (11) 4xZ4 13xZ4 7xQ.3 7xQ.3
actually (165) 165xA5.4 72xA5.4 0 0
(I) think (549) 126xZ4 121xZ4 423xX2.1 319xX2.1
right (114) 55xZ4, 53xA5.3 211xZ4, 98xA5.3 6xT1.1.2 12xN3.8, 16xS7.4,
15xT1.1.2
1 Preliminary Issues: Category Membership, Methodology …
19
20 P. B. Furkó

Chapter 2). For example, the D-value of well is 0.92, and the D-value
of mean is 0.851 in the CI sub-corpus based on automatic annotation
(312 A5.1 tags divided by a total of 339 tokens, 201 Z4 tags divided by
a total of 236 tokens, respectively).
In the second stage of the research, a representative sample of tokens in
the MPI was manually annotated using numeric 1 for discourse marker
tokens and 2 for non-discourse marker uses. With a view to comparing
the results of automatic and manual annotation, all discourse marker-
related tags (Z4 and A5.x) yielded by USAS were re-coded as numeric
1, while non-discourse marker tags (B2, I1.1, T1.3, etc.) were re-coded
as 2. Consequently, the extracted list of the corresponding manual and
automated tags was entered into a reliability calculator (Freelon’s ReCal
2 for 2 coders) in order to calculate inter-annotator agreement statistics.
Table 4 shows the result.
Although the above inter-coder agreement values appear high (cf.
Spooren and Degand 2010), it is important to note that there is a
great degree of variation in the precision with which individual dis-
course markers are tagged by USAS. On the one hand, there are dis-
course markers such as I mean and you know whose discourse marker and
non-discourse marker uses are disambiguated with surprising precision
(resulting in a kappa score of <.98, i.e. close to perfect inter-coder agree-
ment between USAS and the human annotator), cf. (10ab) and (11ab)
below:

(10a) I_Z4[i1.2.1 mean_Z4[i1.2.2 the_Z5 long-term_T1.3+

plans_X7 + or_Z5 Britain_Z2
are_A3+ for_Z5 a_Z5 second_N4 West_M7[i3.2.1 Coast_M7[i3.2.2
mainline_M3
railway_M3c ._PUNC
(10b) Well_A5.1+ ,_PUNC yes_Z4,_PUNC I _Z8mf do_Z5
mean_Q1.1 that_Z8 ._PUNC
(11a) Getting_A9+ crime_G2.1- down_Z5 below_Z5 what_Z8 it_Z8
used_A1.5.1,_PUNC
you_Z4[i1.2.1 know_Z4[i1.2.2,_PUNC otherwise_A6.1- would_A7+
be_A3+ _PUNC
(11b) You_Z8mf know_X2.2+ the_Z5 answer_Q2.2 to_Z5 that_Z5
question_Q2.2 ._PUNC
Table 4 Inter-annotator agreement between automated and manual tagging of discourse marker/non-discourse marker
tokens
Percent N
agreement Scott’s Pi Cohen’s Kappa N agreements disagreements N cases N decisions
Variable 92.75 0.85 0.85 371 29 400 800
(DM/non-DM)
1 Preliminary Issues: Category Membership, Methodology …
21
22 P. B. Furkó

I mean in (10a) qualifies as a discourse marker on grounds of its non-

propositionality and syntactic optionality (cf. Sect. 2.1 above). In (10b),
on the other hand, I mean is not a discourse marker, as it receives an
object complement (that ) and the verbal status of mean is also under-
lined by do-support, which is used for emphasis. Similarly, (10a) contains
a discourse marker use of you know based on the criteria of syntactic inte-
gration and variable scope/discourse function. It is interesting to consider
that you know is correctly tagged despite the fact that the host utterance
does not strictly speak comprise a grammatical sentence, i.e. it does not
have a straightforward syntactic structure that can be parsed and PoS-
tagged in its original formulation. The utterance contains a false start
(below what it used ), a reformulation marker (the discourse marker you
know) and a reformulated segment (otherwise would be). The intended
utterance Getting crime down below what it otherwise would be is only
recoverable if the word used is not considered in the course of syntactic
and semantic parsing. You know in You know the answer to that question
is correctly tagged as a syntactically integrated, i.e. non-discourse marker
token.
The disambiguating precision with respect to you know and I mean
is probably due to two of the disambiguation methods USAS applies:
firstly, its multi-word-expression extraction algorithm and its core com-
ponent of MWE lexicon (cf. Rayson et al. 2004) and secondly, the fact
that POS tagging enables the parser to differentiate between syntactically
integrated tokens that are monotransitive (and are thus followed by their
nominal or clausal complements) and syntactically non-integrated ones
that are marked by the absence of complements.
However, even in the case of multi-word units, we can observe a
bias towards discourse marker-related tags, i.e. the discourse marker use
is overgeneralized and is applied to occurrences of I mean, I say and
you know that are syntactically integrated and have the valency pat-
tern/complement options of other monotransitive verbs, cf. (12) and
(13):

(12) I_Z8mf want_X7+ to_Z5 assure_A7+ you_Z8mf that_Z8

I _Z4[i1.2.1 mean_Z4[i1.2.2
 1 Preliminary Issues: Category Membership, Methodology … 23

what_Z8 I _Z4[i2.2.1 say _Z4[i2.2.2 when_Z5 I_Z8mf tell_Q2.2

you_Z8mf I_Z8mf
appreciate_E2+ your_Z8 contributions_A9- ._PUNC
(13) As_Z5 you_Z4[i1.2.1 know_Z4[i1.2.2,_PUNC the_Z5 Govern-
ment_G1.1c says_Q2.1
it_Z8 ’s_A3+ too_N5.2+ early_T4+ to_Z5 tell_Q2.2 about_Z5
that_Z8 ._PUNC

In (13), you know is not a discourse marker on the basis of the criterion
of compositionality, i.e. by virtue of the fact that it comprises part of
the phrase as you know. As you know, in its entirety, could be considered
a discourse marker, but in that case, the appropriate annotation would
be “as_Z4[i1.3.1 you_Z4[i1.3.2 know_Z4[i1.3.3” rather than “as_Z5
you_Z4[i1.2.1 know_Z4[i1.2.2”, where i1.3.1 marks the first segment in
an idiom comprising three lexical items, i1.2.2 tags the second segment
in an idiom that consists of two lexical items, etc.
On the other hand, there are lexical items that are invariably
tagged with the same (sometimes discourse marker-relevant and other
times non-discourse marker relevant) tags regardless of their syntactic
(non-)integration and functional scope.
An example for discourse marker-relevant invariant tagging is actually,
which might be used as a discourse marker that has the ensuing discourse
unit in its scope (14a) or as an adverbial modifier that has scope over the
verb it modifies as in 14b below (all extracts are from the USAS-tagged
CI corpus, emphases are mine):

(14a) No_Z4,_PUNC that_Z8 was_A3+ n’t_Z6 exactly_A4.2 +

the_Z5 reason_A2.2 ._PUNC Actually_A5.4 + ,_PUNC what_Z8
it_Z8 was_A3+ ,_PUNC is_Z5 I_Z8mf felt_X2.1 that_Z5 films_Q4.3
were_Z5 getting_A9+ they_Z8mfn started_T2+ to_Z5 be_Z5 repeat-
ing_N6+ ._PUNC
(14b) They_Z8mfn ‘re_A3+ one_T3 of_Z5 the_Z5 few_N5-
cats_L2 mfn in_Z5 the_Z5
world_W1 that_Z8 can_A7+ actually_A5.4 + swim_M4
under_M4[i619.2.1
water_M4[i619.2.2
24 P. B. Furkó

An example for non-discourse marker-relevant invariant tagging is now,

which can be used as a discourse marker that marks topic shift (15a)
or as a circumstance adverb (15b). However, USAS does not distinguish
between discourse marker and non-discourse marker uses of now, and
both are labelled as T1.1.2, i.e. as “general terms relating to a present
period/point in time”:

(15a) Good_Z4[i297.2.1 heavens_Z4[i297.2.2,_PUNC such_Z5

an_Z5 intelligent_X9.1+ man_S2.2 m is_Z5 excited_X5.2 +
about_Z5 a_Z5 movie_Q4.3 star_W1 ?_PUNC Now_T1.1.2
what_Z8 about_Z5 her_Z8f and_Z5 the_Z5 Kennedy_Z1mf ’s_Z5
?
(15b) Somebody_Z8mfc explain_Q2.2/A7 + to_Z5 Paris_Z2 and_Z5
Nicole_Z1f,_PUNC live_L1+ means_X4.2 we_Z8 ’re_A3+ on_Z5
television_Q4.3 right_T1.1.2[i7.2.1 now_T1.1.2 [i7.2.2 ._PUNC

Another discourse marker that displays “high frequency and versatility”

(Andersen 2001: 221) is like, whose functional spectrum ranges from
(16a) marking explanation/exemplification, (b) emphasis/new informa-
tion, to (c) quotative, (d) approximative and (e) softening functions. The
list of functions is far from exhaustive, for a more detailed functional
spectrum of like cf. e.g. Beeching (2016: 126ff ).

(16)
(a) My roommate never cleans when I ask him to. Like, I asked him
yesterday to clean, and he never did it. (Like_E2 + ,_PUNC I_Z8mf
asked_Q2.2 him_Z8m yesterday_T1.1.1 to_Z5 clean_B4,_PUNC
and_Z5 he_Z8m never_T1/Z6 did_A1.1.1 it_Z8 ._PUNC)
(b) This guy is so cool. I mean, he’s like the coolest person you could
meet. (I_Z4[i1.2.1 mean_Z4[i1.2.2,_PUNC he_Z8m s_T1.3 like_Z5
the_Z5 coolest_O4.6-person_S2mfc you_Z8mf could_A7+ meet_S3.1
._PUNC)
(c) I went to the clerk to ask him where the beer was, and he’s like,
‘I don’t know, I’m new here’, so I’m like, yeah, sure, like, you should
know this, man! (so_Z5 Im_Z99 like_Z5,_PUNC yeah_Z4,_PUNC
sure_A7+ ,_PUNC like_Z4,_PUNC
 1 Preliminary Issues: Category Membership, Methodology … 25

you_Z8mf should_S6+ know_X2.2+ this_Z8,_PUNC man_S2.2m)

(d) I missed like 40 questions on the exam. (I_Z8mf missed_A5.3-
like_Z5 40_N1 questions_Q2.2 on_Z5 the_Z5 exam_P1 ._PUNC)
(e) Could you, like, loan me $100? (Could_A7+ you_Z8mf,_PUNC
like_Z4,_PUNC loan_A9- me_Z8mf $100_Z99 ?_PUNC)

It is interesting to observe on the basis of the above examples that USAS

tags like as a discourse marker (i.e. Z4) in only one of its five main func-
tions: in (16a), it is incorrectly marked as “emotional state of liking”,
and in 16b, c and d, the tag Z5 indicates a fuzzy boundary between
the prepositional use (“grammatical bin”) and the discourse marker use
(“discourse bin”) of like: while the criteria of syntactic optionality and
non-propositionality would suggest that such uses are marked as Z4, the
grammatical interpretation (Z5) is also plausible in (b) and (d), where
no punctuation marks are used before and after like.
Yet another discourse marker that USAS has a hard time identifying
is the connective use of so. Syntactically integrated tokens receive either
A13.3 tags (“degree: boosters”), N5 tags (“quantities”) as in (17), or Z5
tags (“grammatical bin”), as in (18):

(17) but_Z5 they_Z8mfn have_Z5 wasted_N5.2+ so_A13.3

much_N5+ time_T1 and_Z5
so_N5 [i1.2.1 many_N5[i1.2.2 businesses_I2.1 have_Z5 gone_M1
bust_B1 as_Z5 a_Z5
result_A2.2
(18) But_Z5 why_A2.2 I_Z8mf feel_X2.1 so_Z5 frustrated_S8-
about_Z5 it_Z8 is_Z5
we_Z8 talked_Q2.1 about_Z5 this_Z8 at_Z5 the_Z5 start_T2 +
of_Z5 November_T1.3
._PUNC

The first token of so is most likely categorized as a subjective booster

(A13.3) on the basis of its co-occurrence with the word wasted, which
is associated with emotion (frustration), while so <A13.3> could also
be associated with quantification, similarly to the second token in the
26 P. B. Furkó

same utterance (tagged as N5). The indeterminacy between quantita-

tive and boosting functions of so reveals a fuzzy boundary between pri-
marily propositional (objective) and primarily evaluative/subjective uses
of the lexical items under scrutiny. Naturally, in the course of manual
annotation, the human annotator can have recourse to intonation and
other paralinguistic features of the utterance; however, it is still hard
to draw a boundary between clear-cut cases of non-boosting, quantita-
tive and, conversely, boosting, non-quantitative uses of so as the above
examples also illustrate. Moreover, it can be observed that in (18), so
also co-occurs with words associated with emotions (“feel” and “frustrat-
ed”, respectively), yet it is tagged as a purely grammatical item (Z5, i.e.
“grammatical bin”), rather than a booster.
Disambiguation between grammatical items and discourse markers
can also be an issue in utterances where so has a straightforward con-
nective function, connecting its host unit to the previous utterance and
marking a conclusive, resultative relation between the two discourse seg-
ments, as in (19):

(19) So_Z5 would_A7+ you_Z8mf bail_S8 +[i1.2.1 them_Z8mfn

out_S8+ [i1.2.2 in_Z5
the_Z5 short-term_T1.3- to_Z5 keep_A9+ those_Z5 jobs_I3.1
?_PUNC

Once again, USAS does not make a distinction between such uses and
grammatical uses exemplified by (18) above, marking so as Z5 (gram-
matical bin), rather than Z4 (discourse bin), the latter of which would
be more appropriate based on its syntactic detachment and connective
function.
Well, on the other hand, is a non-multi-word lexical unit whose dis-
course marker and non-discourse marker uses are clearly distinguishable
based on both manual and automated annotation, as (20) and (21) illus-
trate:

(20) But_Z5 that_Z8 goes_E2+[i1.2.1 for_E2+[i1.2.2 British_Z2

and_Z5 American_Z2
 1 Preliminary Issues: Category Membership, Methodology … 27

nuclear_Y1/G3[i2.2.1 weapons_Y1/G3[i2.2.2 as_N5++[i3.2.1

well_N5 ++[i3.2.2
._PUNC
(21) No-one_Z6/Z8cmf asked_Q2.2 them_Z8mfn,_PUNC “_PUNC
Well_A5.1 + what_Z8 does_Z5 that_Z8 mean_Q1.1 with_Z5
the_Z5 things_O2 you_Z8mf ‘re_Z5 not_Z6 going_T1.1.3[i1.2.1
to_T1.1.3[i1.2.2 go_T2++[i2.2.1 ahead_T2++[i2.2.2 with_Z5
?_PUNC “_PUNC

As the examples show, compositional/propositional uses are tagged

as N5 (quantities), especially in the phrase as well, while discourse
marker/interpersonal uses are marked as A5.1 (evaluation).

6 Conclusions, Utility of USAS

as a Heuristic Tool
In the above introduction, I argued that discourse markers are notori-
ously difficult to identify for humans and computers alike, and there are
several borderline phenomena, fuzzy boundaries and cases of ambiguity
resulting from discourse markers’ inherent, criterial features. In answer
to the research questions posed in Sect. 3 above, it can be observed that
the disambiguation methods automatic annotation uses are efficient for
filtering out non-discourse marker tokens of the most frequent discourse
marker types: thus, automatized annotation enables the researcher to
obtain an adequate global picture of the D-values of most of the lexi-
cal items that are frequently used as discourse marker types.
It was also demonstrated that the margin of error reported to apply in
general also applies to the identification of discourse markers collectively
and, in the case of multi-word units such as you know and I mean, indi-
vidually as well. However, we find a great degree of variation in the pre-
cision/margin of error with which non-multi-word discourse markers are
tagged. Such varying precisions are mostly due to discourse markers’ cri-
terial features of source category layering, syntactic non-integration, vari-
able/functional scope, all of which challenge the disambiguation meth-
ods USAS applies, with special reference to part-of-speech tagging, gen-
eral likelihood ranking and multi-word-expression extraction.
28 P. B. Furkó

7 Alternative Perspectives
on Discourse-Pragmatic Devices: Outline
of the Volume
While discourse markers will continue to puzzle humans and computers
alike, we can safely say that automatized methods can open new vistas
of research into the study of discourse markers, i.e. lexical items where
drawing a distinction between propositional and non-propositional, syn-
tactically semantically integrated and discourse-pragmatic uses is espe-
cially relevant. Moreover, discourse markers cannot be studied in isola-
tion, but with reference to other discourse-pragmatic devices (quotatives
in Chapters 2, 3, 8 and 9, contextualization cues in Chapter 3, modal
particles in Chapters 4 and 8 and pragmatic routines in Chapters 7 and
8), while the analysis of non-propositional items needs to be comple-
mented by the study of the contribution of propositional lexical items
that belong to related semantic fields (cf. Chapters 7–9). Secondly, only
a mixed methodology that involves computerized means as well as the
human element, a quantitative analysis of large datasets as well as careful
qualitative analysis of individual examples needs to be undertaken when
analysing genres/sub-genres (Chapter 2) or when reflecting critically on
discourses (Chapters 3 and 4), input for SLA (Chapter 5) as well as the
adequacy of translated texts (Chapters 6–9).
I hope to further illustrate these points in the ensuing chapters with
regard to genre analysis (Chapter 2) and other applied linguistic fields
such as Critical Discourse Analysis (Chapters 3 and 4), second language
acquisition (Chapter 5), translation theory (Chapters 6–9), the analysis
of scripted discourse in contemporary cinematography (Chapters 6 and
7), literary discourse (Chapter 8), as well as Bible studies (Chapter 9).
In the individual chapters, I use my previous research as a starting
point and reconsider previous findings from new methodological per-
spectives as well as new datasets, in the course of which the following
principles have been applied:
Other documents randomly have
different content
 La povertà ci affanna, e la ricchezza
Ci fa odiosi, superbi ed ignoranti:
L'amore ci rïempie di tristezza;
L'ira e lo sdegno ci turba i sembianti:
Un mar turbato sembra giovinezza,
Pieno di rotte sarte e legni infranti:
È la vecchiezza languida e da poco,
E la virilità dura pur poco.

In somma in ogni tempo, in ogni stato

Non ha mai requie e non ha mai conforto:
E quegli al parer mio solo è beato,
Che nato appena, o poco dopo è morto.
Perchè, sebben c'è qualche fortunato
Il cui naviglio già si trova in porto,
Pure in guardando le miserie altrui,
Moveransi a pietà gli affetti sui.

Perchè, siccome le diverse corde

D'uno strumento, se son ben temprate,
Fanno un suono dolcissimo e concorde,
In cotal guisa le genti create
Convien fra loro che natura accorde;
Onde non ponno l'une esser toccate,
Che non rispondan l'altre: e di qua viene
Che abbiam tanto dolor dell'altrui pene.

5
 Che se non fosse questa gran catena,
E si vivesse come querce o abeti
Fissi ad ogn' or su la paterna arena;
Siccome a quei non duol che spezzi e inquieti
La scure l'altre piante, e non han pena;
Così staremmo noi contenti e lieti
Su le miserie di questo e di quello:
Ma natura ci diè senso e cervello;

E ci diede per quello gentilezza,

E per quest'altro senno e intelligenza:
Onde per l'una il male altrui s'apprezza,
E fassi nostra ancor la sua doglienza;
E per l'altro s'accresce l'amarezza:
Chè (come dice il Savio in sua sentenza)
Quei che aggiunge sapere, aggiunge affanno,
E men si dolgon quelli che men sanno.

E oh quanto volentieri or mi porrei

In cotal truppa! e viverei più lieto,
E tra me stesso non maledirei
Il dì ch'io presi in mano l'alfabeto,
Onde a leggere appresi, e m'abbattei
In quel racconto, in quel crudel decreto,
Che, come dissi, per sua dura sorte
Condannava Despina a fiera morte.

8
 Fatto ella dunque con la man di neve
Segno a ognun che tacesse, diede in pria
Un ardente sospiro, e quei fu breve;
Poi disse ad alta voce: Io non son mia,
Nè di quel d'altri disponer si deve
Senza permissïon da chi che sia.
A Ricciardo donai me stessa e il core;
Ond'egli è solo il dolce mio signore.

Ed ho sì gran piacer di questo dono,

Che niun tempo verrà ch'io me ne penta;
E se ben tanto presso a morte io sono,
Che già mi vedo trucidata e spenta,
Odio la vita, e pongo in abbandono
Quanto oggi qui da te mi si presenta,
Principe ingiusto, che discioglier brami
Questi dell'amor mio sacri legami.

10

Serpedonte a quel dir, come mastino

Che veduto abbia la nemica fera,
Con l'aspra mano il collo alabastrino
Le serra, e vuol che onninamente pera.
Ma tante strida il popol Saracino
Diè, che interruppe quell'opera nera;
E colmo d'ira in verso lui si volse,
E in guisa tale la sua lingua sciolse:

11
 Se voi sapeste quale alberga in questa
Donna, anzi furia del tartareo chiostro,
Alma crudele ed agl'inganni presta,
Risparmiato avereste il pianto vostro,
Nè la sua morte vi saría molesta:
Ma voi le bianche perle ed il vivo ostro
Di lei mirando, e i suoi begli occhi neri,
Più là non penetrate coi pensieri.

12

Questa adescommi, un lustro è già compiuto,

Nell'amor suo in maniera sì strana,
Ch'io n'ero morto, e ancor ne son perduto;
Ed al principio mi comparve umana;
Poi di me fece un barbaro rifiuto,
E si fuggì, resa d'amore insana,
Con uno, alla cui morte ella col padre
In Francia andò con tante armate squadre.

13

Ma non rende ragione a' suoi vassalli

Di quel ch'egli opra un supremo signore:
E perchè lieve pena a tanti falli
È presta scure e subito dolore,
Di lunga morte i tormentosi calli
Voglio che prema in un perpetuo orrore.
E qui rivolto alla donzella il viso,
Guardolla con disprezzo e con sorriso:

14
 Ed ordin diede a quattro cavalieri
Che la guardasser dentro d'una tenda
Insino a tanto che de' suoi pensieri
Tutta la somma il fabbro non comprenda,
Che formar deve il misero quartieri
Della donzella, anzi la tomba orrenda:
E perchè questa presto sia finita,
I lavoranti a molto prezzo invita.

15

Nell'isoletta, se ve ne sovviene,
Dove le regie tende egli fa porre,
Vuol che si formi il loco delle pene.
Onde la gente tutta colà corre,
E fan gran fosso nelle asciutte arene:
Nè in questo mentre alcun viene o soccorre
L'innocente fanciulla; e intanto bolle
L'opra, e sul fosso un gran tempio s'estolle.

16

A guisa del famoso Panteonne

Formato sembra; e v'è di più, che attorno
Ci son di nero porfido colonne;
Di neri marmi ancora è tutto adorno
L'infausto tempio: e di abbrunate donne
Un drappel vuol che dentro al suo contorno
Abiti; e questo quasi ogni momento
Mandi fuori un mestissimo lamento.

17
 Poi fa dipinger sopra d'ampie tele
Tutti i casi di donne sventurate,
Ch'ebbero il cor superbo, o pur crudele;
E di queste le mura sono ornate
Della gran volta: e di nere candele
Vuol che arda in esso tanta quantitate,
Che a lui, che il giorno splendido ne adduce,
Soprastar possa la racchiusa luce.

18

Quindi in mezzo del tempio erge un avello

D'un bel dïaspro che ha la porta d'oro;
E d'oro ha pure il grosso chiavistello,
Per cui dal cieco sotterraneo foro
Vassi al carcere iniquo, orrido e fello,
Dove Despina per suo reo martoro
Deve condursi a terminar sua vita.
Ed oh che l'opra omai è già finita!

19

Finita l'opra, d'un gran manto nero

Fanno vestir la povera Despina;
E ogni altra donna, ogni altro cavaliero
Si veste a bruno per quella mattina:
E verso il loco dispietato e fiero
Tacita e pensierosa ella cammina:
Entra nel tempio, e Serpedonte è seco,
Che la riguarda minaccioso e bieco.

20
 Apre un soldato la dorata porta,
E, Qua, le dice, misera fanciulla,
Entrar convienti e rimanerci morta.
Essa lo guarda, e non risponde nulla:
Quand'ecco il vecchio rege che l'esorta
A non passar sì presto dalla culla
A tomba sì crudele e spaventosa,
E ch'esser voglia a Serpedonte sposa.

21

Le dame e i cavalieri a mille a mille

Le son d'intorno, e le stesse preghiere
Le fanno: ed ella in sembianze tranquille
Lor si dimostra, e quelle lusinghiere
Voci non cura; ma con le pupille,
Di cui natura non fe' le più nere,
Si fissa in Serpedonte, e immantinenti
Tali gli vibra al cor detti pungenti:

22

Eccomi giunta alla soglia fatale,

Donde si varca al regno della Morte.
Questo è l'ospizio, uomo micidiale,
Questo è il palazzo e la superba corte
Ove tu alloggi una donna reale?
Or vanne pure, e vantati di forte;
E la fama di te dica, ovunque erri,
Come vive le femmine sotterri:

23
 E le sotterri, perchè troppo fide
Sono agli sposi loro, a' lor mariti.
Africa sola e le spiagge Numide,
E più d'ogni altro della Nubia i liti
Veggon tai cose: altrove sol si uccide
Chi fede rompe per minacce o inviti,
O per forza d'amore al suo consorte;
E qui sol chi è fedel si danna a morte.

24

Crudel, se data t'avess'io parola

D'esser tua sposa, e t'avessi mancato,
Ben mi starebbe addolorata e sola
Viver morendo in loco tanto ingrato;
Nè mi dorrebbe vedermi alla gola
Pungente ferro, o il petto mio piagato;
Chè merita abbreviare i giorni sui
Chi tradisce il suo sposo, e dassi altrui.

25

Ma a voi, donne di Nubia e cavalieri,

I Genj di queste orride contrade,
E su del cielo e degli abissi neri,
E i Numi ancor che le marine strade
Scorrendo vanno placidi e leggieri,
E i gran Numi di fede e di onestade
Parlino a mia difesa; e chiara sia
La sua calunnia e l'innocenza mia.

26
 Nè gran tempo anderà ch'aspra vendetta
Faran di me più spade peregrine:
E forse forse l'amor mio s'affretta
Per ritrovarmi su l'onde marine.
Deh, se prego mortale in ciel s'accetta
Da quelle immense potestà divine,
Fate, gran Dii, che in questa tomba io viva,
Sino a che il mio Ricciardo non arriva;

27

E non ti tragga, traditor, dal petto

L'indegno core, e dica a me: Tel dono.
Quel poi guardando entrambi con diletto,
Diremo entrambi ancor: Quivi ebbe il trono
L'amor da prima, e poi l'ira e il dispetto
Contro una che lasciata in abbandono
Era da tutti; e questo uomo sì forte
La racchiuse tra barbare ritorte.

28

Nè ti allegrar con la vana speranza

Che una lagrima sola, un sol sospiro,
Un pallor breve su la mia sembianza
Abbi a vedere in tanto mio martiro.
Al par di tua ferocia avrò costanza:
E s'egli è ver che, terminato il giro
Di questa vita, ogni anima disciolta
Si trovi con chi ell'ama un'altra volta;

29
 Qual sarà il mio piacere e il mio conforto
Nel ritrovarmi col mio Ricciardetto?
Qual gioja trarrem noi da questo torto,
Da questo sdegno e questo tuo dispetto?
Io lui dirò come in crudele e corto
Carcer fui spenta per l'estremo affetto
Ch'io volli conservargli; e più gradita
Mi fu santa onestà, che lunga vita.

30

Questa sola speranza ella è bastante

A farmi lieta in compagnia di Morte.
Ma tu nulla rispondi, e nel sembiante
Ti cangi, e tieni le tue luci smorte?
Forse ti duol che alla tua gente avante
Spalancate del vero abbia le porte,
Onde veggano a qual tristo signore
Debbano soggettar la roba e il core?

31

Povera Nubia e misere pendici,

Che aspettar vi potete da costui?
Se me distrugge, farà voi felici?
Me che tanto d'amore accesi in lui?
E se chi ama, tratta da nimici
Dannando a morte in luoghi acerbi e bui,
Di color che avverrà ch'egli non cura,
Se non la stessa sorte, e ancor più dura?

32
 Però, s'io mal non veggo, il più beato
Sotto costui è quel che muorsi presto.
Misero certo e doloroso stato
Ad un cor vile che non pensi al resto;
Ma felice, soave e fortunato
A chi il futuro è tutto manifesto,
E che legge ne' fati e nelle stelle
Il gran tragitto alle cose più belle.

33

Però, donne amorose e cavalieri,

Non vi prenda pietà del morir mio.
Ch'oltre ch'io muojo tanto volentieri,
Ch'altro non ho che di morir desìo,
Ho gran piacer che questi si disperi
In non avermi, e sì ne paghi il fio;
E mi diletta più d'ogni altra cosa,
Ch'io muojo onesta, e di Ricciardo sposa.

34

Volea più dir; ma generosa e forte

Varcò la soglia, e con l'eburnea mano
A sè tirò le spaventose porte,
E si riacchiuse nell'oscuro vano,
U' nera face con fiammelle smorte,
Che la luce movea poco lontano,
Le fe' vedere il tenebroso avello,
Più crudo assai di qualunque coltello.

35
 Chiusa Despina, si fece un gran pianto
Dalle abbrunate femmine pietose;
E Serpedonte infurïato intanto
A custodia del tempio mille pose
Uomini d'armi, che famoso vanto
S'acquistaro per opre glorïose:
A guardia poi della tomba spietata
Egli si pone, ed altri non la guata.

36

E vuol, chïunque nel tempio penètra,

Despina rea, e lui giusto confessi;
E chi ciò nega, fa scrivere in pietra,
O che coi mille alla pugna s'appressi;
O se pur grazia dalle stelle impetra,
Essendo ei sol, che quei restino oppressi,
Debba seco pugnar, del cui valore
Libia avvezza ai spaventi n'ha terrore:

37

E chi vinto rimane (odi che furia,

Odi che mostro orribile e spietato!)
Vuol che di tutto patendo penuria,
Sia vivo per tre giorni riserbato;
Poi con affanno e con estrema ingiuria
Sopra l'avello rimanga scannato;
E fuor venga Despina in quei momenti,
Acciò vegga il suo sangue, oda i lamenti.

38
 Ciò decretato, alle femmine impera
Che attorno attorno all'avello funesto
Facciano un tristo canto in su la sera,
Perchè il carcere a lei sia più molesto.
Onde due giovinette in veste nera
Andaro avanti, e in tuon lugubre e mesto
Il canto principiaro; e l'altre appresso
Piangendo ripetevano lo stesso.

39

O verginella, dove mai ti trovi

Separata da' vivi in una oscura
Tomba, ove morte ancor viva tu provi?
Quando nascesti, ogni mala ventura
Teco pur nacque. A pietà mi commovi:
Ma se non eri al signor nostro dura,
Avresti regno e vita lieta e bella.
E il coro rispondeva: O verginella!

40

E quindi in tuono più roco e languente

Seguìano: O d'Amatunta, o di Citera
Leggiadra Dea, che fai bella e ridente
Del terzo cielo la feconda sfera,
Piega la dura ed ostinata mente
Di questa verginella aspra e severa,
Acciò di sè le incresca, e si rivolga
Al nuovo amore, e dal primo si sciolga.

41
 Ma non tardar, se sei così pietosa,
Come fama di te fra noi favella;
Chè dentro all'atra tomba e spaventosa
Potrà poco durar la vergin bella.
Dunque impera alla tua prole famosa,
Che armata di acutissime quadrella
Nel carcere penétri, e il cor le spezzi
Per Serpedonte, e Ricciardetto sprezzi.

42

E mentre quelle cantavan di fuore,

Dalla profonda tomba a lor risponde
Despina, e dice: Del vostro dolore,
Donne, ho pietà; ma pria di sasso l'onde
Del mar faransi, e sentiranno ardore,
O nere si faran le chiome bionde
Del sempre chiaro apportator del giorno,
Ch'io faccia all'amor mio oltraggio e scorno.

43

In questo dir, di guerra aspra nascenza

S'ode fra i mille; onde spezzano il canto
Le meste donne vinte da temenza,
E del gran tempio s'ascondon 'n un canto.
Un guerriero di forza e di potenza
Combatte; e questi è il Cavalier del Pianto,
Il padre della giovine racchiusa,
Che d'uomo ingiusto Serpedonte accusa.

44
 Errò tanto costui per aspri e vari
Luoghi, che giunse a quell'orribil porto,
Dove udì della figlia i casi amari,
E n'ebbe per dolore a restar morto:
E se ben sa che con mille contrari
Vincer non puote e vendicar suo torto,
Pur ama meglio una morte spedita,
Che senza lei più mantenersi in vita.

45

Quindi è che disperato egli si caccia

In mezzo a loro, e col brando tagliente
A questi il collo, a quei tronca le braccia.
Ma or più non è quello Scricca valente,
Ch'allora ei fu che su la fresca faccia
La nera barba ruvida e pungente
Segno faceva e mostra di vigore;
Or ella è bianca, ed egli ha men valore:

46

Ond'è che vinto e prigioniero ei resta,

Ed è condotto al fero Serpedonte;
E l'elmo duro trattogli di testa,
Conobbe ei tosto la real sua fronte,
Che gli era per lungo uso manifesta;
E con parole dispettose e pronte
Gli dice: Gran mercè debbo agli Dei,
Se in questo giorno mio prigion tu sei;

47
 Chè già la legge ed il fatal decreto
Saper ben dêi del tuo prossimo fine.
Ma s'esser tu vorrai uomo discreto,
Questa sventura tua giunta al confine
Non sol farai ch'ella ritorni indreto,
Ma rose diverran tutte le spine
Che or pungono il cor tuo, e quello ancora
Di tua figlia che tanto ti addolora.

48

Io t'aprirò la porta dell'avello,

E tu discendi seco a parlamento;
E se addolcisci lo suo cor rubello
Per me, cangerò teco anch'io talento:
Sarò suo sposo, e non sarò più quello
Che or sono, ad ambo voi tutto spavento;
E queste squadre e il braccio mio saranno
In avvenir de' tuoi nemici in danno.

49

Nè, gran rege de' Cafri, io ti dimando

Ingiusta cosa. Anzi, se t'enno a core
I patrj Dei, a' quali io raccomando
Me stesso e l'opra e il lor macchiato onore.
Dovresti far con paterno comando
Ch'ella spegnesse il mal acceso ardore:
Chè donna saracina ad uom cristiano
Non deve unirsi, o il matrimonio è vano.

50
 E qui raccontò lui di Ricciardetto
E di Despina gli teneri amori;
E come egli rapilla per affetto;
E gli sdegni di lei, l'ire e i furori
Contro di lui per quel suo giovinetto.
S'empie lo Scricca tutto di stupori
A quelle voci, e fassi aprir la porta
Dell'urna, ed alla figlia egli si porta.

51

Ma ritorniamo un poco, se vi piace,

Al nostro Carlo, e partiam da Despina,
Or che col padre suo in santa pace
Si trova dentro a quella sua cantina.
Ma duolmi che ammalato Carlo giace,
Ed ha presa la terza medicina,
E gli han cavato sangue, e messi gli hanno
I vescicanti che gran duol gli fanno.

52

E già s'era ridotto a mal partito,

Quando San Dïonigi di persona
Gli apparve, ed era di bianco vestito,
E disse: Carlo Magno, nuova buona:
Il moccolino tuo non è finito.
Ciò detto, disparisce e l'abbandona.
Carlo s'alza sul letto, per far prova
S'egli è guarito, e sano si ritrova.

53
 Di che si rallegrò tanto Parigi,
Che quasi se ne andò tutto in baldore;
E allor fu fabbricato a San Dionigi
Quell'ampio tempio e di tanto valore,
Di cui ancor si veggono i vestigi,
E di cui Francia non vide il maggiore:
E questa grazia ciaschedun più prezza,
Perch'era presso all'ultima vecchiezza.

54

E mentre si fan feste da per tutto,

Ecco che a mezzodì giunge un corriero
D'Alfonso il casto con vestito a lutto,
Che vien di Spagna, e dice come il nero
Popol di Libia ha il suo signor distrutto;
Onde ha sua speme nel francesco impero;
E prega Carlo con sospiri e pianti,
Che a lui voglia mandar cavalli e fanti:

55

Ma che non ponga punto tempo in mezzo;

Chè qual torrente che rotte ha le sponde,
Va l'Africano a fiere stragi avvezzo
Per le ispane contrade, ove confonde
L'umane e sacre cose, e con disprezzo
Insulta tutti, e niuno a lui risponde:
Cotanto de' Spagnuoli è lo spavento,
Che dieci Mori ne disfanno cento.

56
 Nè tacque i santi letti maritali,
Nè le sacrate a Dio vergini pure,
Fatte trastullo di quegli animali.
Onde mosso a pietà di lor sventure,
Rispose Carlo, che d'aquila l'ali
Avrìa voluto in quelle congiunture,
Per ritrovarsi vie più presto in Spagna,
E dar principio a una crudel campagna.

57

Ma che non averìa troppo indugiato

A mandarvi soccorso e venirvi esso.
E corrieri spedì per ogni lato,
E diede lor comandamento espresso
Di ricercare Orlando suo pregiato,
E il buon Rinaldo che gli andava appresso,
E quale altro trovasser nel cammino
Famoso in armi e chiaro paladino.

58

E volle la fortuna dei Spagnuoli,

Ch'Ulivieri e Dudone, ed altri molti
Bravi soldati, in guerra rari e soli,
Giungessero in quel punto, e insiem raccolti
In Parigi: onde avvien che si consoli
Carlo in vederli; e stampò su i lor volti
Baci di gioja e di allegrezza estrema,
E fa dire ad Alfonso che non tema.

59
 Ed unisce un'armata presto presto
Di trentamila e forse più cavalli,
E pedoni altrettanti; ed esso lesto
Va loro avanti fra trombe e timballi,
E fa il suo ardire a tutti manifesto:
Che non sì corre villanella ai balli,
Com'egli a quella guerra correr sembra,
Col bianco crine e l'invecchiate membra.

60

Ma mentre egli cammina in questa guisa,

Torniamo a Ferraù che pur dimora
Nell'isoletta dal mondo divisa,
Ed ha fatto degli occhi doppia gora
Per lavar l'alma sua di colpe intrisa.
Ma il demoniaccio, che sempre lavora,
Gli guastò tanto il debole cervello,
Che ancor di nuovo a Dio si fe' rubello.

61

Non aspettò che all'isola giungesse

Tornata al mondo qualche nuova Eléna,
Che co' begli occhi e le dorate e spesse
Ricciute chiome, in amorosa pena
Ed in voglie caldissime il ponesse,
Talchè obblïasse desinare e cena;
Ma fece seco in modo che in un mese
D'una donna dell'isola s'accese.

62
 Cosa più brutta certo di costei
Non fe' natura, e farla non la puote.
Di statura simìle era a pigmei,
Con un gran capo, tutta bocca e gote,
Gran ventre, gambe grosse e lunghi pièi,
Le schiene grosse; e l'altre cose ignote
Eran nefande tanto, che mi viene
Stomaco, ognora che me ne sovviene.

63

Gli occhi poi tutti bianchi e infora infora,

Siccome le locuste, e sopra il petto
La lana avea, qual di pecora mora,
Che giù scendeva e s'univa al boschetto,
Che a darle fuoco, certo la baldora
Sarìa durata qualche buon pezzetto:
Stiacciato il naso, e i denti lunghi e storti,
Come si dice che il cinghial li porti:

64

Corte le braccia e grosse, e corta e grossa

La mano: in somma pareva una Furia.
Ma vedi del tristo abito la possa
Ed i prodigj della rea lussuria!
Che siccome fa bere acqua di fossa,
De' fonti e de' ruscelli la penuria
A chi si muor di sete, e di letame
Cibarsi quei che muorsi dalla fame:

65
 Così quando dal senso l'uomo è preso,
Ogni cosa gli piace e gli par bella;
E per tal via il buon romito acceso
Restò di quella cosa trista e fella.
E perchè questo fatto è male inteso
Nell'isola, e mal pur se ne favella,
Un dì con questa strega maladetta
Fuggissi il frate sopra una barchetta.

66

E perchè la sguajata lagrimava

Abbandonando il patrio suo terreno,
Il fraticello stretta l'abbracciava,
E le diceva: Anima mia, pon freno
A questo duol che l'anima ti cava;
Chè se tu miri bene in questo seno,
Vedrai che c'è chi ti porta più amore
Della tua madre e del tuo genitore.

67

A queste voci quella cosa brutta

Rise, qual ciuca in sul fiutar l'orina;
Ed al suo collo gettatasi tutta,
Pian pian gli dice all'orecchia mancina:
Ovunque io sarò mai da te condutta,
Per terra estrania o lontana marina,
Mio cor, mia vita e mia dolce speranza,
Sarà l'usata mia paterna stanza.

68
 Il capitano e la gente di barca,
Ch'erano, se non sbaglio, d'Inghilterra,
Stimaro il frate de' pazzi il monarca,
Mentre sì brutta cosa al sen si serra:
E quinci il ciglio ciascheduno inarca
Per vedere or quel mostro della terra,
Ora quel frate impazzito per lui;
Nè sanno qual più ammirin di que' dui.

69

Ma consolata la sozza piangente,

S'accorse Ferraù come il padrone
Si rideva di lui apertamente;
Onde gli diede un cotal sorgozzone,
Che gli fece inghiottire più d'un dente.
Danno i soldati di mano al bastone
Per castigare il pazzo temerario;
Ma la cosa per loro andò al contrario.

70

Perchè una spada datagli alle mani

La maneggiò sì presto su coloro,
Che li fe' tutti dell'anima vani.
Onde soli rimasero fra loro,
E poi per rabbia si davano a' cani;
Ch'ei non sapeva il nautico lavoro,
Nè quando dare, oppur raccor le vele,
O come governarsi in mar crudele.

71
 Ma tanto egli è il piacer ch'egli risente
Nel rimirarsi l'amor suo sì presso,
Che il mare e l'aura non gli cal nïente,
E non gli cal se in lui rimane oppresso.
O Ferraù briccone veramente,
Deh apri gli occhi omai, torna in te stesso:
L'offender Dio per cosa sì bestiale,
Se tu nol sai, ti fa peggior nel male.

72

La barca intanto su l'onde galleggia;

Chè il vento e la corrente non la move.
Il Sol già cade, e nel cader s'ombreggia
L'aria di nubi, e fra non molto piove,
E con la pioggia tuona e lampaneggia,
E fassi un tempo da spaventar Giove;
Ed ecco cade un fulmin d'improvviso
Della donna bruttissima sul viso;

73

E non contento d'averla bruciata,

Sfonda la barca, e d'acqua è già ripiena,
E già s'affonda, anzi ella è già affondata,
E già si posa su l'ultima arena.
Il frate con la donna fulminata
Sul collo, nuota come una balena.
Cessa la pioggia, e Dori e Galatea
Corron pel mar che placato ridea:

74
 E visto quel bruttissimo romito
Nuotar con peso di tanta bruttezza,
Un Tritone mandâr di lito in lito
Próteo ad avvisar che con prestezza
Dall'orrido suo gregge circuito
Colà venisse; e pieno d'allegrezza
Spediro da per tutto l'Oceàno:
Sì lor sembrò lo spettacolo strano.

75

Nè guari andò che al regnator del mare

Giunse tal voce; onde fe' porre il freno
A due balene; e là si fe' portare,
Ove il romito veniva già meno
Per lo timor di doversi annegare:
E le belle Nerëidi non meno
Quivi n'andaro pe' flutti marini,
Portate da prestissimi delfini.

76

Non tanta festa, non tanta allegrìa

Fanno d'attorno al gufo gli augelletti,
Come di riso e di piacer moría
Nettuno; e vuol che Próteo suo s'aspetti
Con quella d'atri mostri aspra genía;
Chè veder vuol se fra cotanti aspetti
Orridi e spaventosi un se ne veda,
Che la bruttezza della morta ecceda.

77
 Ed ecco il gran pastor del marin gregge,
Che dal Carpazio mar tutte traea
Le foche e l'orche ch'ei governa e regge,
Per ubbidire all'alma Galatea;
Chè per lui ogni sua parola è legge:
Alla cui vista ogni Nume, ogni Dea
Gli andaro incontro, e gli accennâr con mano
Quel nuotator col carico sì strano.

78

Ancorchè avvezzo a cose spaventose,

Próteo s'inorridì per quella vista;
E le sue bestie divennero ombrose,
E fuggîr via: così lor parve trista
Colei che tanto amabil foco pose
Nel romito, che par che ancor persista
In adorarla: e pur questi è quel frate
Che d'Angelica amò sì la beltate.

79

Di che n'ebber trastullo singolare

Que' Numi; e rider Ino fu veduta
La prima volta da che cadde in mare;
E Scilla che crudel tanto è tenuta,
Che fa Triquetra e il mar vicin tremare,
Dall'antro uscita e colà pur venuta,
Non volendo, sorrise; e rise ancora
Cariddi che le navi si divora.

80
 Ma Teti con lo stomaco rivolto,
E perchè gravida era, intimorita
Di non fare un figliuol con simil volto,
In un pesce ordinò che convertita
Fosse colei, e sì gli fosse tolto
Sì strano aspetto e vista sì sgradita.
Fu fatta seppia: indi partissi ognuno;
E del frate pensier n'ebbe Nettuno,

81

Che gli fe' far dugentomila miglia

In una notte, e trasportollo in Francia.
Di che cotanta il prende maraviglia,
Che crede di sognare, e tien per ciancia
Quel che pur vede con aperte ciglia:
E il bello è, che scudo, spada e lancia
Si mira appresso; onde quel più s'imbroglia:
Ma più parlar di lui or non ho voglia.

82

Mi sta nel core il mesto Ricciardetto,

Che chiama l'amor suo, e non l'ascolta.
Oh se sapessi, meschin giovinetto,
Come Despina tua si sta sepolta
Viva dentro un avello oscuro e stretto,
Solo perchè dall'amor tuo disciolta
Esser non vuole; se di duol si muore,
T'ucciderebbe certo il gran dolore.

83
 Come dicemmo, i forti cavalieri,
Ucciso il fiero mostro, s'imbarcaro
Inverso Nubia, dove i suoi pensieri
Avea Ricciardo, chè del furto amaro
Troppo gli duole, e assai mal volentieri
Soffre ogn'indugio; e già col crudo acciaro
Esser vorría con l'empio Serpedonte,
Col suo rivale combattendo a fronte.

84

E già sei volte e sei fuora dell'onde

Il Sole era comparso, ed altrettante
S'era in esse sommerso; e lido e sponde
Non si vedeano ancora: e il fido amante
Se si dispera, e le sue chiome bionde
S'egli si strappa, e Scirocco e Levante
Prega che soffi, ed empia ben le vele,
Sel pensi chi d'Amor servo è fedele.

85

Ma pur l'ottavo giorno in su la sera

Veggon la terra tanto desïata,
E la deserta ed orrida riviera
Sol da lïoni e da tigri abitata,
Dove sepolta viva Despina era:
E quando di bei fiori inghirlandata,
Vergognosetta in ciel splendea l'Aurora,
Toccare il lido con l'acuta prora.

86
 Primiero sul terren Ricciardo scende,
Di poi le donne e i due forti cugini,
E da un vecchio nocchiero i casi intende
Della sua donna, e gli orridi destini.
Pensate voi se d'ira egli s'accende;
E vestiti gli usberghi e gli elmi fini,
S'invìano a gran passo inverso il tempio,
Di far vogliosi un memorando scempio.

87

Il Cavalier del Pianto, l'infelice

Misero padre dell'alma Despina,
Sebbene molto prega e molto dice,
Perchè si tolga da tanta ruina,
E faccia lui e faccia sè felice,
Nulla intanto la smove: e già vicina
È l'ora ch'egli deve in su la tomba
Morire; e roca già suona la tromba.

88

Piange Despina il duro caso acerbo

Del genitore, e vorrebbe morire
In cambio suo; ma il principe superbo
Nulla affatto del cambio vuole udire;
Anzi le dice: In vita ti riserbo,
Perchè mi piace vederti patire.
Ed ecco fuor dell'avello crudele
Son tratti il padre e l'amante fedele.

89
 D'un nero panno ricoperto egli era
L'avello tutto; e la tagliente scure
Teneva in mano un uom d'orrida cera.
Vicine al duro ceppo in vesti oscure
Stavan le donne, che mattino e sera
Piangevan di Despina le sventure;
E in mezzo a loro v'era un basso scanno
Coperto pur d'un nerissimo panno.

90

Quivi fa porre il barbaro Africano

La misera Despina, acciò che veda
Morire il padre, il qual dolce ed umano,
Figlia, diceva, il giusto Dio proveda
Al tuo dolore: il mio fato inumano
E il tuo ci han fatti una misera preda
Di questo mostro, che ragione e Dio
Non cura, e segue solo il suo desío.

91

Un pezzo io ti pregai che tu stringessi

La tua con la sua mano, e in questa guisa
Te alla tomba, e a morte me togliessi:
Ma quanto or lieto nella valle elisa
Vo, perchè dura a' miei comandi espressi,
Figlia, tu fosti! che piuttosto uccisa
Io ti vedrei, che consorte a costui,
Di cui peggior non v'è tra' regni bui.

92
 Segui dunque, dolcissima Despina,
Ad odiar questo mostro: e se riserba
L'alma in passar la stigia onda divina
Il giusto sdegno e la giusta ira acerba,
Temi, ribaldo, pur, temi vicina
La vendetta che Giove a te pur serba.
L'African non risponde, e fa con gli occhi
Cenno al ministro che il gran colpo scocchi.

93

Alza quegli la scure; ma nell'atto

Che vibrar vuole il reo colpo fatale,
Sorge Despina furibonda a un tratto,
E il feritore abbraccia: e tanto vale
Sua forza, che al ministro non vien fatto
Troncar del padre lo stame vitale;
Ma dura gran fatica e stenta molto,
Che il ferro dalla man non gli sia tolto.

94

Or mentre questo succede nel tempio,

Già co' mille attaccata era la mischia
Da' tre guerrieri, che ne fanno scempio.
Tristo è colui che alla pugna s'arrischia;
Chè danno colpi che son senza esempio:
E il rombo delle spade tanto fischia,
Che s'ode dentro al tempio; e d'ira insano
Esce fuor Serpedonte al caso strano.

95
 Despina intanto, generosa e forte,
Discioglie il padre, e intrepida e sicura
Corre del tempio a spalancar le porte;
E già dentro del core si figura
Che il suo Ricciardo per benigna sorte
Il guerrier sia che lei salvar procura;
E gli altri due che pugnano per lui,
Sieno i tanto famosi cugin sui.

96

Ricciardo appena Serpedonte ha visto,

Che lo corre a investir, siccome toro
Il suo rivale, e grida: Iniquo e tristo
E perfido ladrone, ove è il decoro
Di real sangue? per rapina acquisto
Far delle donne, e a forza di martoro,
Di catene, di carceri e di morti
Tentar di superar l'alme più forti?

97

Con questo (che pur anco e fuma e gronda

Del vil sangue de' tuoi) ferro che stringo,
Perchè l'altrui superbia si confonda,
Di trapassarti il core io mi lusingo.
Qual torbido torrente che la sponda
Rompa improvviso, e del villan guardingo
Ogni riparo, e con l'altera fronte
Tutto abbatte; tal féssi Serpedonte.

98
Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.

More than just a book-buying platform, we strive to be a bridge

connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.

Join us on a journey of knowledge exploration, passion nurturing, and

personal growth every day!

ebookmasss.com