Discourse Markers and Beyond Descriptive and Critical Perspectives On Discourse Pragmatic Devices Across Genres and Languages 1St Ed 2020 Edition Peter B Furko Full Chapter

Discourse Markers and Beyond:
Descriptive and Critical Perspectives

on Discourse-Pragmatic Devices
across Genres and Languages 1st ed.
2020 Edition Péter B. Furkó
Visit to download the full and correct content document:
https://ebookmass.com/product/discourse-markers-and-beyond-descriptive-and-critic
al-perspectives-on-discourse-pragmatic-devices-across-genres-and-languages-1st-ed
-2020-edition-peter-b-furko/
POSTDISCIPLINARY STUDIES IN DISCOURSE
SERIES EDITOR: JOHANNES ANGERMULLER
Discourse Markers
and Beyond
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
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
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
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
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
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):
(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.
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.
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.
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
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)
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
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
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
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:
Another random document with
no related content on Scribd:
line the paragaster as fast as its original covering of choanocytes
retreats into the newly formed chambers.
Fig. 81.—S. setosum. Young Sponge, with one whorl of radial tubes. o, Osculum;
p, pore; sp1, monaxon; sp4, quadriradiate spicule. (After Maas.)
With a canal system precisely similar to that of Sycon, Ute (Fig. 83)
shows an advance in structure in the thickening of the dermal layers
over the distal ends of the chambers. The dermal thickenings above
neighbouring chambers extend laterally and meet; and there results
a sheet of dermal tissue perforated by dermal ostia, which open into
the inhalant canals, and strengthened by stout spicules running
longitudinally. This layer is termed a cortex; it covers the whole
sponge, compacting the radial tubes so that they form, together with
the cortex, a secondary wall to the sponge, which is once more a
simple sac, but with a complex wall. The cortex may be enormously
developed, so as to form more than half the thickness of the wall
(Fig. 84). The chambers taken together are spoken of as the
chamber layer.
Fig. 82.—Sycon raphanus. A, Longitudinal section of young decalcified Sponge
at a stage somewhat later than that shown in Fig. 81. B, Transverse section
of the same through a whorl of tubes. d, Dermal membrane; g, gastral
membrane; H, paragaster; sp4, tetraradiate spicule; T, radial tube. (After
Maas.)
Fig. 83.—Transverse section of the body-wall of Ute, passing longitudinally
through two chambers. a.p, Apopyle; d.o, dermal ostium; fl.ch, flagellated
chamber or radial tube; i.c, inhalant canal; p, prosopyle. (After Dendy.)
We have already alluded to the resemblance between a young

Ascon person and a radial tube of Sycon—a comparison which calls
to mind the somewhat strange view of certain earlier authors, that
the flagellated chambers are really the sponge individuals. If now we
suppose each Ascon-like radial tube of Sycon to undergo that same
process of growth by which the Sycon itself was derived from the
Ascon, we shall then have a sponge with a canal system of the type
seen in Leucandra among British forms, but more diagrammatically
shown in the foreign genus Leucilla (Fig. 85). The foregoing remarks
do not pretend to give an account of the transition from Sycon to
Leucilla as it occurred in phylogeny. For some indication of this we
must await embryological research.
In Leucandra the fundamental structure is obscured by the

irregularity of its canal system. It shows a further and most important
difference from Leucilla in the smaller size and rounded form of its
chambers. This change of form marks an advance in efficiency; for
now the flagella converge to a centre, so that they all act on the
same drop of water, while in the tubular chamber their action is more
widely distributed and proportionately less intense (see p. 236).
Fig. 84.—Transverse section through the body-wall of Grantiopsis. d.o, Dermal
ostium; fl.ch, flagellated chamber; i.c, long incurrent canal traversing the
thick cortex to reach the chamber layer; p, apopyle. (After Dendy.)
Fig. 85.—Transverse section through the body-wall of Leucilla. d.o, Dermal

ostium; ex.c, exhalant canal; fl.ch, chamber; i.c, inhalant canal. (After
Dendy.)
Above are described three main types of canal system—that of

Homocoela, of Sycon, and of Leucandra and Leucilla. These are
conveniently termed the first, second, and third types respectively,
and may be briefly described as related to one another somewhat in
the same way as a scape, umbel, and compound umbel among
inflorescences. These types formed the basis of Haeckel's famous
classification.[221] It has, however, been concluded[222] that the
skeleton is a safer guide in taxonomy, at any rate for the smaller
subdivisions; and in modern classifications genera with canal
systems of the third type will be found distributed among various
families; while in the Grantiidae, Ute and Leucandra stand side by
side. This treatment implies a belief that the third type of canal
system has been independently and repeatedly evolved within the
Calcarea—an example of a phenomenon, homoplasy, strikingly
displayed throughout the group. It is, remarkably enough, the case
that all the canal systems found in the remainder of the Porifera are
more or less modified forms of one or other of the second two types
of canal system above described.
The families Grantiidae, Heteropidae, and Amphoriscidae, all

possessing a dermal cortex, are distinguished as follows:—The
Grantiidae by the absence of subdermal sagittal triradiate spicules
and of conspicuous subgastral quadriradiates; the Heteropidae by
the presence of sagittal triradiates; the Amphoriscidae by the
presence of conspicuous subgastral quadriradiates.
Two families of Calcarea, possibly allied, remain for special mention

—the Pharetronidae, a family rich in genera, and containing almost
all the fossil forms of the group, and the Astroscleridae.
The Pharetronidae are with one, or perhaps two exceptions, fossil

forms, having in common the arrangement of the spicules of their
main skeletal framework in fibres. The family is divided into two sub-
families:—
I. Dialytinae.—The spicules are not fused to one another; the exact

mode of their union into fibres is unknown, but an organic cement
may be present.
Lelapia australis, a recent species, should probably be placed here

as the sole living representative. Dendy has shown[223] that this
remarkable species has a skeleton of the same fibrous character as
is found in typical Dialytinae, and that the triradiate spicules in the
fibres undergo a modification into the "tuning-fork" type (Fig. 86, C),
to enable them to be compacted into smooth fibres. "Tuning-forks,"
though not exclusively confined to Pharetronids, are yet very
characteristic of them.
Fig. 86.—Portions of the skeleton of Petrostroma schulzei. A, Framework with
ensheathing pellicle; B, quadriradiate spicules with laterally fused rays; C, a
"tuning-fork." (After Doederlein.)
II. Lithoninae.—The main skeletal framework is formed of spicules

fused together, and is covered by a cortex containing free spicules.
Fig. 87.—A spicule from the skeleton framework of Plectroninia, showing the
terminally expanded rays. (After Hinde.)
The sub-family contains only one living genus and a few recently
described fossil forms. Petrostroma schulzei[224] lives in shallow
water near Japan; Plectroninia halli[225] and Bactronella were found
in Eocene beds of Victoria; Porosphaera[226] long known from the
Chalk of England and of the Continent, has recently been shown by
Hinde[226] to be nearly allied to Plectroninia; finally, Plectinia[227] is a
genus erected by Počta for a sponge from Cenomanian beds of
Bohemia. Doederlein, in 1896, expressed his opinion that fossil
representatives of Lithoninae would most surely be discovered. The
fused spicules are equiangular quadriradiates; they are united in
Petrostroma by lateral fusion of the rays, in Plectroninia (Fig. 87) and
Porosphaera by fusion of apposed terminal flat expansions of the
rays, and in some, possibly all, genera a continuous deposit of
calcium carbonate ensheaths the spicular reticulum. Thus they recall
the formation of the skeleton on the one hand of the Lithistida and on
the other of the Dictyonine Hexactinellida (see pp. 202, 211).
"Tuning-forks" may occur in the dermal membrane.
Fig. 88.—Astrosclera willeyana, Lister. A, the Sponge, × about 3. p, The ostia on

its distal surface. B, a portion of the skeleton showing four polyhedra with
radiating crystalline fibres. C, an ostium; the surrounding tissue contains
young stages of polyhedra. (After Lister.)
The Astroscleridae, as known at present, contain a single genus

and species, apparently the most isolated in the phylum. Astrosclera
willeyana[228] was brought back from the Loyalty Islands, and from
Funafuti of the Ellice group. Its skeleton is both chemically and
structurally aberrant. In other Calcarea the calcium carbonate of the
skeleton is present as calcite, in Astrosclera as aragonite, and the
elements are solid polyhedra, united by their surfaces to the total
exclusion of soft parts (Fig. 88). Each element consists of crystalline
fibres radially disposed around a few central granules, and
terminating peripherally in contact with the fibres of adjacent
elements. Young polyhedra are to be found free in the soft parts at
the surface. The chambers are exceptionally minute, especially for a
calcareous sponge, comparing with those of other sponges as
follows:—
Astrosclera chambers, 10µ × 8µ to 18µ × 11µ.

Smallest chambers in Silicea, 15µ × 18µ to 24µ × 31µ.
Smallest chambers in Calcarea, 60µ × 40µ.
In its outward form Astrosclera resembles certain Pharetronids. The
minute dimensions of the ciliated chambers relegate Astrosclera to
the Micromastictora, and the fortunate fact that the calcium
carbonate of its skeleton possesses the mineral characters not of
calcite, but of aragonite, renders it less difficult to conceive that its
relations may be rather with the non-calcareous than the calcareous
sponges.
BRANCH II. MICROMASTICTORA

All sponges which do not possess calcareous skeletons are
characterised by choanocytes, which, when compared with those of
Calcarea, are conspicuous for their smaller size. The great majority
(Silicispongiae) of the non-calcareous sponges either secrete
siliceous skeletons or are connected with siliceous sponges by a
nicely graded series of forms. The small remainder are entirely
askeletal. All these non-calcareous sponges are included, under the
title Micromastictora, in a natural group, opposed to the
Megamastictora as of equal value.
The subdivision of the Micromastictora is a matter of some difficulty.

The Hexactinellida alone are a well circumscribed group. After their
separation there remains, besides the askeletal genera, an
assemblage of forms, the Demospongiae, which fall into two main
tribes. These betray their relationship by series of intermediate
types, but a clue is wanting which shall determine decisively the
direction in which the series are to be read. The askeletal genera are
the crux of the systematist. It is perhaps safest, while recognising
that many of them bear a likeness of one kind or another to various
Micromastictora, to retain them together in a temporary class, the
Myxospongiae.
CLASS I. MYXOSPONGIAE
The class Myxospongiae is a purely artificial one, containing widely
divergent forms, which possess a common negative character,
namely, the absence of a skeleton. As a result of this absence they
are all encrusting in habit.
One genus, Hexadella, has been regarded by its discoverer

Topsent[229] as an Hexactinellid. The same authority places
Oscarella with the Tetractinellida; it is more difficult to suggest the
direction in which we are to seek the relations of the remaining type,
Halisarca.
Hexadella, from the coast of France, is a remarkable little rose-

coloured or bright yellow sponge, with large sac-like flagellated
chambers and a very lacunar ectosome.
Oscarella is a brightly coloured sponge, with a characteristic velvety

surface; it is a British genus, but by no means confined to our
shores. Its canal system has been described by some authors as
diplodal, by others as eurypylous. Topsent[230] has shown, and we
can confirm his statement, that though the chambers have usually
the narrow afferent and efferent ductules of a diplodal system, yet
since each one may communicate with two or three canals, the canal
system cannot be described as diplodal. The hypophare attains a
great development, and in it the generative products mature. The
pinacocytes, like those of Plakinidae, and perhaps of Aplysilla, are
flagellated.
Halisarca, also British, is easily distinguished from Oscarella by the

presence of a mucus-like secretion which oozes from it, and by the
absence of the bright coloration characteristic of Oscarella. It
naturally suggests itself that the coloration in the one case and the
secretion in the other are protective, and in this respect perform one
of the functions of the skeleton of other sponges. The chambers are
long, tubular, and branched. There is no hypophare.
CLASS II. HEXACTINELLIDA[231]
Silicispongiae, defined by their spicules, of which the rays lie along
three rectangular axes. The canal system is simple, with thimble-
shaped chambers. The body-wall is divided into endosome,
ectosome, and choanosome.
Some authors would elevate the Hexactinellida to the position of a

third main sub-group of Porifera, thus separating them from other
siliceous sponges. In considering this view it is important to realise at
the outset that they are deep-water forms. They bear evident traces
of the influence of their habitat, and like others of the colonists of the
deep sea, are impressed with marked archaic features. Yet they are
still bound to other Micromastictora, first by the small size of their
choanocytes, and secondly by the presence of siliceous spicules.
This second character is really a double link, for it involves not
merely the presence of silica in the skeleton, but also the presence
in each spicule of a well-marked axial filament. Now this axial
filament is a structure which is gaining in importance, for purposes of
classification, in proportion as its absence in Calcarea is becoming
more probable. The Hexactinellida are the only sponges, other than
the bath sponge, which are at all generally known. They have won
recognition by their beauty, as the bath sponge by its utility, and, like
it, one of their number—the Venus's Flower-Basket—forms an
important article of commerce, the chief fishery being in the
Philippine Islands. This wonderful beauty belongs to the skeleton,
and is greatly concealed when the soft parts are present.
We have said that the Hexactinellids are deep-sea forms; they are
either directly fixed to the bottom or more often moored in the ooze
by long tufts of rooting spicules. In the "glass-rope sponge," the
rooting tuft of long spicules, looking like a bundle of spun glass, is
valued by the Japanese, who export it to us. In Monorhaphis the
rooting tuft is replaced by a single giant spicule,[232] three metres in
length, and described as "of the thickness of a little finger"! Probably
it is as a result of their fixed life in the calm waters of the deep
sea[233] that Hexactinellids contrast with most other sponges by their
symmetry. It should not, however, be forgotten that many of the
Calcarea which inhabit shallow water exhibit almost as perfect a
symmetry.
Fig. 89.—Longitudinal section of a young specimen of Lanuginella pupa O.S.,

with commencing formation of the oscular area. × 35. d.m, Dermal
membrane; g.m, gastral membrane; pg, paragaster; sd.tr, subdermal
trabeculae; Sg.tr, subgastral trabeculae. (After F. E. Schulze.)
The structure of the body-wall in Hexactinellida is so constant as to

make it possible to give a general description applicable to all
members of the group. It is of considerable thickness, but a large
part is occupied by empty spaces, for the actual tissue is present in
minimum quantity. In the wall the chamber-layer is suspended by
trabeculae of soft tissue, between a dermal membrane on the
outside and a similar gastral membrane on the inner side (Fig. 89).
Thus the water entering the chambers through their numerous pores
has first passed through the ostia in the dermal membrane and
traversed the subdermal trabecular space; on leaving the chambers
it flows through the subgastral trabecular space and the ostia in the
gastral membrane, to enter the paragaster and leave the body at the
osculum. The trabeculae and the dermal and gastral membranes
together constitute the dermal layer. This conclusion is based on
comparison with adults of the other groups, for in the absence of
embryological knowledge no direct evidence is available. According
to the Japanese investigator, Isao Ijima,[234] the dermal and gastral
membranes are but expansions of the trabeculae, and the
trabeculae themselves are entirely cellular, containing none of the
gelatinous basis met with in the dermal layer of all other sponges.
There is no surface layer of pinacocytes, the cells forming the
trabeculae being all of one type, namely, irregularly branching cells,
connected with one another by their branches to form a syncytium.
In the trabeculae are found scleroblasts and archaeocytes.
The chambers have a characteristic shape: they are variously

described as "thimble-shaped," "tubular," or "Syconate," and they
open by wide mouths into the subgastral trabecular space. Their
walls have been named the membrana reticularis from the fact that,
when preserved with only ordinary precautions, they are seen as a
regular network of protoplasmic strands, with square meshes and
nuclei at the nodes. This appearance recently found an explanation
when Schulze, for the first time, succeeded in preserving the collared
cells of Hexactinellids.[235] Schulze was then able to show that the
choanocytes are not in contact with one another at their bases,
where the nuclei are situated, but communicate with one another by
stout protoplasmic strands. The form of the choanocyte can be seen
in Fig. 91.
Fig. 90.—Portion of the body-wall of Walteria sp., showing the thimble-shaped

flagellated chambers, above which is seen the dermal membrane. (After F.
E. Schulze.)
To Schulze's description of the chamber, Ijima has added the
important contributions that every mesh in the reticulum functions as
a chamber pore or prosopyle; and that porocytes, such as are found
in Calcarea, are wanting. This structure of the chamber-walls, the
absence of gelatinous basis in the dermal layer, and the slight
degree of histological differentiation in the same layer, added to the
more obvious character of thimble-shaped chambers, are the chief
archaic features of Hexactinellid morphology.
Fig. 91.—Portion of a section of the membrana reticularis or chamber-wall of

Schaudinnia arctica, × 1500. (After F. E. Schulze.)
The skeleton which supports the soft parts is, like them, simple and
constant in its main features. It is secreted by scleroblasts, which lie
in the trabeculae, and is made up of only one kind of spicule and its
modifications. This is the hexactine, a spicule which possesses six
rays disposed along three rectangular axes. Each ray contains an
axial thread, which meets its fellow at the centre of the spicule,
where they together form the axial cross. Modifications of the
hexactine arise either by reduction or branching, by spinulation or
expansion of one or more of the rays. The forms of spicule arising by
reduction are termed pentactines, tetractines, and so on, according
to the number of the remaining rays. Those rays which are
suppressed leave the proximal portion of their axial thread as a
remnant marking their former position (Fig. 94). Octactine spicules
seem to form an exception to the above statements, but Schulze has
shown that they too are but modifications of the hexactine arising by
(1) branching of the rays of a hexactine, followed by (2)
recombination of the secondary rays (Fig. 92).
Fig. 92.—A, discohexaster, in which the four cladi a, a', b, b', c of each ray start
directly from a central nodule. B, disco-octaster, resulting from the
redistribution of the twenty-four cladi of A into eight groups of three. (After
Schulze, from Delage.)
The various spicules are named, irrespective of their form, according

to their position and corresponding function. The arrangement of the
spicules is best realised by means of a diagram (Fig. 93).
Fig. 93.—Scheme to show the arrangement of spicules in the Hexactinellid

skeleton. Canalaria, microscleres in the walls of the excurrent canals;
Dermalia Autoderm[alia], microscleres in the dermal membrane; D.
Hypoderm[alia], more deeply situated dermalia; Dictyonalia, parenchymalia
which become fused to form the skeletal framework of Dictyonina; Gastralia
Autogastr[alia], microscleres in the gastral membrane; Gastralia
Hypogastr[alia], more deeply situated gastralia; Parenchymalia Principalia,
main supporting spicules between the chambers; P. Comitalia, slender
diactine or triactine spicules accompanying the last; P. Intermedia,
microscleres between the P. principalia; Prostalia, projecting spicules; P.
basalia, rooting spicules, from the base; P. marginalia, defensive spicules,
round the oscular rim; P. pleuralia, defensive spicules, from the sides. (From
Delage and Hérouard, after F. E. Schulze.)
The deviations from this ground-plan of Hexactinellid structure are
few and simple. They are due to folding of the chamber-layer, or to
variations in the shape of the chambers, and to increasing fusion of
the spicules to form rigid skeletons. A simple condition of the
chamber-layer, like that of the young sponge of Fig. 89, occurs also
in some adult Hexactinellids, e.g. in Walteria of the Pacific Ocean
(Fig. 90). Thus is represented in this order the second type of canal
system described among Calcarea. More frequently, however,
instead of forming a smooth sheet, the chamber-layer grows out into
a number of tubular diverticula, the cavities of which are excurrent
canals; these determine a corresponding number of incurrent canals
which lie between them. In this way there arises a canal system
resembling the third type of Calcarea. By still further pouching so as
to give secondary diverticula, opening into the first, a complicated
canal system is formed, as, for example, in Euplectella suberea.
To return to the skeleton, the most complete fusion is attained by the

deposit of a continuous sheath of silica round the apposed parallel
rays of neighbouring spicules. This may be termed the dictyonine
type of union, for it occurs in all those forms originally included under
the term Dictyonina, in which the cement is deposited pari passu
with the formation of the spicules. In other cases connecting bridges
of silica unite the spicules, or there may be a connecting reticulum of
siliceous threads, or, again, rays crossing obliquely may be soldered
together at the point of contact. These more irregular methods occur
in species where the spicules are free at their first formation.
Spicules originally free may later be united in a true Dictyonine
fashion. The terms Lyssacina and Dictyonina are useful to denote
respectively: the former all those Hexactinellida in which the spicules
are free at their first formation, and the latter those in which the
deposit of the cementing layer goes hand in hand with the formation
of the spicules. But the terms do not indicate separateness of origin
of the groups denoted by them, for there is evidence that Dictyonine
types have been derived repeatedly from Lyssacine types, and that
in fact every Dictyonine was once a Lyssacine.
Fig. 94.—Amphidisc, at a are traces of the four missing rays.
The real or natural cleft in the class lies between those genera
possessing amphidiscs (Figs. 94, 97) among their microscleres, and
all the remainder of the Hexactinellida which bear hexasters (Fig.
96). The former set of genera constitute the sub-class
Amphidiscophora, the latter the Hexasterophora.
Fig. 95.—Portion of body-wall of Hyalonema, in section, showing the irregular

chambers.
Sub-Class 1. Amphidiscophora.—Amphidiscs are present,

hexasters absent. A tuft of rooting spicules or basalia is always
present. The ciliated chambers deviate more or less from the typical
thimble shape, and the membrana reticularis is continuous from
chamber to chamber (Figs. 94, 95, 97).
Fig. 96.—Hexasters. A, Graphiohexaster; B, floricome; C, onychaster.
Sub-Class 2. Hexasterophora.—Hexasters are present,

amphidiscs absent. The chambers have the typical regular form, and
are sharply marked off from one another (Figs. 90, 96).
All the Amphidiscophora have Lyssacine skeletons; in the

Hexasterophora both types of skeleton occur. The subdivision of the
Hexasterophora is determined by the presence or absence of
uncinate spicules. An "uncinatum" is a diactine spicule, pointed at
both ends and bearing barbs all directed towards one end. This
method of classification gives us a wholly Dictyonine order,
Uncinataria, and an order consisting partly of Dictyonine, partly of
Lyssacine genera, which may be distinguished as the
Anuncinataria. Ova have rarely been found, and sexually produced
larvae never; but Ijima has found archaeocyte clusters in abundance,
and his evidence is in favour of the view that they give rise asexually
to larvae, described by him in this class for the first time (see p. 231).
Both sub-classes are represented in British waters: the

Amphidiscophora by Hyalonema thomsoni and Pheronema
carpenteri; the Hexasterophora by Euplectella suberea and
Asconema setubalense, and of course possibly by others.
Hyalonema thomsoni, one of the glass-rope sponges, was dredged

by the Porcupine off the Shetland Islands in water of about 550
fathoms. The spindle-shaped body of the sponge is shown in Fig. 97.
Its long rooting tuft is continued right up its axis, to end in a conical
projection, which is surrounded by four apertures leading into
corresponding compartments of the paragaster.
Fig. 97.—Hyalonema thomsoni. A, Whole specimen with rooting tuft and

Epizoanthus crust; B, pinulus, a spicule characteristic of but not peculiar to
the Amphidiscophora, occurring in the dermal and gastral membranes; C,
amphidisc with axial cross; D, distal end of rooting spicule with grapnel.
(After F. E. Schulze.)
The crust of Anthozoa of the genus Epizoanthus (p. 406) on the

rooting tuft is a constant feature in this as in other species of
Hyalonema. It contributed to make the sponge a puzzle, which long
defied interpretation. The earliest diagnosis the genus received was
the "Glass Plant." Then the root tuft was thought to be part of the
Epizoanthus, which was termed a "most aberrant Alcyonarian with its
base inserted in a sponge"; next we hear of the sponge as parasitic
on the Sea Anemone. Finally, the root tuft was shown to be proper to
the sponge, which was, however, figured upside down, till some
Japanese collectors described the natural position, or that in which
they were accustomed to find it.
Pheronema carpenteri was found by the Lightning off the north of

Scotland in 530 fathoms. The goblet shaped, thick walled body and
broad, ill-defined root tuft are shown in Fig. 98, but no figure can do
justice to the lustre of its luxuriant prostalia and delicate dermal
network with stellate knots at regular intervals. The basalia are two-
pronged and anchor-like.
Fig. 98.—Pheronema carpenteri. × ½. (From Wyville Thomson.)
Both the Hexasterophoran genera were dredged off the north of

Scotland, and both conform to the Lyssacine type without uncinates.
Euplectella suberea is a straight, erect tube, anchored by a tuft of
basalia. The upper end of the tube is closed by a sieve plate, the
perforations in which are oscula, while the beams contain flagellated
chambers, so that the sieve is simply a modified portion of the wall. It
is a peculiarity of this as of one or two other allied genera that the
lateral walls are perforated by oscula. They are termed parietal gaps,
and are regularly arranged along spiral lines encircling the body.
Fig. 99.—Sieve plate of Euplectella imperialis. (After Ijima.)
Ijima, who has dredged Euplectellids from the waters near Tokyo,
finds that in young specimens oscula are confined to the sieve plate;
parietal gaps are secondary formations. The groundwork of the
skeleton is a lattice similar to that shown in Fig. 100. The chamber-
layer is much folded. Various foreign species of Euplectella afford
interesting examples of association with a Decapod Crustacean,
Spongicola venusta, of which a pair lives in the paragaster of each
specimen. The Crustacean is light pink, the female distinguished by
a green ovary, which can be seen through the transparent tissues. It
is not altogether clear what the prisoner gains, nor what fee, if any,
the host exacts.
Ijima relates that the skeleton of Euplectella is in great demand in

Japan for marriage ceremonies. He also informs us that the
Japanese name means "Together unto old age and unto the same
grave," while by a slight alteration it becomes "Lobsters in the same
cell," and remarks that the Japanese find this an amusing pun.
Fig. 100.—Skeletal lattice of Euplectella imperialis. (After Ijima.)

Discourse Markers and Beyond Descriptive and Critical Perspectives On Discourse Pragmatic Devices Across Genres and Languages 1St Ed 2020 Edition Peter B Furko Full Chapter

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Discourse Markers and Beyond Descriptive and Critical Perspectives On Discourse Pragmatic Devices Across Genres and Languages 1St Ed 2020 Edition Peter B Furko Full Chapter

Uploaded by

Copyright:

Available Formats

Discourse Markers and Beyond:

Descriptive and Critical Perspectives

More information about this series at

Postdisciplinary Studies in Discourse

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

Cover illustration: RooM the Agency/Alamy Stock Photo

1 Preliminary Issues: Category Membership, Methodology,

Part I Discourse Markers Across Genres

2 Discourse Markers in Natural Conversations,

3 Discourse Markers from a Critical Perspective:

3 The Political News Interview as a Genre 71

4 Discourse Markers from a Critical Perspective:

Part II Discourse Markers Across Languages

5 The Use of Discourse Markers in Business English

6 Discourse Markers in Scripted Discourse I: Issues

3.2 The Functional Spectrum of I Mean 151

7 Discourse Markers in Scripted Discourse II:

8 Discourse Markers and Their Translation in Literary

9 The Use of Discourse Markers in Bible Translations 219

2 Greek Sentence Conjunctions in John and Galatians

Concluding Remarks 247

Software and Online Tools 251

Data Availability 253

Appendix—List of Abbreviations 255

1 Theoretical and Empirical Approaches

issues include the lack of generally accepted terminology and classifica-

2 Characteristics of Discourse Markers,

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

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

Discourse markers’ extreme context-dependence is frequently identified

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

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

2.4 Weak Clause Association and Variable Scope

It is frequently observed in the literature that discourse markers usually

Conjuncts are more like disjuncts than adjuncts in having a relatively

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

(4a) Ed and Doris loved each other.

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

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

2.5 Procedural Meaning/Non-compositionality

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

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

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

2.6 High Frequency, Orality, Stigmatization

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

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

3 Automatic Semantic Annotation:

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

4 Corpus and Methodology

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

To increase comparability, the downsampling technique (cf.

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

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

(12) I_Z8mf want_X7+ to_Z5 assure_A7+ you_Z8mf that_Z8

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

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

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

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

Another discourse marker that displays “high frequency and versatility”

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

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

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

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