Professional Documents
Culture Documents
Discourse Markers
and Beyond
Descriptive and Critical Perspectives
on Discourse-Pragmatic Devices across
Genres and Languages
Péter B. Furkó
Postdisciplinary Studies in Discourse
Series Editor
Johannes Angermuller
Centre for Applied Linguistics
University of Warwick
Coventry, UK
Postdisciplinary Studies in Discourse engages in the exchange between
discourse theory and analysis while putting emphasis on the intellectual
challenges in discourse research. Moving beyond disciplinary divisions
in today’s social sciences, the contributions deal with critical issues at
the intersections between language and society.
Edited by Johannes Angermuller together with members of
DiscourseNet, the series welcomes high-quality manuscripts in dis-
course research from all disciplinary and geographical backgrounds.
DiscourseNet is an international and interdisciplinary network of
researchers which is open to discourse analysts and theorists from all
backgrounds.
Editorial Board
Cristina Arancibia
Aurora Fragonara
Péter Furkó
Tian Hailong
Jens Maesse
Eduardo Chávez Herrera
Michael Kranert
Jan Krasni
María Laura Pardo
Yannik Porsché
Kaushalya Perera
Luciana Radut-Gaghi
Marco Antonio Ruiz
Jan Zienkowski
Discourse Markers
and Beyond
Descriptive and Critical Perspectives
on Discourse-Pragmatic Devices across
Genres and Languages
Péter B. Furkó
Department of English Linguistics
Károli Gáspár University of the Reformed
Church in Hungary
Budapest, Hungary
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
v
vi Contents
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
IS—interactional signal
PP—pragmatic particle
PFM—pragmatic force modifier
*—categorized according to the position/slot they take in the utterance
3
4 P. B. Furkó
markers as a functional class (cf. e.g. Schourup 1999; Fraser 1999; Beech-
ing 2016; Brinton 2017), but empirical studies rely on different subsets
of such criterial features when identifying particular instances of dis-
course markers in a given corpus (for a detailed discussion, cf. Crible
2017). Naturally, this makes it difficult to compare the results of empir-
ical research even if similar datasets are involved. Table 2 illustrates this
problem.
An even more challenging task is to develop annotation software that
can automatically identify discourse markers in oral discourse and fil-
ter out non-discourse marker tokens of lexical items that are frequently
used as discourse marker types (e.g. adverbial uses of well or now, prepo-
sitional uses of like, etc.). Moreover, to date, few attempts have been
made to use automated means of identification involving semantic crite-
ria and semantic fields, since one of the very criterial features of discourse
markers is their semantic underspecification (cf. Crible et al. 2019), which
is a result of the diachronic process of semantic bleaching (cf. Brinton
2017: 31).
Accordingly, the present introduction will explore the utility of
using an automated semantic tagging software, USAS as a pre-
annotation tool for the identification of oral discourse markers, includ-
ing (inter)subjective as well as textual markers. After an overview of the
formal and functional features that can be used for manual annotation,
and after comparing the results of manual and automatic annotation of
selected discourse markers, I will argue that despite the semantic under-
specification of most discourse markers, automatic semantic annotation
(ASA) can be an effective tool for the disambiguation between discourse
marker and non-discourse marker uses with regard to certain items, but
needs to be complemented by extensive manual error correction and
filtering.
Table 2 Individual authors’ focus on categorial properties discourse markers display
seq. context oral synt. proced poly-func. attitude scope non-prop. inv.
Schiffrin (1987) x x x x (x)
Fraser (1990, 1999) x x x x x
Redeker (1990, x x (x)
1991)
Stenström (1994) x (x)
Kroon (1995) x x x
Knott and Sanders x
(1998)
Andersen (1998) x x x
Hansen (1998) x (x) (x) x x
Risselada and x x
Spooren (1998)
Romaine and Lange x (x) (x)
(1998)
Blakemore (1987, x
2002)
González (2004) x x x
Crible (2017) x x x
Legend
seq.—sequentiality-coherence-connectivity
context—context-dependence—context-coordination
oral.—orality
synt.—syntactic criteria (diversity, non-integration)
proced.—procedural meaning
poly-funct.—poly-functionality
1 Preliminary Issues: Category Membership, Methodology …
attitude—marking attitudes
scope—variable scope, functional scope
non-prop.—non-propositional content
inv.—invariable form
5
6 P. B. Furkó
(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 …
(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
2.3 Multifunctionality
(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:
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:
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.
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
(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)
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?
5 Findings
Table 3 summarizes the raw frequency of the relevant lexical items’ dis-
course marker and non-discourse marker-related USAS tags. Since both
sub-corpora were compiled in a way that they are of the same size of
100,000 words, the raw frequencies can also be compared as normalized
frequencies.
As a first step, the ratio of discourse marker and non-discourse marker
tokens of individual items was compared with the results of previous
research in the course of which discourse markers in the same sub-
corpora were manually annotated (cf. Furkó and Abuczki 2014 and
Sect. 3 in this Chapter). In order to gauge the categorial multifunction-
ality of discourse markers, the measure of D-function ratio or D-value
(a term proposed by Stenström 1990) was used. An individual item’s
D-value is calculated as a quotient of the number of tokens that ful-
fil discourse-pragmatic functions and the total number of occurrences
in a given corpus. The D-value of oh, for example, is 1 (100%) in the
London-Lund Corpus, since it is used exclusively as a discourse marker,
whereas well showed a D-value of 0.86 (86%), as 14% of its tokens serve
non-discourse marker (adverbial, nominal, etc.) functions (ibid.).
If we calculate the D-values of individual discourse markers based on
the above values and compare them to the findings of previous research,
we see that the results of automatic annotation and manual annotation
converge to a great extent. Mean, for example, has a D-value of 0.808 in
the MPI corpus based on automatic annotation (calculated as the num-
ber of Z4 tags divided by all tokens of mean, i.e. 141), while manual
annotation yielded a D-value of 0.797 (cf. Furkó and Abuczki 2014:
50). Similarly, manual annotation yielded a D-value of 0.82 for well in
the MPI corpus (Furkó and Abuczki 2014: 54), while Table 1 yields a
D-value of 0.839 for this lexical item (360 Z4 tags divided by the total
number of tokens, i.e. 429).
The table also correctly predicts that most of the lexical items under
scrutiny have higher D-values in the CI sub-corpus than in the MPI
sub-corpus, which is explained by the fact that there is a higher degree
of conversationalization in celebrity interviews, i.e. they are more similar
to spontaneous, informal, face-to-face conversations (cf. this chapter and
Table 3 Summary of discourse marker and non-discourse marker-related semantic tags assigned to the most frequent
discourse marker types in the MPI and CI sub-corpora
frequency of frequency of frequency of frequency of
DM-related tag in the DM-related tag in the non-DM-related tag non-DM-related tag
lexical item MPI CI in the MPI in the CI
well (429) 360xA5.1 312xA5.1 14xI1.1, 55xN5 1xA7, 2xB2, 24xN5
sort (38) 14xZ4 25xZ4 21xA4.1, 3xA1.1.1 10xA4.1
now (299) 4xZ4 1xZ4 288xT1.1.2, 7xZ5 229xT1.1.2, 6xZ5
(you) know (346) 205xZ4 455xZ4 140xX2.2, 1xZ6 307xX2.2
like (97) 6xZ4 17xZ4 51xZ5, 40xE2+ 238xZ5, 139xE2+
(I) mean (141) 114xZ4 201xZ4 27xQ1.1 30xQ1.1, 5xS2.2.2
(in other) words (11) 4xZ4 13xZ4 7xQ.3 7xQ.3
actually (165) 165xA5.4 72xA5.4 0 0
(I) think (549) 126xZ4 121xZ4 423xX2.1 319xX2.1
right (114) 55xZ4, 53xA5.3 211xZ4, 98xA5.3 6xT1.1.2 12xN3.8, 16xS7.4,
15xT1.1.2
1 Preliminary Issues: Category Membership, Methodology …
19
20 P. B. Furkó
Chapter 2). For example, the D-value of well is 0.92, and the D-value
of mean is 0.851 in the CI sub-corpus based on automatic annotation
(312 A5.1 tags divided by a total of 339 tokens, 201 Z4 tags divided by
a total of 236 tokens, respectively).
In the second stage of the research, a representative sample of tokens in
the MPI was manually annotated using numeric 1 for discourse marker
tokens and 2 for non-discourse marker uses. With a view to comparing
the results of automatic and manual annotation, all discourse marker-
related tags (Z4 and A5.x) yielded by USAS were re-coded as numeric
1, while non-discourse marker tags (B2, I1.1, T1.3, etc.) were re-coded
as 2. Consequently, the extracted list of the corresponding manual and
automated tags was entered into a reliability calculator (Freelon’s ReCal
2 for 2 coders) in order to calculate inter-annotator agreement statistics.
Table 4 shows the result.
Although the above inter-coder agreement values appear high (cf.
Spooren and Degand 2010), it is important to note that there is a
great degree of variation in the precision with which individual dis-
course markers are tagged by USAS. On the one hand, there are dis-
course markers such as I mean and you know whose discourse marker and
non-discourse marker uses are disambiguated with surprising precision
(resulting in a kappa score of <.98, i.e. close to perfect inter-coder agree-
ment between USAS and the human annotator), cf. (10ab) and (11ab)
below:
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):
(16)
(a) My roommate never cleans when I ask him to. Like, I asked him
yesterday to clean, and he never did it. (Like_E2 + ,_PUNC I_Z8mf
asked_Q2.2 him_Z8m yesterday_T1.1.1 to_Z5 clean_B4,_PUNC
and_Z5 he_Z8m never_T1/Z6 did_A1.1.1 it_Z8 ._PUNC)
(b) This guy is so cool. I mean, he’s like the coolest person you could
meet. (I_Z4[i1.2.1 mean_Z4[i1.2.2,_PUNC he_Z8m s_T1.3 like_Z5
the_Z5 coolest_O4.6-person_S2mfc you_Z8mf could_A7+ meet_S3.1
._PUNC)
(c) I went to the clerk to ask him where the beer was, and he’s like,
‘I don’t know, I’m new here’, so I’m like, yeah, sure, like, you should
know this, man! (so_Z5 Im_Z99 like_Z5,_PUNC yeah_Z4,_PUNC
sure_A7+ ,_PUNC like_Z4,_PUNC
1 Preliminary Issues: Category Membership, Methodology … 25
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:
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.)
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.
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.
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.
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 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.
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.