Intercultural Pragmatics 2012; 9(4): 413 – 452
DOI 10.1515/ip-2012-0025 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Anita Fetzer and Augustin Speyer
Discourse relations in English and German
discourse: Local and not-so-local constraints
Abstract: This paper examines the overt and non-overt coding of discourse relations in the argumentative discourse genre of editorial based on a contrastive
study of British English and German editorials. Particular attention is given to the
linguistic coding of discourse relations positioned adjacently and non-adjacently,
and to the question of granularity. The analysis of the German editorials is based
on the syntactic unit of sentence, while their British counterpart is based on the
syntactic unit of clause.
In the data at hand, the two languages code the discourse relation of Contrast
overtly in adjacent and non-adjacent positioning but employ different strategies
as regards the overt coding of the coordinating discourse relation of Continuation
and the subordinating discourse relations of Elaboration, Explanation, and Comment. The rate of overt marking for adjacently positioned coordinating relations
is higher in the German data. In the British data, there is hardly any difference
between the overt marking of adjacently positioned discourse relations holding
between clauses and sentences. The overt marking of subordinating discourse
relations is lower in the German data, and in the British data, there is a clear preference for coding adjacently positioned subordinating discourse relations in an
overt manner on the level of clause.
25
26
27
28
Anita Fetzer: ■■
Augustin Speyer: ■■
29
30
31
32
33
1 Introduction
34 The concept of discourse is used in all kinds of context and discourses, and it is
35 often used in a rather intuitive manner without clear-cut delimitations. So, what
36 is discourse? From a quantity-anchored perspective, discourse analysis examines
37 “language patterns above the sentence” (Widdowson 2004: 3). This implies that
38 discourse is composed of more than one sentence, and that the composition of
39 these sentences needs to be in accordance with some kind of a more general pat-
40 tern, if not a rule. Another premise of the quantity-anchored definition is that the
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 413–452 2655_9-4_01
PMU:(idp) 24/9/2012
(p. 413)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
414 constitutive parts of discourse are sentences (in Widdowson’s terms). But is that
really a felicitous definition of discourse?
To account for the patterned linearization of sentences in a larger frame of
reference, which delimits discourse from context on the one hand, and from arbitrarily concatenated sentences on the other requires the accommodation of a
quality-anchored perspective which may account for (1) the semantics and pragmatics of the joints, metaphorically speaking, connecting the constitutive parts of
discourse, (2) the semantics and pragmatics of the constitutive discourse units,
and (3) the semantics and pragmatics of discourse as a whole. Qualitatively oriented discourse studies generally share the assumption that discourse comes in
with the presumption of being coherent (cf. Bublitz, Lenk, and Ventola 1999;
Gernsbacher and Givón 1995), and it is not the ‘language patterns above the sentence’ and their semantic well-formedness which makes them cohere, but rather
its recipients who construe discourse coherence both locally and globally. Hence,
discourse coherence does not lie in the discourse itself but in the minds of language users and is thus a socio-cognitive construct. This holds for both the constitutive parts of discourse and for discourse-as-a-whole.
Discourse coherence feeds on semantic coherence and on pragmatic coherence (cf. van Dijk 1980). The former captures logical relations between discourse
units as well as lexical coherence holding among lexical units. The latter refers
to language users’ coding and implicating, and decoding an inferring speaker-­
intended meaning in local and global contexts comprising, e.g., relations between speech acts and their constitutive parts, and relations among presuppositions. The construal of semantic coherence is based on logical reasoning – for
instance, deduction and entailment – while pragmatic coherence is construed
through inference and abductive reasoning (Givòn 2005). Different modes of communication, e.g., spoken and written discourse, employ mode- and genre-specific
linguistic means to signal semantic, pragmatic and discourse coherence, such as
meta-communicative comments (“as has been examined thoroughly in the previous section,” “coming back to what I’ve said before”) and discourse connectives
(“and,” “but,” “however”) (cf. Biber 1988). The necessary cognitive operations to
construe discourse coherence, pragmatic coherence, and semantic coherence are
based on directly adjacent discourse units, lexical units and illocutions, on nondirectly adjacent discourse units, lexical units, and illocutions, on discoursegenre1 specific constraints and requirements (cf. Levinson 1979; Thibault 2003),
and on encyclopaedic knowledge.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
1 In this paper, discourse genre is used as a functional hyperonym for communicative genre,
activity type and communicative activity type (cf., Levinson 1979; Linell 1998).
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 414–452 2655_9-4_01
PMU:(idp) 24/9/2012
39
40
(p. 414) (CS4) PMU:
24 September 2012 3:53 PM
415
Discourse relations in English and German discourse 1
The socio-cognitive construct of coherence is connected intrinsically with
2 ­cohesion and cohesive ties, viz. linguistic items which express the nature of
3 the connectedness between clauses and sentences, sentences and paragraphs,
4 and paragraphs and discourse as a whole (Hasan and Halliday 1987; Halliday
5 1994). In general, discourse contains numerous cohesive ties, but there are
6 ­discourses that do not contain any cohesive ties but are considered to be co­
7 herent, and there are discourses that display numerous cohesive ties but are
8 ­considered to be incoherent. Both kinds can be found in literary discourse and
9 are constitutive, e.g., comedy, where discourse coherence is construed on a
10 meta-level. However, there is no coherent discourse without coherence strands,
11 to use a term from Givón (1993), that are referential continuity, temporal continu12 ity, spatial continuity, and action continuity. The communicative value of dis-
13 course relations can be implicit in these coherence strands, and it can be marked
14 overtly by using cohesive ties. It is the linguistic coding of discourse relations
15 with discourse connectives and meta-communicative comments in adjacently
16 and non-adjacently positioned discourse units, which is at the heart of our
17 analysis.
18
The goal of this paper is to analyze the linguistic coding of the discourse rela-
19 tions Continuation, Contrast, Elaboration, Explanation, and Comment in the
20 ­discourse genre of editorial, paying particular attention to their overt coding by
21 discourse connectives and meta-communicative comments in those contexts in
22 which they are positioned adjacently and others in which they are positioned
23 non-adjacently.2 To avoid possible overgeneralizations, a comparative analysis
24 of German and British English editorials has been undertaken as the linguistic
25 coding of discourse relations may well be language-preferential, if not language26 specific.
27
The discourse genre of editorial is an argumentative genre par excellence. In
28 argumentation theory, argumentation is assigned a dual status. It refers to the
29 process of calculating intra-subjective meaning (Anscombe and Ducrot 1983),
30 and it refers to an intersubjective activity, in and through which situated com­
31 municative meaning is negotiated, and discourse coherence is construed accord-
32 ingly. Argumentation is thus assigned a key function in the internal and external
33 relationships between premises and conclusions. Moreover, the discourse genre
34 of editorial is also a persuasive genre and that is why we expect its authors
35 to strategically employ various cohesive ties that may signify their preferred
36 interpretations. The overt representation of discourse connectives and meta-­
37 communicative comments is thus expected to be of great importance, as they
38
39
40 2 These relations are the most frequent ones in the data under investigation.
. 414) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 415–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 415)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
416 make the intersubjective processes of reasoning explicit, signaling how the
­author intends her/his local contributions and the overall editorial to be taken
and how the reader is intended to interpret it. Since the object languages English
and German of our contrastive analysis are both Germanic languages and thus
related quite closely, we may expect the use of similar cohesive ties for the overt
coding of discourse relations in similar contexts, yet in a different fashion (but
see Clyne 1987; Fetzer 2005, 2008; House 1996).
The methodological framework of our contrastive discourse-based corpus
analysis is an integrated one, supplementing the Segmented-Discourse-­
Representation–based definition of discourse relation (Asher and Lascarides
2003) with the Systemic-Functional-Grammar concepts of multiple themes and
thematic progression, and applying them to a quantitative and qualitative corpus analysis using the pragmatic tools of inference and implicature, and the
­discourse-analytic tools of sequencing and coherence. Context is accommodated
explicitly in the analysis: Social context is accounted for through the discourse
genre of editorial, linguistic context is accounted for through adjacency, and cognitive context is accounted for through inference.
The paper is organized as follows: Section 2 examines discourse relations
and thematic progression, section 3 contextualizes the concept of adjacency, and
section 4 presents the contrastive analysis. Section 5 summarizes the most important findings.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
2 Discourse relations and thematic progression
Discourse relations are of key importance for the construal of discourse coherence as they do not only express the nature of the connectedness between the
constitutive units of discourse but also signal their sequential ordering with
­respect to chronology and/or logic. The relation between discourse units may
be represented overtly through discourse connectives or meta-communicative
comments, and it may be represented non-overtly through coherence strands,
such as referential continuity, temporal continuity, spatial continuity, and action continuity. Discourse semantics distinguishes between coordinating discourse relations and subordinating discourse relations whose definition is not
based on syntax but rather on their semantics and the pragmatics of information
packaging. Prototypical coordinating discourse relations are, e.g., Narration,
Continuation, and Contrast, and prototypical subordinating relations are Elab­
oration, Explanation, and Comment, to name the most prominent relations as
put forward by Segmented Discourse Relation Theory (Asher and Lascarides
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 416–452 2655_9-4_01
PMU:(idp) 24/9/2012
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
(p. 416) (CS4) PMU:
24 September 2012 3:53 PM
417
Discourse relations in English and German discourse 1 2003). Segmented Discourse Relation Theory (SDRT) is anchored firmly to the
2 framework of Discourse Semantics and bases its definitions on the unit of seman-
3 tic proposition and its representation in discourse, viz. language-use-based
4 ­utterance.
In Systemic Functional Grammar (SFG), discourse relations have been exam-
5
6 ined in the framework of thematic progression, considering the structured inter7 play of theme and rheme, and their progression in discourse (Bloor and Bloor
8 1995; Halliday 1994). Theme and its refinement as multiple themes are anchored
9 firmly to the clause, and it is that unit of investigation that is going to be our bridg-
10 ing point between SFG and SDRT. In SFG every clause has thematic structure, and
11 theme is defined as its initial position, while the remainder of the clause is called
12 rheme. Initial positions are of key importance to the analysis of texture and dis-
13 course coherence. The initial position as the “point of departure of the message”
14 (Halliday 1994: 38) signifies how a preceding clause is to be taken and how the
15 discourse is to proceed. Put differently, themes express a connectedness between
16 what has just been said/written, thus realizing anaphoric reference, and at the
17 same time they expresses connectedness with what is going to be said/written
18 thus fulfilling cataphoric reference.
19
20
21
22
2.1 Discourse relations
23 Discourse relations (or, rhetorical relations) have been the subject of several dy24 namic semantic models, such as Rhetorical Structure Theory (Mann and Thomp25 son 1987), Discourse Representation Theory (Kamp and Reyle 1993) and, more
26 recently, SDRT (Asher and Lascarides 2003; Asher and Vieu 2005; Benz and
27 ­Kühnlein 2008). It is the latter framework by which our research is informed.
28 In SDRT, a Discourse Relation is a function that takes two propositions as its
29 ­arguments. A Discourse Relation is thus the logical connection between a propo30 sition π1 as part of a discourse D and some other proposition π2 in D. The prop­
31 ositions π1 and π2 stand in the Discourse Relation R iff the inferences the
32 hearer/reader makes and the logical connection s/he draws between π1 and π2
33 are in accordance with the ones defined for R. The definitions of the relations
34 relevant to our investigation are going to be illustrated with the following exam35 ple (1):
36
37 (1) a.π1: Uller came home early. π2: He called Ingbert.
38 39 40 b.π1: Uller came home early. π2: He was unusually tired.
c.π1: Uller came home early. π2: His meeting ended earlier than expected.
d.π1: Uller came home early. π2: That surprised his neighbors.
. 416) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 417–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 417)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
418 e.π1: Uller came home by car. π2: The car was an old Rover. π3: Uller was 1
very fond of it. π4: The first thing he did at home was drain a jug of 2
water.
3
Sequence (1a) exemplifies the relation Narration. The defining conditions are that
both utterances share a common topic and that there is temporal succession between π1 and π2; (Asher and Lascarides 2003: 162ff.). A relation is veridical if the
conjunction of both propositions involved in R does not lead to a contradiction,
as is the case here (Asher and Lascarides 2003: 56, 157). Narration is not very
­common in argumentative discourse, but it is considered to be the archetypical
coordinating relation and serves as the background against which Continuation,
which is a very prominent relation in argumentative discourse, is defined. Narration is illustrated by (1e).
(1b) is an example for Elaboration. In an Elaboration relation, π2 offers
­additional information about one of the referents in π1. This might be the topic
of π1, as in (1b), but need not be. There is no temporal sequence between π1
and π2, rather, π2 is temporally included in π1 (Asher and Lascarides 2003:
159ff.).
The sequence (1c) is an example for Explanation. This relation is a special
case of Elaboration. Here, π2 provides the cause or reason for π1, or at least for
a part of π1. As with Elaboration, π2 is temporally included in π1 (Asher and Lascarides 2003: 159ff.).
In (1d) the relation Comment is illustrated. It is defined as π2 selecting π1 as
topic. It is referred to with the symbol “→” in Asher and Lascarides (2003: 146ff.).
In our research, we use the term Comment for this relation.
In (1e), we have a larger chunk of discourse. π1 and π2 are in an Elaboration
relation. The topic of π2 is not the same as in π1, but that does not cause any problems for an Elaboration relation. Looking further into the discourse, the relation
between π2 and π3 is not one of a temporal sequence, nor is π3 an Elaboration of
π2. Rather, π3 continues the sequence of action opened up by π2, and both elaborate on π1. Such a relation is called Continuation (cf. Asher and Lascarides 2003:
461). Continuations share a common topic, as do Narrations, but they do not involve a temporal sequence.
The sequence (1e) shows that SDRT is based on the premise that discourse
has a multi-layered structure. In (1e), π1 and π4 share the same topic, they involve
a temporal consequence, and they are veridical. Hence, they are in a Narration
relation and thus on the same level of discourse. This is not the case with π2 and
π3, which do not develop the story line of the discourse any further. Rather, they
suspend the main narrative by adding information about Uller’s car. They have
an insertion-like status, and the inserted information constitutes a sub-discourse,
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 418–452 2655_9-4_01
PMU:(idp) 24/9/2012
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
(p. 418) (CS4) PMU:
24 September 2012 3:53 PM
419
Discourse relations in English and German discourse 1 which depends on the main discourse but not vice versa. This can be represented
2 graphically as in (2):
3
4 (2) π1 Narration
π4
5
6
Elaboration
7
8 9
π2 Continuation
π3
10 In SDRT, and in other theories of discourse, discourse is thus not a one-­
11 dimensional string of utterances, which progresses thematically, but rather a
12 ­hierarchically organized system, as has already been pointed out by Grosz and
13 Sidner (1986), for instance. This is of prime importance to our contrastive analysis
14 of the overt and non-overt representation of discourse relations in adjacent and
15 non-adjacent positioning, as it entails that discourse relations do not only hold
16 between directly adjacent propositions, but also between non-adjacently posi-
17 tioned propositions, as has been the case with the Narration relation between two
18 distant propositions, namely π1 and π4 in (1e).
19
There is also ample evidence for non-adjacently positioned discourse rela-
20 tions in naturally occurring discourse. From a theory-driven perspective it would
21 be reasonable to assume that non-adjacently positioned coordinating and sub­
22 ordinating discourse relations are represented overtly by discourse connectives
23 or meta-communicative comments to ensure felicitous communication. As for
24 (1e), the Narration relation holding between π1 and π4 could be represented
25 ­overtly by the discourse connectives then, and then or simply and with the impli26 cature “chronological concatenation.” The subordinating relation Elaboration
27 holding between π1 and π2 could be represented overtly with a non-defining rela-
28 tive clause and the pronoun which or that, and the Continuation relation between
29 π2 and π3 could be represented by the discourse connective and. As the discourse
30 relation of Narration is positioned non-adjacently, a discourse connective or
31 ­meta-communicative comment would facilitate the construal of discourse coher32 ence. The non-adjacently positioning is indicated by the cohesive link “the first
33 thing,” implying a chronological sequence that is a necessary condition for the
34 definition of Narration.
35
The semantics of discourse relations can be represented overtly through dis-
36 course connective or meta-communicative comments, and it can be represented
37 non-overtly. In the latter case, the relation needs to be inferred. Sometimes, how-
38 ever, the speaker /writer provides explicit hints, or contextualization cues in
39 ­interactional-sociolinguistic terminology, how s/he intends the reader/hearer to
40 connect the propositions. These “hints” can be represented through word order,
. 418) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 419–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 419)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
420 especially the sequential organization of the “theme zone” (Fetzer 2008), or
through lexical means, for instance discourse connectives; particles or adverbs,
such as denn, aber, dazuhin, etc. in German; or because, but, moreover, etc. in
English, which are often represented in the initial position or theme zone, as discussed below.
1
2
3
4
5
6
7
2.2 Multiple themes and thematic progression
8
Discourse relations have been examined from a both structural and discoursesemantic perspective in systemic functional grammar (SFG) as regards cohesion
and thematic progression (Bloor and Bloor 1995; Halliday 1994). SFG is anchored
to a tripartite system of experiential, interpersonal, and textual metafunctions.
The experiential metafunction looks upon the clause as representation and is
based on its semantic representation within a system of transitivity. The interpersonal metafunction considers the clause as exchange and is based on its modal
representation within a system of mood, and the textual metafunction looks
upon the clause as message and is based on its bipolar conception as theme
and rheme and their structured interplay within a system of thematic structure.
While the experiential and interpersonal metafunctions are primarily discoursesemantic in nature, the textual metafunction is both syntactic and discourse-­
semantic considering continuative, structural, and conjunctive phenomena.
In SFG every clause has thematic structure, and theme is defined as its initial
position, while the remainder of the clause is called rheme.3 In line with the three
metafunctions, the theme has been further categorized as topical (or experiential) theme, interpersonal theme and textual theme, which are subsumed under
the header of multiple themes. A topical theme is a necessary constituent in the
configuration of a clause. It is defined as the first element in the clause carrying
ideational meaning and can be seen as functionally equivalent to topic in the
topic-comment paradigm. Textual and interpersonal themes are optional elements in the configuration of a clause. Regarding their status in a discursive
frame of reference, however, they need to be considered as necessary parts.
Hence, topical themes, which are underlined in (3) and (4), need to be repre­
sented overtly, while textual themes printed in bold, and interpersonal themes
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
3 The definition of theme and rheme in SFG is based on syntax only. For this reason, it is not a
functional synonym for the information-structure and information-packaging based concept of
topic, which encodes a relation of aboutness in discourse, and its counterpart, the comment.
Theme and topic, and rheme and comment may conflate, but they do not need to (cf. GómezGonzález 2001; Krifka and Féry 2008).
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 420–452 2655_9-4_01
PMU:(idp) 24/9/2012
36
37
38
39
40
(p. 420) (CS4) PMU:
24 September 2012 3:53 PM
421
Discourse relations in English and German discourse 1 printed in bold italics, that is the textual themes but, so that, and, that, and
2 while in (3) and (4), and the interpersonal themes naturally and in reality in (4),
3 can be represented overtly or non-overtly, as is illustrated with the following ex4 amples (3) and (4) from the Corpus of British editorials. If the textual and inter-
5 personal themes were represented non-overtly, the propositional content of the
6 clauses would not change:
7
8 (3) But the reality is that Mr Blair’s eyes are not set on 2016 but on getting past
9
10
11
12
13
the conference season and surviving into October. He appears prepared to
pay almost any price to retain his vestigial authority so that he can leave at a
moment of his own choosing next year. This may look reasonable to Mr Blair
and to those around him. But it makes precious little sense any longer to
anyone else. (blair)
14
15 (4) Naturally, Drax’s scheme is foiled by James Bond. The paradox is that in re16
17
18
ality the Drax that is doing the most to destroy the world is the giant coal-fired
power station of the same name in Yorkshire – while it is nuclear technology
that to many people’s minds stands ready to save the day (environment)
19
20 Based on the structured interplay of themes and rhemes, thematic progression
21 has been further refined with respect to more linear and more hierarchical order-
22 ings of discourse, viz. constant theme patterns, linear theme patterns, split rheme
23 patterns, and derived themes (Bloor and Bloor 1995). Constant-theme-patterned
24 discourse and linear-theme-patterned discourse are straightforward unfolding
25 types of discourse with chronological and logically-ordered story lines, and split26 rheme-patterned discourse and derived-theme-patterned discourse are more
27 complex types of discourse and may display non-chronological story lines with
28 sub-discourses. In real-world discourse, the two types of thematic progression
29 need to obtain a scalar interpretation with more or less chronologically and logi-
30 cally ordered discourses, as is the case with excerpts (3) and (4), and more com31 plex types of discourse.
32
33
34
35
36
37
38
39
40
Thematic structure refers to the structured interplay between theme and
rheme, which is based on direct adjacency, while multiple themes and thematic
progression feed on both directly adjacent and non-adjacent relations. The latter
provide another bridging point between SDRT and its conception of Discourse
Relations, and is thus of relevance to our empirical analysis of British and ­German
editorials.
In the following the fundamental concept of adjacency, which has so far been
used in its structural meaning only, will be further refined for our contrastive
analysis.
. 420) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 421–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 421)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
422 3 Adjacency
1
In a pragmatics-based theory of discourse, adjacency is one of the most fundamental discursive relations, which is necessary for accounting for inference processes of local and not-so-local discourse units, whose order of inclusion corresponds to the order of accessibility (cf. Sperber and Wilson 1986). At first sight,
adjacency seems to be a fairly straightforward notion, which is indispensible to
linearization in general and to the linearization of discourse in particular. From
a context-based perspective, however, adjacency is rather complex comprising
adjacency position, adjacency relation and adjacency expectation (cf. Levinson
1983; Schegloff 1995).
Adjacency position is a structural notion that occurs at any stage in the process of linguistic linearization. It has been analyzed thoroughly in the research
paradigm of ethnomethodological conversation analysis with respect to the sequential organization of conversation from both global and local perspectives
(cf. Sacks 1995), and it can also be applied to written discourse, such as the ones
investigated here. The former describes conversational patterns in adjacently
­positioned opening sections and topical sections, and topical sections and closing sections, and the latter is anchored to the local turn-taking mechanism and
the basic unit of adjacency pair, that is patterned co-occurrences of two social
actions produced by different speakers, such as greeting and greeting, request
and acceptance/refusal, offer or invite and acceptance/refusal, assessment and
agreement/disagreement, and question, and expected answer/unexpected answer or non-answer (cf. Levinson 1983: 336). The second parts of the adjacency
pairs just listed are not of equal standing, as one of them is preferred, and the
other is dispreferred, as has been examined in the framework of preference organization by Pomerantz (1984), for instance. The classification as preferred or dispreferred second is not based on the interlocutors’ psychological disposition, but
rather on structural and distributional features and hence closely connected with
the linguistic concept of markedness (cf. Levinson 1983: 307).
Adjacency relation goes beyond structure-based positioning. It is a pragmatic
concept that may be encoded in discourse and thus made explicit, or it may be
assigned a presuppositional status and thus needs to be inferred. Adjacency relation may have a narrow scope and be assigned the status of a local constraint, as
is the case with adjacency pairs and their preferred and dispreferred seconds, and
it may have a wider scope and be assigned the status of a less-local constraint, as
is the case with insertion sequences and topical digression, and pre- and postsequences in conversation. Adjacency relation is of great importance for the dynamics of discourse as well as for thematic progression. It is of prime importance
for the differentiation between coordinating and subordinating discourse rela-
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 422–452 2655_9-4_01
PMU:(idp) 24/9/2012
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
(p. 422) (CS4) PMU:
24 September 2012 3:53 PM
423
Discourse relations in English and German discourse 1 tions in SDRT, as has been shown in the analysis of example (1). Closely related to
2 the concept of adjacency relation is the notion of adjacency expectation.
3
The cognitive concept of adjacency expectation is a discourse notion par ex-
4 cellence. It is the foundation against which two adjacent utterances may be clas5 sified as a particular adjacency pair with preferred and dispreferred seconds, or
6 against which the second social action may be assigned the status of the first
7 move of an insertion sequence. Moreover, in the discourse genre of interview so-
8 cial actions performed by the interviewer tend to count as questions and social
9 actions performed by the interviewee tend to count as answers. Adjacency expec-
10 tation is also of importance to discourse relations and their concatenation, and,
11 as we will show, to the overt and non-overt representation of discourse connec12 tives and meta-communicative comments.
13
Adjacency does not only comprise the conversation-analytic conception of
14 adjacency holding between turns, that is adjacency pair/position/relation/­
15 expectation. It may also refer to the turn-internal organization or discourse-­
16 internal sequential organization between discourse units (or parts), and between
17 discourse units and discourse as a whole. It is that discourse-internal anchored
18 conception of adjacency, which is of key importance to our analysis of discourse
19 relations. This is because syntax-based adjacency, that is adjacency position,
20 does not only open up a structural slot for a prior discourse unit and for a suc21 ceeding discourse unit. It also signals a discourse-semantic type of connected22 ness, as is reflected in the discourse relations of Continuation or Contrast, for in-
23 stance. The relation of Contrast can be made explicit by the overt representation
24 of an argumentative move, e.g., in the following I am going to deconstruct your ar­
25 gument, it can be represented by a discourse connective, e.g., but, and thus left
26 underspecified, and it can be left empty. As regards the latter two modes of repre-
27 sentation, the discourse relation needs to be pragmatically enriched through
28 ­inference.
29
Building on the tripartite differentiation of adjacency as adjacency position,
30 adjacency relation and adjacency expectation, we hypothesize that there are pre31 ferred contexts in which the semantics of a discourse relation is made explicit by
32 the overt representation of an argumentative move or by a discourse connective.
33 Those contexts, we assume, are defined by the constraint of structural non-­
34 adjacency. Spelling out the nature of the connectedness between non-adjacently
35 positioned discourse relations facilitates discourse production and discourse
36 processing. Against that background, discourse relations, which are anchored to
37 two directly adjacent discourse units and in which adjacency position and adja-
38 cency relation conflate, tend to be a straightforward matter with respect to pro39 duction and processing. They can generally be processed without the accommo-
40 dation of extra contextual information, and the information contained in them
. 422) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 423–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 423)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
424 and communicated through them can be attributed directly to discourse common
ground. In that scenario, the type of discourse relation is usually not coded
­overtly but rather is implicit. It thus needs to be inferred from the local linguistic
context coded in the semantics of the lexical units and the syntactic configuration
of the discourse unit.
In discourse, it is also possible that adjacency position neither conflates with
adjacency relation nor with adjacency expectation. In that kind of scenario, we
assume that discourse relations tend to be represented overtly in order to facilitate discourse production and discourse processing (cf. Liedtke 1997). Against
that background, discourse connectives may be assigned the status of some kind
of indirect directive, requesting the hearer/reader to perform inferences of a certain kind. For instance, the discourse connective but may signify an upcoming
contrastive context and request the hearer to perform the corresponding inferencing processes to accommodate the incongruent information introduced by “but.”
Or, the discourse connective in addition may signify another argument in a line of
arguments with a stronger force, requesting the hearer to perform the corresponding processes of inferencing.
To shed more light on the theoretical concepts of discourse and discourse
relation, and on their representation in particular discourses, a contrastive analysis of editorials is undertaken in order to avoid a possible bias resulting from
language-specific preferences for coding discourse relations overtly and nonovertly. We argue that the overt marking of a discourse relation does not only depend on its semantics but also on its locality, as is reflected in its structural positioning as directly adjacent and as non-directly adjacent.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
4 Contrastive analysis of British and
German editorials
27
28
29
In contrastive analysis, “any two objects can be compared with respect to various features, and they may turn out to be similar in some respects but different
in ­others” (Krzeszowski 1989: 60). For instance, the contrastive conjuncts but
and its German counterpart aber may have similar sentential positions in English and German, but they may be different in their distribution in negative
and non-­negative contexts. To be compared in a felicitous manner, the phe­
nomena at hand need to have at least some features of similarity (Chesterman
1998).
This study of discourse relations in English and German argumentative discourse starts off with the quantitative analysis of the two sets of data. Adopting
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 424–452 2655_9-4_01
PMU:(idp) 24/9/2012
30
31
32
33
34
35
36
37
38
39
40
(p. 424) (CS4) PMU:
24 September 2012 3:53 PM
425
Discourse relations in English and German discourse 1 the three classical steps of description, juxtaposition and comparison (Krzes­
2 zowski 1989: 57) to identify cross-linguistic similarities and differences, we addi-
3 tionally examine the embeddedness of the discourse relations in their local and
4 not-so-local contexts, paying particular attention to adjacency. The two sets of
5 data share similar contextual features: They are instances of written argumenta6 tive media discourse.
7
The data under investigation comprise a corpus of 24 written editorials: nine
8 British editorials adopted from the quality newspaper The Guardian with 4,826
9 words, 192 sentences, and 596 clauses, and 15 German editorials taken from the
10 quality newspaper Frankfurter Rundschau with 4,784 words in 258 sentences. The
11 data are manually tagged with regard to the relevant linguistic structures and
12 then hand-counted in order to capture subtle aspects of analysis. We fully agree
13 with Hunston (2007: 28) that
14
15
16
17
18
19
20
quantitative methods are not irrelevant to discourse studies, in the sense that recurring instances of a phenomenon are noted, the explication of a single instance normally implies
that a pattern has been identified, and that the explanation would hold true for similar instances. This is the case even when the amount of data collected is relatively small; quanti­
tative does not mean huge, but simply that statements of the type “this is a demonstrably
typical occurrence” are worth making. The second assumption is that, on the other hand,
research in the area of discourse will never be wholly quantitative.
21
22 In spite of the two sets of data under investigation being not excessively “huge,”
23 we expect patterned co-occurrences of discourse connectives depending on
24 (1) the semantics of the discourse relations, and (2) adjacent and not-adjacent
25 positioning.
26
27
28
29
4.1 The British editorials
30 The British data comprise nine editorials with an overall of 192 sentences (S) with
31 a mean of 21.33 sentences per text. They contain an overall of 596 clauses (C) with
32 a mean of 66.22 clauses per text. The quantitative analysis of the British editorials
33 has focused on the identification of the coordinating discourse relations of Con­
34 tinuation (contin) and Contrast (contr), and on the subordinating discourse rela-
35 tions of Elaboration (elabor), Explanation (explan), and Comment (comm). For the
36 unit of investigation of S, the following results summarized in Table 1 have been
37 obtained. The most frequent relations are printed in bold.
38
In six of nine editorials, the discourse relation of Continuation is the most
39 frequent one, while in the other editorials the subordinating discourse relation of
40 Elaboration is more frequent in two texts, and the discourse relation of Comment
. 424) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 425–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 425)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
426 Text
S
Lebanon
Iran
Turkey
Katrina
Ceasefire
Environment
Obesity
France
Blair
∑
Contin
Contr
Elabor
Explan
Comm
1
15
15
16
28
26
25
24
16
27
6
5
6
9
9
15
11
11
6
3
1
2
1
4
2
3
1
3
4
4
3
14
10
6
6
2
6
0
2
3
0
0
0
0
0
3
1
2
1
3
2
1
3
1
8
2
192
78
20
55
8
22
10
3
4
5
6
7
8
9
11
12
Table 1
13
14
in the third text. Explanations are used in three editorials only. As regards distribution across the British data, there are 40.6% Continuations, 28.6% Elaborations, 10.4% Contrasts, 11.4% Comments, and 4.1% Explanations, which will be
discussed below.
The distribution of discourse relations across the editorials provides some
interesting tendencies indicating a preference of the discourse relations of Continuation and Elaboration. However, it is their overt (ov) and non-overt (non-ov)
representation that is of prime interest to our contrastive analysis of British and
German texts. This is systematized in Table 2:
15
16
17
18
19
20
21
22
23
24
25
Text
Lebanon
Iran
Turkey
Katrina
Ceasefire
Environment
Obesity
France
Blair
∑
%
S
Contin
Contin
Contr
Elabor
Elabor
Explan
Comm
Ov
Non-ov
Ov
Ov
Non-ov
Non-ov
Ov
15
15
16
28
26
25
24
16
27
2
2
1
1
1
6
0
1
0
192
14
17.9
4
3
5
8
8
9
11
10
6
64
82.1
3
1
2
1
4
2
3
1
3
20
100
Comm
26
27
Non-ov 28
0
2
0
5
3
2
1
1
1
4
2
3
9
7
4
5
1
5
0
2
3
0
0
0
0
1
3
0
0
0
2
0
0
1
0
2
1
2
1
1
2
1
2
1
6
29
15
27.2
40
72.8
8
0
5
22.7
17
77.3
37
30
31
32
33
34
35
36
38
39
Table 2
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 426–452 2655_9-4_01
PMU:(idp) 24/9/2012
40
(p. 426) (CS4) PMU:
24 September 2012 3:53 PM
427
Discourse relations in English and German discourse 1 Research on the primarily overt representation of the discourse relation of
2 ­Contrast in English discourse (Doherty 2003; Fetzer 2008) is confirmed by
3 our analysis, where not a single occurrence of non-overt representation has
4 been found. All of the other discourse relations under investigation are repre-
5 sented more frequently in a non-overt manner, and the discourse relation of
6 ­Explanation with its function of signifying causality is represented non-overtly
7 only.
8
9
10
11
12
13
In extract (5) the discourse relation of Continuation is represented overtly by
the adverbial too signifying a continuation of the discourse topic British govern­
ment and its particularization Tony Blair especially taken up by the second political force the Conservatives, while a continuation of the discourse topic the role of
local council is represented implicitly in extract (6) with co-reference encoded in
the pronoun they:
14
15 (5) The British government, and Tony Blair especially, may talk a better game in
16
17
18
19
20
tackling climate change than its final actions reveal, but no one can seriously
accuse it of not appreciating the danger. The Conservatives too are at pains to
demonstrate that they are concerned – even displaying the rare sight of a Tory
shadow chancellor offering to raise taxes, as George Osborne did yesterday, to
aid the environment. (Environment)
21
22 (6) This is where local councils can help. They already consider applications for
23
24
alcohol licences carefully, and will soon have greater powers to reject poorly
designed planning applications. (Obesity)
25
26 Extract (7) illustrates the discourse relation of Contrast, which is represented
27 overtly throughout the data. Here it is represented by the discourse connective
28 but which indicates a contrast between Mr Bush’s visit and the success of his
29 government in rebuilding the region indexed by little sense:
30
31 (7) Yesterday, with midterm elections in the offing, Mr Bush spent rather more
32
33
34
35
time in New Orleans than he managed to do a year ago during the disaster itself. But there is little sense in most parts of the afflicted Gulf coast, and in
New Orleans itself in particular, that public officials from the president down
have yet got a grip of the situation that confronts them. (Katrina)
36
37 In extract (8) the discourse relation of Elaboration is represented non-overtly, and
38 the reader needs to infer the relation between the discourse topic European Union
39 and its role in the Middle East, and the EU member Germany, and its offer to send
40 troops. In extract (9), the Elaboration of the discourse topic Hizbullah as the victor
. 426) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 427–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 427)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
428 is represented overtly by the adverbial as such which particularizes the role of 1
the Bekaa raid from the perspective of Israel:
2
(8) The European Union has performed better in the peace than it did during the
fighting, overcoming its reticence and divisions to pledge thousands of troops
for an expanded UN force that will be deployed along the dangerous border
with Israel. Germany’s offer of 1,500 personnel, albeit not ground forces, is a
remarkable advance. (Lebanon)
(9) There is no doubt that Hizbullah sees itself as the victor in the conflict and this
in itself is a provocation to an Israeli government which is under pressure
at home to show it can still protect its territory. As such, the Bekaa raid
may have been a symbolic exception to a policy of compliance with the UN
resolution, rather than a disturbing indication of flagrant breaches to come.
(Lebanon)
Extracts (10) and (11) are illustrations of the discourse relation of Comment. In
(10) the whole discourse unit is referred to by the pronoun this, which provides
the point of departure for the writer’s comment that the scenario described is very
tempting. In (11) the Comment is represented overtly by lexical coherence coded
in the heavy NP halting climate change, which comments on the whole discourse
unit:
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
(10) The journey home from work or school is fraught with unhealthy temptations:
24
crisps from the corner shop, a fried chicken takeaway, a call to the pizza
25
company. This is when resolve is at its weakest. (Obesity)
(11) The 1987 Montreal protocol helped ban damaging halogenated hydrocar­
bons, and Nasa says that the layer is mending itself. Halting climate
change may not be as simple, but it is easier when the political will is there.
(Environment)
Extract (12) exemplifies the discourse relation of Explanation, which is repre­
sented non-overtly throughout the data. The pronominal reference it, which indexes Iraq in the prior context, is taken up by Mahmoud Ahmadinejad, the “hardline president,” and the predication has been conducting and has been paid
represented in the perfective and progressive aspect, implicating temporal precedence, which is a constitutive feature of Explanation:
26
27
28
29
30
31
32
33
34
35
36
37
38
(12) It is certainly operating, as Chatham House puts it in a timely new report, in 39
an atmosphere of “confident ease.” Mahmoud Ahmadinejad, the hardline 40
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 428–452 2655_9-4_01
PMU:(idp) 24/9/2012
(p. 428) (CS4) PMU:
24 September 2012 3:53 PM
429
Discourse relations in English and German discourse 1
2
3
president, has been conducting an energetic charm offensive in which much
attention has been paid to his twinkling eyes and domestic popularity.
(Iran)
4
5 As regards distribution across the British editorials, there are 17.9% overt and
6 82.1% non-overt Continuations, 27.2% overt and 72.8% non-overt Elabora-
7 tions, and 22.7% overt and 77.3% non-overt Comments with a mean of 1.55 overt
8 and 7.11 non-overt Continuations, 2.22 Contrasts, 1.66 overt and 4.44 non-overt
9 Elaborations, 0.88 non-overt Explanations, and 0.55 overt and 1.88 non-
10 overt Comments.
11
The primarily quantity-based analysis of the overt and non-overt representa-
12 tion of discourse relations is refined by the explicit accommodation of context,
13 investigating the question whether directly adjacent (ADJ) or non-directly adja14 cent (-ADJ) positioning of sentences connected with a particular discourse rela-
15 tion has a decisive influence on its overt or non-overt representation. The results
16 obtained are systematized in Table 3.
17
In the data at hand only the coordinating discourse relation of Continuation
18 is positioned non-adjacently across all of the editorials. There are 38.4% non-­
19 adjacent and 61.6% adjacent Continuations with 75% of the non-adjacent Con-
20 tinuations represented non-overtly. The subordinating discourse relations of
21 Comment and Elaboration are positioned non-adjacently in one editorial each. As
22 regards their distribution across the editorials, there are 4.5% non-adjacent and
23 95.5% adjacent Comments with none of the non-adjacent Comments represented
24 overtly, and 1.8% non-adjacent and 98.2% adjacent Elaborations with 100% of
25 the non-adjacent Elaborations represented overtly.
26
For Continuation and Comment, the non-overt representation is preferred,
27 and for Elaboration, the overt representation is preferred. Analogously to the pre28 ferred non-overt representation of the discourse relations of Continuation (82.1%)
29 and Comment (77.3%), Continuations and Comments are not signaled by the
30 overt representation of a discourse connective when positioned non-adjacently
31 either. Their kind of connectedness is rather signaled by lexical coherence, as
32 is the case in (14). In (13), the non-adjacent discourse relation is represented
33 overtly.
34
In extract (13), the first discourse unit dealing with the slow progress of re-
35 building New Orleans and possible reasons for the non-success is elaborated on
36 in the immediately following units providing exact numbers and reasons for the
37 non-progress, viz. the corrupt government. The discourse topic about possible
38 reasons for the non-success in (#1) is continued in the non-adjacent unit (#4)
39 which is introduced by the discourse connective nevertheless realizing ana­
40 phoric reference with the discourse topic and its elaborations:
. 428) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 429–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 429)
24 September 2012 3:53 PM
Table 3
∑(%)
∑
Obesity
France
Blair
Environment
Ceasefire
Turkey
Katrina
Iran
Lebanon
Text
192
24
16
27
25
26
16
28
15
15
S
17.9% ov
78
11
11
6
15
9
6
9
5
6
Contin
24 non-ov
6 ov
25% ov
2 non-ov
1 ov
1 non-ov
1 ov
2 non-ov
5 non-ov
1 ov
4 non-ov
1 ov
2 non-ov
2 ov
4 non-ov
2 non-ov
2 non-ov
-ADJ
Contin
22.7% ov
22
3
1
8
1
2
1
3
2
1
Comm
0% ov
1 non-ov
1 non-ov
-ADJ
Comm
55
6
2
6
6
10
3
14
4
4
Elabor
100% ov
1 ov
1 ov
-ADJ
Elabor
8
0
0
3
0
3
2
0
Explan
20
3
1
3
2
4
2
1
1
3
Contr
430 Anita Fetzer and Augustin Speyer
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 430–452 2655_9-4_01
PMU:(idp) 24/9/2012
38
39
40
(p. 430) (CS4) PMU:
24 September 2012 3:53 PM
431
Discourse relations in English and German discourse 1 (13) (#1) In a devastated city and a region that are crying out for steady incremen2
3
4
5
6
7
tal progress and planning, improvements have been painfully slow and much
has already stalled. (#2) Of the much touted $110bn of federal aid to the region, only $44bn has yet been handed over. (#3) Louisiana and New Orleans
are bywords for corrupt government and failed politics, so not all of this can
be laid at Mr Bush’s door. (#4) Nevertheless, conspiracy theories abound.
(Katrina)
8
9 In extract (14), the non-adjacent discourse relation of Continuation, which holds
10 between discourse unit (#1) and (#4), is represented non-overtly. While dis-
11 course unit (#2) elaborates on (#1), and (#3) comments on its preceding unit,
12 (#4) picks up the discourse topic (#1) and continues with the story line. Because
13 of the non-overt representation of the relation holding between (#1) and (#4),
14 the reader needs to infer the type of connectedness which is signified by lexical
15 coherence, viz. the meronomy relation holding between anti-obesity drive and
16 start:
17
18 (14) (#1) “There’s only so much the government can do,” the health secretary said
19
20
21
22
23
last week as she revealed the alarming scale of British obesity. (#2) “People
need to want to change their lifestyles and take responsibility for their
health.” (#3) Yet there is a great deal the government can do to help them
without falling foul of the familiar charge of nannying. (#4) Labelling super­
market food is a start. (Obesity)
24
25 Functional-grammar–based research on discourse (Givón 1993; Halliday 1994)
26 has informed us that the unit of investigation of sentence is rather controversial
27 for a discourse-based analysis of English. This is why a textual-metafunction
28 based analysis of thematic progression and of the linguistic representation of the
29 theme zone is anchored to the clause. In both functional and descriptive research
30 paradigms of English, “sentence” is more of an orthographic unit than a syntactic
31 or semantic one. Against this background, the data at hand are further analyzed
32 with respect to discourse relations holding between the syntactic units of clause
33 (c). The results are systematized in Table 4.
34
In the sentence-based analysis, the discourse relation of Continuation has
35 been the most frequent one in six of nine editorials. In the clause-based analysis,
36 it is the discourse relation of Elaboration in all nine editorials. As regards its dis-
37 tribution, there are 54.5% Elaborations, 26.6% Continuations, 6.7% Contrasts,
38 6.7% Explanations, and 4.5% Comments. As above, it is their overt and non-overt
39 representation that is of prime importance to our contrastive analysis of British
40 and German texts. The results are presented in Table 5.
. 430) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 431–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 431)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
432 Text
Lebanon
Iran
Turkey
Katrina
Ceasefire
Environment
Obesity
France
Blair
∑
C
Contin
Contr
Elabor
Explan
Comm
1
56
57
50
70
84
79
68
44
88
10
17
10
19
19
32
19
12
21
5
2
2
4
5
7
3
3
9
28
30
31
40
53
38
42
16
47
11
3
4
2
1
1
0
12
6
1
4
3
4
5
1
3
1
5
2
596
159
40
325
40
27
10
4
5
6
7
8
9
11
12
Table 4
13
Analogously to the sentence-based analysis of the discourse relation of Contrast above, Contrasts are represented only overtly on the level of clause, while
almost all of the other discourse relations under investigation are represented
more ­frequently in a non-overt manner. Only the discourse relation of Elaboration is represented more frequently in an overt manner. Across the editorials,
there are 76.9% overt and 23.1% non-overt Elaborations, 26.4% overt and 73.6%
non-overt Continuations, 37.5% overt and 62.5% non-overt Explanations, and
14.8% overt and 85.2% non-overt Comments. As regards their overall mean, there
are 4.66 overt and 13 non-overt Continuations, 4.44 Contrasts, 27.77 overt and 8.33
non-overt Elaborations, 1.66 overt and 2.77 non-overt Explanations, and 0.44
overt and 2.55 non-overt Comments.
The primarily quantity-based–analysis of the clauses and their functions in
discourse has been further refined with respect to their status as finite (F) and
non-finite (NF) in order to identify possible patterned co-occurrences of discourse
relations in finite, respectively non-finite contexts. Furthermore, the question
whether directly adjacent positioning or non-directly adjacent positioning of finite and non-finite clauses connected with a particular discourse relation has a
decisive influence on their overt or non-overt representation has been examined.
Table 6 systematizes the results obtained for non-adjacent positioning as well
as for the distribution of discourse relations across finite and non-finite contexts. The discourse relations of Continuation, Comment, Elaboration, Explanation, and Contrast have been distributed across finite and non-finite contexts,
but only Continuations, Explanations, and Elaborations occur in non-adjacent
positioning.
As has been the case with the unit of investigation of sentence, only the coordinating discourse relation of Continuation is positioned non-adjacently across
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 432–452 2655_9-4_01
PMU:(idp) 24/9/2012
3
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
(p. 432) (CS4) PMU:
24 September 2012 3:53 PM
30
31
32
34
35
37
38
39
40
Table 5
28
29
40
100%
26
117
73.6%
27
36
596
25
∑
∑(%)
24
5
2
2
4
5
7
3
2
9
23
8
12
8
15
13
16
18
10
17
42
26.4%
21
2
5
2
4
6
16
1
2
4
20
56
57
50
70
84
79
68
442
88
19
Lebanon
Iran
Turkey
Katrina
Ceasefire
Environment
Obesity
France
Blair
250
76.9%
24
22
28
27
42
26
33
14
34
17
Ov
16
Ov
75
23.1%
4
8
3
13
11
12
9
2
13
Non-ov
14
Non-ov
13
Ov
12
15
37.5%
0
2
2
1
0
1
0
7
2
Ov
25
62.5%
11
1
2
1
1
0
0
5
4
Non-ov
Explan
9
Explan
10
Elabor
8
Elabor
4
14.8%
0
0
0
2
0
0
1
0
1
Ov
Comm
6
Contr
5
Contin
23
85.2%
1
4
3
2
5
1
2
1
4
Non-ov
Comm
3
Contin
2
C
1
Text
Discourse relations in English and German discourse 433
4
7
11
15
18
22
33
. 432) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 433–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 433)
24 September 2012 3:53 PM
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 434–452 2655_9-4_01
PMU:(idp) 24/9/2012
NF
F
Ceasefire
NF
F
Katrina
NF
F
Turkey
NF
F
Iran
NF
F
Lebanon
Text
84
70
50
57
56
C
F
57
53
30
38
31
27
17
20
19
25
NF
19
19
10
17
10
Contin
10 non-ov
5 ov
7 non-ov
1 ov
5 non-ov
2 ov
3 non-ov
2 ov
3 non-ov.
2 ov
-ADJ
Contin
1 ov
4 ov
4 ov
2 ov
2 ov
5 ov
Contr
1 non-ov
1 non-ov
1 ov
2 non-ov
2 ov
1 non-ov.
2 overt
11 non-ov
Explan
-ADJ
Explan
6 non-ov
22 ov
5 non-ov
20 ov
7 non-ov
17 ov
6 non-ov
10 ov
11 ov
3 non-ov
17 ov
6 non-ov
10 ov
2 non-ov
12 ov
2 non-ov
12 ov
2 non-ov
12 ov
Elabor
-ADJ
Elabor
5 non-ov
1 ov
1 ov
2 non-ov
3 non-ov
2 non-ov
2 non-ov
1 non-ov
Comm
434 Anita Fetzer and Augustin Speyer
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
(p. 434) (CS4) PMU:
24 September 2012 3:53 PM
37
38
39
40
. 434) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 435–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
Table 6
22
100% ov
17
7 non-ov (F )
6 overt (F )
18 non-ov (NF )
9 overt (NF )
37.5% ov
100% ov
1 ov (F )
14
39 ov (F )
3 ov (NF )
1 ov
53 non-ov (F )
118 ov (F )
21 non-ov (NF )
130 ov (NF )
76.9% ov
17 ov
10 non-ov
10
28.2% ov
66 non-ov (F )
25 ov (F )
1 ov (NF )
3 non-ov
1 ov
1 non-ov
66.6% ov
2 ov (F )
1 non-ov. (NF )
1 ov
1 non-ov
1 ov
5
26.4% ov
159
1 ov
8 ov
9 ov
2 non-ov
4 non-ov
10 ov
5 non-ov
23 ov
21 non-ov (F )
2 ov (F )
2 ov (NF )
2 non-ov (NF )
14.8% ov
4 non-ov
1 ov
1 non-ov
1 ov
2 non-ov
1 non-ov
1
∑(%)
393
203
596
∑
25
14 non-ov
3 ov
16
24
21
2 ov
1 non-ov
5 ov
4 non-ov
15
NF
64
3 ov
0
0
7
F
88
9 non-ov
2 ov
3 ov
19
14
12
7 non-ov
6
Blair
30
29
19
18
NF
36
1 ov
9
F
35
44
34
2 non-ov
10 ov
10 non-ov
16 ov
8
France
33
NF
32
39
31
F
24
68
21
1 ov
20
1 ov
4
Obesity
30
1 ov
29
1
28
28
23
6 ov
3
NF
27
5 non-ov
8 ov
26
31
13
51
2
F
12
79
11
Environment
Discourse relations in English and German discourse 435
(p. 435)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
436 all of the editorials. There are 57.8% non-adjacent and 42.2% adjacent Con­
tinuations with 72.6% of the non-adjacent finite Continuations represented nonovertly and 27.4% represented overtly. In non-finite contexts, the coordinating
discourse relation of Continuation positioned non-adjacently has occurred overtly only once. As regards the distribution of Comments and Explanations across
the editorials, both occur more frequently non-overtly in finite contexts with
91.3% non-overt and 8.7% overt Comments, and 53.8% non-overt and 46.2%
overt Explanations. In non-finite contexts, by contrast, there are 66.6% nonovert and 33.3% overt Explanations. Elaborations are more frequently repre­
sented overtly in both finite and non-finite contexts. There are 69.0% overt and
31.0% non-overt Elaborations in finite contexts, and 86.0% and 14.0% in non-­
finite contexts. Analogously to the unit of investigation of sentence, the con­
tinuative, explanatory, and comment discourse relations are more frequently
­signaled by lexical coherence, while contrastive and elaborative discourse relations are signaled more often through discourse connectives, as is discussed
­below.
The preferred mode of coding the discourse relation Elaboration on the level
of clause is overt for both finite and non-finite contexts, while it is non-overt for
the coordinating relation of Continuation. In extract (15), the elaborative relation
is represented overtly in the non-finite contexts of (#2) and (#4), and of (#3) and
(16). In extract (16), the continuative relation in (#1) is signaled by lexical coherence connecting the generous commitments in (#1) with prior units in which financial support by the European Union, Germany and Italy have been mentioned;
the relation of Explanation (#2 and #4) adheres to the preferred mode of nonovert coding in both finite and non-finite contexts:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
(15) (#1) Iran took its time (#2) in responding to the incentives (#3) it was offered 27
in June (#4) to halt nuclear research work. (Iran)
28
29
(16) (#1) The generous commitments (#2) will go some way (#2 made yesterday in 30
Stockholm) (#3) to financing emergency and reconstruction aid, with $940m 31
(#4) being nearly twice the targeted amount. (Lebanon)
32
The relation of Contrast is only coded overtly for both sentences and clauses,
while Commentaries are coded non-overtly for both units of investigations, as is
illustrated in (17) and (18). In extract (17), the Elaborations are coded overtly in
(#2) and (#3), and in (#2) of (18), and the Contrast (#1) is also coded overtly by the
discourse connective but, which realizes an anaphoric relation with a prior discourse unit. In (18), the Comment is coded non-overtly by the deictic term this
indexing a discourse content which has been made explicit in the prior unit,
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 436–452 2655_9-4_01
PMU:(idp) 24/9/2012
33
34
35
36
37
38
39
40
(p. 436) (CS4) PMU:
24 September 2012 3:53 PM
437
Discourse relations in English and German discourse 1 namely unhealthy temptations, such as crisps from the corner shop or pizza from
2 the pizza company:
3
4 (17) (#1) But the force’s responsibilities under the ceasefire extend well beyond
5
6
7
8
9
the suppression of Hizbullah, (#2) which anyway will only cooperate with
the UN and the Lebanese army (#3) if it sees the peace process as even-­
handed. (Ceasefire)
(18) (#1) This is (#2) when resolve is at its weakest. (Obesity)
10
In the following, the coding of discourse relations is examined in the German
data, focusing on particular on their overt and non-overt representations as well
12
as on the constraint of adjacency.
11
13
14
15
16
17
4.2 The German editorials
18 The German data consist of 15 editorials of varying length from the Frankfurter
19 Rundschau. The size of the corpus is 317 sentences in total. In German, the most
20 frequent indicative sentence type for the marking of discourse relation is the de21 clarative verb-second main clause, and that is why we confine the investigation to
22 the 265 V2-sentences, of which 258 are in a discourse relation to some sentence
23 in the previous discourse. The mean number of V2-sentences per text is 17.67, and
24 the mean sentence number is 21.13. The quantitative analysis of the German edi-
25 torials is given in Table 7. It focuses on the same discourse relations as in the
26 English case study (see Table 1).
27
For the analysis of discourse relations, only main clauses are taken into con-
28 sideration because the marking of discourse relations between subordinate
29 clauses and their matrix clause is obligatorily in German where it is coded by dif30 ferent complementizers. The syntactic unit of sentence as unit of investigation
31 has ­never been controversial in any analysis of German. Besides, subordinate
32 clauses do not play a very important a role in the German data, as can be seen in
33 the ratio in the randomly chosen text “Menetekel für die Zukunft,” which con-
34 tains 25 main clauses, 17 subordinate clauses, of which two are control infinitives,
35 10 are relative clauses, and two are adverbial participle constructions. Hence, it
36 is not necessary to determine the discourse relation between clauses, which are
37 embedded that deeply. The only class of subordinate clauses in which dis38 course relations might be determined are adverbial clauses, but they tend to be
39 deeply embedded in German as well. In the text at hand, there are only three
40 ­adverbial clauses (two temporal, one conditional), and they are all embedded. As
. 436) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 437–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 437)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
438 Text
Assad
Blair
Bleiben
Ökologie
Vertrauen
Gespalten
Panik
Knapp
Kunden
Kuscheln
Markig
Menetekel
Milderung
Optionsspiele
Risikofaktor
∑
Sent.
V2Sent.
Continua‑
tion
Contrast
Elabora‑
tion
Explana‑
tion
Comment 1
11
17
35
39
46
13
14
16
13
14
15
25
31
14
14
7
10
27
33
39
13
11
14
10
11
13
23
31
10
13
3
1
5
3
8
1
0
3
1
0
1
2
1
1
1
0
2
5
3
4
4
4
4
2
1
1
5
5
2
0
1
1
6
8
9
4
1
3
2
2
5
5
10
4
4
1
3
2
4
5
1
0
0
3
3
1
2
5
0
0
0
0
0
3
2
0
1
0
0
1
0
0
3
1
0
3
317
265
31
42
65
30
11
16
2
5
6
7
8
9
10
11
12
13
14
15
17
18
Table 7
19
embedded clauses in German cannot form a separate focus unit, they are part of
the information unit of their matrix clause. They cannot assign a nucleus of their
own and therefore are part of the intonation contour of the matrix clause (cf., e.g.,
Reis 1997). This shows that in Modern German, the sentence as a whole functions
as a prototypical information unit. In earlier stages of German, when adverbial
clauses were not embedded as deeply as they are in Modern German, clauses
were independent informational units (cf. Speyer 2010). English seems to share
the patterns rather with the earlier stages of German in that respect, where
­clauses, not sentences, are considered to be prototypical information units.
If we look at the frequency of the discourse relations, we can see that Elaboration is either the most frequent relation or is in a tie for first place with other relations in nine of the 15 texts. Explanation is the most frequent relation in three
texts. As regards coordinating relations, Contrast is the most frequent relation in
four of the 15 texts, or it is in a tie, whereas Continuation is the most frequent relation in one text only. The overall distribution is 24.5% Elaborations, 15.8% Contrasts, 11.7% Continuations, 11.3% Explanations, and 4.2% Comments.
As in the analysis of the British data, we are interested in the question whether these relations are represented overtly or non-overtly, and whether there are
preferred ways of coding discourse relations. The data, corresponding to Table 2
in the English analysis, is systematized in Table 8.
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 438–452 2655_9-4_01
PMU:(idp) 24/9/2012
4
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
(p. 438) (CS4) PMU:
24 September 2012 3:53 PM
27
28
30
31
32
34
35
37
38
39
40
Table 8
26
265
6
20.0
24
∑
% overt
23
0
2
4
3
3
2
3
3
2
1
1
4
3
2
0
33
78.6
24
25
2
1
5
1
7
1
0
3
1
0
1
2
1
0
0
21
1
0
0
2
1
0
0
0
0
0
0
0
0
1
1
20
7
10
27
33
39
13
11
14
10
11
13
23
31
10
13
19
Assad
Blair
Bleiben
Ökologie
Vertrauen
Gespalten
Panik
Knapp
Kunden
Kuscheln
Markig
Menetekel
Milderung
Optionsspiele
Risikofaktor
17
26.2
0
1
3
0
0
2
0
1
2
1
2
0
2
2
1
17
Ov
16
Ov
48
1
0
3
8
9
2
1
2
0
1
3
5
8
2
3
Non-ov
14
29
Non-ov
13
Ov
12
11
36.7
1
2
1
2
1
0
0
0
2
2
0
0
0
0
0
Ov
19
0
1
1
2
4
1
0
0
1
1
1
2
5
0
0
Non-ov
Explan
9
Explan
10
Elabor
8
Elabor
9
81.8
0
0
0
2
1
0
1
0
0
1
0
0
3
1
0
Ov
Comm
6
Contr
5
Contin
2
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
Non-ov
Comm
3
Contin
2
V2S
1
Text
Discourse relations in English and German discourse 439
4
7
11
15
18
22
33
36
. 438) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 439–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 439)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
440 Contrast and Comment display a strong tendency to be marked overtly, while the
others tend to be realized non-overtly. The preference for Contrast to be marked
overtly is in line with the English data – an example of Contrast marked overtly by
the connecting particle aber ‘but’ is (19) – although it is much less categorical; we
also find examples of Contrast without overt marking, as in (20).
(19) (Context: Die Bundes-SPD tut dabei so, als würde sie die Farbenlehre in den
Provinzen herzlich wenig interessieren.)
(The federal Social Democratic Party acts in that matter as if they would not
really be interested in the science of colors in the provinces.)
Aber in Wahrheit bastelt die Parteizentrale unter Kurt Beck vielmehr sehr
bewusst am Eindruck der Bündnisoffenheit. (Optionsspiele)
But in truth works the party headquarter under Kurt Beck rather very con­
scious at the impression of the coalition openness.
(But in reality the party headquarter under Kurt Beck rather works very
consciously to create an impression of being open for a coalition.)
(20) (Context: Das Gesetz sollte das Ende der Kettenduldungen besiegeln.)
(The law should mark the end of the policy of chain tolerances)
In der Praxis ist das genaue Gegenteil eingetreten. (Bleiben auf Bewährung)
In the practice is the exact opposite happened.
(In reality, the exact opposite has happened.)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
The preference for Comment to be marked overtly – usually by means of a dis24
course anaphor, as in (21), is different to the English data, where the relation is
25
marked overtly only in about a quarter of cases. In German, Comments can, how26
ever, be left unmarked as well, as in (22).
27
(21) (Context: Auch Sigmar Gabriel spielt virtuos diese Klaviatur einer auf ökono­
mische Vernunft setzenden Umweltpolitik.)
(Sigmar Gabriel, too, plays this keyboard in a virtuous manner based on an
environmental policy anchored to economy-based reason.)
Das ist klug. (Die neue Macht der Ökologie)
That is clever.
(22) (Context: Im Kanon mit Schröder postulierte Sigmar Gabriel eine „ökologische
Industriepolitik“, einen „New Deal“ von Umwelt-, Wirtschafts- und Beschäfti­
gungspolitik, eine „dritte Industrielle Revolution“.)
(Together with Schröder, Sigmar Gabriel postulates an “ecological industry
policy,” a “New Deal” based on ecologic, economic and employment policies, a “third industrial revolution.”)
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 440–452 2655_9-4_01
PMU:(idp) 24/9/2012
28
29
30
31
32
33
34
35
36
37
38
39
40
(p. 440) (CS4) PMU:
24 September 2012 3:53 PM
441
Discourse relations in English and German discourse 1 Die Schlagworte sind neu, die Erkenntnisse nicht. (Die neue Macht der
2
­Ökologie)
3 The slogans are new, the knowledge not.
4 The slogans are new, the knowledge is not.
5
6 Another difference is reflected in Explanations. In German, we can find overt
7 marking quite frequently, whereas Explanation is represented non-overtly
8 throughout the English data. The marking is usually done by the discourse con-
9 nective denn (‘for’), as in (23). But there are also various examples where Expla-
10 nation is represented non-overtly, as in (24).
11
12 (23) (Context: Die Schere zwischen Einsicht und Handeln klafft auf internationaler
13
Ebene noch weiter auseinander.)
15
context.)
17
forcieren keineswegs das politische Tempo. (Die neue Macht der Ökologie)
19
force not at all the political speed.
21
of political action either.
14 (The cut between insight and action is even further apart in an international
16 Denn auch die jüngsten Erkenntnisse über die Dramatik des Klimawandels
18 Because also the latest knowledge about the drama of the climate change
20 For the latest facts about the dramatic climate change do not push the speed
22
23 (24) (Context: Die Schlagworte sind neu, die Erkenntnisse nicht.)
24 Grüne, Ökologen, Klimaforscher, ja auch Ökonomen predigen seit Jahren,
25
26
27
28
29
30
31
32
dass kluge Wirtschaftspolitik auch Umweltpolitik sein muss – und
­umgekehrt. (Die neue Macht der Ökologie)
Greens, ecologists, climate scientists, even also economists preach for years,
that intelligent economic policy also environmental policy be must – and the
other way round.
Greens, ecologists, climate scientists, and even economists have preached
for years that an intelligent economic policy has to be an environmental
policy as well – and the other way around.
33
34 The following excerpts exemplify Continuation and Elaboration. In both cases,
35 we do not find overt marking very frequently, viz. between 20 and 24 percent. The
36 marking of continuation is usually done by means of the coordinating discourse
37 connective und ‘and,’ sometimes also by adverbials like jetzt ‘now’ as in (25), but
38 most frequently the relation is not represented overtly, as in (26), but relies solely
39 on means of lexical coherence, such as the use of the demonstrative dieser ‘this’
40 in (26). Things are very similar for the discourse relation of Elaboration: Although
. 440) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 441–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 441)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
442 we do find the odd example represented overtly – mostly in non-adjacent relations, as in (27), where the marking is done by the modal particle ja – it is represented non-overtly most of the time as in (28), where discourse coherence is secured by the anaphoric expression dafür ‘for that.’ As a change of topic tends to
occur with Elaboration (cf., e.g., Hinterhölzl and Petrova 2009), it is not so important to mark the relation itself, but rather give hints regarding the thematic structure. Besides, Elaboration is the most common discourse relation of the argumentative texts under investigation. It can even be regarded as the default discourse
relation in argumentative genres, which is a further reason why it can be left unmarked: Explicit marking seems to imply that a deviation from the expected pattern is occurring, and the lack of explicit marking would suggest that the textual
progression follows the expected pattern.
(25) (Context: Vor fünf Wochen schon, als Hisbollah-Chef Nasrallah mit seinen
Waffen protzte, um auch die deutsche Kanzlerin Merkel im fernen Berlin
einzuschüchtern.)
(Five weeks ago Hizbullah-chief Nasrallah showed off displaying his
­weapons to intimidate even the German chancellor Merkel in far away
­Berlin.)
Der von den UN bestätigte Waffenfund im Libanon wirkt jetzt wie ein Beleg
dafür, dass man die Organisatoren des Terrors doch ernst nehmen sollte.
(Assads Grenzen)
The by the UN confirmed weapons discovery in the Lebanon looks now like a
proof for it, that one the organizers of terror but serious take should.
The discovery of weapons in the Lebanon confirmed by the UN at this stage
looks like some evidence that the organizers of terror should be taken more
seriously.
(26) (Context: Ganz so, wie es Nasrallah, Exponent einer Quasi-Nebenregierung im
Libanon, schätzt).
(Absolutely the way Nasrallah, exponent of a quasi-extra government in
Lebanon likes it.)
Nicht erst dieser Waffenfund, [. . .] verweist auf mächtige regionale Kräfte
hinter der Hisbollah. (Assads Grenzen)
Not only this weapons discovery (. . .) refers to mighty local forces behind the
Hizbullah.
Not only this discovery of weapons (. . .) refers to mighty local forces backing
Hizbullah.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
(27) (Context: Die SPD, so soll es aussehen, kann es mit allen: Große Koalition in 39
Schwerin, Rot-Rot in Berlin.)
40
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 442–452 2655_9-4_01
PMU:(idp) 24/9/2012
(p. 442) (CS4) PMU:
24 September 2012 3:53 PM
443
Discourse relations in English and German discourse 1 (The Social Democratic Party, that’s the way it is supposed to appear,
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
can cooperate with anybody: a big coalition in Schwerin, a red-red one in
Berlin.)
Vielleicht geht das ja auch deshalb so geräuschlos nebeneinander, weil
­Landespolitik trotz Föderalismusreform kaum mehr nennenswerten Handlungsspielraum hat. (Optionsspiele)
Maybe works that ptc also because of that so quietly side by side, because
federal policy despite federalism reform hardly more remarkable maneuvering
space has.
Maybe it has also worked in such a quiet manner because despite reforms
federal policy has hardly had any significant space for maneuvering.
(28) (Context: Aber in Wahrheit bastelt die Parteizentrale unter Kurt Beck vielmehr
sehr bewusst am Eindruck der Bündnisoffenheit.)
Was in Schwerin passiert, war dafür weit weniger wichtig als die Entwicklung in der Hauptstadt, [. . .] (Optionsspiele)
What in Schwerin happens, was for that far less important than the develop­
ment in the capital, (. . .)
What has happened in Schwerin had been far less important for that matter
than the development in the capital, (. . .)
As has been the case with the British data, there is a strong correlation between
the overt representation of a discourse relation in a discourse unit and the locality of the discourse unit standing in relation to the discourse unit under consideration. Plainly speaking, if a discourse unit A, represented (in German at least) by
a sentence SA, stands in a relation to a discourse unit B, expressed by a sentence
SB which does not immediately precede SA, but is separated from SA by at least one
discourse unit, the readiness to mark the relation overtly is much more developed. Table 9 gives the relevant data, for all relation types lumped together.
30
31
overt
Nonovert
total
Rate of overt
marking
coord
subord
tot
57
28
85
67
68
135
124
96
220
46%
29%
39%
coord
subord
tot
17
11
28
8
2
10
25
13
38
68%
85%
74%
32
33 adjacent
34
35
36 Non-adjacent
37
38
39
40 Table 9: Adjacency and overt marking of discourse relations in the corpus.
. 442) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 443–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 443)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
444 It is obvious that there is indeed a correlation between the distance between
­relation pairs and overt marking: Whereas the rate of overt marking is 39%
with adjacent relations (and the rate is only so high because some relations,
such as contrast, are usually marked overtly, no matter whether they are adjacent or not), it is almost twice as high, 74%, with non-adjacent relations. The
­difference is even more striking if we look at subordinating relations alone;
under adjacency, sub­ordinating relations tend not to be marked overtly: Only
in 29% of the cases we find overt lexical marking. If the partner is distant, however, overt marking occurs in 85% of the cases, which almost triplicates the
rate.
Let us have a closer look to some relations. In Table 10, we confined ourselves
to a detailed presentation of the relations that are investigated in the English part,
Continuation and Comment, but added Elaboration, since many non-adjacent
passages in the German data are Elaborations and since the difference is visible
very nicely.
Let us concentrate on Elaboration since it is the most frequent relation. If we
compare the rate of overt marking in non-adjacent relations to that of Elaboration
in total, we see that the rate is almost four times as high: 24% with Elaboration
total, 83% with only non-adjacent Elaborations. We can calculate the rate of overt
marking in only adjacent Elaborations as well. The number of adjacent Elaborations is 53, the number of overt marked sentences among them is five. This gives
us a rate of 9%. This means that the rate of overt marking in non-adjacent Elaborations is almost nine times as high as the rate in adjacent Elaborations. In (29) an
example is given of a non-adjacent Elaboration, marked by the adverbial nun
(‘now’). In sentences 7–11, not represented here, the Preise ‘prices’ of sent. Six are
elaborated upon, giving some examples of demands by the politicians. In sentence 12, the discourse refers back to sent. Six, taking another referent as starting
point of an evaluation, namely Verhandlungen ‘negotiations,’ which appears in
sent. Twelve in a slightly more pointed formulation as den Knoten der politischen
Selbstfesselung ‘the node of political self-bondage.’ The sub-discourse of sen­
tences 7–11 is finished and a new sub-discourse sets in, taking the same sentence
6 as starting point. The discursive meaning of the adverbial nun (‘now’), which is
polysemous and not used in its basic, temporal meaning ‘now’ here, indicates a
break in the discourse coherence, which manifests itself in the beginning of a
new sub-discourse, the starting point of which has to be inferred from the nonlocal context.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
(29) Context: sent. 6: Einigen in der Ministerrunde dürfte die Schlagzeile zumind­ 38
est willkommen gewesen sein, um in den Verhandlungen über ein Bleiberecht 39
für Flüchtlinge die Preise hoch zu treiben. [. . .]
40
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 444–452 2655_9-4_01
PMU:(idp) 24/9/2012
(p. 444) (CS4) PMU:
24 September 2012 3:53 PM
445
Discourse relations in English and German discourse 1 Text
S
Contin
total
2
3 Assad
4
7
3
5 Blair
6
10
Bleiben
-ADJ
Contin
0 non-ov
3 ov
Comm.
Total
-ADJ
Comm
Elabor
Total
- ADJ
Elabor
0
1
1
0
1
0 non-ov
1 ov
27
5
0
6
0 non-ov
1 ov
Ökologie
33
3
3
8
1 non-ov
0 ov
Vertrauen
39
8
2
9
Gespalten
13
1
0
4
16 Panik
11
0
1
1
17 Knapp
14
3
0
3
Kunden
10
1
0
2
Kuscheln
11
0
1
2
13
1
0
5
0 non-ov
2 ov
23
2
0
5
1 non-ov
0 ov
31
1
3
10
0 non-ov
2 ov
10
0
1
4
0 non-ov
1 ov
13
0
0
4
0 non-ov
1 ov
265
31
65
2 non-ov
10 ov
7
8
9
10
11
12
13
14
15
18
19
20
21
22 Markig
23
24 Menetekel
2 non-ov
0 ov
25
26 Milderung
27
28 Optionsspiele
29
30 Risikofaktor
31
32 ∑
33
34
35
36
37
% overt
2 non-ov
3 ov
60
11
0 non-ov
1 ov
0 non-ov
1 ov
100
0 non-ov
1 ov
0 non-ov
1 ov
83
Table 10
38
39
40
. 444) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 445–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 445)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
446 (At least some members of the minister round might have welcomed the
headline in order to raise the prices in the negotiations about a residence
right for refugees.)
Sent. 12: Nun könnte die große Koalition den Knoten der politischen Selbstfesselung durchschlagen.
Now could the big coalition the knot of political self bondage hit through.
Now the big coalition could get out of the political self-bondage.
The use of this discourse relation in context demonstrates a notable difference
between British and German editorials with respect to the relevant unit of description. Whereas Elaboration in the English data is mostly a relation between
clauses, it becomes an important relation between sentences in German, where
the sentence as a hierarchical structure composed of clauses is a more fundamental unit; a relation that can even hold between two non-adjacent sentences.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
4.3 Contrastive analysis
17
The British and German data do not only differ with respect to the discoursegenre specific distribution of discourse relations across the genre as a whole but
also with respect to their distribution across sentences and clauses. While the
most frequent relation is Continuation in the British sentence-anchored analysis,
it is less frequent than Contrast and only about as frequent as Explanation in the
German data, where the most frequent relation is Elaboration. However, the two
sets of data have similar preferences for the overt coding of the discourse relation
of Contrast, and the preferred non-overt representation of Continuation, Elaboration, Comment and Explanation, as is systematized in Table 11.
On the level of clause, however, Elaboration is also the most frequent discourse relation in the British data. The use of this discourse relation demonstrates
a decisive difference between British and German editorials with respect to the
relevant unit of description. Whereas Elaboration in the English data is mostly a
relation between clauses, it becomes an important relation between sentences in
German, where the sentence as a hierarchical structure composed of clauses is
a more fundamental unit. Here, Elaboration can even hold between two non-­
adjacent sentences.
The differences between the preferred overt and non-overt realization of discourse relations across the two sets of data does not really show significant differences. As regards the overt and non-overt representation of discourse relations in
non-adjacent contexts, however, there are some more pronounced differences, as
systematized in Table 12.
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 446–452 2655_9-4_01
PMU:(idp) 24/9/2012
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
(p. 446) (CS4) PMU:
24 September 2012 3:53 PM
447
Discourse relations in English and German discourse 1 British data (sentence)
2 40.6% continuation
3 82.1% non-overt
4
28.6% elaboration
5 72.8% non-overt
6
10.4% contrast
7
100% overt
8
11.4% comment
9
77.3% non-overt
10
11 4.1% explanation
100% non-overt
12
13
14
British data (clause)
German data (sentence)
54.5% elaboration
23.1% non-overt
24.5% elaboration
73.8% non-overt
26.6% continuation
73.6% non-overt
15.8% contrast
78.6% overt
6.7% contrast
100% overt
11.7% continuation
80% non-overt
6.7% explanation
62.5% non-overt
11.3% explanation
63.3% non-overt
4.5% comment
85.2% non-overt
4.2% comment
81.8% non-overt
British data (clause)
German data (sentence)
26.6% continuation
26.4% overt: 56.4% non-adj.
73.6% non-overt: 61.9%
non-adj.
11.7% continuation
20% overt: 100% non-adj.
80% non-overt: 16.7%
non-adj.
Table 11
15
16
17
British data (sentence)
18 40.6% continuation
19 17.9% overt: 42.8% non-adj.
82.1% non-overt: 37.5%
20
non-adj.
21
22 11.4% comment
22.7% overt: 0% non-adj.
23
77.3% non-overt: 5.8%
24 non-adj.
25
26
27
28
4.2% comment
81.8% overt: 22.2% non-adj.
18.2% non-overt: 0%
non-adj.
54.5% elaboration
76.9% overt: 0.8% non-adj.
23.1% non-overt: 1.3%
non-adj.
24.5% elaboration
26.2% overt: 58.8% non-adj.
73.8% non-overt: 4.2%
non-adj.
29
30 Table 12
31
32
33 While all of the overtly coded Continuations are non-adjacent in the German
34 data, only 46.2% are non-adjacent in the British data. As regards Comments, ap35 proximately a quarter of the discourse relation is coded overtly, which is not the
36 case in the British data. There does not seem to be any connectedness between
37 the overt coding of Comment and its positioning in discourse. As regards Elabora-
38 tions, there is a similar pattern in the German data: Almost 60% of the non-­
39 adjacent Elaborations are coded overtly. Does that pattern also hold for the
40 clause-based analysis of the British data?
. 446) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 447–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 447)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
448 There is a difference between the overt coding of Continuations as regards
discourse relations holding between sentences and clauses in the British data. On
the level of clause, more than half of the overtly coded Continuations are nonadjacent, and 59.5% of the overtly coded Continuations occur in finite contexts.
Only 2.3% of overtly coded Continuations occur in non-finite environments, as is
systematized in Table 13:
1
2
3
4
5
6
7
8
British data (clause)
9
26.6% continuation
26.4% overt: 59.5% finite
26.4% overt: 2.3% non-finite
73.6% non-overt: 41.5% finite
73.6% non-overt: 0% non-finite
10
11
12
13
14
15
Table 13
16
17
The overall rate of overt marking across the two sets of data is systematized in 18
Table 14:
19
20
Adjacent
Non-adjacent
Rate of overt
marking (S/C) British
Rate of overt
marking (S) German
Coord
Subord
Tot
34.6%/41.2%
21.2%/67.7%
28.6%/54%
46%
29%
39%
Coord
Subord
Tot
6.1%/13.0%
1.0%/0.7%
3.6%/4.8%
68%
85%
74%
22
23
24
25
26
27
28
29
30
Table 14
31
The rate of overt marking for adjacently positioned coordinating relations is
­higher in the German data, where almost half of the coordinating relations are
marked overtly. The rate is lower in the British data, and there is hardly any difference between the overt marking of adjacently positioned discourse relations
holding between clauses and sentences. As regards adjacently positioned sub­
ordinating discourse relations, different patterns surface: The overt marking of
subordinating discourse relations is lower than the overt coding of coordinating
discourse relations in the German data.
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 448–452 2655_9-4_01
PMU:(idp) 24/9/2012
21
32
33
34
35
36
37
38
39
40
(p. 448) (CS4) PMU:
24 September 2012 3:53 PM
449
Discourse relations in English and German discourse 1
In the British data, there is a significant difference between the overt coding
2 of subordinating discourse relations anchored to clauses and sentences. There is
3 a clear preference for coding adjacently positioned subordinating discourse rela4 tions in an overt manner on the level of clause.
5
The rate of overt marking for non-adjacently positioned discourse relations
6 shows a clear preference for both coordinating and subordinating relations for
7 the German data, and a less clear preference for the coordinating relations in the
8 British data for clauses. Only a quarter of the non-adjacently positioned coordi9 nating relations are coded overtly for discourse relations across sentences, and a
10 third are coded overtly for coordinating discourse relations across clauses. Non11 adjacently positioned subordinating relations are not coded overtly but rather
12 through lexical coherence.
13
14
15
16
5 Conclusions
17 This investigation of discourse relations in English and German discourse is
18 based on the premise that discourse comes in with the presumption of being
19 ­coherent as regards its constitutive discourse units, and as regards the dis-
20 course as a whole. Furthermore, discourse is seen as hierarchically structured,
21 as is reflected in the classification of discourse relations as coordinating and
22 ­subordinating relations. To capture language-specific units of investigation, viz.
23 clause in English and sentence in German, this contrastive analysis is based
24 on the units of sentence and clause. The discourse relations under investiga-
25 tion are the coordinating relations of Continuation and Contrast, and the sub­
26 ordinating relations of Elaboration, Explanation and Comment, and their overt
27 and non-overt representation in adjacently and non-adjacently positioned
28 propositions.
29
The analysis of the British data is based on the units of sentence and clause,
30 and their finite and non-finite contexts, whereas the analysis of the German data
31 is based on the unit of sentence. This is because “sentence” is a logic-based unit
32 in German, while it is more of an orthographic, and less of a logic-based unit in
33 English. An analysis of clauses in German is not considered to be appropriate
34 because of their rather high degree of embeddedness. In the German data, the
35 subordinating discourse relations are very frequent and they are positioned both
36 adjacently and non-adjacently. They tend to be represented overtly only in the
37 latter case. This is especially true for Elaboration, which tends to be a relation
38 between clauses in the British data, but a relation between sentences in the Ger39 man data, where the sentence as a hierarchical structure composed of clauses is
40 a more fundamental unit and can even hold between non-adjacent sentences. In
. 448) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 449–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 449)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
450 the British data, only Continuation is positioned non-adjacently in clauses and
sentences.
In both sets of data, there is a strong correlation between the overt marking of
a relation in a clause/sentence and the locality of the clause/sentence standing in
relation to the clause/sentence under consideration. More precisely, if a sentence
SA in German stands in a relation to a sentence SB that does not immediately precede SA, but that is separated from SA by at least one sentence, the readiness to
mark relations overtly is much more developed. For the British data, the situation
is different. Here, the readiness to mark subordinating discourse relations between directly adjacent clause-anchored discourse units is much more developed
than for sentence-anchored non-adjacently positioned discourse units.
A contrastive analysis of discourse relations in a pragmatic theory of discourse has the potential of identifying language-preferential patterns for coding
coordinating and subordinating relations. It would be of interest for future studies whether these differences also hold for spoken discourse, and whether there
are similar differences in other languages. Being aware of language-preferential
strategies for the overt and non-overt representation of discourse relation could
not only refine research educational L1 and L2 discourse but also in the field of
intercultural communication.
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
References
21
22
Anscombe, Jean-Claude & Oswald Ducrot. 1983. L’Argumentation dans la Langue. Mardaga:
Bruxelles.
Asher, Nicholas & Alex Lascarides. 2003. Logics of conversation. Cambridge: CUP.
Asher, Nicholas & Laure Vieu. 2005. Subordinating and coordinating discourse relations.
Lingua 115. 591–610.
Benz, Anton & Peter Kühnlein. 2008. Constraints in discourse: An introduction. In Anton Benz &
Peter Kühnlein (eds.), Constraints in discourse, 1–26. Amsterdam/Philadelphia: John
Benjamins.
Biber, Douglas. 1988. Variation across Speech and Writing. Cambridge: Cambridge University
Press.
Bloor, Thomas & Meriel Bloor. 1995. The Functional Analysis of English: A Hallidayan approach.
London: Arnold.
Bublitz, Wolfram, Uta Lenk, & Eija Ventola (eds.). 1999. Coherence in spoken and written
discourse. Amsterdam: John Benjamins.
Chesterman, Andrew. 1998. Contrastive functional analysis. Amsterdam/Philadelphia: John
Benjamins.
Clyne, Michael. 1987. Cultural differences in the organization of academic texts. Journal of
Pragmatics 11. 211–247.
Doherty, Monika. 2003. Discourse relators and the beginnings of sentences in English and
German. Languages in Contrast 3. 223–251.
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 450–452 2655_9-4_01
PMU:(idp) 24/9/2012
1
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
(p. 450) (CS4) PMU:
24 September 2012 3:53 PM
451
Discourse relations in English and German discourse 1 Fetzer, Anita. 2004. Recontextualizing context: Grammaticality meets appropriateness.
Amsterdam/Philadelphia: John Benjamins.
2
Fetzer, Anita. 2008. Theme zones in English media discourse: Forms and functions. Journal of
3
Pragmatics 40(9). 1543–1568.
4
Gernsbacher, Moton-Ann & Talmy Givón (eds.). 1995. Coherence in spontaneous text.
5
Amsterdam: John Benjamins.
6 Givón, Talmy. 1993. English Grammar: A function-based introduction. Amsterdam/Philadelphia:
John Benjamins.
7
8 Givón, Talmy. 2005. Context as other minds. Amsterdam: John Benjamins.
Gómez-González, Maria. 2001. The theme–topic interface: Evidence from English. Amsterdam:
9
John Benjamins.
10 Grosz, Barbara, Aravind Joshi, & Scott Weinstein. 1995. Centering: A framework for modelling
11
the local coherence of discourse. Computational Linguistics 21. 203–225.
12 Grosz, Barbara & Candace Sidner. 1986. Attention, intentions and the structure of discourse.
Computational Linguistics 12. 175–204.
13
Halliday, Michael & Ruqaiya Hasan. 1976. Cohesion in English. London: Longman.
14
Halliday, M. A. K. 1994. Introduction to English functional grammar. London: Arnold.
15
House, Juliane. 1996. Contrastive discourse analysis and misunderstanding: The case of
16
German and English. In M. Hellinger and U. Ammon (eds.), Contrastive Sociolinguistics,
17
345–361. Berlin/New York: Mouton de Gruyter.
18 Hunston, Susan. 2007. Using a corpus to investigate stance quantitatively and qualitatively.
In R. Englebretson (ed.), Stancetaking in Discourse, 27–48. Amsterdam/Philadelphia:
19
John Benjamins.
20
Kamp, Hans & Uwe Reyle. 1993. From discourse to logic: Introduction to modeltheoretic
21
semantics of natural language, formal logic and discourse representation theory.
22
Dordrecht: Kluwer.
23 Krifka, Manfred & Caroline Féry. 2008. Information structure: Notional distinctions, ways of
expression. In Piet van Sterkenburg (ed.), Unity and diversity of languages, 123–136.
24
Amsterdam: John Benjamins.
25
Krzeszowski, Tomas. 1989. Towards a typology of contrastive studies. In W. Oleksy (ed.),
26
Contrastive Pragmatics, 55–72. Amsterdam/Philadelphia: John Benjamins.
27 Levinson, Stephen C. 1979. Activity types and language. Linguistics 17. 365–399.
28 Levinson, Stephen C. 1983. Pragmatics. Cambridge: Cambridge University Press.
29 Liedtke, Frank. 1997. Gesagt? getan: Über illokutionäre Indikatoren. In Rolf Eckhard (ed.),
Implikaturen und Sprechakte. Linguistische Berichte, Sonderheft 8. 189–213.
30
Mann, W.C. & S.A. Thompson. 1987. Rhetorical Structure Theory: Description and construction
31
of text structures. In G. Kempen (ed.), Natural language generation. Den Haag: Martinus
32
Nijhoff.
33 Pomerantz, Anita. 1984. Agreeing and disagreeing with assessments: Some features of
preferred/dispreferred turn shapes. In J. Atkinson & J.M. Heritage (eds.), Structures of
34
social action, 57–101. Cambridge: Cambridge University Press.
35
Reinhart, Tanya. 1982. Pragmatics and linguistics: An analysis of sentence topics. Philosophica
36
27. 53–94.
37
Reis, Marga. 1997. Zum syntaktischen Status unselbständiger Verbzweit-Sätze. In Christa
38
Dürscheid, Karl-Heinz Ramers, & Monika Schwarz (Hgg.): Sprache im Fokus. Festschrift für
39
Heinz Vater zum 65. Geburtstag, 121–144. Tübingen: Niemeyer.
40 Sacks, Harvey. 1995. Lectures on conversation. Oxford: Blackwell.
. 450) (CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 451–452 2655_9-4_01
3 PM
PMU:(idp) 24/9/2012
(p. 451)
24 September 2012 3:53 PM
Anita Fetzer and Augustin Speyer
452 Schegloff, Emanuel. 1995. Discourse as an interactional achievement III: The omnirelevance of
action. Research on language and social interaction 28(3). 185–211.
Sperber, Dan & Deirdre Wilson. 1986. Relevance: Communication and cognition. Oxford:
Blackwell.
Speyer, Augustin. 2010. Die Markierung von Diskursrelationen im Frühneuhochdeutschen.
Sprachwissenschaft 35. 409–442.
Thibault, Paul. 2003. Contextualization and social meaning-making practices. In Susan L.
Eerdmans, Carlo L. Prevignano, & Paul J. Thibault (eds.), Language and interaction:
Discussions with John J. Gumperz, 41–62. Amsterdam: John Benjamins.
van Dijk, Teun A. 1980. Textwissenschaft. München: dtv.
Widdowson, Henry. 2004. Text, context, and pretext: Critical issues in discourse analysis.
Oxford: Blackwell.
Acknowledgements
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
We are deeply grateful to our reviewers and to the editor for helpful comments on
16
the first version of this article.
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
(CS4) WDG (155×230mm) DGMetaScience J-2655 IP 9:4 pp. 452–452 2655_9-4_01
PMU:(idp) 24/9/2012
(p. 452)
24 September 2012 3:53 PM