204 Kerstin Kunz and Erich Steiner
this ‘inside-out’, or ‘grammar-based’, perspective (Dik 1978; Givón 1979; Halliday
& Matthiessen 2004).
A level-based perspective posits a linguistic level additional to (lexico)gram-
mar (as in Martin 1992). It is within this perspective that the question of that
level’s precise nature, and of the distinction between ‘cohesion’ and ‘coherence’,
usually arises. In fact, in some of the literature a threefold distinction is made into
‘Konnexität’ (syntactic), ‘Kohäsion’ (semantic), and ‘Kohärenz’ (pragmatic), which
in Halliday and Hasan’s (1976) terms would be “lexico-grammar vs. cohesion vs.
coherence”. ‘Lexico-grammar/Konnexität’ refers to the domain of grammaticaliza-
tion, ‘cohesion/Kohäsion’ to semantic relationships across grammatical domains
(phrases/clauses), and ‘coherence/Kohärenz’ to contextually motivated semantic
enrichment of symbolic patterns of the first two types. The latter two are conceived
of as being ‘transphrastic’ (Schreiber 1999: 16ff ).
The view adopted in our research is closest to the formulation in Halliday and
Matthiessen (2004: 579), though in a modified re-formulation which we believe
is consistent with their overall view of language: the textual resources of English
(and German) are (a) structural (thematic, information structure, focus, clause-
complexing) and (b) cohesive, with those under (a) engendering grammatical
structure, and those under (b) creating cohesive structure across grammatical
and lexical units.
In summary, comprehensive discussions of the systemic resources of cohesion
exist for English and to a lesser extent for German. However, no contrastive work
is available comparing a broad range of cohesive resources in the two language
systems as well as their instantiations in texts across registers. An attempt will be
made here towards narrowing this gap for the area of substitution.
1.3 Outline of the GECCo corpus
The following brief remarks about the corpus architecture outline the empirical
base of our project GECCo (German English Contrasts in Cohesion, see Kunz &
Lapshinova-Koltunski 2011; Amoia et al. 2011).
1.3.1 Corpus design
The GECCo corpus is an extended and modified version of the former CroCo cor-
pus, which was developed in the frame of the CroCo project (Linguistic Properties
of Translations; see Hansen-Schirra et al. 2005 and Culo et al. 2011).
GECCo is structured as shown in Table 1 and consists of six subcorpora:
First, the corpora in written language EO-WRITTEN (English originals),
GO-WRITTEN (German originals), ETRANS-WRITTEN (English Translations)
and GTRANS-WRITTEN (German Translations).