Coherence and Cohesion in English Discourse

5.2 Lexical cohesion
As mentioned above, lexical cohesion, i.e. “cohesion which is realised by the 
selection of vocabulary” (Tanskanen 1995: 531), is demonstrated in newspaper 
discourse more prominently and more ‘visibly’ than grammatical cohesion, 
because it is mainly the choice of vocabulary that helps to position the event as 
well as the reader in a particular social and cultural context, which is necessary for 
deriving coherence from discourse. On the other hand, this is not to suggest that 

---

116
grammatical and lexical devices are separate phenomena; it must be emphasized 
that cohesion for the reader follows from their interplay in the first place. 
Before proceeding to concrete examples of lexical cohesion in the material 
under investigation, it is useful to sum up at least the basic approaches to lexical 
cohesion within the models proposed by Halliday and Hasan (1976), Hasan (1984), 
Hoey (1991) and Tanskanen (1995). Halliday and Hasan originally distinguished 
only two categories of lexical cohesion (i.e. ‘reiteration’ and ‘collocation’), 
which later proved insufficient and led to the modification of lexical cohesion 
relations by Hasan (1984: 2002), who proposed two completely new categories of 
‘general’ and ‘instantial’ relations; the former included ‘repetition’, ‘synonymy’, 
‘antonymy’, ‘hyponymy’ and ‘meronymy’, the latter ‘equivalence’, ‘naming’ 
and ‘semblance’. Still, Halliday and Hasan consider cohesion and coherence as 
interdependent phenomena. 
According to Hoey (1991), lexical cohesion, due to its high frequency of 
occurrence, is the main contributor to the creation of texture and is stronger than 
grammar; lexical relations may create cohesion between neighbouring sentences 
(i.e. local cohesion) as well as across larger chunks where cohesion operates 
between sentences separated by several other sentences (i.e. distant cohesion). 
The connections between the sentences, which Hoey labels as ‘bonds’, play a 
key role in discourse coherence in that they ensure “topic continuity” and activate 
“relevant discourse-processing knowledge” (Dontcheva-Navratilova 2011: 42).
Tanskanen (1995: 533) aims to define cohesion in more general terms in 
order to provide a model which could be used for analysis of cohesion in different 
types of discourse and under different conditions. Therefore, she proposes two 
general categories: ‘reiteration’ and ‘collocation’. Unlike other researchers 
she abandons traditional terms from lexical semantics, such as synonymy or 
antonymy. Reiteration within her model comprises a) repetition (
map
– 
map
; 
a plan
– 
planned
); b) equivalence (
establish
– 
set up
); c) generalisation (
buses
– 
public transport
); d) specification (
public transport
– 
underground
); e) co-
specification (
buses
– 
underground
) and f) contrast (accept – refuse). Collocation 
comprises a) ordered set (
Monday
– 
Thursday
) and b) implication (
winter
– 
cold
; 
armies
– 
war
).
Tanskanen’s research into lexical cohesion shows that reiteration is
 
more 
frequent than collocation and “dominates especially under more difficult 
communicative conditions” where cohesion is relatively dense (i.e. greater 
proximity between items in pairs); collocation “is used to a greater extent in less 
demanding conditions”, where the cohesion is less dense (i.e. there are greater 
distances between items in pairs) (ibid.: 537).
The repertoire and frequency of occurrence of lexical cohesive devices 
depends at least to a certain extent on the discourse type, i.e. the range and types 
of cohesive devices employed in a narrative, for example, will differ considerably 

---

117
from those in an advertisement or a newspaper report. Also, various discourse 
types allow inner variation, i.e. newspaper reports can be further sub-divided into 
a number of sub-types, usually depending on the main theme. Within the sub-
type of crime reports these include, for example, reports on robberies, accidents, 
murders, verdict reports, etc. In order to depict the variation in the newspaper 
reports under investigation and map principal differences between the serious and 
popular British press it appears particularly suitable to deal with lexical cohesive 
devices in terms of general categories as discussed by Tárnyiková (2002), i.e. 
‘lexical replacement’ and ‘lexical relationships’, which are closely intertwined 
with naming strategies in crime reports.
All the crimes reported in the reports under investigation are high-profile 
murder cases of very young children killed by a parent, or murders of teenagers 
who have been killed in a particularly violent and abhorrent way; as such they 
have enormous potential as ‘human stories’, which seem to be replacing hard 
news such as foreign news and political or investigative reporting in the modern 
press (Franklin 2008). The victim and the killer(s) are thus central to the reports 
since the events happened in real life to real people. From the very beginning 
the main participants are clearly identified not only as ‘
victim’ vs. ‘killer’, but 
also as ‘good’ vs. ‘evil’, ‘innocent’ vs. ‘vicious’, ‘helpless’ vs. ‘sadistic, cruel’. 
The participants are identified as individuals and at the same time via their 
social roles, for example, 
a mother/ father 
vs. 
a child
, 
an altar boy
vs. 
a school 
dropout
, etc. These labels help to express the contrast between the participants 
and also accentuate their positive and negative qualities, as well as to reinforce