Coherence and Cohesion in English Discourse

No. of words 
in all texts 
57,819
Non-native 
speakers 
77,533
Native
speakers 
Types of 
conjuncts
No.
Norm.
frequency
No.
Norm.
frequency
e.g.
63
1.09
80
1.03
for example
25
0.43
58
0.75
for instance
9
0.16
12
0.15
i.e.
107
1.85
7
0.09
that is
4
0.08
8
0.10
what is more
1
0.02
in other words
19
0.33
2
0.03
more specifically
11
0.19
1
0.01
namely
28
0.48
10
0.13
specifically
3
0.05
4
0.05
to be specific
1
0.02
TOTAL
271
4.69
182
2.35
Table 2-3: Appositive conjuncts in both corpora

---

46
Another type of appositive conjuncts which is worth commenting on is 
what 
is more
, since it is the only type in my data realized by a finite clause. However, 
with its single occurrence in the NNSC, it is rather exceptional. In addition to this 
conjunct, Example (9) comprises two other appositive conjuncts – 
i.e.
and 
namely
– both used for reformulation and restatement, especially in non-native speakers’ 
writing (cf. Table 2-3). The abbreviation 
i.e.
, whose origin is the Latin 
id est
, 
can be viewed as an alternative form of the conjunct 
that is
, which is, however, 
much less common in both corpora (0.08-0.10). The reformulatory conjunct 
i.e.
, 
with a frequency of occurrence of 1.85 in the NNSC, is, as stated above, the 
most common conjunct of all in the RAs written by non-native speakers in my 
material. It remains to be added that Example (9) also comprises two tokens of 
conjuncts from semantic classes that are typical in particular of academic texts 
written by non-native speakers of English, namely the listing (additive) conjunct 
moreover 
(0.36) and the contrastive/concessive 
however
(1.57), the latter ranking 
among the most frequent conjuncts of all found in academic texts in general (cf. 
Biber et al. 1999: 885).
(9) NNSC, Text 2A
Examining first the average number of U falling to a paragraph, we 
notice a striking variability in paragraph lengths, ranging from one to as 
many as 29 U. It should be noted, however
, that paragraphs containing 
one to three U made up 60.34 per cent of the overall corpus. What is 
more
, paragraphs containing one to six U amounted to 88.31 per cent of 
the investigated corpus. It follows that the relatively longest paragraphs, 
i.e. those ranging from seven to 29 U, were clearly marginalized, covering 
only 11.69 per cent overall. Interestingly, the extreme length of 29 U was 
detected only in one source, 
namely
in F5, and, 
moreover, in a single 
paragraph.