Coherence and Cohesion in English Discourse

4 Material and method
This investigation into the role of indexical expressions in the perception of 
cohesion and coherence focuses on the use of personal and possessive pronouns 
and demonstratives, both as pronouns and determiners, in written academic 
discourse. The study considers the frequency and function of the target items 
and explores how they establish cohesive links and contribute to the perception 
of coherence at the local and global level of discourse. The analytical approach 
adopted in this study combines quantitative and qualitative analyses, since the use 
of quantification as a starting point of investigation allows us to highlight general 
tendencies in the occurrence of the phenomenon and motivate the selection 
of representative discourse samples which can then be explored qualitatively 
(Hunston 2007: 46).
The analysis is carried out on a specialized corpus of research articles in the 
field of linguistics. The corpus comprises twelve 
research articles (six single-
authored and six co-authored) published in the journal 
Applied Linguistics
in 
the period 2000-2010; the size of the material totals 90,500 words. Obviously, 
the corpus used in this study is rather small. However, since small specialized 
corpora, which allow for “top-down, qualitative, contextually-informed 
analyses” (Flowerdew 2004: 18), are regarded as convenient for the investigation 
of academic and professional discourse, it is considered that it yields enough 
grounds for conclusions. Before proceeding to the analysis, the corpus was 
cleaned to eliminate block quotes and long examples; however, integral citations 
and integral examples were not deleted in order to preserve the coherence of the 
texts. The corpus was searched for the target pronouns (
I/me/my
, 
you/your
, 
he/
him/his
, 
she/her
, 
it/its
, 
we
/
us/our
) and demonstratives (
this/these
and 
that/those
) 
using the freeware Antconc concordance programme. After the rate of occurrence 
of the target items was ascertained, the raw data was normalized to frequencies 
per 1,000 words to allow for comparison with previous and further studies. The 
concor dance lists were then checked manually to exclude occurrences of target 
structures in integral citations and integral examples and, most importantly, to 
identify the functions of the target items in context. Finally, a qualitative analysis 

---

15
of selected representative examples was carried out to serve as a basis for a 
discussion of the contribution of pronominal reference and demonstratives to the 
construal of cohesion and coherence in academic articles.