This thesis presents a comparative empirical study of fixed collocational patterns in written academic English. The fixed collocational pattern is a continuous string of words which is found to occur frequently in language data. This study uses lexical analysis software to study an electronic corpus of academic research articles in an attempt to identify and compare the forms and discourse functions of fixed collocational patterns in different disciplines.
In comparative studies of the language of different academic disciplines there are two ways of collecting comparable amounts of corpus data, both of which are problematic. One approach is to subdivide existing normative (Sinclair 2005) corpora in order to allow comparisons to be made between different disciplines. The amount of data in each resulting subcorpus is often
unequal, however, and results might be biased in favour of the subcorpus with the greatest number of texts or tokens. The other approach is to balance the number of tokens in each subcorpus by using incomplete text samples. This can mean that individual subcorpora do not completely represent all areas of the discourse, and some fixed collocational patterns which perform discourse functions relating to these areas may as a result be missed by the researcher.
This study attempts to establish what might be a comparable amount of data by investigating fixed collocational patterns in two different comparative corpora. First it identifies fixed collocational patterns in an equal number of tokens in each discipline, i. e. an isolexical comparison. It then identifies fixed collocational patterns in an equal number of texts in each discipline, i. e. an isotextual comparison. The findings indicate that the same fixed collocational patterns are frequent in both versions of the corpus, and so what is frequent isolexically is also
frequent isotextually. This suggests that an isotextual corpus is more suitable for comparative studies of the discourse functions of fixed collocational patterns, since it allows their functions to be investigated across similar numbers of communicative acts rather than across similar
amounts of language.
The thesis then compares these isotextual fixed collocational patterns with the results from two previous studies of an isolexical collocational pattern, the lexical bundle (Biber et al. 1999), one of which (Biber 2006) used data from a different academic genre, the other (Hyland 2008) data from three academic genres. There then follows a case study of the relationship between the lexical, semantic, and textual environments of the fixed collocational pattern in the case of the and its discourse functions. The thesis concludes by outlining areas of future research into fixed collocational patterns which have been suggested by the results of this study.