A corpus has been constructed and annotated for categories of speech and
thought presentation (also known as speech and thought reporting) using a
tagset which has been developed by the authors and other researchers at
Lancaster. This tagset is based on a model of speech and thought
presentation (S&TP) proposed in Leech and Short (1981) which posits a
continuum of categories along an axis  representing degrees of narratorial
intervention. The corpus contains 250,000 words is representative of three
genres of modern English narrative texts: - prose fiction, - newspaper news
reports - biography and autobiography Within each genre there is a division
between 'serious' and 'popular' texts, so comparisons between these two
types of text are possible as well as comparisons between genres. Due
chiefly to the fact that a corpus-based approach has been adopted, this
project has been able to do research which has not, as far as the authors
are aware, previously been done in stylistics in a systematic fashion,
despite the close attention that linguists and literary theorists have paid
to STP for many decades. The features that the corpus-based approach has
brought to this area of linguistic enquiry are: - the development of a
tagset for S&TP and a well-defined formalism for the annotation using this
tagset; - the testing of a set of categories of S&TP to see if they can
adequately account for ALL occurences of S&TP in a text, rather than just
the interesting examples identified by the researcher; - the analysis and
cross-genre comparison of S&TP in fiction and non-fiction;  - the
identification and classification of complex mixed forms of  S&TP in fiction
and non-fiction; - results relating to the frequency of occurrence of the
various categories and the extent of ambiguity.