A corpus-based investigation into key words and key patterns in post-war fiction

This study is an exploratory investigation into lexico-grammatical items specific to a large corpus of English-language post-war novels, as compared to corpora of conversation, news and academic English. Its overall aim is threefold: first, to show how the subjective impression of ‘literariness’ arising from fictional works is at least partly based on the statistically significant use of highly specific words and lexico-grammatical configurations; second, to attempt a broad classification of key words and patterns; third, to illustrate the fiction-specific patterns formed by three key words. Analysis proceeded in three steps. First, a key word analysis was performed. In the second step, all two-to-five word strings contained in the English corpus were generated. In the third step, multi-word strings, collocations and colligations associated with three English key words (‘thought’, ‘sun’ and ‘jerk’) were analysed. Results indicate that post-war fiction is characterized by the dense use of specific sets of key words and key patterns, such as multi-word strings (must have been), phrase frames (like a + NP, there was a + NP) colligations (PossDet thoughts were on NP), collocations (the strengthening sun) and lexically specific narrative patterns (PossDet thoughts were interrupted when/as + time clause). The patterns in question are shown to be interconnected through a complex web of analogical creations. Implications are discussed for theories of literature, lexicology and translation.