Unformatted text preview: 5
3918 These key words are used as the query items in the BNC. As many
theoretical works in Conceptual Metaphor Theory seem to refer to language in general rather than to any specific genre, the BNC as a whole is
consulted. The analysis follows the Pragglejaz procedure. Words from the lexical field of war and their metaphoric potential 203 3. The British National Corpus
The British National Corpus is a 100 million words corpus of contemporary British English created between 1991-1994. As with many corpora it
is biased towards the written mode of the language (90%), but has a spoken part as well (10%). The selection criteria devised during the design
stage were to ensure that within the written part 75% of the texts were informative, 25% were fiction. As for the types of media, 60% are books,
25% are newspapers and journals, between 5-10% are miscellaneous published materials (leaflets, brochures), between 5-10% are miscellaneous
unpublished materials (letters, notes) and up to 5% texts written to be
spoken. Texts could not be older than 1975 with an exception of a number
of fiction texts which go back to 1964.
The spoken part is divided into a demographic section and a context-governed part. The demographic part consists of transcriptions of recordings made by 124 volunteers, both men and women from 38 different
locations in the UK and from various social groups, who recorded their
conversations for three days. The context-governed part consists of transcriptions falling into four categories: educational and informative contexts, business events, institutional and public events and leisure events.
More detailed information is available on the BNC website at
BNC can be accessed through a dedicated concordancer SARA.
The concordancer allows the researcher to look up the investigated word
or phrase. It returns the matches in the form of concordance lines. If the
context in the line is not sufficient it is possible to consult the text from
which the example was derived through the Browse function. The returns
can be sorted by left and right context, so that the same phrases preceding
or following the node word can be grouped together, which facilitates the
interpretation. SARA also offers collocability tests: z-score and MI (Mutual Information) statistical tests can be used for a predefined window
size, i.e. a selected number of words to the left and right of the node word.
an attempt to utilize z-score for the task in hand has been made, but the
interpretation of results was as time consuming as interpreting...
View Full Document