Insistent voices - government messages

This chapter investigates some patterns in the way speakers represent speech, thought, and writing across the CorDis Corpus with particular reference to varieties of voices in the texts. Many frequently occurring lexical items in the word lists of the corpus belong to the semantic fi elds of speech, thought, and writing (e.g., talking, conversation, thought, writes, dossier, page, line). In order to investigate how discourse representation is carried out in the corpus, a number of these were chosen and examined for the way they exemplify different text voices. These words often need lexicalization, as do general nouns such as point, thing, way, and, also like general nouns, they have a discourse organizing function (see Francis 1994). There are differences in their distribution and use in the subcorpora which make up the CorDis Corpus. Such differences refl ect not just the varieties of discourse type but also diverse interests in the representation of voice and different levels of take-up of these insistent messages from government and administration voices. The methodology of corpus-assisted discourse studies is used, as set out in the introduction to this volume. I will fi rst deal with questions of speech and thought representation and corpus methodology, with a short overview of previous research. Using Thompson’s (1996) framework for my description I will give some examples of the key features of the data related to the representation of thought, speech, and writing to be found in the CorDis Corpus. Finally I will go on to a consideration of language awareness and language management in politics, and the way in which the relationship between politicians and the media is sometimes played out via the manipulation of speech and thought representation.i