2. COMPOSITION OF CORPUS

The final total number of words collected was 52,637. In line with the conventions used in the LOB corpus project, each sample text is assigned to an overall category (indicated by a letter) and identifled by a "part number" (indicated by a digit). In addition, each text is given an absolute number to indicate its position in the corpus as a whole.

The composition of the corpus is shown in Figure 1, with total number of words in each category, and this figure as a percentage of the total number of words in the corpus.

2.1 Breakdown into categories

Category A - Commentary

News reports on events happening around the world. The texts are more informal than those in category B. In Perspective covers a religious topic, the From our own Correspondent reports all deal with overseas news events:

Category B - News Broadcasts

News reports of current and recent events in Great Britain and abroad. B04 contains one speaker, B01-B03 contains a main newsreader and addifional reporters. The style of the main newsreaders is more formal than that of the reporters.

Category C - Lecture type l

A lecture on economics addressed to the general public and entitled "Needs, Centralism, and Autarchy".

Category D - Lecture type II

Lectures designed to be used as part of an Open University course. D01 covers the Berlin Dada movement in Germany, and so contains some German names and words. D02 is a discussion of theology and science in 18th century France. D03 describes the development in the notation used in the representation of fractions, and contains some simple mathematical formulae.

Category E - Religlous Broadcast

Religlous services (the hymns have been edited out).

Category F - MagazIne-style reporting

Magazine-style in-depth reporting of financial news. Topics covered are: The perks of owning shares; the upgrading of state benefits; listeners' trusts; and Building Society rates.

Category G - Fiction (general)

All these texts are general fiction. G01 is a story aimed at an adult audience - "Through the Tunnel", by Doris Lessing. G02 is aimed at children aged between 5 and 10 - "Lion at School", by Philippa Pearce. G03 and G04 are stories taken from the ELT course book "Streamline English". G05 is aimed at adults - 'What shall we do if it rains?", by Graham Seal.

Category H - Poetry

Poetry readings of John Betjeman and Sir Henry Newbolt's poetry. H01-H03 are John Betjeman's readings of his own poems: Eunice, The Last of Her Order, and Harrow-on-the-Hill. H04 and H05 are actors reading Sir Henry Newbolt's poems: The Linnet's Nest, and The Nightjar.

Category J - Dialogue

Dialogues of varying degrees of informality. J01 consists of a radio discussion of notable sports events of 1986. J02-J05 are dialogues contrived to illustrate a particular facet of English (although this is not immedlately noticeable) for the Streamline English ELT course. J06 is an informal dialogue between two MA students about working abroad.

Category K - Propaganda

Charity appeals.

Category M - Miscellaneous

M01 is a sample of John Betjeman reading a section of prose: An Unpleasant Nursemaid. M02 and M07 consist of reports on road conditions. M03 and M08 are weather forecasts. M04 and M09 give details of forthcoming programmes on Radio 4. M05 and M06 are speeches delivered at degree ceremonies before the presentation of honorary degrees to Nelson Mandela and Tom Stephenson.

2.2 Speakers in the Corpus

When selecting material for the corpus, we chose only those speakers whose accent was as close to RP as possible. This was relatively easy where material from the BBC programmes From our own Correspondent and the News was concerned, as the BBC themselves require similar standards from their presenters of news or news commentary programmes. Most of the speakers in the corpus have an accent which -is close to RP; if a speaker used a particularly strong definable regional accent, they were not included in the corpus.

Of the 53 texts in the corpus, 17 contain female speakers. This represents 30 per cent of the corpus. We have tried where possible to achieve a balance between male and female speakers, and in the highly-stylised texts - poetry, religious broadcast, propaganda, and dialogue, we have managed to do this. The higher percentage of male speakers in the News and Commentary categories reflects the tendency of the BBC to use mainly male speakers in these types of programmes.

The following lists the speakers in the corpus by category.

2.3 Dates of Texts

For material obtained from BBC radio programmes the date of broadcast is given:

The samples from "Betjeman Reads Betjeman" are all dated as "slnce 1954":

For material prepared at the Media Services Unit, University of Lancaster, actual recording date is given:

The Streamline English texts were irst published in 1982. The relevant categories are:

The material obtained from the Open University unfortunately contained no information on date of composition or publication. The categories are:

2.4 Duration of extracts

The total duration of the corpus is 339 minutes 18 seconds. The average length of a sample is 6 minutes, but an individual text may vary greatly from this. Texts are not of equal length as we wanted to study complete sections of speech - having a predetermined cut-off point based on number of words or length of extract wouid have resulted in an unnatural-sounding endpoint to a sample of speech. The following gives details of extract lengths in the corpus.

2.5 Source of Material

The following samples were obtained with the permission of the BBC and are covered by a contract with them covering copyright permission

In addition to the copyright agreement with the BBC, permission to use material had to obtained from any individuals not employed by the BBC.

The Open University Educational Enterprises Ltd gave permission for the inclusion of samples D01-D03.

Oxford University Press gave permission for the Streamline English material to be included - texts G03, G04, J02, J03, J04, and J05.

Decca International gave permission for the samples of John Betjeman's work to be included - texts H01, H02, H03 and M01.

The remaining material was recorded at the University of Lancaster's Media Services Unit using volunteer speakers - M05 and M06 speeches, and the J06 dialogue.

2.6 SEC Text Details

The following table gives details of programme title (where relevant), speaker(s), and number of words for each text in the corpus

Column A = Absolute number of text in corpus

¹Date of composisition:
H04 The Linnett's Nest - May, 1924
H05 The Nightjar - May, 1925