MANUAL OF INFORMATION

TO ACCOMPANY

THE AUSTRALIAN CORPUS OF ENGLISH (ACE)

MACQUARIE UNIVERSITY

BY

PAM PETERS

WITH THE ASSISTANCE OF

ADAM SMITH

MANUAL TO ACCOMPANY THE AUSTRALIAN CORPUS OF ENGLISH (ACE)

Introduction

The Australian Corpus of English (ACE) was compiled in the department of Linguistics at Macquarie University NSW Australia, from 1986 on. It was supported by a small grant 1988-9 from the Australian Research Grants Council, and by a series of grants from Macquarie University. Other support came from the National Languages and Literacy Institute of Australia and the University of New South Wales. The project was conceived by Pam Peters, Peter Collins and David Blair, and was carried through with the help of a number of research assistants, notably Alison Moore, Elizabeth Green, Robert Jenkins, Catherine Martin, Diana Grace, Heather Middleton, Wendy Young and Adam Smith. Computational help and advice was provided by Harry Purvis and Steve Cassidy, and the project enjoyed continuous infrastructure support from Macquarie's Speech, Hearing and Language Research Centre.

Contents

Introduction
Rationale
Sampling procedures : overview
Subjects and genres included within categories
Sample size and Coding
Appendix 1 : Corpus versions
Appendix 2 : Published papers associated with the Corpus
Textextracts

Rationale

ACE was the first systematically compiled heterogeneous corpus in Australia, designed to support a variety of linguistic research. Interest in the differentiation between Australian, British and American English meant that a corpus modeled on the Brown and LOB corpora would provide ready comparisons. It would also serve as a strategic sample of current Australian English, and as a reference corpus for comparisons with more specialised, homogeneous corpora in Australia.

ACE matches the Brown and LOB corpora in most aspects of its structure and constituency, so that direct interdialectal comparisons can be made on a comparable range of printed genres. (The few small points of difference are outlined below pp. 3 to 8.) Yet the desire to create an up-to-date corpus of Australian English prompted the decision not to match Brown and LOB chronologically, ie. with data drawn from publications of the early 1960s. Instead, ACE consists of material from 1986. A time difference is therefore inherent in any regional intercomparisons with Brown and LOB, though that may itself be of considerable interest in showing the direction of influence in the latter part of this century. The twenty-five year difference in fact allowed us to match rather more categories of publishing than would have been possible had we attempted to create a retrospective corpus of Australian publications of the 60s (as LOB did). Independent southern hemisphere publishing has increased steadily since World War II, yet even in 1986 the range of locally published novels was limited and insufficient for the quota required by the Brown/LOB model. It was topped up with a higher proportion of extracts from short stories than were used in the model corpora. (See below, Table 4, p.5.)

Sampling procedures : overview

The prime objective in compiling ACE was to match the balance of genres represented in Brown and LOB, and to create a more or less equivalent set of 2000-word samples in each category. This provided quantitative targets in each of the fifteen categories of Brown and LOB, and the number of samples in the ACE categories A to J are closely matched with them, as shown below in Table 1. The fiction categories in ACE are slightly different in their constituency, for reasons explained below, p.8, but the total of fiction samples remains the same.

Table 1. Makeup of the three corpora

Within each corpus category, the sampling procedures were mostly strategic rather than random, because of the felt need to match subgenres and subject areas where possible. In some categories, e.g. fiction, the corpus requirements were such that we sampled almost every Australian monograph published in that year, and so the representation in ACE is almost total. Where there was a choice, as with the selection of monographs in some nonfiction categories, we gave preference to those which were held in multiple libraries in several states, and therefore probably had more readers and more impact. Among the serials, both popular and scholarly, the selection was usually dictated by subject, to insure a spread of interests and disciplines like the broad range captured by our predecessors.

• Newspaper samples (categories A - C) were taken from the full range of broadsheets and tabloids published in each capital city, and the limited range of national newspapers. The count of independent newspapers was in fact larger in 1986 than at any time since, what with closures and amalgamations. The dearth of national papers (only 4) has the ACE collection weighted towards the metropolitan, and more like that of Brown than LOB. The relative circulations of individual papers was taken into account in deciding whether to take one or more samples from each. Thus the Melbourne Sun-News Pictorial with over half a million copies sold provided five samples of reportage to the one from the Canberra Times with only 45 000. By the same token we wanted some representation of newspapers in all capital cities, and so made a point of sampling from Darwin’s Northern Territory News with its tiny circulation of 18 000.

Table 2. Sampling of Australian newspapers for categories A,B,C

( * indicates tabloid format, but not necessarily low-brow journalism.)

We also targeted both Sunday and weekly papers , but the predominance of Sunday papers in Australia means that ACE is closer to LOB in this respect, as shown in Table 3.

Table 3 Sampling of reportage, editorial matter and reviews from daily, weekly and Sunday newspapers in the three corpora.

• Magazine samples for categories E, F and G were selected with the aid of the biannual Margaret Gee's Media Guide, which provides circulation figures for the benefit of prospective advertisers and other interested parties like language and media researchers. The Media Guide also provided a useful overview of subject areas from which we might hope to match those of LOB and Brown, or consider alternatives.
• Government publications (category H) were sampled with the aid of the Australian Government Publishing Service's quarterly catalogues, which facilitated the selection of parliamentary and government reports. Some samples chosen to represent state government issues, as well as those of federal government. Other institutional publications were taken ad hoc, though with the aim of including material from beyond New South Wales.
• Learned and scientific monographs (category J) were identified with the help of the Australian National Bibliography, and selected according to their representation in multiple libraries. For serials, the catalogues of the NSW State Library and university libraries in Sydney provided listings from which the Australian-based journals could be selected. The shortage of academic journals in some fields (eg. music, philosophy) required us to use a slightly higher ratio of monographs to journals than in Brown and LOB (see Table 4).
• Fiction samples (categories K - W) were identified from books listed in the Australian National Bibliography, though the unsystematic annual reporting to it meant that it had to be supplemented by (Australian) items featured in the commercial magazine Australian Bookseller and Publisher. Short story samples were selected from literary magazines, or by tracing the work of identified Australian authors through to their vehicles of publication. They make up the numbers where there was a shortfall of fiction monographs, there being limited amounts of publishing in this genre. The rationale for adding categories S and W is explained below, p.8.

The overall balance of samples from books/monographs to articles/short stories is shown in Table 4. Further details on sampling are discussed with the individual categories below.

Table 4 Monographs v. articles/short stories

Subjects and genres included within categories

• Category A (Newspaper reporting) In Brown and LOB, the 44 samples in this category were spread over six types of reporting: political, sports, financial, spot news, society and cultural. The first four of these were paralleled in the Australian press, but the lack of obvious counterparts to the last two forced us to seek a substitute. This was found in the regular articles (indeed whole supplements) on material culture (e.g. home decor), material which we called "Living". This type of reporting took the place of the two we could not match from the Brown/LOB inventory. The proportions of each type of reporting are shown in Table 5.

Table 5: types of reporting represented in the three corpora

• Category B The collection of editorials from Australian newspapers is necessarily biased towards the broadsheets, because of the brevity or absence of regular editorials from some of the tabloids. Letters to the editor were usually available in both types of paper, but the tabloid-style personal editorials, like the main editorial, were often very brief if there at all! Several composite samples had to be made up, normally with material collected from other tabloids in the same state.
• Category C Reviews were obtained from both broadsheets and tabloid-style papers. As in Brown and LOB, individual samples contain several reviews on diverse topics. Many were concerned with the performing arts (drama, classical and contemporary music); others reviewed TV programs, compact disks, movies and books.
• Category D Most of the material in this category is devotional rather than theological (this being included in category J). Some of it came from tracts, as well as periodicals and books. The category contains samples from all the major Australian denominations, Catholic and Protestant, as well as the Bahai faith.
• Categories E and F All samples in category E came from popular (ie. mass circulating) magazines, which are the typical vehicle for instructional texts. The fields represented range from music and food to pets and farming. For category F, texts could be extracted from both magazines and monographs. They are typically informative, with nontechnical accounts of issues in many fields, as shown in table 6

Table 6 Subjects represented in Categories E and F

• Category G Belles lettres, biography Discursive non-technical prose is the substance of this category, but like the creators of LOB, we found challenges in distinguishing raw material for this and several other categories. The subject matter of Category G samples, especially the "general essays", sometimes seemed to approach that of Category F, but we took as a criterion for Category G the principle that its non-biographical samples should treat their subjects thematically, ie. not as narrative. Occasionally too the subject matter resembled those of Category J samples. Here the intended audience and degree of specialisation served to discriminate between them (a J sample being written for academic readers, published in a specialist journal and often technical in its lexicon). Another close frontier to discriminate was between Category G samples concerning the arts, and the reviews of Category C. For G we required discussion of a broader range of objects than was typical of a review.

Table 7 Genres included in Category G

• Category H Government and Institutional This category contains 5 samples of Hansard and 20 from a variety of government publications put out by departments and parliamentary committees. There are in addition 2 samples from university publications, and 3 from other institutions - Red Cross, CSIRO and a museum.

• Category J Learned and scientific This large category ranges over many academic fields, matching Brown and LOB where possible, but allowing that the academic fields which are the focus of publishing in Australia are not as many as those in the northern hemisphere. There are nevertheless particular strengths in agricultural and environmental science, and in social and local history. Certain fields of inquiry have no doubt developed since the 1960s (eg. semiotics, information technology), and on this account too it was inevitable that the ACE collection should be a little different in its detail from its predecessors. In broad disciplinary coverage, meanwhile, ACE maintains the same proportions, as Table 8 shows.

Table 8. Academic disciplines of Category J

• Categories of fiction K-W Matching the fiction categories of Brown and LOB posed a number of problems. Certain kinds of fiction were underrepresented owing to their scarcity on the Australian publishing scene in 1986, including mystery/detective (category L) and science fiction (category M). Some types of genre fiction, eg. western, romance of the Mills and Boon variety, are not published at all in Australia. There are Australian authors of such things, but their publishers are in the US and the UK, and being edited outside Australia meant that their work was not suitable for inclusion in ACE. To cope with this, and to include the nearest kinds of fiction which are written and edited/published in Australia, we allowed that bush narratives might be legitimately added to Category N (adventure and western); and that "romance" could include writing concerned with fantasy as well as novels focusing on human relationships whether romantic or not. The category of humor (R) was very well represented, and we took in extra samples there to make up the shortfall elsewhere. The shortfall was also balanced by the addition of two further categories of fiction: historical sagas (Category S) and Women's writing along feminist lines (Category W). These two new domains of fiction were blossoming in Australia in the mid-80s, as no doubt in other parts of the world, and it seemed important to add them as distinct categories, rather than merge them in an enlarged category of "general fiction" (Category K). In ACE, Category K is much like that of Brown and LOB, though the samples are more highbrow because the majority come from literary magazines.

Sample size and Coding

Each sample is notionally 2000 words, the counts being done via WORD for WINDOWS 6, with the coding excluded. The samples contain a minimum of 2000 words, though most are a little more than that in order to conclude the sentence. A few, especially from the disciplines of mathematics and science, have a larger buffer because of the high proportion of formulae in them, which tended to fragment the discourse.

Texts are in the ASCII format, with each category and each sample prefaced by coded identification. The samples carry details of the sources from which they were obtained and individual headings or titles. Within the texts there is a limited amount of markup, for certain discrete elements such as bylines or formulae, and for certain nonalphabetic symbols, both of them in SGML-style codes. Mathematical and scientific symbols beyond those of the Greek alphabet were covered by a generic annotation (&symbol;). The markup <note></note> was used for a variety of extra corpus material, both editorial comment and components of the text itself which stood outside the ongoing discourse, such as extended quotations, graphs or tables.

Format/Comment Coding

<section></section> at the start and end of each category

<title></title> around the title of each category

<sample></sample> around each sample

<subsample></subsample> around any subsample

<id></id> around the sample number

<source></source> around the name of the source from which the sample was taken

<h></h> around the heading or title of each sample/subsample <bl></bl> around any bylines

<list></list> around extended lists

<note></note> to enclose any additional comments or text not be included within the wordcount

to <misc></misc> around unpunctuated or irregularly punctuated sections

* replacement of typographic or spelling errors in original e.g. assessment*assesement

+ replaces hyphen at line-break e.g. proces+sors &formula; replacing any complex formula

&symbol; replacing any symbol not listed below individually

Symbol Coding

&amp; & &epsilon; e

&pound; £ &theta; q

&bullet; • &eta; h

&deg; ° &zeta; z

&reg; ® &lambda; l

&para; ¶ &caplambda; L

&ohm; W &mu; m

&alpha; a &rho; r

&beta; b &sigma; s

&gamma; g &upsilon; u

&capgamma; G &psi; y

&delta; d &capomega; W

&capdelta; D

APPENDIX I

Corpus versions

ACE exists in two versions:

ACE I This is the full version, containing all 500 samples, available for interrogation via CD ROM or Internet connection

ACE II This reduced version includes 75% of ACE I, that is 375 samples available for unrestricted use. (The remaining 25% could not be copyright-cleared for use throughout the world.) The samples excluded are listed below:

APPENDIX II

Published Papers Associated With The ACE Corpus

1. Peters, P. Towards a corpus of Australian English. ICAME JOURNAL No.11 (1987), 27-38. (ICAME = International Computer Archive of Modern English).

2. Collins, P. and Peters, P. The Australian corpus project in Corpus linguistics, hard and soft, ed. M Kyto et al. Amsterdam: Rodopi (1988), 103-120.

3. Collins, P. Computer corpora in English language research: a critical survey. AUSTRALIAN REVIEW OF APPLIED LINGUISTICS 10 i (1987), 1-19.

4. Peters, P., Collins, P., Blair, D. and Brierley, A. The Australian corpus project, findings on some functional variants in the Australian press. AUSTRALIAN REVIEW OF APPLIED LINGUISTICS 11 i (1988), 22-33.

5. Collins, P. The semantics of some modals in contemporary Australian English. AUSTRALIAN JOURNAL OF LINGUISTICS 8 (1988), 233-258.

6. Peters, P. and Fee, M. New configurations: the balance of British and American English features in Australian and Canadian English. AUSTRALIAN JOURNAL OF LINGUISTICS 9 (1989), 135-147.

7. Peters, P. The Australian corpus project: word punctuation in newspapers, in Frontiers of style: proceedings of Style Councils 87 and 88, ed. P.H. Peters. Sydney: Dictionary Research Centre, Macquarie University (1990) 72-79.

8. Peters, P., Purvis, H., Martin, C. and Jenkins, R. Word frequencies from the Macquarie corpus: the newspaper files. WORKING PAPERS OF THE SPEECH, HEARING AND LANGUAGE RESEARCH CENTRE, MACQUARIE UNIVERSITY (1990) 13-92.

9. Green, E. and Peters, P. The Australian corpus project and Australian English. ICAME JOURNAL no.15 (1991) 37-53.

10. Collins, P. The modals of obligation & necessity in Australian English, in English Corpus Linguistics, edd. Aijmer and Altenberg. London: Longman (1991) 145-165.

11. Peters, P. American & British English in Australian Usage, in Style on the move: proceedings of Style Council 92, ed. P.H. Peters. Sydney: Dictionary Research Centre, Macquarie University (1993) 20-27.

12. Peters, P. Corpus evidence on some points of usage, in J. Aarts et al. edd. English language corpora: design, analysis and exploitation Amsterdam: Rodopi (1993) pp. 247-256

13. Peters, P. American and British influence in Australian verb morphology, in U. Fries et al. edd. Creating and Using English Language Corpora Amsterdam: Rodopi (1994) pp. 149-158

14. Collins P. Get- passives in English World Englishes 15:1 (March 1996) pp. 43-56

15. Peters, P. Comparative insights into comparison World Englishes 15:1 (March 1996) pp.57-68

16. Peters, P. and Delbridge, A. Fowler’s Legacy in E. Schneider ed. Englishes Around The World vol. 2 Amsterdam, John Benjamins (1997) pp. 301-318

Textextracts

A B C D  E F G  H J K L  M-N P R S  W