COMPANION documents





Electrical & Information Engineering, Language Centre


  • Julie Vonwiller (Electrical & Information Engineering, APPEN Speech Technology)
  • Inge Rogers (Electrical & Information Engineering)
  • Chris Cleirigh (Electrical & Information Engineering)
Associated Colleagues
  • Brian Taylor (Language Centre)
  • Geraint Evans (Language Centre)
  • Richard Baldauf (Language Centre)
  • Keir Vaughan-Taylor (Electrical & Information Engineering)
  • Paul Meares (Web Site Design)

Pro Vice Chancellor's Office, APPEN Speech Technology

Associated Institutions

National Film & Sound Archives Canberra


A.G. Mitchell and Arthur Delbridge were both staff in the English Department of The University of Sydney. At the beginning of the project, Alex Mitchell was the McCaughey Professor of Early English Literature and Language. Soon after the commencement of the project, he became the Deputy Vice Chancellor of The University of Sydney. Arthur Delbridge was also a Lecturer in the Department of English and had at one time been a student of Mitchell. Both men had a major interest in Australian Speech and this study of theirs was very important for a number of reasons. This was the first large socio-linguistic study undertaken in the world, and was the inspiration for William Labov to undertake a similar study in the USA. The study also made use of taped recordings of speech - it did not rely exclusively on the ear of one researcher thus ensuring a greater degree of stability in the analysis. It made use of the then relatively new process of spectrographic analysis in analysing vowel data. It made use of the power of computer based statistical analysis- the analysis of the data was performed by SILLIAC - one of the world's first major stored program computers.

The best description of the reasons for the study and the outcome is to be found in the words of Mitchell & Delbridge themselves.

Mitchell A. & Delbridge A. (1946) "The Pronunciation of English in Australia" Angus & Robertson (pages 12 & 13)

Page 12

"In 1959 and 1960 a collection was made at Sydney University of the recorded voices of more than nine thousand secondary school students in the final year of their schooling, most of them about to sit for the Leaving Certificate or Matriculation Examinations. The age range was from sixteen to eighteen years. There were roughly equal proportions of boys and girls, and they were drawn in due proportion from State Schools, Catholic Independent Schools and Non-Catholic Independent Schools. All States of Australia were represented except the Northern Territory. The 327 participating schools were widely scattered throughout Australia, though they were mostly to be found in cities, or in towns of considerable size serving large rural areas. Nearly half of the students attended schools in the capital cities, though some of the pupils of city schools belonged to families normally residing in the country.

Of each Pupil it was known where he went to school, and to what type of school, his birthplace and those of his parents, and his father's occupation.

The recordings were made in each school by one of the teaching staff, who was asked to engage each of the students in conversation for about 3 minutes. The intention was to have the student talking easily about familiar matters with a person with whom he was sufficiently at ease to speak comfortably, in the manner that he would normally adopt with such a person."


Page 13

"The intention was to arrive at a classification of the speech differences within Australia, to study the regional distribution of the varieties of Australian English, and the relations between speech type in individuals and the complex of socio-economic information available."


"The general conclusion we arrived at was that among Australians of this age and level of education there are several varieties of spoken English, varieties distinguished chiefly by differences in the vowels used. These varieties hardly constitute dialects since they are not structurally diversified, except perhaps occasionally in minor and limited features. They have no geographical or even cultural boundaries firmly drawn, and speakers of the main varieties might be found anywhere within the same city or town, or even within the same family."


"These results point to the conclusion that Australia is, generally speaking, linguistically unified."

The material gathered by Mitchell & Delbridge was stored on reel-to-reel tapes, and the historical records supplied by the teachers was stored in paper form. Both were housed for many years in the Sydney University Phonetics Laboratory under the roof of the Main Quadrangle in the corner above the familiar Jacaranda tree. This was not the most ideal place for them, but when the Phonetics Laboratory moved to be with the rest of Linguistics Department, they were housed in even worse conditions under glass in a very hot corridor and the paper records disappeared into boxes and were lost. Brian Taylor, Director of the Language Centre at the time, by chance noticed the tapes in their glass case and immediately took steps to remove them to an air-conditioned room. At the same time, Julie Vonwiller a linguist working in Electrical & Information Engineering, who had seen and used the tapes and records in the old Phonetics Labs, began to search for them. Both people recognised that the material on the tapes was a valuable linguistic, social and historical record of the speech of Australians from 40 years ago. However, initially because of its location, and later because of its condition, the material has not been used since the original work.

This was a pity because at the very least the material could be used for longitudinal linguistic studies. It would be of value for social and historical research. As it stood, this material was an example of research which would not reach its full potential unless it was converted into a more readily accessible form.

Taylor and Vonwiller met and decided to raise awareness of the state of the very valuable tapes and asked the Pro Vice Chancellor Research, Roger Tanner, for assistance. Funds were made available for a project to preserve and re-issue the tapes. As a first urgent step, the reel-to-reel tapes were sent immediately to the Sound Preservation Section of the National Film & Sound Archives (NFSA) in Canberra for re-spooling and repair. At the same time the NFSA also copied the material from the tapes to DAT recordings. The reel-to-reel tapes returned to their air conditioned space, and the second phase of the project commenced when the DATs were sent to Sydney University Electrical & Information Engineering (SUEE) where the Speech Group proceeded to transfer the students speech records and the historical records to computer.


The first task was to check what exactly was in the collection of tapes. After this it was necessary to decide how much of the material would be transferred to computer for future use. Here the decision was guided by where best to put effort to gain the maximum result.


  • Historical records for most of the schools (including Aboriginal schools) on the tapes
  • Transcriptions of a large number of narratives from the University and Colleges as well as the schools
  • Some of the original analysis of the student speech
  • Records of the regional schools division
  • Records of a Value Judgement Survey
  • Instructions to research assistants
  • Notes for the first daft of the resulting book
  • Letters to institutions & schools
  • School Contact Persons list
  • Master Tape details
A most useful piece of information found in the paper records was a copy of the letter sent to the schools. It is quoted here verbatim as the instructions supplied to the teachers would be an important factor in moulding their perceptions and the outcome of the project.
Department of English
University of Sydney
June 1960
We are carrying out a piece of research by which we hope to discover with some precision the pattern of speech variation in Australia. We are also hoping to find out what the answer might be to questions such as these:
  • Are there any regional variations in Australia?
  • Do variations in speech habits correspond with social status and home background?
  • Our aim is to assemble a body of spontaneous speech from an unselected sample of pupils of equivalent age and educational grading and from every high school in Australia. We are anxious to have samples of speech from pupils in the fifth (or final) year of every high school in Australia.

    The recorded speech needs to be accompanied by information about the sex and place of birth of the pupil, the place of birth of the pupil's mother and father and the father's occupation. In order that the pupil's privacy may be respected we suggest that they be identified on the tape by number and not by name. We are attaching a sample sheet containing this information supplied to us from a school in Tasmania and we shall later send a roneoed form on which the information may be recorded.

    The method of recording is to be as follows:

    1. The sampling of pupils must be quite at random. The ideal method of ensuring a random sampling is to go down the list of pupils in the class, selecting every fifth name and continuing to go through the class list this way until the required number (of about twenty-five) is reached. It is most important that the speakers should not be selected according to the teacher's knowledge of their ability as speakers.

    2. We want samples of about twenty-five pupils from each school. If the fifth year is a mixed class the sex difference should be ignored in making the sample. If there are separate classes of boys and girls, we should like to have recordings of twenty-five chosen by the same method from each class.

    3. The recorded speech should be spontaneous and unprepared. It is our experience that the simplest way of ensuring this is the unprepared discussion. The teacher supervising the recording should simply engage the pupil in conversation for about two minutes (no more is needed). The teacher might make the discussion go more smoothly and fluently by thinking beforehand of questions and topics on which the pupil is likely to be able to speak freely without preparation.

    4. At the end of his or her interview the pupil is to be asked

    (1) to say the following words - beat boot say so high how

    (2) to say the following sentences;

    (a) Let's pick a good spot near the water and pass the morning surfing and relaxing in the sun.

    (b) The plane flew down low over the runway, then increased speed and circled the aerodrome a second time.

    5. The pupil should identify himself by the number corresponding to the cyclostyled form.

    6. We have found by experience that, after recording the first pupil in the series, there is an advantage in playing the recording back, as a check on loudness and the position of the microphone, before going on with the rest of the series.

    Business Arrangements

    Schools that are willing to co-operate will have a reel of tape posted to them. They are asked to return the completed recording by post, along with the information sheet, to Mr. A. Delbridge, Department of English, University of Sydney. The expenses of postage will be refunded by the University.

    Technical Arrangements

    So that we might know how many reels are required for the fifty minutes recording involved, we should be glad if the schools participating could inform Mr Delbridge;

    a. whether their machine is capable of recording in double tracks

    b. The speed or speeds at which their machine may be operated.


    A.G. Mitchell

    A. Delbridge

    Additional material added for Catholic Schools

    The Diocesan Director of Schools, The Reverend J.F. Slowley, has suggested to us that in writing to the Principals we may like to assure them that the project has the full approval and recommendation of the Catholic Education Office, Sydney.


    1. Aboriginal Primary School Tapes

    There are a number of tapes of Aboriginal children of primary school age. Delbridge reported that they were planning to research Aboriginal English at one time but the plan did not eventuate. These have not been extensively investigated, other than to record that they are there, and that there is one photograph of a school group.

    Current Status:

    The original reel-to-reel tapes are stored in an air conditioned room in the Language Centre. Digitised versions (DATs) were made and are stored in the Language Centre. No historical records were transferred to computer because the recordings were of very poor quality from both acoustic and linguistic information perspectives. The historical records and the tape recordings also carry the children's names.

    Aboriginal primary school tapes: 7

    Reel-to-Reel: Language Centre

    DAT: Language Centre

    Historical Records: paper only - Language Centre

    2. NZ tapes

    Arthur Delbridge advised that there were also some tapes of NZ English recorded by W.S. Ramson, the father of an ANU linguist, Bill Ramson. Ramson Senior was teaching in NZ at the time. These may be of interest to NZ researchers.

    NZ tapes = none identified.

    We did not find tapes that matched this description, but as our primary goal was to collate the Australian data, these tapes may actually be there, possibly stored with the University & College material. We did find paper records referring to the recordings.

    Current Status:

    Reel-to-Reel: Language Centre (possibly)

    3. Bernard tapes

    Dr. John Bernard recorded Australian English speakers for his Doctoral Thesis and these tapes are stored with the M&D tapes. These tapes were not digitised. Bernard tapes = identified but not counted.

    Current Status:

    Reel-to-Reel: Language Centre

    4. Primary School Children's Tapes

    Paper records were found consisting of a letter from an official in the Education Department in Brisbane about recording children, with a list of speakers, their ages, fathers occupation etc. These children seem to be 10 year olds. Arthur Delbridge has no recollection of the reason for the recording of the 10 year old QLD children. This may have been part of the earlier plan to record children - viz. the Aboriginal Children's recordings - possibly as a study into the development of the Australian accent.

    Primary Children Tapes = not found at this point.

    5. University and Teachers College tapes

    There are some recordings taken of speakers from University and Teachers Colleges which were used by the M&D team to establish their criteria for the dialect sampling for the larger student project. These were not digitised.

    University and Teachers College tapes = possibly mixed in with the Bernard tapes

    Current Status:

    Reel-to-Reel: Language Centre (possibly)

    6. Master Tapes

    The Master Tapes are collections of samples of speech from the recordings of the student dialect variants. These were not digitised.

    All of the above material was secondary to the hard core of tapes actually used in the research conducted by Mitchell and Delbridge. We recorded their existence and took no further action. The most important tapes and records for our purposes were the next item, the actual school recordings. Conservation efforts were focused on these.

    7. M&D project tapes

    SA tapes = 40 tapes of 35 schools (39 recordings)

    QLD tapes = 69 tapes of 58 schools (68 recordings)

    NSW tapes = 107 tapes of 160 schools (186 recordings)

    VIC tapes = 70 tapes of 57 schools (89 recordings)

    TAS tapes = 14 tapes of 12 schools (12 recordings)

    WA tapes = 10 tapes of 8 schools (10 recordings)

    Schools were sometimes recorded twice or the recording was split across two tapes. Ultimately there were 402 separate recordings of 330 schools on 297 tapes.

    Additional Notes:


    Despite the careful instructions, not all recordings have the requested materials. Mitchell & Delbridge established the set of 6 vowels that they considered would demonstrate divergence in speaker accents, if an accent existed. These 6 vowels were elicited through the 6 words - "beat boot say so high how". Two sentences were also designed to elicit dialect type differences in a larger context. In the digitised versions the set of words was labelled "s1" - the sentences "s2" & "s3".- and the continuous speech "n."

    For example, the recordings for student number 7736 would be labelled





    Repeat Recordings:

    Some schools were recorded twice. Several schools in the Brisbane area used a recording facility at the QLD University. Unfortunately, there was a problem as the recording supervisors did not use the script sent by M&D and thus the students did not produce a set of materials comparable to that produced by other pupils from other schools. Instead, these QLD students recorded monologues. They were re-recorded later at their home school. From examination of the paper records, it appears that the second recording did not always include the same students. As there are no names or identifying features in the records, there is no way of knowing which students were recorded twice, so each student recording of this type was given a separate UID. If it seemed that a match between recordings of a student could be achieved (based on rather unique historical records matching), then a note about the possible match was made in the miscellaneous column.

    Tape Numbering and Naming in the M&D project:

    All the tapes were numbered. A copy of the list of these tapes for all the States is in the Appendix. There are missing 'numbers' in the tape lists. Delbridge reported that some schools recorded on the wrong side of the tape, or the recording was so low in volume, or of such poor quality, that it was rendered unusable. The numbers for these tapes will be missing as the tapes were then reused and re-numbered.

    It is not uncommon for several schools to be on the one tape, especially in Queensland. This arrangement is reflected in the numbering of the DATs.

    Numbering of tapes will be, for example, n79; q60_61; s35a; s35b; s35c. The alphabet character refers to the State, n=NSW q=QLD v=VIC s=SA w=WA t=TAS. Tape number "X_Y" indicates the material for this school carries across 2 tapes; Tapes numbered 'X' 'Xa' may refer to a repeated recording. Alternatively, this may refer to the fact that there is more than one school on a tape. The name of the school will decide this.

    Paper Records:

    The historical information was investigated and a consistent set of criteria was identified for the project. This was entered onto a table with 13 fields. Paper records were produced with these in table form. In the electronic version of the records the material is kept in a relational database with a graphic user interface to facilitate interrogation of the data holdings.

    Field 1 UID-unique identification number e.g.1210
    Field 2 State e.g. Qld
    Field 3 Town/City eg Quirindi
    Field 4 School e.g Quirindi State High School
    Field 5 Date of Recording e.g 1959
    Field 6 Tape number & Speaker number on tape
    Field 7 Sex e.g. m
    Field 8 Place of Birth (of student)
    Field 9 Father's Place of Birth e.g. Somewhere in China
    Field 10 Father's Occupation e.g. storekeeper
    Field 11 Mother's Place of Birth e.g. China
    Field 12 miscellaneous
    Field 13 environment e.g. Recorded at QLD Uni

    The tape number and student number was retained in the historical record to enable users who will be working from the DATs to know which DAT has the speaker they want and where that speaker is on the tape. Numbering of the tapes for each State began at '1' in the original tapes, so the first letter of the State name was use to make each tape unique.

    Some very basic figures were extracted to illustrate some of the features of this collection. Note that there will be differences between these numbers and those given by Mitchell & Delbridge in their publications as these figures include the repeated schools and do not include recordings which have deteriorated too much to be digitised.

    A total of 7736 student records were created.

    Number State
    3885 NSW
    1291 Vic
    1234 Qld
    847 SA
    289 Tas
    189 WA

    Of these, The gender division was- 3648 male and 4087 female students.

    The recordings were made over three years, the vast majority made in 1960.

    Number  Date
    106 1958
    1742 1959
    5823 1960
    39 Not Known

    A comparison of birth places reveals that fathers were more likely to come from diverse birth places, and it was the impression of the staff that more fathers than mothers came from overseas.

    Total number of different student birth places 1174

    Total number of different father's birth places 1721

    Total number of different mother's birth places 1559

    There are 2076 separate fathers occupations listed. The largest category

    was for "farmer".

    The paper records were re-printed and published in 6 volumes for which this book is a companion.