Aims of the Ethnographic

1

• Aims of the Ethnographic
• Eresearch project
• Nick Thieberger
• University of Melbourne
• Sponsored by the Pacific And Regional Archive for
  Digital Sources in Endangered Cultures
  (PARADISEC)
• Funded by the Australian Research Council
• Supported by HCSNet

2
ARC Grant SR0566965 Eresearch ethnographic media

• Summary of aims
• 1 Develop a methodology for collaborative
  research based on large digital media datasets,
  particularly video, using highspeed networks and
  large data repositories.
• 2 Ensure that digital research collection
  management is able to provide access and ensure
  longevity and reusability of research data by
  conforming to relevant standards.
• 3 Enable access via authentication to
  distributed research datasets using embedded
  rights metadata.
• 4 Ensure that the participant testbeds allow
  interoperability of data models and provide
  training so that future such datasets can
  similarly interoperate.
• 5 Provide the means to distribute richly
  annotated ethnographic media not only to
  researchers but also to its source communities in
  rural, regional and remote locations.

3
Participants

• Melbourne Thieberger (lead), Wigglesworth,
  Nordlinger, Evans, Hajek
• Sydney Simpson, Barwick, Marett, Corn, Foley
• ANU Rumsey, Bowden, Buchhorn, Hungerford
• Macquarie Johnston, Schembri
• CSIRO Pfeiffer
• DSTC/Qld Hunter
• AIATSIS McConvell
• School of Oriental and African Studies, London
  Austin
• University of Alaska at Fairbanks Holton, Alaska
  Native Language Center
• University of Texas at Austin Johnson, Archive
  of the Indigenous
• Languages of Latin America
• Galiwinku Indigenous Knowledge Centre Gumbula

4
Projects

• 1. Pacific And Regional Archive for Digital
  Sources in Endangered Cultures (PARADISEC)
• 2. Aboriginal Child Language Acquisition (ACLA)
• 3. Corpus of grammar and discourse strategies of
  deaf native users of Auslan (Australian Sign
  Language) (ELDP Auslan)
• 4. Melbourne University Reciprocals Project
  (MURP)
• 5. Waima'a (East Timor) documentation project.
• 6. National recording project for indigenous
  performance in Australia

5
Rationale

• The proposed research will implement and, if
  necessary, customise tools for collaboration and
  access to audiovisual data of the testbed
  projects using tools and software developed by
  the ANU Internet Futures project (AccessGrid),
  CeNTIE (Centre for Networking Technologies for
  the Information Economy) (Annodex) and UQ
  (Vannotea). These innovative applications will
  take advantage of Australian higher education's
  world-class storage and networking capacity
  (CeNTIE, GrangeNet, APAC).
• Free and open source software
• Archival data

6
Initial program (as per the application)

• Transcoding of existing annotations in ELAN,
  Transcriber, and CLAN schemas (currently used by
  participating projects) to common Continuous
  Media Markup Language (CMML).
• Batch transcoding of media stored in APAC
  repositories to open media formats Ogg vorbis
  (audio) and Ogg theora (video), the streaming
  formats addressed by CSIROs Annodex software.
• Establishment of streaming architecture and
  choice of appropriate transport for implementing
  streaming of media files amongst participants.
• Testing of appropriate telecollaboration
  environments for annotation Vannotea and Access
  Grid
• Staged implementation of text-based browsing and
  streamed excerpt delivery for participating
  projects via Annodex.
• Authorisation and authentication of access to
  annotation.

7
Issues

• Location of test datasets (online server)
• Implies persistent identification and location
• Use of standard metadata repositories to enable
  location of data
• Interoperability with existing systems (metadata
  and data structure)
• Ease of use is critical otherwise practitioners
  won't use it

8
Issues

• Meta-ethnographic observation
• Solutions exist and are widely known in computing
  disciplines
• Implementation in a way that practitioners can
  access is yet to be achieved
• Aim of this project to overcome this problem

9
The development of linguistic documents entails
the prior creation of a corpus of primary
material for which relationships need to be
tracked
Grammar
Dictionary
Texts
Media - audio - images - video
transcripts historical sources maps etc .
10
Typical linguistic fieldwork

• Media recordings
• Transcribed in a standard schema (e.g. using
  Elan, Clan, ITE, or Transcriber)
• Output as time-aligned text to Toolbox for
  interlinear annotation
• Output from Toolbox as interlinear version in XML

11
Ideally

• Ability to have multiple files in one of e.g.
  four schemas all viewable by the same browser (ie
  no transcoding)
• Persistent file id allows citation and linking
• Analytical apparatus built on corpus of
  annotations
• Concordance
• Playlists
• Copying segments from the corpus with their
  citation form (e.g. media name, start time, end
  time)

12
Standalone model

• Audiamus allows the user to interact with
  linguistic data
• Need to make a more generic tool with the same
  functions

13
Sample fragment

• lt?xml version"1.0" encoding"UTF-8"?gt
• lt!DOCTYPE TEXT PUBLIC "-//CNRS-LACITO//DTD
  Archivage//EN" "http//lacito.Archivage.vjf.cnrs.f
  r/archives/All/Archive.dtd"gt
• ltTEXT id"ERKRIRIAL" xmllang"x-sil-ERK"gt
• ltHEADERgt ltTITLE xmllang"en" /gt
• ltSOUNDFILE href"RIRIAL" /gt lt/HEADERgt
• ltSgt
• ltFORM kindOf"sentence"gtIpiatlak nmatu
  iskei,lt/FORMgt
• ltAUDIO start"0" end"2.3" /gt
• ltWgtltFORM kindOf"morph"gtilt/FORMgt
• ltTRANSL xmllang"en"gt3sgRSlt/TRANSLgtlt/Wgt
• ltWgtltFORM kindOf"morph"gtpiatlaklt/FORMgt
• ltTRANSL xmllang"en"gthavelt/TRANSLgtlt/Wgt
• ltWgtltFORM kindOf"morph"gtnmatult/FORMgt
• ltTRANSL xmllang"en"gtwomanlt/TRANSLgtlt/Wgt
• ltWgtltFORM kindOf"morph"gtiskeilt/FORMgt
• ltTRANSL xmllang"en"gtonelt/TRANSLgtlt/Wgt
• ltTRANSLgtThere was this womanlt/TRANSLgt lt/Sgt
  .....

14

• http//lacito.archivage.vjf.cnrs.fr/servlet/

15

• Annotation as a means of establishing
  relationships between existing objects
• These objects need to be citable
• Persistent identification
• Persistent location
• Standard cataloging information - metadata

16
Wurm collection, Solomon Islands, 1979.
Digitised cassette tape with page image of
transcript, and Wurms language map
17
Relationship between objects needs to be tracked
PARADISEC archival objects
Media files
Timecodes
Image file 1
Image file 2
Image file 3
Image file 4
Image file 5
Image file 1 is a transcript of offset points 10
seconds to 2 minutes 30 seconds of media file
SAW2-011
18
Relationship between objects needs to be tracked
Field recordings and associated items
Media files
Timecodes
Story 1
Story 2
Story 3
Story 4
Story 5
Each story has interlinear versions, with
relations required to speaker information (name,
age, sex, clan etc.) and other metadata
19
Summary

• Need to
• capture and store relations between objects
  (Resource Description Framework, RDF)
• fix objects with persistent identification
• allow users to locate objects via sufficient
  metadata
• authenticate authorised users
• capture and store annotations made by authorised
  users

20

• Sponsored by the Pacific And Regional Archive for
  Digital Sources in Endangered Cultures
  (PARADISEC)
• Funded by the Australian Research Council
• Supported by HCSNet