Using Virtual Reality for Interpreter-mediated Communication and Training

Cyberworlds International Conference – September 2012
Ysgol Gwyddorau Cyfrifiadurol
School of Computer Science

Using Virtual Reality for Interpreter-mediated
Communication and Training

Panagiotis D. Ritsos1, Robert Gittins1, Sabine Braun2, Catherine Slater2 and Jonathan C. Roberts1

[1] School of Computer Science, Bangor University, UK - {p.ritsos, r.gittins, j.c.roberts}@bangor.ac.uk
[2] Centre for Translation Studies, University of Surrey, UK – {s.braun, c.slater}@surrey.ac.uk

Lifelong Learning Programme – Project 511862-LLP-2010-1-UK-KA3-KA3MP - The IVY project has been funded with support from the European
Commission. This publication reflects the views only of the authors, and the Commission cannot be held responsible for any use which may be made
of the information contained therein.

Presentation Outline

 To present purpose of project IVY – Interpreting in Virtual Reality and the

IVY Virtual Environment (IVY-VE)

 To present the strategic decisions, resulting design and implementation progress
to date, towards the creation of a prototype

 To provide an overview of the main features of our prototype

 Comment on preliminary evaluation and pinpoint identified limitation

 To allow for discussion on future development

Ysgol Gwyddorau Cyfrifiadurol The Project IVY Consortium

University of Surrey (UK)

Uniwersystet im. Adama Mickiewicza (Poland)

University of Cyprus (Cyprus)

Steinbeis GmbH & Co. KG für Technologiertransfer (Germany)

Bangor University (UK)

Eberhard Karls Universität Tübingen (Germany)

Bar Ilan University (Israel)

Project IVY – Scope

 The rise of migration and multilingualism in Europe requires professional

interpreters in business, legal, medical and many other settings.

 Future interpreters need to master an ever broadening range of interpreting
skills and scenarios – training for which is often difficult to achieve with
traditional teaching methods.

 Project IVY employs 3D virtual world technology to create an educational
space that supports the acquisition and application of skills required in
interpreter-mediated communication.

 Project IVY uses existing interpreter resources – audio and video material from
previous video conferencing research.

IVY VE – Requirements

 To provide an intuitive, easy to use interface within a Virtual World for

accessing multimedia material created for interpreting training and simulation.

 To allow easy scenario management by users who often are not experts in
commuting (i.e., can not/should not write code)...

 … meaning the creation and modification of existing scenarios in terms of their
multimedia content, requiring basic CRUD functionality.

 To enable limited dialogue and monologue synthesis, resulting in the
enrichment of the corpora with different language combinations of existing
scenarios.

IVY Virtual Environment (IVY-VE) in a nutshell

 A dedicated, adaptive 3D virtual environment for

o interpreting students and
o future clients of interpreters

 Supports range of virtual interpreting scenarios (e.g., ‘business meeting’) that can
be run in different modes:
o Interpreting (& Learning Activity) mode, where students can practice using
dialogues and monologues
o Exploration mode, where clients can learn about interpreting
o Live interaction mode, where both groups can engage in role plays

 Uses multilingual video/audio-based content for interpreting scenarios, by
adapting existing multimedia corpora from the LLP project BACKBONE and the
ELISA corpus, and creating three new corpora in Greek, Russian and Hebrew;

 Supported by two sets of pedagogical material for interpreter students and
(future) ‘clients’, e.g. awareness-raising and interpreting exercises.

Ysgol Gwyddorau Cyfrifiadurol Project IVY – Scenario forms

Dialogue

Monologue

IVY Virtual World – Second Life

 Second Life was chosen as the Virtual World for our first prototype.

 Exploration of alternatives, such as OpenSim, WebGL and Unity will follow in
the future.

 Second Life compared to alternatives (OpenSim, ActiveWorls etc) offers:

o Large community, various add-ons, plugins and examples of customisations.

o A platform for social interaction and education, used by numerous institutions,
colleges, universities – thus increasing chances of exposure.

o Accessibility via public servers and it does not require that you run the VW yourself.

IVY VE – Lack of Instancing & Scalability

 Due to limits on the number of primitives available to the IVY Island and the

lack of instancing mechanisms in SL, IVE-VE uses a collection of unique, in-
world locations for each type of scenario (e.g., Classroom, Meeting Office).

 Therefore, a scenario may share its location with another, being carried out in a
similar setting.

 In order to maintain consistency in the virtual world only one scenario can be
played per location at a given time.

 Upon a scenario launch by a user, all scenarios sharing the same location
become unavailable for other users.

 Once the user exits the selected scenario, all scenarios sharing the same
location become available again.

IVY VE – Design Strategies

 Script things in Second Life using LSL

+ Comparable aesthetics and better integration with the native SL GUI
+ Graphics appear crispier, clearer

- Scripting can be fairly static and requires coding
- Communication with DBs is fairly limited in terms of size of info that can be
transmitted.

 Rely on web application technologies as interface creation, database
connectivity and overall flexibility surpasses LSL. Therefore…

 …either embed SL in a web application
o Popular notion, no web viewer supported from Linden Labs.
o Only Canvas from Tripodean Technologies appears to exist at the moment.

 …or ‘embed’ the web application within SL, using HTML on a Prim.

IVY VE – Implementation

 Our chosen implementation strategy aspires to merge useful features from both

alternatives, resulting into a hybrid solution.

 One module consists of a web application with two entry points. One entry point
remains independent of SL and is used by content managers to manage scenarios
and users’ information, offering basic CRUD functionality.

 The other end is viewable within Second Life, in the form of a ‘Heads-up-
Display’ (HUD), populated from a database, showing available scenarios to the
users, having a player functionality and initiating in-world teleport events.

IVY VE – User Classes and Characteristics

 Separating user roles allows controlled access to different parts of the application

as well as means of monitoring scenario selection and execution.

 The system’s user classes and their respective role descriptions are:

o Interpreters/Users, whose purpose is to explore, participate and exercise with the
scenarios in Second Life.

o Observers, whose purpose is to observe other users in Second Life.

o Content Managers who are responsible for user and scenario management.

IVY VE – The IVY Island

 Locations created according to the corpus requirements, trying to keep prim-count

to a minimum.

 Reception serves as a central hub and a HUD pick-up point.

IVY VE – Scenario Actors

 We populate the scenarios with additional ‘actors’. Currently these actors are

manually placed in each location, using Second-life ‘robots’ (bots) from
Pikkubot and Thoys.

 Bots are controlled either by using in-world chat to issue commands directly to
the bots or through a dedicated server’s telnet prompt.

 We currently use animation overrides, to make them appear life-like.

 Our aspiration is to create a service that talks to the bots and relays scenario
specific information and teleport commands.

 However… sound does not appear to originate from the bots.

IVY VE – Web Application

 The IVY web application is build using the Appfuse 2 open source project

(appfuse.org).

 Appfuse, built on the Java platform, uses industry-standard features, such as
Apache Maven integration, JPA support for database operations and popular
web frameworks such as Spring and Apache Struts, employed in this example.

 Appfuse comes with out of the box features, needed in IVY-VE such as:

o Generic CRUD backend
o Authentication and authorization,
o User management
o Strong Internationalization support

 Our prototype is deployed using Apache Tomcat 6.x and uses the MySQL 5.x
database.

IVY VE – Audio File Management

 The IVY-VE uses audio extracts (segments), in MPEG-2 Audio Layer III

format, from the LLP project BACKBONE, wrapped in XSPF (XML Shareable
Playlist Format) play-lists (scripts) and played within pre-fabricated scenes.

 Audio segments are uniquely named and can be interchanged — within each
script — to derive further language combinations of that scenario.

 Actors may speak for more than one consecutive segments, allowing greater
flexibility in creating scripts where one talks for extended periods.

 However, it is assumed that only one actor talks per audio segment and there is
no overlap between actor’s speech.

 Each script has textual information associated with it, such as brief content
information, scene description and domain keywords.

IVY VE – Heads Up Display I

 The HUD is built using the jQuery JavaScript library, displaying the list of

scenarios in the database as a drill-down menu.

 It is normally attached to the bottom left corner of the user’s viewport.

 Audio is being played by means of a Flash player, parsing the XSPF playlists
upon scenario selection.

 Navigation through the island is performed using slurls, providing direct url-
like teleport links to locations within the virtual world.

 Each slurl is being called upon scenario launch, triggering the native SL-client
teleport interface.

IVY VE – Heads Up Display II

 Login, form selection and language combination selection views

 Free Scenarios views – Scenario Info & Player

 Locked Scenarios views

IVY VE – Administrator’s Panel I

 The management console allows Content Administrators to easily populate the

scenario database and create XSPF playlists.

 New scenarios are created through a form, where administrators provide textual
information (title, language combos, participants gender etc.) as well as an
ordered list of the audio tracks in each scenario.

 Scenario textual info is stored in the DB and XSPF playlists on a separate playlist
inventory.

 A separate listing of all the scenarios in the system allows administrators to see
which user is currently working on each scenario and also fire-up their SL client
and teleport to that location.

IVY VE – Administrator’s Panel II

 Scenario Upload Form

 Scenario Listing
(with Teleport functionality)

IVY VE – Preliminary Functional Evaluation

 A preliminary evaluation of the current prototype was done by nine

interpreting and two virtual world experts, using talk-aloud, try-out sessions
in Second Life, followed by a discussion with the assessor.

 The evaluation focused on the HUD functionality and in-world locations.

 The admin panel was not evaluated at this stage; however it has been
successfully used, by content administrators, for the past four months to upload
scenarios in our system.

 Overall opinion was quite positive and users with very limited experience in
virtual worlds, gaming or similar environments felt comfortable using IVY-VE.

 However, a series of limitations of the current system where pinpointed…

IVY VE – Current Limitations

 Some users focused too much on the HUD, not paying attention to the world.

 No sound directionality and no visual cues on who is speaking.

 Interpreters were enthusiastic they have a tool – but does that generates bias?

 Lots of comments regarding aesthetics of locations – all participants observed
and commented on the quality of locations using photorealistic graphics.

 No zoning, instancing and replication like in games.

 Limitation on available ‘prims’ affects world scalability

 … and consequently scenario management and execution, actor placement.

 Currently bots are manually placed to required scenarios – nowadays we use
many bots and manage locations to cover all gender combinations in our corpus.

 Sound is heard only from the player controller and is not broadcast back to the
world – hence observers do not hear it.

IVY VE – Future Work

 Two evaluation cycles will follow, focused on interpreting students and potential

clients of interpreters during the autumn of 2012 and will attempt to get some
feedback on the IVY-VE usability and resulting sense of immersion.

 HUD is currently being updated to accommodate the additional modes, namely
exploration and live interaction.

 Interpreting mode will also be enhanced with a series of exercises, both generic to
dialogues and monologues, as well as specific to particular scenarios which have
an inherit challenge in interpreting practice.

 Enhancement of the current system with dedicated service components to allow
puppeteering of bots,

 Use student/client feedback and experience from using IVY-VE to design a
new bespoke system…

 ..while exploring alternative technologies that allow tighter integration with
current web-based scenario system, e.g., Unity and WebGL.

Ysgol Gwyddorau Cyfrifiadurol IVY VE – Forthcoming Events

 Virtual Learning Technologies (VLT) 2012, Bangor, Wales, 31st October 2012

 Exploiting Emerging Technologies to Prepare Interpreters and their Clients
for Professional Practice, London, 23rd November 2012

For more info visit:
Consortium website http://www.virtual-interpreting.net/Seminar.html

…or the Bangor IVY partner website http://www.vmg.cs.bangor.ac.uk/IVY/


Thank you!

Using Virtual Reality for Interpreter-mediated Communication and Training

More Related Content

What's hot (6)

Similar to Using Virtual Reality for Interpreter-mediated Communication and Training (20)

Recently uploaded (20)

Using Virtual Reality for Interpreter-mediated Communication and Training

Editor's Notes