In this newsletter:
   

-   High School students use LDC data  -
   

New publications:
   

-  GALE Phase 2 Chinese Broadcast Conversation       Parallel Text Part 1   -
   

-   Greybeard   -
   

-   Manually Annotated Sub-Corpus Third       Release   -

High School students use LDC data
     
       A team of students at       Thomas Jefferson High School for Science and Technology in       Alexandria, VA, USA, have used an LDC database for the development       of a device to help autistic children recognize emotions.  This team was funded by a       grant from the Lemelson-MIT InvenTeam
        Initiative Program.  InvenTeams are
      groups of high school students, teachers, and mentors that receive       grants up to US$10,000 each to invent       technological solutions to real-world problems.

The team set out to invent an emotive aid in       the form of a bracelet that uses a computational algorithm to       extract emotional signatures from speech and display expressed       emotions in real-time during a conversation. Potential       beneficiaries include children with autism, Asperger’s syndrome,       or similar diseases that impair the ability to detect emotion.  The algorithm employed       machine learning and neural network-based techniques to improve       accuracy and efficiency relative to current methods.
     
      The students used speech samples from the LDC database, Emotional Prosody Speech and           Transcripts (LDC2002S28) as well the Berlin Database
        of Emotional Speech for training and testing their       algorithm. Although the samples proved to be too small to produce       an algorithm with a high degree of accuracy, the team's algorithm       did demonstrate some degree of success.  The students will present their       results at Eurekafest       at MIT in June.

LDC thanks the InvenTeam’s teacher, Mark       Hannum, and group leader, Suhas Gondi, for contributing to this       article.

New publications

(1) GALE Phase
        2 Chinese Broadcast Conversation Parallel Text Part 1 was       developed by LDC. Along with other corpora, the parallel text in       this release comprised training data for Phase 2 of the DARPA GALE       (Global Autonomous Language Exploitation) Program. This corpus       contains Chinese source text and corresponding English       translations selected from broadcast conversation (BC) data       collected by LDC in 2006 and 2007 and transcribed by LDC or under       its direction.

This release includes 21 source-translation       document pairs, comprising 146,082 characters of Chinese source       text and its English translation. Data is drawn from seven       distinct Chinese programs broadcast in 2006 and 2007 from the       following sources -- China Central TV, a national and       international broadcaster in Mainland China and Phoenix TV, a Hong       Kong-based satellite television station. Broadcast conversation       programming is generally more interactive than traditional news       broadcasts and includes talk shows, interviews, call-in programs       and roundtable discussions. The programs in this release focus on       current events topics.

The data was transcribed by LDC staff and/or       transcription vendors under contract to LDC in accordance with       Quick Rich Transcription guidelines developed by LDC. Transcribers       indicated sentence boundaries in addition to transcribing the       text. Data was manually selected for translation according to       several criteria, including linguistic features, transcription       features and topic features. The transcribed and segmented files       were then reformatted into a human-readable translation format and       assigned to translation vendors. Translators followed LDCs Chinese       to English translation guidelines. Bilingual LDC staff performed       quality control procedures on the completed translations.

GALE Phase 2 Chinese Broadcast Conversation       Parallel Text Part 1 is distributed via web download.

2013 Subscription Members will automatically       receive two copies of this data on disc. 2013 Standard Members may       request a copy as part of their 16 free membership corpora.        Non-members may license this data for US$1750.
   

(2) Greybeard       was developed by LDC and is comprised of approximately 590 hours       of English telephone conversation speech collected in October and       November 2008 by LDC. The goal was to record new telephone       conversations among subjects who had participated in one or more       previous LDC telephone collections, from Switchboard-1 (1991)       through the Mixer studies (2006).

A total of 172 subjects were enrolled in the       Greybeard collection, all of whom had participated in one of the       following:

Switchboard-1 (LDC97S62)       1991-1992: 2 subjects

Switchboard-2 (LDC98S75,       LDC99S79,       LDC2002S06)       1996-1997: 16 subjects

Mixer 1 and 2 2003-2005: 103 subjects

Mixer 3 2006: 51 subjects

Most Greybeard participants completed 12 calls.       Some subjects completed up to 24 calls. Calls were made or       received via an automatic operator system at LDC which connected       two participants and announced a topic for discussion.

This release consists of 4680 calls -- the       complete set of calls recorded during the Greybeard collection       (1098 calls) as well as all calls from the legacy collections that       involved the Greybeard speakers.

The audio from each call was captured digitally       by the operator system and stored in a separate file as raw mu-law       sample data. As the recordings were uploaded daily from the robot       operator to network disk storage, automated processes reformatted       the audio into a 2-channel SPHERE-format file for each       conversation and queued the recordings for manual audit to verify       speaker identification and to check other aspects of the       recording. Auditors provided impressionistic judgments on overall       audio quality, presence of background noise and cross-channel echo       and any other technical difficulty with the call, in addition to       confirming the speaker-ID on each channel.

Greybeard is distributed on five DVD-ROM.

2013 Subscription Members will automatically       receive two copies of this data on disc. 2013 Standard Members may       request a copy as part of their 16 free membership corpora.        Non-members may license this data for US$7500.

(3) Manually Annotated
        Sub-Corpus Third Release (MASC) was developed as part of The American National Corpus       project and consists of approximately 500,000 words of       contemporary American English written and spoken data annotated       for a wide variety of linguistic phenomena.

The MASC project was established to address, to       the extent possible, many of the obstacles to the creation of       large-scale, robust, multiply-annotated corpora of English       covering a wide range of genres of written and spoken language       data. The project provides appropriate data and annotations to       serve as the base for a community-wide annotation effort, together       with an infrastructure that enables the incorporation of       contributed annotations into a single, usable format that can then       be analyzed as it is or transduced to any of a variety of other       formats. Further information about the project is available at the       MASC website.

The source texts were drawn from the open       portion of the American National
        Corpus Second Release, and from the Language Understanding
        Annotation Corpus.  MASC Third
      Release includes the contents of MASC First Release (LDC2010T22)       (82,000 words) which is also available from LDC. There is no       second release.

All data in this release was annotated for       logical structure (paragraph, headings, etc.), token and sentence       boundaries, part of speech and lemma, shallow parse (noun and verb       chunks) and named entities (person, organization, location and       date). Portions of the corpus were also annotated for FrameNet       frames (40k full text), Penn Treebank syntax (82k) and opinion       (50k).

Manually Annotated Sub-Corpus Third Release is       distributed via web download.

2013 Subscription Members will automatically       receive two copies of this data on disc. 2013 Standard Members may       request a copy as part of their 16 free membership corpora.        Non-members may request this data by submitting a signed copy of LDC User
        Agreement for Non-members.  This data
      is available at no-cost.