TalkBank Data Access Levels

In accord with the data management and Open Science standards of NIH/s Office of Data Science strategy (ODSS) CHILDES data is available through these four levels, as defined by ODSS. Specifically:
  1. Open access: The web pages at https://childes.talkbank.org describing the corpora and analysis programs such as CLAN, Chatter, and Batchalign are open with no access restrictions or registration required.
  2. Registration required: CHILDES transcript and media data on this level are open to all, but users need to be signed in or registered. To do this, users provide their email and create a password. They then receive an email message asking them to confirm registration and then they are registed. Once this is done, nearly all of CHILDES data are available for access through download or browsing in the TalkBank browser.
  3. Approved access: This level of access is only provided to academic researchers with a verified university position who send us an email request in which they agree to follow the Ground Rules for data usage at https://talkbank.org/0share/rules.html. This form of access is used for all the banks with clinical data (AphasiaBank, DementiaBank, FluencyBank, TBIBank, and RHDBank) as well as for most of HomeBank.
  4. Controlled access: Controlled access for specific more highly protected corpora requires verification of requestor identity, and committee approval of the proposed research use. The GlobalTales narrative corpus is the only CHILDES transcript corpus that requires this level of access. However, controlled access is also required for the video (but not the transcripts) of the DeHouwer/Bornstein, Koch, Leo, Rigol, Lund, Zlatev, and Nieva corpora. Corpora in HomeBank and PsychosisBank also require this level along with a special interview process and proof of CITI training.

We use the TalkBank AdminView system at https://sla2.talkbank.org/AdminView to control access. The creation of the AdminView system was funded by a FAIR grant in 2023 from ODSS.

It is also possible to set up a system in which the data contributor implements a data use agreement (DUA) which then allows TalkBank to provide access. However, largely because of the extra administrative burden and the fact that this system is not approved by NIH, we have not yet been using this.