TalkBank Code of Ethics

This document presents the proposed TalkBank Code of Ethics. This document is intended to supplement, but not replace the following generally accepted professional codes:

TalkBank is an international database of audio, video, and transcript recordings of vocal interactions from both humans and animals. The following standards apply only to the human component of the database. The purpose of these standards is to guarantee that the dignity of all the human participants in TalkBank interactions is fully respected. Below, we give illustrations of ways in which TalkBank data could conceivably be misused. No such misuses have ever occurred (at least between 1984 and 2021). However, our purpose here is to guard proactively against possible misuses. There are three groups of individuals involved:

The goal of theTalkBank Code of Ethics is to assure that users respect the rights of both the participants and the contributors. The basic framework for data included in TalkBank assumes the following model.

  1. When the data were collected, participants signed an Informed Consent release like the TalkBank sample. This consent form was approved by the local IRB or Institutional Review Board.
  2. The IRB or Institutional Review Board is a panel of researchers at a university that reviews all research conducted at the institution. Each IRB sets up its own procedures. However, we assume that these standards are in general agreement with those suggested by TalkBank.
  3. Some data will be collected outside of the IRB model. These include data collected before the IRB system was established, data from non-funded work, and data from speakers of indigenous and endangered languages. Data collected in these other ways are still subject to the same ethical concerns as for data collected with IRB-approved informed consent forms.
  4. Users must be members of TalkBank. Membership is open to all scholars and students. For password-protected corpora, members are given passwords on request. Membership can be revoked if there is evidence that a user has violated this Code of Ethics.

The TalkBank Code of Ethics is composed of seven basic principles:

  1. Responsibilities for Data-Sharing. All TalkBank members accept the responsibility to contribute the results of their studies to TalkBank and to participants to grant access to the data whenever possible.
  2. Appreciation to Contributors. The contributors of TalkBank data have spent many long hours collecting and transcribing data. Although data-sharing is a scientific responsibility, not all scientists have learned to accept this responsibility. Often scientists refrain from data-sharing because they believe that, by sharing their data, they will advance the careers of their competitors. Therefore, when scientists decide to share their data publicly, we owe them both respect and appreciation. It would be wrong, for example, to publish criticisms of the transcription or data collection methods involved in the corpora you analyze. In addition, as noted in the Ground Rules page, Researchers who use TalkBank data must carefully cite their sources. We at TalkBank are happy to acknowledge these contributions by writing letters of support to departmental review committees and granting agencies explaining the important of data contribution to the advance of science.
  3. Respect for Participants. The participants in TalkBank interactions have made an important contribution to scientific research by permitting scholars to access and analyze their data. Both the contributors and the users of language data need to recognize the generosity of participants in making their data publicly available. It would be wrong to repay this generosity by making any open public criticism of these individuals. In particular, commentary on participants needs to avoid comparisons between speakers in terms of personal characteristics such as intelligence, verbal facility, social skills, or physical appearance. Comparisons can often be interpreted as invidious and should be generally avoided. If properly expressed, it is still possible to analyze speech features in a positive fashion. Examples of ways of dealing with this are given below.
  4. Confidentiality. No personal data about individual subjects will be made available to users apart from video and audio records and basic facts such as age or location of the taping. Recordings must be done in accord with legal principles barring surreptitious recording. When participants grant permission to include their data, they often assume that the interactions are being analyzed by researchers with whom they are not personally acquainted. If a user happens to discover that they are personally acquainted with a particular participant, and if they have not received direct permission from the participant to examine their data, they must refrain from further analysis of the data, since use of such data would then violate assumptions about confidentiality.
  5. Confidentiality in Publications. When publishing segments of conversations or images from videos, it is important to maximize anonymity. If a transcript includes either a first or last name or an address, that name or address should be replaced with a pseudonym in publication. If an image of a person is included in a publication, that image should be blurred unless the participant provides explicit permission for inclusion of their image in the publication.
  6. Respect for Groups. TalkBank data is contributed by individual contributors and participants. However, these individuals are also members of groups. Many of these groups have developed a well-justified sensitivity to criticism from the wider society. Generally, analyses of TalkBank data should avoid making between-group comparisons that impact core features of social identity and worth.
  7. Sanctions. If users violate this Code of Ethics, it becomes the responsibility of the whole TalkBank community to counter the misuse in public forums and through personal contact. TalkBank cannot block the publication of disrespectful or irresponsible analyses. Nor can it block the publication of analyses without proper citation of the data sources. However, when such violations might occur, it can call the community's attention to these abuses when they occur. Between 1984 and 2021, the only violations that were noted were two cases in which graduate students published conference papers without proper citation of sources. In both case, we contacted their faculty advisors who explained to them that this was unacceptable. In addition, we then posted message to the GoogleGroups mailing lists reminding researchers of the need to properly cite sources.