The ACLEW-DARCLE Annotation Scheme

A central component of the project was the development of a unified annotation system (now called the ACLEW DAS*, or AAS) that meets several desiderata. First, it needed to be compatible with existing annotation systems and software. Second it needed to allow for broad class annotation of segments of the waveform (e.g. speaker types like "female adult" or "child", characteristics like "child-directed" or "adult-directed"), that can be applied consistently across different cultures and languages to allow for direct comparisons across datasets. Third, it must be flexible to allow for use on very different kinds of research questions related to child language development. Fourth, it must create output that can easily be used to develop tools for automated speech processing.

You can find a series of tutorials for the ACLEW DAS here. We welcome feedback and questions. The current version is focused on the needs of the ACLEW project, but it is designed to be flexibly adapted to other research needs, and other versions of the DAS developed by other researchers can be found here. The DAS is designed for annotation within the ELAN transcription software.

*ACLEW emerged from a larger network of researchers interested in longform audio known as DARCLE.

Below is a schematic of key elements in the ACLEW DAS.