The abbreviation EAF stands for ELAN Annotation Format, also known as the EUDICO Annotation Format. It was developed by The Language Archive at the Max Planck Institute for Psycholinguistics. The ELAN (EUDICO Linguistic Annotator) Tool stores complex multi-layered annotations of audio and video recordings in EAF. The annotations in EAF are time-aligned and linked to a single point in the timeline of the audio or video recordings.

The EAF specification defines an XML schema and describes the elements and attributes of the schema. It is based on a tier-based data model. A tier is a kind of annotation layer that groups annotations with same characteristics in user-defined tier objects. For example, a tier may contain the orthographic transcription, and another tier the Part of Speech annotation of the items corresponding to the former tier. In this way the tiers can be represented independent, or can be hierarchically connected and associated with a parent tier. Within a tier, annotations can not overlap, but the annotations on one tier can overlap with annotations on other tiers.

Similar to LAF, CES and XCES, EAF uses stand-off annotation. This means, that the primary data and annotation are stored in separate files.

EAF is a multi-purpose format. It can be used not only for linguistic transcription and annotation, but also for the annotation of gestures and sign languages. That’s why it is more and more recognized, and has been used in a variety of disciplines such as linguistics, gesture research, sign language research and behavioural studies.