liminfoELAN Reference
Free reference guide: ELAN Reference
About ELAN Reference
The ELAN Annotation Reference is a searchable guide to ELAN (EUDICO Linguistic Annotator), the multimedia annotation tool developed by the Max Planck Institute for Psycholinguistics. It covers seven categories: basic interface operations, tier structure and linguistic types, annotation workflows, import/export formats, keyboard shortcuts, analysis features, and file format specifications, providing researchers with quick access to the features they need for time-aligned language annotation.
The tier structure section explains the four ELAN stereotype types: Independent tiers with no parent dependency, Time Subdivision for splitting parent intervals into time-based child segments, Symbolic Subdivision for dividing parent content into evenly-timed symbolic units (useful for phonemic segmentation), and Symbolic Association for one-to-one mappings like glosses or part-of-speech tags. The annotation section covers Segmentation Mode for real-time boundary creation during playback, Transcription Mode with Tab/Shift+Tab navigation, regular expression search, and Tokenize Tier for automatic word-level splitting.
For data exchange, the reference details import capabilities for Praat TextGrid files, CSV/tab-delimited text with column mapping, and SRT subtitle files, plus export options including tab-delimited text for statistical analysis and interlinear gloss format for linguistic publications. Analysis features include inter-annotator agreement calculation with Cohen's Kappa, annotation statistics (frequency, duration), and multiple file batch processing. The EAF XML format specification and template system (.etf) are also documented for advanced workflow customization.
Key Features
- Tier structure guide: Independent, Time Subdivision, Symbolic Subdivision, and Symbolic Association stereotypes
- Linguistic Type definitions with stereotype constraints for hierarchical annotation layers
- Controlled Vocabulary (CV) setup for closed-set annotation tagging on specific tiers
- Annotation mode reference: Segmentation Mode (Enter-key boundary), Transcription Mode (Tab navigation)
- Import guides for Praat TextGrid, CSV/tab-delimited text, and SRT subtitle files
- Export formats including tab-delimited text, interlinear gloss, and EAF template (.etf)
- Essential shortcuts: Alt+Wheel (zoom), Ctrl+Shift+A (new annotation), Shift+Space (selection playback)
- Analysis tools: inter-annotator agreement (Cohen's Kappa), annotation statistics, and batch processing
Frequently Asked Questions
What are the four tier stereotypes in ELAN?
ELAN uses four stereotypes for tier relationships: (1) None/Independent for tiers with their own time alignment, (2) Time Subdivision for splitting a parent tier's time interval into smaller time-aligned child segments, (3) Symbolic Subdivision for dividing parent content into symbolic units with automatically even time distribution (e.g., phoneme segmentation), and (4) Symbolic Association for one-to-one mappings where the child inherits the parent's time range (e.g., translations, POS tags).
How do I import Praat TextGrid files into ELAN?
Go to File > Import > Praat TextGrid File and select your .TextGrid file. ELAN automatically converts Praat interval tiers into ELAN tiers with matching time alignment. Point tiers are converted as well. This allows researchers who began transcription in Praat to continue multi-layered annotation work in ELAN without data loss.
What is the difference between Segmentation Mode and Transcription Mode?
Segmentation Mode lets you create time boundaries in real-time by pressing Enter during media playback, which is efficient for initial utterance-level segmentation. Transcription Mode is designed for text entry after boundaries are set, using Tab to move to the next annotation and Shift+Tab for the previous one, with Enter confirming the annotation value. Choose Segmentation for boundary creation and Transcription for text input.
How do I calculate inter-annotator agreement in ELAN?
Use File > Compare Annotators and select two EAF files containing annotations of the same media by different annotators. ELAN calculates agreement metrics including Cohen's Kappa coefficient and percentage agreement. A Kappa value of 0.80 or higher typically indicates strong agreement, while values between 0.60-0.80 indicate moderate agreement.
What is a Controlled Vocabulary (CV) and how do I use it?
A Controlled Vocabulary is a closed set of predefined labels that restricts what values can be entered on a tier. Create one via Edit > Edit Controlled Vocabularies, define entries (e.g., POS tags: N, V, ADJ, ADV), then link it to a tier. When annotating, instead of free text input, you select from the CV dropdown, ensuring consistency across annotators and sessions.
How do I export ELAN annotations for statistical analysis?
Go to File > Export As > Tab-delimited Text. The output includes columns for Begin Time, End Time, Tier name, and Annotation Value, which can be directly imported into R, Python, Excel, or SPSS for quantitative analysis. For multiple files, use File > Multiple File Processing to batch export all .eaf files in a folder to a single CSV output.
What is the EAF file format?
EAF (ELAN Annotation Format) is an XML-based file format that stores all annotation data. It contains TIER elements with ANNOTATION children, each having ALIGNABLE_ANNOTATION elements that reference TIME_SLOT entries for precise time alignment. You can save tier structures as .etf templates (File > Save as Template) and reuse them across projects for consistent annotation schemas.
Does the ELAN reference work on mobile devices?
Yes, this reference is fully responsive and works on desktop, tablet, and mobile browsers. All 28 entries across seven categories are searchable and filterable in your browser with no installation needed. It supports dark mode and provides instant keyword search across all annotation commands, shortcuts, and format specifications.