Time Group Analyser (TGA) - Quick User Guide

The Time Group Analyser (TGA) is an open-access, browser-based utility developed by Dafydd Gibbon (Bielefeld University). It provides rapid annotation mining, statistical evaluation, and visual modeling for speech timing and rhythm research without requiring local spreadsheet tools or programming scripts.

1. Supported Input Formats

The TGA application primarily processes phonetic duration data from speech recordings. You can provide input in two main formats:

Praat TextGrid files: Standard annotation files generated by Praat. The tool parses interval tiers automatically.
CSV / Text tables: Flat text data containing labeled columns with timestamp or duration values.

2. Step-by-Step Workflow

Prepare Annotations: Open your audio in Praat, create a TextGrid, and annotate your target tiers (e.g., syllables, words, interpausal groups, or feet). Save the TextGrid as a text file.
Upload or Paste Data: Navigate to the online TGA interface. Upload your .TextGrid file or copy-paste the raw annotation text directly into the designated browser input area.
Configure Tier Settings: If utilizing a multi-tier TextGrid, type the exact name of the specific tier you want analyzed (e.g., syllable) into the settings panel.
Process: Click the analysis/submit button to let the client-side JavaScript extract the timestamps.

Pre-validation Note: For batch processing or checking file formats for tier errors before running a full evaluation, use the auxiliary TGAinspector web tool.

3. Key Output Metrics & Visualizations

Once compiled, TGA automatically outputs descriptive statistics and rhythmic mappings directly onto the web interface:

Rhythm Metrics: Computes standard pairwise variability metrics—specifically raw PVI (rPVI) and normalized PVI (nPVI)—to evaluate syllable-timed vs. stress-timed linguistic properties.
Duration Modeling: Computes the local timing trends, including the mean, median, slope, and intercept functions of speech events within interpausal phrase groups.
Dynamic Diagrams: Generates immediate, browser-side graphical layouts representing timing structures, including hierarchical Time Trees, duration box plots, and Wagner Quadrant charts.