Conducting Keylogging Experiments with Translog-II¶

This guide provides a step-by-step walkthrough for using Translog-II to conduct a basic keylogging experiment. It covers the complete workflow — from project configuration to data collection — using an English-to-Chinese translation-from-scratch task as a worked example.

The guide is based on Zou & Carl (2026)¹.

1. Overview of Translog-II¶

Translog-II is a specialized tool for recording user activity data (UAD) during translation, writing, and reading tasks. It records keystrokes and, when connected to an eye-tracker, gaze movements during sessions.

Architecture¶

Translog-II consists of two main components:

• Translog-II Supervisor — used to create project files that define the experimental setup and to replay recorded sessions.
• Translog-II User — used to run text production experiments in which a participant reads, writes, or translates a text.

During a session, Translog-II produces log files containing UAD, including keystrokes and (optionally) gaze movements.

What Translog-II logs¶

Translog-II classifies keystroke data into several categories:

For each keystroke event, Translog-II records the event type, the value of the keystroke, the cursor position, and the timestamp in milliseconds.

2. Setting Up an Experiment¶

Translog-II provides three main functions for conducting experiments: project creation, session recording, and replay/analysis.

Step 1 — Create a project¶

1. Open Translog-II Supervisor.
2. Select Project → Create Project.

1. In the "New Project" dialogue box, click Configure Experiment to define the experimental parameters.

A project file can be configured for different experimental setups:

The source and target windows can be oriented horizontally or vertically and positioned left-right or top-bottom.

Step 2 — Configure the experiment¶

For our English-to-Chinese example:

1. Enter the English source sentences into the upper window.
2. Select Translation and Target window at bottom.
3. Leave the bottom window empty for the participant's translation.
4. Adjust font (sans-serif is recommended), text alignment, and layout as needed.
5. Save the configuration as a .project file.

Step 3 — Record a session¶

1. Open the .project file in Translog-II User.
2. Click Start Logging to begin the session. This opens the configured interface displaying the source text and an empty target window.
3. The participant translates the text.
4. When the participant finishes, click Stop Logging. The session is saved as an XML file.

Step 4 — Replay and analyze (optional)¶

Translog-II includes a replay tool with three visualization modes:

• User view — Real-time replay of the logged data, showing typing activity as it occurred. Playback controls allow researchers to accelerate, decelerate, pause, rewind, or fast-forward.
• Linear view — Displays each key and mouse activity in a linear textual sequence. Pauses between successive actions are shown as asterisks or numeric values. The pause granularity can be adjusted from 1ms upward.
• Pause plot — A 2D graph where keyboard activities appear on the X-axis and accumulated time (pauses) on the Y-axis.

All three views can be opened simultaneously and synchronized so the cursor in all windows is positioned at the same point in time.

Next steps: Processing your data¶

The logged data collected by Translog-II can be uploaded to the CRITT Translation Process Research Database (TPR-DB) for post-processing and analysis. See the Automatic Processing documentation for details.

3. Tips for IME-Mediated Translation Experiments¶

When conducting experiments with languages that require IMEs (Chinese, Japanese, Korean, etc.), keep the following in mind:

• Review longer deletions manually — the automated keystroke-to-word mapping heuristic becomes less reliable as deletion length increases, particularly for multi-word IME conversions. Use the TPR-DB's semi-automatic adjustment tool if precise deletion analysis is needed.
• Be aware of timing ambiguity — for IME input, there are at least two possible measures of production time: the duration of the keystrokes typed, and the time until the character appears on screen. The delay between triggering IME conversion and character appearance reflects both technical processing time and the time the typist spends evaluating and selecting among candidate characters.

4. Further Resources¶

• Download Translog-II: CRITT website
• TPR-DB 3.0 data tables: Features table
• Auxiliary tools: Semi-automatic adjustment tool
• Progression graphs: Shiny R interface

Citation¶

If you use this guide or the Translog-II workflow in your research, please cite:

@inproceedings{carl2012translog,
  title={Translog-II: A Program for Recording User Activity Data for Empirical Translation Process Research},
  author={Carl, Michael},
  booktitle={Proceedings of the 8th International Conference on Language Resources and Evaluation (LREC 2012)},
  year={2012},
  address={Istanbul, Turkey}
}

@inproceedings{carl2012critt,
  title={The CRITT TPR-DB 1.0: A database for empirical human translation process research},
  author={Carl, Michael},
  booktitle={Workshop on post-editing technology and practice},
  year={2012}
}

@incollection{carl2016critt,
  title={The CRITT translation process research database},
  author={Carl, Michael and Schaeffer, Moritz and Bangalore, Srinivas},
  booktitle={New directions in empirical translation process research: Exploring the CRITT TPR-DB},
  pages={13--54},
  year={2016},
  publisher={Springer}
}

@article{zou2026using,
  title={Using Translog-II for Conducting Keylogging Experiments},
  author={Zou, Longhui and Carl, Michael},
  year={2026},
  publisher={Preprints}
}