README.md

---
dataset_info:
  features:
  - name: date
    dtype: string
  - name: speaker
    dtype: string
  - name: chamber
    dtype: string
  - name: reference
    dtype: string
  - name: source
    dtype: string
  - name: party
    dtype: string
  - name: content
    dtype: string
  - name: dog_whistle
    dtype: string
  - name: ingroup
    dtype: string
  splits:
  - name: train
    num_bytes: 450414528
    num_examples: 1096682
  download_size: 135137623
  dataset_size: 450414528
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

# Silent Signals | Formal Potential Dogwhistles (Potential Instance Dataset)

**A dataset of *potential* dogwhistle use cases in formal discourse.** A dogwhistle is a form of coded communication that carries a secondary meaning to specific audiences and is often weaponized for racial and socioeconomic discrimination. Dog whistling historically originated from United States politics, but in recent years has taken root in social media as a means of evading hate speech detection systems and maintaining plausible deniability.

**This dataset contains over 1 million excepts from U.S. Congressional Records that contain a dogwhistle term, created via keyword search on terms in the [Allen AI Dogwhistle Glossary](https://dogwhistles.allen.ai/)**. Therefore, many of the excerpts are likely using these terms in a non-coded sense. This dataset was used as input to our approach for word-sense disambiguation of dogwhistles from standard speech using Large Language Models (LLMs). Please see the paper linked below for more details.

*Although we collect over 7 million potential dogwhistle instances, due to resource constraints, we only sample 100,000 instances for the creation of the Silent Signals dataset. **Therefore, we release the Potential Dogwhistle Dataset to enable the open sourced expansion of the Silent Signals dataset.***

<p style="color:red;">Please note, this dataset contains content that may be upsetting or offensive to some readers.</p>

**Published at ACL 2024!**

📄 **Paper Link** - [Silent Signals, Loud Impact: LLMs for Word-Sense Disambiguation of Coded Dog Whistles](https://aclanthology.org/2024.acl-long.675/)<br>
🗂️ **Disambiguated Dogwhistles Dataset** - [Silent-Signals on HuggingFace](https://huggingface.co/datasets/SALT-NLP/silent_signals)<br>
💻 **Dataset webpage** - Coming soon 🚀

<!-- ![head_figure_large.png](https://cdn-uploads.huggingface.co/production/uploads/632d02054a4991e711591c34/m70hfQTN2Aw7t3Ilkga4u.png) -->
<centering><img src="https://cdn-uploads.huggingface.co/production/uploads/632d02054a4991e711591c34/m70hfQTN2Aw7t3Ilkga4u.png" alt="head_figure" width="400"/></centering>


## Dataset Schema ##

| <nobr>Field Name </nobr>| <nobr>Type</nobr> | <nobr>Example</nobr> | <nobr>Description</nobr> |
|:------------|:------|:---------|:-------------|
| <nobr>**dog_whistle**</nobr> | <nobr>str</nobr> | <nobr>"welfare"</nobr> | <nobr>Dogwhistle word or term.</nobr> |
| <nobr>**ingroup**</nobr> | <nobr>str</nobr> | <nobr>"welfare"</nobr> | <nobr>The community that uses the dogwhistle.</nobr> |
| <nobr>**content**</nobr> | <nobr>str</nobr> | <nobr>"You wonder how long can we keep spending money on <br>all kinds of pension and welfare programs and feel-good <br>programs and reward-people-for-not-working programs <br>and food stamp programs."</nobr> | <nobr>Text containing the dogwhistle.</nobr> |
| <nobr>**date**</nobr> | <nobr>str</nobr> | <nobr>"05/26/2010"</nobr> | <nobr>Date of comment, formatted as `mm/dd/yyyy`.</nobr> |
| <nobr>**speaker**</nobr> | <nobr>str</nobr> | <nobr>W. AKIN</nobr> | <nobr>Speaker, included for U.S. Congressional speech <br>excerpts and Null for Reddit comments.</nobr> |
| <nobr>**chamber**</nobr> | <nobr>str</nobr> | <nobr>H</nobr> | <nobr>Chamber of Congress, 'S' for Senate, <br>'H' for House of Representatives, and <br>Null for Reddit comments.</nobr> |
| <nobr>**source**</nobr> | <nobr>str</nobr> | <nobr>"Stanford Congressional Record"</nobr> | <nobr>The source or method of data collection.</nobr> |
| <nobr>**party**</nobr> | <nobr>str</nobr> | <nobr>R</nobr> | <nobr>The political party affiliation of the speaker, <br>available only for U.S. Congressional excerpts.</nobr> |
| <nobr>**reference**</nobr> | <nobr>str</nobr> | <nobr>CREC-2010-05-26-pt1-PgS11-8</nobr> | <nobr>The reference code for the Congressional record utterance, <br>available only for U.S. Congressional excerpts.</nobr> |
> NOTE: The dogwhistles terms and definitions that enabled this research and data collection were sourced from the [Allen AI Dogwhistle Glossary](https://dogwhistles.allen.ai/).

<br>

# Citations #

### MLA ###

Julia Kruk, Michela Marchini, Rijul Magu, Caleb Ziems, David Muchlinski, and Diyi Yang. 2024. Silent Signals, Loud Impact: LLMs for Word-Sense Disambiguation of Coded Dog Whistles. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 12493–12509, Bangkok, Thailand. Association for Computational Linguistics.

### Bibtex ###

```
@inproceedings{kruk-etal-2024-silent,
    title = "Silent Signals, Loud Impact: {LLM}s for Word-Sense Disambiguation of Coded Dog Whistles",
    author = "Kruk, Julia  and
      Marchini, Michela  and
      Magu, Rijul  and
      Ziems, Caleb  and
      Muchlinski, David  and
      Yang, Diyi",
    editor = "Ku, Lun-Wei  and
      Martins, Andre  and
      Srikumar, Vivek",
    booktitle = "Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = aug,
    year = "2024",
    address = "Bangkok, Thailand",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2024.acl-long.675",
    pages = "12493--12509",
    abstract = "A dog whistle is a form of coded communication that carries a secondary meaning to specific audiences and is often weaponized for racial and socioeconomic discrimination. Dog whistling historically originated from United States politics, but in recent years has taken root in social media as a means of evading hate speech detection systems and maintaining plausible deniability. In this paper, we present an approach for word-sense disambiguation of dog whistles from standard speech using Large Language Models (LLMs), and leverage this technique to create a dataset of 16,550 high-confidence coded examples of dog whistles used in formal and informal communication. Silent Signals is the largest dataset of disambiguated dog whistle usage, created for applications in hate speech detection, neology, and political science.",
}
```