Conversational Uptake

This repository contains data for the paper:

Alic, S., Demszky, D., Mancenido, Z., Liu, J., Hill, H., & Jurafsky, D. (2022). Computationally Identifying Funneling and Focusing Questions in Classroom Discourse. In Proceedings of the 17th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2022) (pp. 224-233).

@inproceedings{alic2022computationally,
  title={Computationally Identifying Funneling and Focusing Questions in Classroom Discourse},
  author={Alic, Sterling and Demszky, Dorottya and Mancenido, Zid and Liu, Jing and Hill, Heather and Jurafsky, Dan},
  booktitle={Proceedings of the 17th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2022)},
  pages={224--233},
  year={2022}
}

Annotated Funneling vs Focusing Dataset

The annotated dataset contains a sample of 2348 exchanges extracted from a dataset of anonymized 4-5th grade US elementary math classroom transcripts collected by the National Center for Teacher Effectiveness (NCTE) in New England schools between 2010-2013. These exchanges are turns by students (with at least 5 words), followed by teacher turns in a classroom conversation.

Each exchange is coded for one of three categories, by three raters:

• 1 = funneling
• 2 = focusing
• 0 = neither (exchane is not related to math and/or it does not include a follow-up prompt) The coding details are explained in greater detail in our paper.

There are two subsets of the data. In the unfiltered subset, we include all utterances, regardless of whether they have a follow-up prompt. Use the filtered set if you want to train a classifier to identify funneling and focusing questions among a dataset of various types of exchanges (not just questions). In the filtered subset, exchanges without a follow-up prompt have missing values. Use the filtered subset if you want to train a classifier to distinguish different question types within a dataset of questions.

The data is in the comma-separated file funneling_focusing_dataset.csv.

The file includes the following columns:

• obs_id: Observation ID, mappable to unique transcripts in the NCTE dataset.
• exchange_idx: ID of the exchange within the transcript.
• student_text: Student utterance.
• teacher_text: Teacher utterance (following the utterance in student_text).
• follow_up_unfiltered_majority: The majority rating for funneling and focusing across the three raters -- unfiltered subset. Missing values indicate no majority label
• follow_up_filtered_majority: The majority rating for funneling and focusing across the three raters -- filtered subset. Missing values indicate no majority label.
• follow_up_unfiltered_zscore: Average of raters' z-scored judgments for funneling and focusing across the three raters -- unfiltered subset. Missing values indicate no majority label
• follow_up_filtered_zscore: Average of raters' z-scored judgments for funneling and focusing across the three raters -- filtered subset. Missing values indicate that an exchange is not part of the filtered subset.

Each example can be uniquely identified with the combination of the obs_id and exchange_idx columns.

Please contact Dora (ddemszky@stanford.edu) for questions.