GLAM Special Collection: A Generalized Framework for Participatory Transcription: Lessons, Challenges, and Interdisciplinary Impact from the Zooniverse Platform

CHALLENGE	GUIDING QUESTION(S)
Variety	What are the properties (layout, form, genre, and any important features the transcription workflow needs to accommodate) of the texts to be transcribed? What transcription conventions will capture the desired data? Is it important to capture text exactly as it appears in the original documents or standardize spellings, punctuation, etc. to align with disciplinary standards? Is additional markup necessary, for example, to indicate supplied or deleted text or formatting like bold or italics? What are the edge cases in which a particular approach that works for the bulk of the material might not work for a subset?
Units of transcription	What is an appropriate unit of transcription? A whole document, a page, a line, a word, a single character? How does the unit of transcription translate to time spent on a task? Are there tradeoffs between unit size and task flow or immersion for participants? What workflow design best supports participants in contributing the unit of transcription?
Single-track transcription or multi-track transcription with aggregation	Who will transcribe and how will their transcriptions be vetted, if at all? Single-track: Can take many forms, for example, one or more participants transcribe, and other participants edit and/or an expert editor may finalize. Multi-track: Can take many forms, for example, N-blind keyings combined algorithmically, with or without expert review. What aggregation methods work best for combining multi-track transcriptions? What resources (e.g., expertise, time) are needed to conduct expert review and/or aggregation? Do I/we have this expertise?
Bias	Is data quality higher when multiple independent transcriptions are compared and combined into a single reading, or when transcribers are allowed to collaborate? Can collaborative methods be designed that retain the quality control metrics of independent transcription?