Attribute typology
Attributes are metadata about individual transcripts or participants in APLS, such as when the interview was recorded or the participant’s gender. You can use attributes as filters when browsing transcripts/participants or searching the corpus, you can view a transcript’s attributes on its transcript page, and you can export attributes alongside search matches.
On this page
Scope
Attributes can have one of three scopes:
- Transcript
- Participant
- main_participant
For example, recording_date is a transcript attribute and occupation is a participant attribute. main_participant is unique; it identifies the primary participant in a transcript, so it’s a property of both the participant and the transcript.
Display title
Each attribute has a longer display title. Display titles show up on a transcript’s or participant’s attributes page (or the transcript attributes dropdown on a transcript page) in the left-hand column. If the attribute is filterable, the display title shows up in the column headers on the transcripts or participants page (see Browsing transcripts, Browsing participants). For example, the transcript attribute duration shows up on the transcripts page as Duration (sec). If you hover over an attribute’s display title in APLS, you can see the attribute’s name and short description.
Export name
For most attributes, when you export attributes to a CSV file, either on their own or in search matches, their column names will have transcript_ or participant_ added as a prefix. For example, education gets exported as participant_education, while corpus just gets exported as corpus. This is the attribute’s export name.
Export names are helpful given that some attribute names are shared between scopes: transcript type vs. participant type, transcript neighborhood vs. participant neighborhood.
Five attributes (corpus, episode, transcript, participant, and main_participant) don’t have a prefix for their export name. These all have something in common: APLS actually stores them internally as layers, not attributes! Despite this, we think it makes better sense to think of them as “attributes” because of how users interact with them. Unlike most layers, they can’t be directly viewed in a transcript or searched; when exporting a formatted transcript or search results, they also show up in the same column as transcript/participant attributes.
Filterability
Some attributes are filterable: they’re in columns on the transcripts or participants page that can be used to filter by that attribute (see Browsing transcripts, Browsing participants). Other attributes are left off the transcripts or participants page to save room on the page; those attributes are viewable on a transcript’s or participant’s attributes page. (Some attributes aren’t viewable on any attributes page: corpus, episode, transcript, and main_participant.)
episode is a transcript attribute, but it is filterable on the participants page.
Multiple values
Most attributes have a single value per transcript/participant (e.g., each participant has just one year_of_birth). Two attributes can take on multiple values: participant ethnicities and transcript transcription_ai_tools. These are similar to layers that allow vertical peers.
Option codes and descriptions
Some attributes have options: a pre-defined set of possible values. For example, each participant represents one of four Pittsburgh neighborhoods, so neighborhood has options; but each participant’s relationship to other APLS participants is more open-ended, so relationships doesn’t have pre-defined options. In some cases, these options are option codes that are abbreviated and/or need further explanation, so these option codes have option descriptions. For example, the neighborhood CB is short for Cranberry Township.
On the participants page, the Neighborhood, Education, and Speaker type columns all have option codes; if you hover over an option code, a tooltip will pop up with the option description. On a transcript’s or participant’s attributes page, attributes with option codes and descriptions are displayed as <code> (<description>) in the right-hand column. If you export attributes to CSV (either on their own or in search matches), the file will contain the option codes, not the option descriptions.