CECSpecialistI
commented
2 years ago Adam's explanation 2022-01-18 @mcm104 @AdamSchiff : Do I understand timespan? Hmmm, sort of! In MARC, I think we are recording timespan sometimes as a relationship and sometimes as an attribute. Timespan can go in many different fields: 033, 045, 046, 388, 3XX $s and $t, as well as 5XX (note in bib; related entity in authority), and 6XX (subject) in bib.
date of expression (https://access.rdatoolkit.org/Content/ContentById/a4d33934-9633-4fa2-80d6-28e7acf81078), "A timespan that is the earliest associated with an expression, is a subelement of related timespan of expression (https://access.rdatoolkit.org/Content/ContentById/d119eeb3-2187-4ce9-8bdf-fe97cade8c6a), "A timespan that is associated with an expression." Note the difference from the definitions. In practice, I doubt we will use the broader element much if ever. I'm having trouble even thinking of a case where we'd know a date for an expression that isn't the date of expression. Maybe: an expression with date of expression 1965 has the related timespan of expression 1960s, or 20th century. Why would we bother to record that though?
I think how to choose which property to use depends on what data is recorded. In an authority record for an expression, 046 $k will map to date of expression. So would 1XX $f. In a bib record, which combines all or some of WEMI, 046 $k can't reliably be mapped unless an indicator value for type of entity is included. In absence of an indicator, the general assumption is that the date in $k is date of work. 1XX and 240 $f in a bib would always map to date of expression. 388 was only defined for work creation date controlled vocabulary.
date of capture is a subelement of date of expression, so 033 second indicator 0 will map to it. Second indicator 1 (date of broadcast) and 2 (finding) don't exactly match expression. Unless, perhaps, date of broadcast is the same as the date of expression (e.g. a live broadcast).
The RDA Toolkit has a whole section on Entity Boundaries: https://access.rdatoolkit.org/Guidance/GuidanceById?id=5eeaa822-e21c-4576-b728-0e7d0035d5ca The toolkit is not working very well at the moment however. I keep getting a system error when I try to view that guidance chapter. But the entity boundaries are provided to give help in determining when something has changed from one entity to another (such as where the boundary is between works versus one work with two expressions; or when two people with the same name are different entities; there will be boundaries for timespans as well (any time/date between 1901 and 2000 is in the 20th century; if the date is 23:59 hours of 1900, that is across the boundary and in the 19th century).
I don't know if any of this helps. Dates are complex and the new timespan elements are new. The LC/PCC Metadata Guidance Documents on timespan ought to help once they are issued
Email TG about timespans CEC