"unininomial" would equal "canonicalName" for ranks subgenus and above, but not for species and below, while canonicalName (or scientificNameCanonical if you prefer) covers all cases, which is why I thik it is preferable, especially as the majority of names in circulation are at species level and below I think...
Atomising further i.e. a binomial or poynomial into genus, species, infaspecies is actually a separate activity with its own rationale, I would say.
Just my personal view, of course...
The cleanest way to do it is to simply have Rank, NameElement and parentNameUsageID, and be done with it (maybe with the addition of verbatimNameString for purists). But that assumes that providers have parsed data, which they often do not. Maybe with services like those associated with GNI, the time of databases with unparsed names data are drawing to a close. Or, maybe if GNUB gets a foot-hold, we'll solve all the problems via a simply actionable persistent identifier.
But until that time, dwc needs to find a balance between users who want pre-parsed data, and providers who do not have pre-parsed data.
I think dwc *almost* accomodates both worlds, as long as scientificName is defined as "the complete set of textual elements useful for recognizing a unique scientific name"; which is either concatenated by the provider with parsed data, or simply "provided" by the provider with unparsed data.
What we seem to be arguing about now is how many different forms of a "formatted" name do we want?
With or without authorship?
With or without year?
With or without infraspecific prefixes ("var.", "f." etc.)?
With or without infrageneric name(s)?
With or without italics codes?
With or without qualifiers like "cf.", "aff.", etc.?
Etc.
Etc.
Etc.
There are potentially dozens of different terms we could define to accommodate every particular niche-need.
Personally, I think that the existing "scientificName" should be split into two different terms:
fullScientificNameStringWithAuthorship And verbatimNameString
The first would be a concatenated text string assembled from parsed bits, according to a community standard concatenation form.
The second would be the literal text string as it appeared in the original source.
Otherwise, we could argue forever about which of the dozen possible forms we think DwC needs a term for.
Rich