In my ancestral town, the records used the variants interchangeably. RM needs a feature to allow addition of alternate names FOR SORTING PURPOSE ONLY when the end user has no source for the alternate name. ARCIERI and ARCIERO are used interchangeably as are LO TITO and LOTITO which should be sorted the same. Distinguishing LO TITO from LOTITO defeats the purpose of arranging names in an alphabetical list because the reader might not find the entry.
Anyway, even when all the records on an individual say ARCIERO it might be helpful to file the entry also under ARCIERI, FOR SORTING PURPOSE ONLY.
Sorting two-word names as one word would put Rose Mary with Rosemary, making the entry easier to find; but would also put Smith, Josephine between Smith, Joseph Ian and Smith Joseph Isaiah.
Allowing the alternate name FOR SORTING PURPOSE ONLY would also give the user more leeway when handling a given name such as Carminella, allowing the user to find her under Carmina without suggesting there is any authority for the given name Carmina.
This common need has been discussed many times including the 10 year old link below, many researchers employ work around methods using Alternate Name and I include myself in that group. Early indications are that RM8 will emplot different partial finding and filtering of names and we will have to wait and see how this works in practice.
Many years ago I suggested an option to sort the Index pane by Soundex but neither Soundex or Metaphone systems are the perfect solution and will contain false positives, my own name alternatives are covered by more that one Soundex.
What I currently do is apply an Alternate Name of type Other Spelling using the most prolific and common spelling version of the name, this also often turns out to be the most modern spelling version.
Since discussing directly with RM many years ago without a satisfactory answer I have new thoughts on this ever present issue which is really what is being sought here, a custom Soundex code contained within the NameTable or PersonTable. In it's simplest form you would you could select all the individuals similar to Marking for a Group and click on a button to say, in your opinion, all these individuals are the same family name, so not very dynamic and what about when you enter a new person tomorrow?
The more workable solution would be for RM to maintain a MatchTable where the user is allowed to populate the table with Surname Variant spellings and RM assigns a custom Sort Code. New individual entries to the database would then not be a problem as the MatchTable would be a lookup table for the custom Sort Code and a simple routine to rebuild those custom Sort Codes when a new Name Variant is added to the MatchTable by the user really covers all.
SURNAME Spelling Variation Work Around