Forbidden repetitions of identical labels and exceptions

[johentsch]

At the moment, annotators are allowed to repeat a label to mark a phraseending, e.g. I (beginning of harmony) I\ (end of phrase). Instead, we could allow for \ to stand as a separate symbol, which is something that annotators sometimes do intuitively and has to be corrected so far. The change can be easily made by updating the regex with a ( )? . Would you agree on that, @fabianmoss ? This change would avoid to have the exception from forbidden repetition of identical labels.

**However, ** there would still be a similar exception for organ points. We could at the same time consider to allow for organ point notation to occur independently of the onsets of harmonies. This would be quite a big change, however:

• It would even out the discrepancy between the time points designated by \\ and ]
  • The phraseend marks this time point directly, but the end of the organ point occurs only when the harmony designated with ] ends.
  • Allowing for ] to stand alone would make it possible to have it designate the ending of the pedal note directly instead.
  • When parsing this label, a row would be created without new chord features, showing that the previous harmony is still on. The countermodel would be that the same harmony is repeated, resulting in a new harmonic segment. This is actually a strong argument for allowing for the exception, considering that a harmony over a pedal is different from the same harmony without the pedal.

[johentsch]

Maybe one guiding principle to keep in mind for the discussion could be: Since every label represents a segmentation of the score, it should represent a change in harmony. The alternative would mean to allow for different levels of segmentation, e.g. allowing for singular \\ alone would designate the end of a phrase segment but not of a harmonic segment. But this is happening at present, in any case, with the difference that the same label is repeated, e.g. I I\\.

[fabianmoss]

I am not quite sure if I follow your considerations. What are circumstances where you would annotate a \\ or ] without a "supporting" harmony?

I don't see a problem with repeating harmonies at phrase boundaries but wouldn't it be rather I\\ I instead of I I\\? Meaning: First I conclude a phrase and then I continue with the same harmony? I can't imagine a situation in which the latter makes sense but I might just not think of the right thing.

[johentsch]

No, in I\\ I the second I is superfluous because under the current semantics (labels valid until the next label) it does not add new information. I I\\, however, says "here, the harmony starts, and here is where the phrase ends".
This is a very extreme case where both occur very close to each other:

(Corelli op. 1/7 I mm. 1-4)

Suggestions to discuss:

• Permit \\ to stand alone
• Permit repetition of identical labels in order to perform a lower level segmentation.

[johentsch]

The new version will allow for repetition of identical labels under special circumstances and for having phrase annotations occur separately.

The solution we discussed was in favor of a possibility to annotate phrase interlocking. The natural solution seems to be to annotate both phrase endings and beginnings in a way, that both symbols can be combined in one. The first suggestion was to use \ for endings, / for beginnings and \/ as combination. However, since the expressions continue to end labels, / would make it impossible to detect mistakes where an applied chord is missing it's secondary key, as in V/ instead of V/V.

Solution: We suggest a solution with { for beginnings, } for endings, and }{ for the combination. In order to distinguish phrase interlocking from "normal" phrase transitions we ask annotators to put } on the beat where the phrase ends structurally (e.g. the 1 of a cadence's ultima). This way, the phrase length is still calculated until the next phrase's beginning, but we will know how much of the phrase is just a 'codetta'. As an example, here is the previous one with the suggested new notation: