I want a third, preferrably referenced opinion on a terminological dispute here. The problem is the following (though I am avoiding the actual example to avoid unnecessary complication):
In German phonetics, consonant duplication indicates a shortening of the preceding vowel. So, e.g., Schafe is pronounced [ʃaːfə] and schaffe is pronounced [ʃafə]. In the latter example, I would call ff a digraph, as it encodes a single phoneme, namely [f].
Now it may happen that two identical consonants stand next to each other due to word composition, e.g., Schaf ([ʃaːf]) and Fell ([fɛl]) are composed to Schaffell. Now if you ask native speakers to very clearly prounce the latter word or if they are dictating it and want to make sure to be understood, they will probably say [ʃaːffɛl], but in natural pronuciation, the two [f] sounds will merge and the pronunciation will be [ʃaːfɛl] or [ʃaːfːɛl]. Note that the first vowel would never be shortened. (Also, experts would avoid to typeset Schaffell without an ff ligature.)
Finally, the controversy is this: Would ff in Schaffell be called a digraph? Or, more abstractly: If two phonemes are merged in pronunciation across a morpheme boundary, would the corresponding letters be called a digraph? The arguments so far boil down to:
- Yes, it would be called a digraph, because ff corresponds to a single sound.
- No, it would not be called a digraph, because in the relevant phonetics, ff does not encode a single sound in this word (though it does elsewhere).
I am looking for answers supporting either of these sides (or a third alternative) that are backed up with further arguments or references (other than Wikipedia).