Italian 'gn' vs Spanish 'ñ' - Why does their use differ intervocalically and word internally?

Question

First some examples. Here are the IPA transcriptions of 'bathroom' in Italian and Spanish:

bagno /baɲɲo/

baño /baɲo/

As you can see in Italian 'gn' becomes a geminate between vowels in the middle of a word but remains singular word initially as in:

gnocchi /ɲɔkki/

My question is why does this gemination occur in Italian? Is it related to the larger presence of gemination and even hidden gemination (fonosintatic doubling) present in Italian? or is it due to the way Italian developed from Latin?

Is it present in other Italian dialects?

Thanks

Answer 1

The most obvious reason for the difference is that Spanish, like English, does not have geminate consonants (aside from occasional “fake” geminates that arise from the same consonant occurring on either side of a morpheme or word boundary). Consonants that were geminate in Latin were simplified to singletons in Spanish but not in Italian.

Interestingly, there are some possible arguments for considering Spanish palatal consonants (ñ, ll, y, ch) as abstractly geminate on a "structural" level, presented in "Palatal Phenomena in Spanish Phonology", by Gary Kenneth Baker (2004). Baker states that antepenultimate stress appears to be forbidden in Spanish when the final syllable starts with one of the palatal consonants; similarly, heterosyllabic consonant clusters and the rhotic trill /r/ (sometimes analyzed as underlyingly geminate in Spanish) also attract stress to the penultimate syllable (page 4).

The historical reason for the gemination seen in Italian is the origin of the palatal nasal in the Latin heavy consonant cluster /gn/ or a /nj/ sequence. Both the nasal and palatal gestures were combined, but the length ended up being that of two consonants—approximately, and only in some varieties. As mentioned in the comments, since there is no contrast, duration may vary even in Italian.

Fonosintatic doubling affects a different set of consonants: the Italian always-geminate consonants are geminated after any vowel, whereas fonosintatic doubling occurs only after stressed word-final vowels or certain lexically specified vowel-final function words.

Answer 2

In The Phonology of Italian [2009, Martin Krämer], page 47, the author states that

Word-initially, all consonants can only be short ... The consonants [ts, dz, ʃ, ɲ, ʎ] are short (like all the other consonants) word-initially, but are always long word-internally. In word-initial position they are realized as long whenever preceded by a vowel. The voiced alveolar fricative [z], on the other hand, does not occur in long form.

{My italics and bold}

In the introduction to the book, the author says the book deals with the spoken standard, or Italiano, which is often influenced by the local Italian dialects and other Romance languages of Italy.