English syllabification discussion

[Qwynegold]

syllables exist on a phonetic level

I've been wondering about this lately. Can you see syllable boundaries on a spectrogram? If not, how can they be something that exist in reality and not just in our minds?

Intuitively, I feel like two phones that have a syllable break between them are "less connected" than two phones that belong to the same syllable. If this is the case, surely it should be somehow observable?

And for those who want minimal pairs, here's one that I've mentioned once a long time ago: The streetname [tiː.li.kɑn.o.jɑ] ("tiilikka" ditch, though I have no idea what tiilikka means) vs. [tiː.li.kɑ.no.jɑ] (brick hen), though there's probably a difference in secondary stress here somehow.

[gach]

And for those who want minimal pairs, here's one that I've mentioned once a long time ago: The streetname [tiː.li.kɑn.o.jɑ] ("tiilikka" ditch, though I have no idea what tiilikka means) vs. [tiː.li.kɑ.no.jɑ] (brick hen), though there's probably a difference in secondary stress here somehow.

There's a stress pattern difference (/ˈti:likanˌoja/ vs. /ˈti:liˌkanoja/, corresponding to the stem/word boundaries) but that's by no means the only difference. In the first example the third syllable is clearly closed with an /n/ coda that also easily extends as the phonetic onset of the following syllable, hence being pronounced as longer than the corresponding /n/ in the second example. In the second example all the consonants are pronounced short.

A classic syllabification minimal pair with no stress pattern differences is hauin = /hau.in/ ("with pikes", from NOM SG hauki) vs. hauin = /ha.uin/ ("with applications", from haku). The difference is whether the intensity peaks fall on the vowels /a/ and /i/ or on /a/ and /u/ and whether it's the /u/ or /i/ that's treated as a glide.