- Definitely difficult to syllabify -- the patterning like onsets part is what I mean by having a coda-onset profile. You can have it look like both in the same coda (ex. /rbr/ is coda+onset), though agreed it can be solely onset as well (ex. /pr/), like it can be solely coda (ex. /rb/). Coda-onset reflect the 'maximal' complexity that occurs (and fairly commonly).Sumelic wrote: - The literature I've seen suggests that French word-final clusters are complicated to syllabify (example: http://www.cog.brown.edu:16080/People/d ... eJCatL.pdf). In terms of permissible consonant clusters, they pattern like onsets. One the other hand, they do affect the quality of preceding vowels and count as codas for the "loi de position."
- [wa] and [ɥi]] can be analyzed as diphthongs (I've read that one reason to do so is in fact that they can occur after plosive + liquid clusters; ordinarily, semivowels are not allowed in this position in French)
- I'm pretty sure they are only classifying languages based on phonemic clusters, not phonetic clusters that arise due to vowel reduction. This is common: we don't say English has /dn/ as a possible onset cluster even though some people might pronounce "deny" as [dnaɪ].
- Agreed that they're diphthongs (that's what I called them too!), but that does seem like an added level of complexity (just like a diphthong having an off-glide being followed by a coda -- especially a complex one -- seems like it could be more complex than simply having a [short] monophthong in that position).
- I wouldn't expect them to go to the level of reduced pronunciations, especially as reduced as my example, but it is an example of added syllable structure complexity in the input children receive. (Just like I didn't assume getting into linking and resyllabification, though they're potentially even more relevant!)
The goal was to raise points of "how did they and should we evaluate syllable structure complexity?" since the article didn't specify what they did (at least based on a very quick read-through). Having syllable structure complexity only be binary seems particularly surprising, but it looks like they were really just saying "Romance" vs. "Germanic" as the main goal, without necessarily specifying actually syllable structure motivations of the given languages. They noted a tendency towards CV in Romance, to paraphrase, but no specifics of how well that did apply to Romance or did/didn't apply to the others. I partly wanted to see how how well their classification matched what people thought for other languages as well, since I'm not familiar enough with all of them!