solpahi
commented
2 years ago The reason behind the "high"/"low" naming convention is that low adverbs attach low (under FP) and high adverbs attach high (just below CP).
When I made the split, I had two proposals for high adverbs. I went with the simpler one for the refgram (although the difference wouldn't have been visible there). However, I should probably explain both proposals here.
The simple one looks like this (I'm using bu as a stand-in for (0) verbs here):
(right-attachment:)
(left-attachment:)
This is as simple as can be. Some drawbacks include:
- unlike low adverbs, where :t6: and :t7: are interpreted as ∧, high adverbs need a special rule to compose correctly (basically, they act like full verbs, taking the phrase they are adjuncts of as argument). This is really not a major issue, though.
- there is a requirement to distinguish high-verbs from low-verbs, and determining the attachment type of a serial appears to be non-trivial (there'd have to be a hierarchy of features, and when serials form their V-clusters, the strongest feature in the cluster determines the height)
The other proposal is more complicated. It takes the insight about high adverbs acting like verbs and realizes it fully. The complexity comes from the fact that it requires inserting invisible phrases:
bũ here is a full verb. Whether it is base-generated with :t7: or whether :t7: is added on is not essential.
If bũ follows the post-field, it involves raising of the CP:
This approach requires no new composition rules. But as it stands it has the same problem of needing to determine the "height" of a serial.
One of the ideas I had next was to make lu a (or the only) high adverb. Then any verb/serial could be used as a high adverb by using it in a lu clause:
[CP Fả jí Fáraqse ] lũ dảqfaı hóa "[that] I go to France, that happens often." "I go to France often."
But this option scared me. Like Pandora's box. Its potential, negative and positive, felt too hard to gauge, so I left it and went with the simpler option.
With this sort of grammar, you can retroactively demote material into CPs, repeatedly.
It also removes the whole high/low split (everything is low except one word).
And it's probably hell when it comes to binding resolution.
Fả súq Fáraqse lũ shảo súq hóa lũ bủ hóa "[that] you go to France, [that] you want that, that's not the case" "It's not the case that you want to go to France"
🤷
Anyway, the refgram was written with the simple version in mind.
I expect that there are ways to implement the complex proposal in a slightly simpler way. I just haven't given it much thought for the reasons stated above.
Also, thinking about this appears to be triggering some of that stress and anxiety I used to get from working on the grammar of Old Adverbs, so maybe someone else could take this one (no rush, tho). Fortunately, it's not nearly as complicated as Old Adverbs.
Or we just stick to the simple version.
Not yet handled by zugai at all.
How do we detect that something like jãq le is high? What about kũqnu le, is that high or low?