The obligatory-branching parameter in metrical theory

In the preceding sections, I have argued that metrical theory should be revised to incorporate revised obligatory-branching feet. This argument was based on the descriptive necessity of bounded ROB feet for Lenakel and unbounded ROB feet for Klamath. Neither of these languages are describable as simply using unrevised OB feet. One nice consequence of ROB footing is that it allows us to constrain the types of restrictions that are put on feet. Quantity-insensitive foot construction puts no constraints on what may qualify as terminals. Quantity-sensitive foot construction limits nonheads of feet. ROB foot construction limits heads of feet. Feet can therefore have unconstrained terminals, or one type of terminal, either heads or nonheads, can be constrained. Old-style OB feet constrained both head and nonhead, and one was left with a curious gap. There were no foot types where only the heads were restricted. Now, by adopting ROB feet, the gap is much more reasonable. Feet where both head and nonhead are constrained do not occur. What have we learned about stress? What insight is expressed by ROB feet? In one sense, this is a meaningless question. ROB feet are a descriptive device that accounts for a range of otherwise difficult to account for facts. In another sense, though, theoretical devices may express some insight, that is, satisfy linguists conceptually in addition to accounting for the data in as simple a fashion as possible. If so, then it is reasonable to ask what intuition is captured by ROB feet.Thanks to an anonymous reviewer for reminding me of this. The intuition is that although heavy syllables tend to attract stress, they do not always do so. In particular, ROB feet enable us to capture the fact that in certain positions (nonhead positions) under certain types of iteration or scanning (ROB types), heavy syllables are treated just like light syllables. With ROB feet, heavy syllables just like light syllables can be nonheads and hence may be unstressed. In this light, ROB feet can be seen as intermediate between quantity-insensitive and quantity-sensitive feet. Quantity-insensitive feet scan ignoring syllable weight. Quantity-sensitive feet scan allowing heavy syllables to interrupt the iteration at any time. ROB feet allow heavies to interrupt the iteration some of the time. Thus, ROB should not be seen as a bizarre foot type tacked onto the system, but as a natural, expected foot type intermediate between quantity-sensitivity and quantity-insensitivity and expressing the same basic intuitions about stress systems.