Pater (2016): Substance matters: A reply to Jardine (2016)

Update – this paper will appear in Phonology in 2018. The final manuscript version is available here: Comments are still welcome, and those on the earlier version by Jardine and McCollum are still applicable.

Pater, Joe. 2016. Substance matters: A reply to Jardine (2016). Ms, University of Massachusetts Amherst.

Comments welcome!

Abstract. Jardine (2016) points out an interesting shared formal property of two types of phonological pattern, termed “Unbounded Plateauing” and “Sour Grapes”: for a given segment, whether or not it undergoes a featural change can only be determined by examining a potentially unbounded string both before, and after that segment.  In this short paper, I point out that Jardine’s tonal Sour Grapes example, Copperbelt Bemba unbounded tonal spreading (Kula and Bickmore 2015), is in crucial respects unlike the hypothetical pattern that Bakovic (2000: 217ff.), Wilson (2003), McCarthy (2011) and others have pointed out is generally unattested in unbounded spreading. It appears that “true” Sour Grapes is just as much unattested for tone as it is for other features. The generalizations that only “apparent” Sour Grapes is attested, and that it is attested only for tone, can be captured by OT, but not by Formal Language Theory as presented in Jardine (2016).


4 thoughts on “Pater (2016): Substance matters: A reply to Jardine (2016)

  1. Adam Jardine

    Thanks again for taking the time to write up a response to this, and for sharing it here. I agree that substance is important in phonological explanation, and that the state of the art of the application of FLT to phonology does not currently have a straightforward way of stipulating substantive restrictions.

    That being said, I have a few thoughts about this, which may or may not form a coherent response to your response. The main one is that while Copperbelt Bemba and Central Veneto are substantively similar, as you point out, they are formally different. In Copperbelt Bemba, the prominent position is fixed but the spreading feature (tone in this case) can freely appear, whereas in Central Veneto stress is restricted to at its leftmost the antepenultimate position, and the spreading feature is somewhere in between that and the end of the word. This means Central Veneto metaphony is bounded, and thus subsequential (even though it’s not myopic). I’d argue that that the particular confluence in Central Veneto of substantive factors matches computational restrictions–and thus the weakly deterministic hypothesis in my paper explains why this process, as opposed to something unbounded like in Copperbelt Bemba, is attested in segmental phonology.

    Consider a hypothetical Veneto^\prime where stress is instead always initial, but a [high] feature in the word will spread to it, given that the stressed vowel is neither [low] or [-ATR] (this condition on spreading is the same as in Central Veneto, as per Kimper 2012). This has the same substantive motivation as Central Veneto—the [high] feature wants to be in a prominent position, as long as it does not get blocked. It’s also similar to Copperbelt Bemba in that the feature wanting to spread can occur anywhere in the word.

    However, Veneto^\prime is (as far as I know) unattested, which is correctly predicted by the weakly deterministic hypothesis. Veneto^\prime is unbounded circumambient (and thus not weakly deterministic), because any any potential target vowel following a nonlow, [+ATR] initial stressed vowel needs to look unboundedly far to the right to see if there is a [high] vowel, and any potential target vowel preceding a [high] needs to look unboundedly far to the left to check if the initial stressed vowel is nonlow and [+ATR].

    Thus, even though Veneto^\prime has reasonable substantive motivation based on our current understanding of Central Veneto and Copperbelt Bemba, the weakly deterministic hypothesis correctly predicts that it shouldn’t exist.

  2. Joe Pater Post author

    Finally getting around to answering this more than a year later (in conjunction with finally doing the revisions to my paper!). Sorry for the delay on both fronts. Yes, indeed, OT as I’ve presented it in my paper predicts Veneto-prime. I wonder how worried we should be about that – these spreading by licensing systems in general seem rare, so it could be an accident that we don’t find it with fixed initial stress. My bigger worry in this context is that it feels to me like we don’t have a good handle on which prominent positions we spread into, and which always act as the sources of spreading, though I have a vague memory that someone (Rachel Walker maybe?) has said something about this. And also why spreading by licensing is rare, if it indeed is.

  3. Adam McCollum

    The prediction that a Veneto-prime type of harmony should/could exist, as in Joe’s paper, is borne out in Tutrugbu and Tafi, two Kwa languages. See the thread on Sour Grapes and Use it or Lose it here: This work has also been presented at this year’s WCCFL and NELS.

    The basics of Tutrugbu and Tafi are: ATR spreads regressively from a root to prefixes, but fails to spread iff a [-hi] prefix is present & the initial prefix is [+hi]. These two, as in [I-tI-ka-a-ba-ba-wu] ‘1S-NEG-PFV-PROG-VENT-VENT-climb’ can interact at a distance (here: 4 intervening syllables) to block ATR harmony. Harmony here would occur fully if the initial syllable were [-hi], like [e-ti-ke-e-be-be-wu] ‘3S-NEG-PFV-PROG-VENT-VENT-climb’ so the dependency is very very similar to unbounded tone plateauing and tone spreading in Copperbelt Bemba.

    As for prominence, Walker (2011) is a great resource. Kaplan (2015) discusses prominence-targeting systems, like Central Veneto, and which prominent positions may license the Veneto-type of harmony (see also his 2008 dissertation). I have a manuscript with James Essegbey currently under revision that addresses this topic, if you’re interested.


Leave a Reply

Your email address will not be published. Required fields are marked *