Post comments on the readings for Tues. Oct.25th here.
5 thoughts on “Cyclicity comments”
Robert Staubs
Kiparksy:
– Footnote 4:
The most cutting criticism (“no non-lexical stress”) of footnote 4 is directly addressed by footnote 4 of Pruitt (2010): “I assume that underlying foot structure is not at issue here. See McCarthy & Pruitt (forthcoming) for arguments that lexical foot structure is not a viable way of encoding contrastive stress in HS and for an alternative proposal.”
p. 17:
– “Since each of these constraint systems is fully parallel, the grammar is formally well-behaved and computationally tractable.”
Where did that computational tractability bit come from? Is there any reason to suspect parallelism per se will make the difference?
– Objection 4 on p. 18:
OI already has to do away with Prec referring only to Faithfulness (that is, insertion operations are referred to directly). If what we really want is to refer to operations rather than Faithfulness contraints, that doesn’t seem like an insurmountable obstacle. Two immediate ideas: 1) operations explicitly present in chains (similar to the approach of OI for morphological operations) 2) most operations should be recoverable by comparing two steps in a chain (exceptions being whether they have morphological significance). This seems to be an unnecessary objection, overall.
– p. 19
Kiparsky argues that OT-CC can’t do the Finnish compensatory lengthening/gemination example in part because it doesn’t use Dep-Mora and Max-Mora. Is this necessarily so? That seems pretty orthogonal to the problem of architecture, but maybe I’m missing something.
Specifically about this, he claims an OT-CC analysis is impossible because it requires both faithfulness and a lack of faithfulness to syllable weight:
But with Dep/Max-Mora this isn’t a paradox at all…
Finally, he points to some kind of parodox with footing that I don’t understand. Vowel lengthening would make foot structure worse, while gemination makes it better. Couldn’t it be that footing occurs in between the two?
Brame raises the question as to why “stress-assigning rules are typically those that can most convincingly be demonstrated to be cyclic” (p 58). His answer is that “all and only the rules mentioning brackets apply cyclically”, (p 59)and since the definitions of stress-assignment rules make reference to bracketing, stress-assignment rules will apply cyclically.
Theoretical Question: Is it just me, or is this answer somewhat circular? Since the cycle itself is “dependent on bracketings” (p. 56), doesn’t Brame’s claim that stress-assigning rules are cyclic because they refer to brackets amount to saying that stress-assignment rules are cyclic because they are cyclic?
Empirical question: Stress-assignment isn’t cyclic/bracket-dependent in all languages, right?
attempts to answer the question of why “stress-assigning rules are typocally those that can most convincingly be demonstrated to be cyclic” (p 58). His answer is that “all and only the rules mentioning brackets apply cyclically” (p 59), and since
I have two little questions. The first is about the notion of ‘equipotency’. I’m not sure I get it. Is it true that two strings are equipotent iff they are phonologically identical, and one is a whole word in the lexicon?
The second is about the hypothesis that it’s just rules that refer to bracketing that are cyclic. Is it just rules that *crucially* refer to bracketing, or can it be rules that just happen to refer to bracketing?
1. Brame argues that a cyclic account of the facts he presents reaches explanatory adequacy, because the “Principle of Cyclicity” helps children acquire these rules. However, as Claire pointed out, the “Principle of Cyclicity” appears to be at least partially circular. Have there been any other attempts to connect cyclicity to learnability or acquisition?
2. It’s interesting that the process of guttural metathesis leads to rule ordering paradoxes in the analysis of both the Maltese data from this paper and the Palestinian Arabic data in JJM’s 2007 book. Could it be that metathesis has some special status that permits it to somehow occupy conflicting positions in derivations? Could the reanalysis of metathesis as two steps (insertion and deletion) help with this?
1. p6: “But even such a modularized OI would still predict weird global long-distance effects within words, as well as within sentences” – in his dissertation Wolf says that he has not developed a fully formalized account how OI would deal with sentence-level phenomena. Also, while “playing” with a theoretical stratal OI, Kiparsky seems to assume that its PREC constraints would be the same for each stratum, while, as we’ve discussed in our earlier classes, it does not have to be so (since the PREC constraints will have to be tailored for each type of morpho-phonological phenomenon).
2. p.7, example (9): “A theory which allows the cyclicity of a stem depend on whether some lengthening process is going to take effect five morphemes away (let alone five words away) is not credible.” – I am not sure whether the cyclicity that Kiparsky is assuming is the same thing as the actions of PREC constraints, since what they assume is that some process has to happen before the other, in this case, obj suffixation before prosody. Therefore, i am not sure whether the phenomena are comparable.
Kiparksy:
– Footnote 4:
The most cutting criticism (“no non-lexical stress”) of footnote 4 is directly addressed by footnote 4 of Pruitt (2010): “I assume that underlying foot structure is not at issue here. See McCarthy & Pruitt (forthcoming) for arguments that lexical foot structure is not a viable way of encoding contrastive stress in HS and for an alternative proposal.”
p. 17:
– “Since each of these constraint systems is fully parallel, the grammar is formally well-behaved and computationally tractable.”
Where did that computational tractability bit come from? Is there any reason to suspect parallelism per se will make the difference?
– Objection 4 on p. 18:
OI already has to do away with Prec referring only to Faithfulness (that is, insertion operations are referred to directly). If what we really want is to refer to operations rather than Faithfulness contraints, that doesn’t seem like an insurmountable obstacle. Two immediate ideas: 1) operations explicitly present in chains (similar to the approach of OI for morphological operations) 2) most operations should be recoverable by comparing two steps in a chain (exceptions being whether they have morphological significance). This seems to be an unnecessary objection, overall.
– p. 19
Kiparsky argues that OT-CC can’t do the Finnish compensatory lengthening/gemination example in part because it doesn’t use Dep-Mora and Max-Mora. Is this necessarily so? That seems pretty orthogonal to the problem of architecture, but maybe I’m missing something.
Specifically about this, he claims an OT-CC analysis is impossible because it requires both faithfulness and a lack of faithfulness to syllable weight:
al.kun -> a.l:un (mora conservation)
a.lu:m -> al.lu:n (mora insertion)
But with Dep/Max-Mora this isn’t a paradox at all…
Finally, he points to some kind of parodox with footing that I don’t understand. Vowel lengthening would make foot structure worse, while gemination makes it better. Couldn’t it be that footing occurs in between the two?
Regarding Brame:
Brame raises the question as to why “stress-assigning rules are typically those that can most convincingly be demonstrated to be cyclic” (p 58). His answer is that “all and only the rules mentioning brackets apply cyclically”, (p 59)and since the definitions of stress-assignment rules make reference to bracketing, stress-assignment rules will apply cyclically.
Theoretical Question: Is it just me, or is this answer somewhat circular? Since the cycle itself is “dependent on bracketings” (p. 56), doesn’t Brame’s claim that stress-assigning rules are cyclic because they refer to brackets amount to saying that stress-assignment rules are cyclic because they are cyclic?
Empirical question: Stress-assignment isn’t cyclic/bracket-dependent in all languages, right?
attempts to answer the question of why “stress-assigning rules are typocally those that can most convincingly be demonstrated to be cyclic” (p 58). His answer is that “all and only the rules mentioning brackets apply cyclically” (p 59), and since
about Brame:
I have two little questions. The first is about the notion of ‘equipotency’. I’m not sure I get it. Is it true that two strings are equipotent iff they are phonologically identical, and one is a whole word in the lexicon?
The second is about the hypothesis that it’s just rules that refer to bracketing that are cyclic. Is it just rules that *crucially* refer to bracketing, or can it be rules that just happen to refer to bracketing?
Brame 1974:
1. Brame argues that a cyclic account of the facts he presents reaches explanatory adequacy, because the “Principle of Cyclicity” helps children acquire these rules. However, as Claire pointed out, the “Principle of Cyclicity” appears to be at least partially circular. Have there been any other attempts to connect cyclicity to learnability or acquisition?
2. It’s interesting that the process of guttural metathesis leads to rule ordering paradoxes in the analysis of both the Maltese data from this paper and the Palestinian Arabic data in JJM’s 2007 book. Could it be that metathesis has some special status that permits it to somehow occupy conflicting positions in derivations? Could the reanalysis of metathesis as two steps (insertion and deletion) help with this?
Kiparsky:
1. p6: “But even such a modularized OI would still predict weird global long-distance effects within words, as well as within sentences” – in his dissertation Wolf says that he has not developed a fully formalized account how OI would deal with sentence-level phenomena. Also, while “playing” with a theoretical stratal OI, Kiparsky seems to assume that its PREC constraints would be the same for each stratum, while, as we’ve discussed in our earlier classes, it does not have to be so (since the PREC constraints will have to be tailored for each type of morpho-phonological phenomenon).
2. p.7, example (9): “A theory which allows the cyclicity of a stem depend on whether some lengthening process is going to take effect five morphemes away (let alone five words away) is not credible.” – I am not sure whether the cyclicity that Kiparsky is assuming is the same thing as the actions of PREC constraints, since what they assume is that some process has to happen before the other, in this case, obj suffixation before prosody. Therefore, i am not sure whether the phenomena are comparable.