Review:21.A good OT research topic contains alternations or INPUT-OUTPUTirregularities.2.Format for a good descriptive generalization: [prohibition] +[restriction]Example:Consonants with the different POA are prohibited. This isenforced by assimilation.3. Direct Ranking Argument:
Review:34. Stringency Relationship•More stringent = more general•Since a SUPERSET is more stringent, it contains more violations.•Unrankable•Rankable only through TRANSITIVITY (If A>> B and B>> C then A>> C)
2.5 Candidates in Ranking Arguments:4Figuring out which losing candidates need to be considered isprobably the hardest thing about doing analysis in OTTwo Stages:•In the early stages, losing candidates are needed for the rankingarguments that emerge from our descriptive generalizations.•In the later stages, the analysis needs to be challenged with losingcandidates that might reveal its inadequacies.
2.5 Candidates in Ranking Arguments:5Technique for finding losers:Starting from [w], eliminate one or more Const2 violations and add one or more Const1violations without adding violations of any other constraints, except for those that arealready known to be ranked below CONST1.The result is *[l].
2.5 Candidates in Ranking Arguments:6Technique for finding losers:“Without adding violations of any other constraints, except thosethat are already known to be ranked below Const1” is such acriterion.
2.5 Candidates in Ranking Arguments:7•The unwanted losers *[la:n.hi.n], *[l.a:n.hin], and *[la:n.hin.?] violatevarious combinations of *Complex-Syllable, Onset, and Dep, and none ofthese constraints is already known to be ranked below *Cunsyll.•The loser we need, *[la:.n.hin], adds no violations other than *Cunsyll
2.5 Candidates in Ranking Arguments:8Practice:•Eliminating a Const2 violation is thereforea matter of restoring whatever property ofthe underlying representation the winnerhas lost – in this case, vowel length.•Since the loser has to add a violation of*Complex-Syllable but no other constraintsthat could dominate *Complex-Syllable, itmust be *[la:n.hin].
2.5 Candidates in Ranking Arguments:9Practice:•The loser must eliminate the winner’s Maxand Ident(long) violations, gain a *V#violation, and avoid acquiring violations ofother constraints that could dominate *V#.•The loser that does all this is *[ta.xa:.k?a].
2.5 Candidates in Ranking Arguments:10Practice:•Starting from the winner [xat.k?a], weneed to eliminate the *V# violation andadd a violation of *Complex-Syllable.•The loser that fulfills this requirement is[xatk?].•It adds a Max violation as well, and thiscould make it unsuitable for the rankingargument, but we have already establishedthat *V# dominates Max, so there is noproblem.
2.5 Candidates in Ranking Arguments:11Practice:•Removing the Ident(long) violation andadding another Max violation would yield*[ta.xa:]. This is unsuitable, however,because it also adds a violation of *V#, and*V# is already known to dominate Max.