Finnish OT Prosody

The analysis given in Finnish OT Prosody is slightly flawed. It gives incorrect results for inputs such as {kalasteleminen} 'fishing' and {järjestelmällisyydelläni} 'my systemaciticy':

(ká.las).te.(lè.mi).nen
(jä́r.jes).tel.(mä̀l.li).syy.(dèl.lä).ni

The desired outputs are:

(ká.las).(tè.le).(mì.nen)
(jä́r.jes).(tèl.mäl).li.(sỳy.del).(lä̀.ni)

The problem lies with the StressToWeight constraint that requires stressed syllables to be heavy. The two undeserving winners have fewer violations of this constraint than the correct alternatives:

(ká*.las).te.(lè*.mi).nen      2 violations
(ká*.las).(tè*.le).(mì*.nen)   3 violations

(jä́r.jes).tel.(mä̀l.li).syy.(dèl.lä).ni       0 violations
(jä́r.jes).(tèl.mäl).li.(sỳy.del).(lä̀*.ni)   1 violation

The 1999 analysis of Finnish Prosody given by Nine Elenbaas and René Karger ("Ternary rythm and the lapse constraint," Phonology 16, 273-329) does not postulate a general StressToWeight constraint but proposes a more specific restriction, *(L'H), that prohibits stress on a light syllable if the following syllable is heavy.

The tasks for this assignment are:

1. Formulate the *(L'H) constraint in finite-state terms, use it in place of StressToWeight, and check the Elenbaas&Karger analysis with the modified Finnish OT Prosody script. Are the results now correct?

2. If there results are not correct, can you suggest other changes, new constraints, some reranking or reformultation of existing constraints that would give the desired output in the two cases above, of course, without getting something else wrong?

Note that the non-OT analysis has no problem with these words provided that we consider a ternary foot such as (tèl.mäl.li) as equivalent to a binary foot followed by an extrametrical syllable: (tèl.mäl).li.

(jä́r.jes).(tèl.mäl.li).(sỳy.del).(lä̀.ni).