prosody meets paradigm uniformity michael wagner, mit. [email protected] web.mit.edu/~chael/www
Post on 22-Dec-2015
218 views
TRANSCRIPT
Prosody meetsParadigm Uniformity
Michael Wagner, [email protected]/~chael/www
Prosody meets Paradigm Uniformity
What is Paradigm Uniformity?
Paradigm Uniformity (PU):A segmental or prosodic property of an output form is ‘imported' from aparadigm which is defined based on a particular set of morpho-syntacticfeatures.
This could be based on a particular privileged form in theparadigm (the ‘base') or by organizational principles affectingphonological properties of entire paradigms.
• Any real case of PU would constitute evidence for the notion paradigm to play a role in the theory of grammar
• This would be a case of phonological as opposed to morphological evidence
(as in, say, Carstairs-Mccarthy’s work and elsewhere).
Prosody meets Paradigm Uniformity
If this is the correct generalization, this would constitute evidence for paradigmatic effects.
• `Underapplication of Assibilation in first part of the word `imported’ from other forms in the paradigm.
• In longer stems, the change introduced by assibilation is irrelevant since stem forms overlap sufficiently anyway.
Prosody meets Paradigm Uniformity
Anttila 2004: Prosodic Approach.
• Reference to Paradigms unnecessary.
• Laalo’s PU-Approach empirically incorrect.
Preview of Comments: Attempt to Generalize Anttila’s Argument
A. Predictive Value of PU within a theory of morpho-phonology
B. Look at another case where PU fails, and Prosody does the job
A. Predictive Value of Paradigm Uniformity
For PU-Effects, there is (to my knowledge) no theoretical expectation with respect to the following questions:
(1) Which phonological properties are likely to be `imported' from the paradigm?
(2) Which particular morpho-syntactic features span out the paradigms that induce PU effects?
At least if PU effects are simply posited by the linguist as a last resort where
(i) other analyses fail to predict a property of an output,
(ii) and importing the property form the paradigm is possible since it IS somewhere in the paradigm in the first place,
…(weak theory of PU) then calling these cases PU effects is rather post-hoc and descriptive at best. Discussion only on Case-by-Case basis possible.
A. Predictive Value of Paradigm Uniformity
Anttila’s Analysis is couched in the theory of Stratal OT (Kiparsky 2003) that makes very explicit predictions:
(1) All phonological `paradigmatic’ effects are only apparent---no PU
(2) There is a three step derivation (stem level > word level > post-lexical level), which is the only source of opacity
(3) Morphological evidence for ordering processes/affixes at certain levels (featural composition, ordering, reference to declension class only at stem level)
(4) Phonological Evidence for ordering processes/affixes at certain levels (same level = transparent interaction, ordering = different derivational steps)
(5) Derivational Effects for ANY phonological property---but independent evidence for ordering of these processes at the respective level is available.
Stratal OT tries to relate phonology and morpho-syntax in a predictive and insightful way, and is thus motivationally similar to DM.
A. Predictive Value of Paradigm Uniformity
A strong theory of PU effects could attempt to replace alternative tools (e.g. level ordering in Stratal OT). But:…
Anttila gives evidence for Opacity Effects that are not amenable to a PU analysis (for more evidence for a derivation and against PU and other Output-Output Correspondence-Tools see Kiparsky (in progress)):
Apocope counter-feeds Consonant Gradation, Apocope counter-bleeds Assibilation. Consonant Gradation counter-feeds Assibilation
Conclusion so far: PU as it stands fails to give an insight on phonology/morphology interaction, fails to capture morphological effects, and cannot replace derivational approaches to opacity.
The lack of a similarly restrictive theory makes weak PU unattractive.
A. Predictive Value of Paradigm Uniformity
Two outstanding puzzles for Anttila’s Approach:
(1)Are ordering predictions really borne out (Kenstowicz 94, Anttila p.c.)?
• e -> i raising feeds assibilation (vete -> vesi)
• i raising is a word level process (vete -> vetena essive (`water’))
• Assibilation is a stem-level process (ordering arguments)
Is Stratal OT wrong?
Maybe it needs modification---given the otherwise correct predictions, it seems rash to throw out everything unless there is a good alternative…
(2) Is prosodic story really correct? What about NDEB? (Anttila 2002, Nels)
• Many of old NDEB effects are explained prosodically (vesi vs. koti `home’))
• But: there are still NDEB effects that are unaccounted: stem > 2mora: suunti
• Is prosody the wrong theory after all? Is NDEB sufficient?
Given correlation between apocope and assibliation, prosodic component clearly necessary.
B. Another case for Prosody and against PU
Lenis obstruents become fortis
- Before a voiceless obstruent
- “Finally” (to be specified)
Approach in Steriade 1997: Licensing by Cue + Paradigm Uniformity
If there was a wide array of PU effects, we could start developing a theory of which types of phonological properties are `imported’---but are there many convincing cases? Here’s a case involving derivational morphology and compounding:
Final Laryngeal Neutralization in German (FLN).
Discussed in Wagner (2002): http://web.mit.edu/~chael/www/wagner2002MITWPL-Paper.pdf
B. Another case for Prosody and against PU
B. Another case for Prosody and against PU
“Finally”: e.g. phrase finally, realized as fortis:
B. Another case for Prosody and against PU
“Finally”: e.g. phrase finally, realized as fortis:
B. Another case for Prosody and against PU
Problem:
Neutralization “finally” even if sonorant or vowel follows---which should be a good environment to realize the contrast. Example here: Compounds
B. Another case for Prosody and against PU
Consider:
/ta:g/ -> ta:k `day’
Sie /ta:g/ + /en/? -> ta:gN `they convene’
*ta:kN
Vs. /ma:g/ -> ma:k `likes 3rd personSie /ma:g/ + /ihn/ -> Sie makN `she likes him’
*Sie magN (unless: she stomach)
There does not seem to be a phonetic motivation here.
B. Another case for Prosody and against PU
Proposed Solution: Paradigm Uniformity
Voicelessness imported from other forms in the Paradigm.
B. Another case for Prosody and against PU
But: This doesn’t work! …as can be illustrated by looking at different derivatives:
B. Another case for Prosody and against PU
(Adjective derives from NOUN!)
B. Another case for Prosody and against PU
But: This doesn’t work!
(both derive from NOUN!)
B. Another case for Prosody and against PU
(not neutralized in any paradigm---but still in derivative it’s neutralized!!)
B. Another case for Prosody and against PU
….still, even though nowehere to `import’ from, Neutralization before Affix.
B. Another case for Prosody and against PU
Alternative Solution: Prosodically Conditioned Neutralization.
(Note: Notion `Syllable’ is irrelevant---there is no coda-devoicing (at least in German and Dutch!). Arguments see Wagner (2002).
B. Another case for Prosody and against PU
B. Another case for Prosody and against PU
B. Another case for Prosody and against PU
Again an account that
(i) makes no reference to paradigms, but only to prosody;
(ii) arguments that PU-Approach is empirically incorrect;.
(iii) The opacity is as expected in Stratal OT
(word level process of neutralization, counter-bled by phrasing with following words and clitics, but bled by stem-level affixation)
(alternative derivational approaches could do the job as well…)
Prosody meets Paradigm Uniformity
Conclusion: Based on the phenomena discussed here,
• no evidence for PU, henceforth no evidence of Paradigms.
• some doubts have been raised whether PU can compete as a theory of morpho-
phonology.
A. Predictive Value of PU within a theory of morpho-phonology
B. Look at another case where PU fails, and Prosody does the job
Prosody meets Paradigm Uniformity
…maybe paradigms exist not in nature but only in the eye of the beholder,
i.e. the lepidopterist: