| ROA: | 104 |
|---|---|
| Title: | The Initial Ranking of Faithfulness Constraints in UG |
| Authors: | Mark Hale, Charles Reiss |
| Comment: | |
| Length: | |
| Abstract: | The Initial Ranking of Faithfulness Constraints in UG ROA-104 19p. initrank.ps, --.wrd Mark Hale and Charles Reiss Concordia University, Montreal hale1@alcor.concordia.ca, reiss@alcor.concordia.ca This paper argues that it is a necessary assumption of the theory of the acquisition of OT phonological systems that faithfulness constraints be ranked above well-formedness constraints in UG. In Part I we sketch our conception of what the study of phonological acquisition entails, namely the acquisition of a language specific ranking of constraints in each OT component, and the acquisition of a (language specific) lexicon. In Part II we argue for the strict separation of the study of grammar from the study of other factors influencing speech production, and thus propose that childrens speech perception abilities provide much better evidence for the state of their grammars than their production abilities. In Part III we argue that the acquisition of a lexicon and the acquisition of the correct ranking for the target language are intimately related processes, since positing an underlying representation for a given output entails positing a ranking of relevant constraints, and the only evidence for reranking comes from a comparison of surface forms with potential underlying representations. For this reason, any theory of ranking acquisition which assumes that the lexicon has already been acquired is deficient. After showing that neither an assumption of no ranking of constraints within UG, nor the widely-held view that all well-formedness constraints are ranked above all faithfulness constraints will allow us to posit a UG which will let infant learners converge upon a lexicon, numerous possibilities persist. A very large set of possible preset rankings, interspersing faithfulness and well-formedness constraints could be entertained, for example. We argue in Part IV, however, that a compelling case can be made for the assumption that all faithfulness constraints are ranked, in UG, above all well-formedness constraints. ::::::::::::::::::::::::::::::::::::::::::::::::::::: |
| Type: | Paper/tech report |
| Area/Keywords: | |
| Article: | Version 1 |