13.07.2015 Views

Dissertation - Michael Becker

Dissertation - Michael Becker

Dissertation - Michael Becker

SHOW MORE
SHOW LESS
  • No tags were found...

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

(103) LOCAL(o) 27.9% ≫ φ-MATCH ≫ LOCAL(o) 72.1%There is another, perhaps simpler way of projecting the relative strength of the twoclones of LOCAL(o) onto novel items. Given a novel item, the speaker can decide thatthe behavior of the novel item mimics the behavior of some given noun, chosen at randomfrom the lists of nouns associated with the clones of LOCAL(o). If such a word is chosenat random, there is a 27.9% chance of that word being associated with the higher rankingclone, thus giving the novel item a 27.9% chance of being an ot-taker. Either way, the resultis the same: The relative strength of the trend created by the existing nouns of the languageis built into the grammar, and then can be projected onto novel items.The use of markedness constraints in this analysis builds into the grammar onlythose generalizations that can be expressed with plausible universal constraints, such asconstraints on the licensing of [o], which is seen cross-linguistically. The lexicon maycontain further generalizations that cannot be expressed in terms of plausible universalconstraints, such as the fact that among the nouns that have an [o] in their final syllable,ot-takers with [i] in their penultimate syllable (e.g. cinór ‘tube’) are more common thanthose with [a] in their penultimate syllable (e.g. xalón). In the experiment presented in§3.3, speakers did not project this trend onto novel nouns, suggesting that they have neverlearned it. If only root [o]’s are relevant for taking –ot, it is expected that other vowelswould be ignored. Note that the speaker cannot simply ignore any vowel that is in thepenultimate syllable, since having an [o] in the penult is conducive to more –ot.To summarize, this section presented a mechanism that detects inconsistent rankingarguments between lexical items, and resolves the inconsistency by cloning a constraint.Once a constraint is cloned, lexical items are associated with different clones, assuring thatthey surface as intended. Additionally, the difference in size between the lists of associatedlexical items is available to the learner, so that the learner can project the relative strengthof lexical trends onto novel items.103

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!