User:Alek.storm/Notes

From Glottopedia
< User:Alek.storm
Revision as of 18:07, 17 March 2010 by Alek.storm (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search
  • flapping, releasing plosives, aspiration
  • make a better IPA - /b/, /p/, and /p^h/ are *the same* - VOT of following vowel is affected! write only the changes in articulators from one sound to the next. this should also show causes of changes better.
  • rhyming
  • phonemes underspecified - no "base" form like /n/; place features aren't filled in yet
  • two steps of changes - *actual* change (always articulators?), other features are changed to conform to phonotactics and sympathetically make the sound *acoustically* as similar as possible
  • rules and constraints are equivalent *and come in pairs*! - allophony is the speaker fitting mental model onto idiolectal phonotactics (there is a constraint and a repairing rule), just like trying to pronounce a second language - it gets filtered through our phonotactics! we already know that phonemic distinctions in other languages that aren't in our own (even if we have both sounds, but they're allophones) are hard to both detect and produce
  • every change has a cause. find it and describe it. the range of possible changes is not infinite. look at enough cross-linguistic data to find all possible changes. figure out why it's just them.
  • figure out the *minimum* kinds of sounds (formants, etc) hearers use to differentiate sounds - some elements will be more important than others. find only the absolutely necessary ones. use computers to analyze phonetic data, and try to recreate segments completely artificially, see if speakers accept it.
  • features are really thresholds. we can model a sound system with n features as rectangles in n-dimensional space - bounded by low and high thresholds on each axis (feature)
  • why isn't syllabization phonemic? (changes too weak) Semitic triconsonantal roots are kind of similar. It looks like prosody happens completely separately, after all morpho and phono changes. But the hearer's data is distorted by it.
  • change = allophones or sound change. the only difference is restructuring
  • what everybody thinks of as "syllable-"/"stem-"/"word-"/"phrase-"/"utterance-" bounding are really completely separate systems on the way to the articulators. There is no symmetry to them; they shouldn't generalize.
  • how the HELL do allomorphs work?
  • two parts of language and one process: UG, lexicon; feature spreading through analogy. there is no language-specific grammar or parameters; all features are in bundles in the lexicon and UG handles them. what looked like language-wide parameters was really just widespread analogy. languages have different syntax on the surface -> they have different features (no language has all features). takes care of diachronic syntax changes, analogy, partial analogy, exceptions, new words (/pentium/ couldn't have its 't' dropped at first; now it can - feature spreading!. so phonological rules don't "target" a specific environment, they spread through *analogy*, which does not occur at utterance-time)
  • [+syllabic] is not a feature - it's added on later, during syllabification (includes some nasals, liquids, etc). for sonority hierarchy, use [+vowel] for vowels
  • morphemes are injected, *then* ordered. pieces getting swapped means they were in the same place, being processed at the same time. once they're injected, they go through an orderer, which sometimes makes mistakes