New version page

MIT 24 962 - The representation of exceptions

Upgrade to remove ads
Upgrade to remove ads
Unformatted text preview:

24.962 Advanced phonology 2 May, 2005 The representation of exceptions (1) What kinds of data should a theory of exceptions explain? Some possibilities: (not all of these may be important) • The existence of exceptions – How do some words manage to avoid the regular processes of the language? • Limits on possible exceptions – Turkish has a few words like [etyd] that do not undergo final devoicing, but there are no words with “anti-devoicing” (hypothetical [kod] ∼ [kotW]) – Should we also rule out other irregular changes (hypothetic [kot] ∼ [kopW])? • Distributional facts about exceptions – Minority status: most Turkish words do devoice (just a handful like [etyd]) – Frequency: exceptional words often tend to have high token frequency Things are different in cases of learned exceptions, or fancy loanwords ◦ • Productivity – Is the regular pattern also the default for novel items? – Are speakers willing to extend exceptional patterns, given the right circumstances? (e.g., a strong subregularity) • Direction of historical change – Exceptions introduced by incomplete sound change – Exceptions introduced by loanwords – Exceptions introduced as a phonological process breaks down – Exceptions eliminated by regularization over time • Direction of errors (child & adult) (2) A classic example from English: exceptions to trisyllabic shortening • Trisyllabic shortening: div[aI]ne div[I]nity sal[aI]ne sal[I]nity obsc[i:]ne obsc[E]nity ser[i:]ne ser[E]nity extr[i:]me extr[E]mity ins[eI]ne ins[æ]nity prof[aU]nd prof[2]ndity verb[oU]se verb[a]sity • Exceptions to trisyllabic shortening: ob[i:]se ob[i:]sity (*ob[E]sity) — pr[oU]bity (3) A traditional approach to exceptions: diacritics • Lexical diacritic prevents application of rule, even though structural description is met • Obesity: negative input exception (-ity ordinarily provides context for TSS, but this root is immune) OBESE: /@bi:s/[−Trisyllabic Shortening]24.962—2 May, 2005 p. 2 (4) Predictions of this theory: • Suffixation of -ity should either cause TSS or not (two possible patterns) • Morphemic consistency: roots like obese should never undergo TSS (even if they happen to occur with other suffixes that cause TSS, like -acy) • New/unknown words: no intrinsic prediction – Obvious extension: default/redundancy rule marking all words as [+TSS] unless they are specifically known to be exceptions – In principle, either value could be default; in this case, most -ity words do trigger TSS, so exceptions are the minority pattern • Historical change and frequency – If a word is too infrequent, learners may never encounter the -ity form that would reveal a morpheme’s [−TSS] status – So, if we assume [+TSS] is the default, then predict regularization to [+TSS] (if learners fail to learn that a particular root is [−TSS]) – This would affect primarily low frequency words; only exceptions that remain over time are high frequency words • Child errors: more complex prediction – Interaction of two factors: what the learner knows about individual morphemes, and what the learner knows about TSS in general – Errors could come about from incorrect formulation of TSS, or incorrect assignment of [±TSS] diacritics – Need a better theory of how TSS is learned before we can make precise predictions (5) How do these predictions stack up empirically? • Just two patterns (TSS or not): not quite true – Insufficient lowering: ant[i:]que ∼ ant [I]quity (*ant [E]quity) – Too much lowering: cl[i:]r ∼ cl[æ]rity (*cl[E]rity) – Vowel deletion: en[@]my ∼ en∅mity *en[E]mity – Other oddities: p[oU]pe, p[eI]pacy (synchronically unjustified) Wang and Derwing (1994): under certain conditions, speakers even volunteer “reverse TSS” (Trisyllabic Lengthening?) on wug words • Morphemic consistency: seems to be false – Another suffix that can cause TSS: -((a)c)y bur[oU]crat bur[a]cracy t[aI]rant t[I]ranny supr[i:]me supr[E]macy consp[aI]re consp[I]racy Many exceptions (see SPE, p. 181); e.g. p[aI]racy, pr[aI]macy, dipl[oU]macy, r[i:]gency; in fact, there are rather few [+TSS] words in -y (words like bureaucracy and tyranny in the minority) – Standard American pronunciation: pr[aI]vate pr[aI]vacy pr[I]vity [−TSS]? [+TSS]? A couple other potential cases (rare but occurring forms; these are just my own intuitions about how they would be pronounced) 1660 R. SHERINGHAM King’s Suprem. Asserted viii. (1682) 70: ◦ “He grants him a primity of share in the supreme power.” ◦ “Locke saw that the extremacy of religious sects was not attributable to piety, but rather to the lust for power.”24.962—2 May, 2005 p. 3 My intuitions: extr[i:]me extr[i:]macy?? extr[E]mity pr[aI]me pr[aI]macy pr[I]mity/pr[aI]mity?? ➢ Likelihood of TSS depends on both affix and stem, but neither can be marked [±TSS] Similar to Spanish diphthongization case discussed previously; each suffix has its ◦ own likelihood to cause the alternation Difficult to test in this case; few roots appear with more than one TSS-inducing suffix ◦ • New/unknown words: a surprising effect – Novel formations are generally [−TSS], even for -ity (where most existing words are [+TSS]) Comedity: web comic ◦ om"¯ e)”“Com·e·dit·y: n. (k˘e-d˘ıt"¯Profoundity: id/893037◦ 3/3/05 2:35pm “Ooo, kudos for the profoundity of that statement. (made up a new word!)” 3/3/05 10:55pm “‘ that I think about it, I think ‘profoundity’ is actually a word . . . hahah, silly ol’ me . . . But if it’s not, go me!” (Lexicalized profundity uncertain/unknown to this particular speaker) – “Wug tests”: native speakers tend not to apply TSS (Jaeger 1983, Wang & Derwing 1983) ➢ Most existing Level 1 formations that could undergo TSS in fact do (especially with -ity); why would [−TSS] be the default for new words? (Caveat: -ity is, in general, not productive;1 whatever process leads to the creation of novel -ity forms goes beyond normal, unconscious application of the rules of English. We should have a better theory of the creative/humorous use of unproductive processes before making too strong a claim based on novel uses of unproductive affixes. Errors:

View Full Document
Download The representation of exceptions
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...

Join to view The representation of exceptions and access 3M+ class-specific study document.

We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view The representation of exceptions 2 2 and access 3M+ class-specific study document.


By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?